N<

Bulletin 42
(Pert 1 of 5 Parts)

THE
SSHOCK AND VIBRATION
BULLETIN
Parti
Invited Papers, Specifications, Mechanical
Impedance, Transportation and Packaging

JANUARY 1972

A Publication of
THE SHOCK AND VIBRATION
INFORMATION CENTER
Naval Research Laboratory, Washington, D.C.

DDC

APR
n.P1rnflTM
5 I'9

Office of
The Director of Defense
NATIONAL TECHNICAL
RproducAd by Research and Engineering "
INFORMATION SERVICE
Springfield, Va. 22151

This document has been approved for public release and sale; its distribution is unlimited.
 !I

SYMPOSIUM MANAGEMENT

THE SHOCK AND VIBRATION INMORMATION

CENTER
William W. Mutch, Director
Henry C. Pusey, Coordinator
Rudolph H. Volin, Coordinator
Edward H. Schell, Coordinator

Bulletin Preduction
Graphic Arts Branch, Technical Information
Division,
Naval Research Laboratory

DOCGEwe own
t¢i".i ..........

Mia mi mi;

.r
____-- ________ _____
 Bulietin 42
(Part 1 of 5 Parts)

THE
SHOCK AND VIBRATION
BULLETIN

JANUARY 1972

A Publication of
THE SHOCK AND VIBRATION
INFORMATION CENTER
Naval Research Laboratory, Washington, D.C.

The 42nd Symposium on Shock and Vibration was

held at the U.S. Naval Station, Key West, Florida,
on 2-4 November 1971. The US. Navy was host.

Office of
The Director of Defense
Research and Engineering
I
 Best
Available
Copy
 CONTENTS

PAPERS APPEAR.LNG IN PART 1

Invited Papers

SMALL SHIPS-HIGH PERFORMANCE .......................................

Rear Admiral H. C. Mason, Commander, Naval Ship Engineering Center, Washington, D.C.

Specifications

SURVEY OF VIBRATION TEST PROCEDURES IN USE BY THE AIR FORCE ............. 11

W. B. Yarcho, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force
Base, Ohio

SPECIFICATIONS- A PANEL SESSION ................................ 19

SOME ADMINISTRATIVE FACTORS WHICH INFLUENCE TECHNICAL APPROACHES

TO SHIP SHOCK HARDENING ......................................... 33
D. M. Lund, Naval Ship Engineering Cente:, Hyattsville, Marylnd

Measurement and Application of Mechanical Impedance

FORCE TRANSDUCER CALIBRATIONS RELATED TO MECHANICAL IMPEDANCE
MEASUREMENTS ........................ .... # .................... 43
E. F. Ludwig, Assistant Project Engineer, and N.D. Taylor, Senior Engineer, Pratt & Whitney
Aircraft, Florida Research & Development Center, West Palm Beach, Florida
THE MEASUREMENT OF MECHANICAL IM]EDANCE AND ITS USE IN
VIBRATION TESTING .......... # .... # ........................ 55
N. F. Hunter, Jr., and J. V. Otts, Sandia Corporation, Albuquerque, New Mexico
TRANSIENT TEST TECHNIQUES FOP MECFANICAL IMPEDANCE AND MODAL
SURVEY TESTING .......... . . . ........
.......... ................... 71
J. D. Favour, M. C. Mitchell, N. L. Olson, The Boeing Company, Seattle, Washington

PREDICTION OF FORCE SPECTRA BY MECHANICAL IMPEDANCE AND ACOUSTIC

MOBILITY MEASUREMENT TECHNIQUES .................................. 83
R. W. Schock, NASA/Marshall S.,-ce Flight Centr, Huntsville, Alabama and G. C. Kao,
Wyle Laboratories, Huntsville, Alabama

DYNAMIC DESIGN ANALYSIS VIA THE BUILDING BLOCK APPROACH ............... .97
A. L. Klosterman, Ph.D. and J. R. Lemon, Ph.D., Structural Dynamics Research
Corporation Cincinnati, Ohio

MOBILITY MEASUREMENTS FOR THE "UBRATION ANALYSIS OF CONNECTED

STRUCTURES ........................... *................#.......... 105
D. J. Ewins aaxd M. C Sainsbury, Imperial College of Science and Technology,
London, England

iii

_________o___
LIQUID-STRUCTURE COUPLING IN CURVED PIPES -II ......................... 123
L. C. Davidson and D. R. Samsury, Machinery Dynamics Division, Naval Ship Research and
Development Center, Annapolis, Maryland

Transportation and Packaging

A SURVEY OF THE TRANSPORTATION SHOCK AND VIBRATION INPUT TO CARGO ....... 137
F. E. Ostrem, General American Research Division, General American Transportation
Corporation, Niles, Illinois

THE DYNAMIC ENVIRONMENT OF SELECTED MILITARY HELICOPTERS .............. 153

M. B. Gens, Sandia Laboratories, Albuquerque, New Mexico

HIGHWAY SHOCK INDEX ......................... # ..................... 163

R, Kennedy, U. S. Army Transportation Engineering Agency, Military Traffic Management
and Terminal Service, Newport News, Virginia

DEVELOPMENT OF A ROUGH ROAD SIMULATOR AND SPECIFICATION FOR TESTING

OF EQUIPMENT TRANSPORTED IN WHEELED VEHICLES ................... 169
H. M. Forkois and E. W. Clements, Naval Research Laboratory, Washington, D.C.

LABORATORY CONTROL OF DYNAMIC VEHICLE TESTING .................... ,.. 191

J. W, Grant, U. S. Army Tank-Automotive Command, Warren, Michigan

IMPACTVULNERABILITY OF TANK CAR HEADS ............................... 197

J. C. Shang and J. E. Everett, General American Research Division,
General American Transportation Corporation, Niles, Illinois

A STUDY OF IMPACT TEST EFFECTS UPON FOAMED PLASTIC CONTAINERS ........... 211
D. McDaniel, Cround Equipment and Materials Directorate, Directorate for Research,
Development, Engineering and Missile Systems Laboratory, U. S. Army Missile Command
Redstone Arsenal, Alabama, and R. M. Wyskida, Industrial and Systems Engineering
Department, The University of Alabama in Huntsville, Huntsville, Alabama

DEVELOPMENT OF A PRODUCT PROTECTION SYSTEM ............................ 223

D. E. Yound, IBM General Systems Division, Rochester, Minnesota, and
S. R. Pierce, Michigan State University, East Lansing, Michigan

MOTION OF FREELY SUSPENDED LOADS DUE TO HORIZONTAL SHIP MOTION IN

RANDOM HEAD SEAS .................................................. 235
H. S. Zwlbel, Naval Civil Engineering Laboratory, Port Hueneme, California

PAPERS APPEARING IN PART 2

Ground Motion

SINE BEAT VIBRATION TESTING RELATED TO EARTHQUAKE RESPONSE SPECTRA

E. G. Fischer, Westinghouse Research Laboratories, Pittsburgh, Pennsylvania

SEISMIC EVALUATION OF ELECTRICAL EQUIPMENT FOR NUCLEAR POWER STATIONS

R. H. Prause and D. R. Ahlbeck, BATTELLE, Columbus Laboratories, Columbus, Ohio

SHOCK INPUT FOR EARTHQUAKE STUDIES USING GROUND MOTION FROM UNDERGROUND
NUCLEAR EXPLOSIONS
D. L. Bernreuter, D. M. Norris, Jr., and F. J. Tokarz, Lawrence Livermore Laboratory,
University of California, Livermore, California

iv
ROCKING OF A RIGID, UNDERWATER BOTTOM-FOUNDED STRUCTURE SUBJECTED TO
SEISMIC SEAFLOOR EXCITATION
J. G. Hammer and H. S. Zwibel, Naval Civil Engineering Laboratory, Port Hueneme,
California

DEVELOPMENT OF A WAVEFORM SYNTHESIS TECHNIQUE-A SUPPLEMENT TO RESPONSE

SPECTRUM AS A DEFINITION OF SHOCK ENVIRONMENT
R. C. Yang and H. R. Saffell, The Ralph M. Parsons Company, Los Angeles, California

THE RESPONSE OF AN ISOLATED FLOOR SLAB-RESULTS OF AN EXPERIMENT IN

EVENT DIAL PACK
J. M. Ferritto, Naval Civil Engineering L'.aboratory, Port Hueneme, California

A SHOCK-ISOLATION SYSTEM FOR 22 .FEET OF VERTICAL GROUND MOTION

E. C. Jackson, A. B. Miller and D. L. Bernreuter, Lawrence Livermore Laboratory,
University of California, Livermore, California

THE COMPARISON OF THE RESPONSE OF A HIGHWAY BRIDGE TO UNIFORM GROUND

SHOCK AND MOVING GROUND EXCITATION
N. E. Johnson and R. D. Galletly, Mechanics Research, Inc., Los Angeles, California

DEFORMATION AND FRACTURE OF TANK BOTTOM HULL PLATES SUBJECTED

TO MINE BLAST
D. F. Haskell, Vulnerability Laboratory, U.S. Army Ballistic Research
Laboratories, Aberdeen Proving Ground, Md.

THE IMPULSE IMPARTED TO TARGETS BY THE DETONATION OF

LAND MINES
P. S. Westine, Southwest Research Institute, San Antonio, Texas

CIRCULAR CANTILEVER BEAM ELASTIC RESPONSE TO AN EXPLOSION

Y. S. Kim and P. H. Ukranetz, Department of Mechanical Engineering, University
of Saskatchewan, Saskatoon, Canada

MEASUREMENT OF IMPULSE FROM SCALED BURIED EXPLOSIVES

B. L. Morris, U.S. Army Mobility Equipment Research and Development Center,
Fort Beiroir, Virginia

Dynamic Analysis
THE EFFECTS OF MOMENTUM WHEELS ON THE FREQUENCY RESPONSE
CHARACTERISTICS OF LARGE FLEXIBLE STRUCTURES
F. D. Day II1 and S. R. Tomer, Martin Marietta Corporation, Denver, Colorado

INTEGRATED DYNAMIC ANALYSIS OF A SPACE STATION WITH CONTROLLABLE

SOLAR ARRAYS
J. A. Heinrichs and A. L. Weinberger, Fairchild Industries, Inc., Germantown, Maryland,
and M. D. Rhodes, NASA Langley Research Center, Hampton, Virginia
PARAMETRICALLY EXCITED COLUMN WITH HYSTERETIC MATERIAL
PROPERTIES
D. T. Mozer, IBM Corporation, East Fishkill, New York, and R. M. Evan-lwanowski,
Professor, Syracuse University, Syracuse, New York

DYNAMIC INTERACTION BETWEEN VIBRATING CONVEYORS AND

SUPPORTING STRUCTURE
M. Paz, Professor, Civil Engineering Department, University of Louisville,
Louisville, Kentucky, and 0. Mathis, Design Engineer, Rex Chainbelt Inc.,
Louisville, Kentucky

v
RESPONSE OF A SIMPLY SUPPORTED CIRCULAR PLATE EXPOSED TO THERMAL
AND PRESSURE LOADING
J. E. Koch, North Eastern Research Associates, Upper Montclair, N.J., and M. L. Cohen,
North Eastern Research Associates, Upper Montclair, N.J., and Stevens Institute of
Technology, Hoboken, N.J.

WHIRL FLUTTER ANALYSIS OF PROPELLER-NACELLE-PYLON SYSTEM ON LARGE

SURFACE EFFECT VEHICLES
Yuan-Ning Liu, Naval Ship Research and Development Center, Washington, D.C.

THE DYNAMIC RESPONSE OF STRUCTURES SUBJECTED TO TIME-DEPENDENT

BOUNDARY CONDITIONS USING THE FINITE ELEMENT METHOD
G. H. Workman, Battelle, Columbus Laboratories, Columbus, Ohio

VIBRATION ANALYSIS AND TEST OF THE EARTH RESOURCES

TECHNOLOGY SATELLITE
T. J. Cokonis and G. Sardella, General Electric Company, Space Division,
Philadelphia, Pennsylvania
FINITE AMPLITUDE SHOCK WAVES IN INTERVERTEBRAL DISCS
W. F. Hartman, The Johns Hopldns University, Baltimore, Maryland

ACCELERATION RESPONSE OF A BLAST-LOADED PLATE

iI
L. W. Fagel, Bell Telephone Laboratories, Inc., Whippany, New Jersey

EFFECT OF CORRELATION IN HIGH-INTENSITY NOISE TESTING AS INDICATED

BY THE RESPONSE OF AN INFINITE STRIP
C. T. Morrow, Advanced Technology Center, Inc., Dallas, Texas

PAPERS APPEARING IN PART 3

Test Control
ON THE PERFORANCE OF TDM AVERAGERS IN RANDOM VIBRATION TESTS
A. J. Curtis, Hughes Aircraft Company, Culver City, California

A MULTIPLE DRIVER ADMITTANCE TECHNIQUE FOR VIBRATION TESTING OF

COMPLEX STRUCTURES
S. Smith, Lockheed Missiles & Space Company, Palo Alto Research Laboratory,
Palo Alto, California, and A. A. Woods, Jr., Lockheed Missiles & Space Company,
Sunnyvale, California

EQUIPMENT CONSIDERATIONS FOR ULTRA LOW FREQUENCY MODAL TESTS

R. G. Shoulberg and R. H. Tuft, General Electric Company, Valley Forge,
Pennsylvania
COMBINED-AXIS VIBRATION TESTING OF THE SRAM MISSILE
W. D. Trotter and D. V. Muth, The Boeing Company, Aerospace Group,
Seattle, Washington
SHOCK TESTING UTILIZING A TIME SHARING DIGITAL COMPUTER
R. W. Canon, Naval Missile Center, Point Mugu, California

A TECHNIQUE FOR CLOSED-LOOP COMPUTER-CONTROLLED REVERSED-

BENDING FATIGUE TESTS OF ACOUSTIC TREATMENT MATERIAL
C. E. Rucker and R. E. Grandle, NASA Langley Research Center,
Hampton, Virginia

vi
PROGRAILMING AND CONTROL OF LARGE VIBRATION TABLES IN UNIAXIAL
AND BIAXIAL MOTIONS
R. L. Larson, MTS Systems Corporation, Minneapolis, Minnesota
A DATA AMPLIFIER GAIN-CODE RECORDING SYSTEM
J. R. Olbert and T. H. Hammon.d, Hughes Aircraft Company, Culver
City, California
STABILITY OF AN AUTOMATIC NOTCH CONTROL SYSTEM IN SPACECRAFT
TESTING
B. N. Agrawal, COMSAT Laboratories, Clarksburg, Maryland
Test Facilities and Techniques

SINUSOIDAL VIBRATION OF POSEIDON SOLID PROPELLANT MOTORS

L. R. Pendleton, Research Specllst, Lockheed Mieseiles & Space Company,
Sunnyvale, California
CONFIDENCE IN PRODUCTION UNITS BASED ON QUALIFICATION VIBRATION
R. E. Deitrick, Hughes Aircraft Company, Space and Communications Group,
El Segundo, California
SIMULATION TECHNIQUES IN DEVELOPMENT TESTING
A. Hammer, Weapons l.aboratory, U. S. Army Weapons Command, Rock
Island, Illinois

A ROTATIONAL SHOCK AND VIBRATION FACILITY

R. T. Fandrich, Jr., Radiation !ncorporated, Melbourne, Florida

THE EFFECTS OF VARIOUS PARAMETERS ON SPACECRAFT SEPARATION SHOCK

W. B. Keegan and W. F. Bangs, NASA, Goddard Space Flight Center, Greenbelt,
Maryland

NON-DESTRUCTIVE TESTING OF WEAPONS EFFECTS ON COMBAT AND

LOGISTICAL VEHICLES
R. L. Johnson, J. H. Leete, and J. D. O'Keefe, TRW Systems Group, Redondo
Beach, California, and A. N. Tedesco, Advanced Research Projects Agency,
Department of Defense, Washington, D.C.

THE EFFECT OF THE FIN-OPENING SHOCK ENVIRONMENT ON GUIDED MODULAR

DISPENSER WEAPONS
K. D. Denton and K. A. Herzing, Honeywell Inc., Government and Aeronautical
Products Division Hopkins, Minnesota
DEVELOPMENT OF A FLUIDIC HIGH-INTENSITY SOUND GENERATOR
H. F. Wolfe, Air Force Flight Dynamics Laboratory, Wright-Patterson
Air Force Base, Ohio
DEVELOPMENT OF A LIGHTWEIGHT, LINEAR MECHANICAL SPRING ELEMENT
R. E. Keeffe, Kaman Sciences Corporation, Colorado Springs, Colorado

TECHNIQUES FOR IMPULSE AND SHOCK TUBE TESTING OF SIMULATED

REENTRY VEHICLES
N. K. Jamison, McDonnell Douglas Astronautics Company, Huntington
Beach, California

VIBRATION FIXTURING - NEW CELLULAR DESIGN, SATURN AND ORBITAL

WORKSHOP PROGRAMS
R. L. Stafford, McDonnell Douglas Astronautics Company, Huntington Beach,
California

vii
WALL FLOW NOISE IN A bUBSONIC DIFFUSER
E. F. Timpke, California State College, Long Beach, California, and R. C. Binder
University of Southern California, Los Angeles, California

PAPERS APPEARING IN PART 4

Isolation and Damping

TRANSIENT RESPONSE OF REAL DISSIPATIVE STRUCTURES

R. Plunkett, University of Minnesota, Minneapolis, Minnesota

DYNAMIC RESPONSE OF A RING SPRING

R. L. Eshleman, 1IT Research Institute, Chicago, Illinois

SHOCK MOUNTING SYSTEM FOR ELECTRONIC CABINETS

W. D. Delany, Admiralty Surface Weapons Establishment, Portsmouth, U.K.

METHODS OF ATTENUATING PYROTECHNIC SHOCK

S. Barrett and W. J. Kacena, Martin Marietta Corporation, Denver, Colorado
ENERGY ABSORPTION CAPACITY OF A SANDWICH PLATE WITH
CRUSHABLE CORE
D. Krajcinovic, Argonne National Laboratory, Argonne, Illinois

ON THE DAMPING OF TRANSVERSE MOTION OF FREE-FREE BEAMS IN

DENSE, STAGNANT FLUIDS
W. K. Blake, Naval Ship Research and Development Center, Bethesda, Maryland

OPTIMUM DAMPING DISTRIBUTION FOR STRUCTURAL VIBRATION

R. Plunkett, University of Minnesota, Minneapolis, Minnesota

A LAYERED VISCOELASTIC EPOXY RIGID FOAM MATERIAL FOR

VIBRATION CONTROL
C. V. Stahle and Dr. A. T. Tweedie, General Electric Company, Space
Division, Valley Forge, Pa.
OPTIMIZATION OF A COMBINED RUZICKA AND SNOWDON VIBRATION
ISOLATION SYSTEM
D. E. Zeidler, Medtronic, Inc.: Minneapolia, Minnesota, and D. A. Frohrib,
University of Minnesota, Minneapolis, Minnesota

TRANSIENT RESPOFT- OF PASSIVE PNEUMATIC ISOLATORS

G. L. Fox, and E. _teiner, B.trry Division of Barry Wright Corporation,
Burbank, California

EXPERIMENTAL DETERMINATION OF STRUCTURAL AND STILL WATER DAMPING

AND VIRTUAL MASS OF CONTROL SURFACES
R. C. Leibowltz and A. Kilcullen, Naval Ship Research and Development Center,
Washington, D.C.
DAMPING OF A CIRCULAR RING SEGMENT BY A CONSTRAINED
VISCOELASTIC LAYER
Cpt. C. R. Almy, U.S. Army Electronics Command, Ft. Monmouth, New Jersey,
and F. C. Nelson, Department of Mechanical Engineering, Tufts University,
Medford, Mass.
DYNAMIC ANALYSIS OF THE RUNAWAY ESCAPEMENT MECHANISM
G. W. Hemp, Department of Engineering, Science and Mechanics, University
ot Florida, Gainesville, Florida

viii
 Prediction and Experimental Techniques
A METHOD FOR PREDICTING BLAST LOADS DURING THE DIFFRACTION PHASE
W. J. Taylor, Ballistic Research Laboratories, Aberdeen Proving Ground, Maryland
DRAG MEASUREMENTS ON CYLINDERS IN EVENT DIAL PACK
S. B. Mellsen, Defence Research Establishment Suffield, Ralston, Alberta, Canada
DIAL PACK BLAST DIRECTING EXPERIMENT
L. E. Fugelso, S. F. Fields, and W. J. Byrne, General American Research
Division, Niles, Illois

BLAST FIELDS ABOUT ROCKETS AND RECOILLESS RIFLES

W. E. Baker, P. S. Westine, and R. L. Bessey, Southwest Research Institute,
San Antonio, Texas

TRANSONIC ROCKET-SLED STUDY OF FLUCTUATING SURFACE-PRESSURES

AND PANEL RESPONSES
E. E. Ungar, Bolt Beranek and Newman Inc., Cambridge, Massachusetts, and H. J.
Bandgren, Jr. and R. Erwin, National Aeronautics and Space Administration, George
C. Marshall Space Flight Center Huntsville, Alabama

SUPPRESSION OF FLOW-INDUCED VIBRATIONS BY MEANS OF BODY

SURFACE MODIFICATIONS
D. W. Sallet and J. Berezow, Naval Ordnance Laboratory, Silver Spring, Maryland

AN EXP1ratIMENTAL TECHNIQUE FOR DETERMINING VIBRATION MODES OF

STRUCTURES WITH A QUASI-STATIONARY RANDOM FORCING FUNCTION
R. G. Chrstiansen and W. W. Parmenter, Naval Weapons Center, China Lake,
California

RESPONSE OF AIR FILTERS TO BLAST

E. F. Witt, C. J. Arroyo, and W. N. Butler, Bell Laboratories, Whippany, N.J.

PAPERS APPEARING IN PART 5

Shock and Vibration Analysis

BANDWITH-TIME CONSIDERATIONS IN AUTOMATIC EQUALIZATION

C. T. Morrow, Advanced Technology Center, Inc., Dallas, Texas

A REGRESSION STUDY OF THE VIBRATION RESPONSE OF AN

EXTERNAL Si'ORE
C. A. Golueke, Air Force Flight Dynamics Laboratory, Wright-Patterson
Air Force Base, Ohio
FACTOR ANALYSIS OF VIBRATION SPECTRAL DATA FROM MULTI-LOCATION
MEASUREMENT
R. G. Merkle, Air Force Flight Dynamics Laboratory, Wright-Patterson Air
Force Base, Ohio

RESPONSES OF A MULTI-LAYER PLATE TO RANDOM EXCITATION

H. Saunders, General Electric Company, Aircraft Engine Group, Cincinnati, Ohio
RESPONSE OF HELICOPTER ROTOR BLADES TO RANDOM LOADS
NEAR HOVER
C. Lakshmikantham and C. V. Joga Rlao, Army Materials and Mechanics Research
Center, Watertown, Massachusetts

ix
INSTRUMENTATION TECHNIQUES AND THE APPLICATION OF SPECTRAL
ANALYSIS AND LABORATORY SIMULATION TO GUN SHOCK PROBLEMS
D. W. Culbertson, Naval Weapons Laboratory, Dahlgren, Virginia, and
V. F. DeVost, Naval Ordnance Laboratory, White Oak, Silver Spring, Maryland

THE EFFECT OF -Q" VARIATIONS IN SHOCK SPECTRUM ANALYSIS

M. B. McGrath, Martin Marietta Corporation, Denver, Colorado, and W. F. Bangs,
National Aeronautics and Space Administration, Goddard Space Flight Center, Maryland

RAPID FREQUENCY AND CORRELATION ANALYSIS USING AN ANALOG COMPUTER

J. G. Parks, Research, Development and Engineering Directorate, U.S. Army Tank-
Automotive Command, Warren, Michigan

INVESTIGATION OF LAUNCH TOWER MOTION DURING AEROBEE 350 LAUNCH

R. L. Kinsley and W. R. Case, NASA, Goddard Space Flight Center, Greenbelt, Maryland

ON THE USE OF FOURIER TRANSFORMS OF MECHANICAL SHOCK DATA

H. A. Gaberson and D. Pal, Naval Civil Engineering Laboratory, Port Hueneme,
California
WAVE ANALYSIS OF SHOCK EFFECTS IN COMPOSITE ARMOR
G. L. Filbey, Jr., USAARDC Ballistic Research Laboratories, Aberdeen Proving
Ground, Maryland

STATISTICAL LOADS ANALYSIS TECHNIQUE FOR SHOCK AND HIGH-FREQUENCY

EXCITED ELASTODYNAMIC CONFIGURATIONS
K. J. Saczalski and K. C. Park, Clarkson College of Technology, Potsdam, New York

Structural Analysis

NASTRAN OVERVIEW: DEVELOPMENT, DYNAMICS APPLICATION, MAINTENANCE,

ACCEPTANCE
J.P. Raney, Head, NASTRAN Systems Management Office and D. J.Weidman, Aerospace
Engineer, NASA Langley Research Center, Hampton, Virginia

EXPERIENCE WITH NASTRAN AT THE NAVAL SHIP R&D CENTER AND OTHER
NAVY LABORATORIES
P. Matula, Naval Ship Research & Development Center, Bethesda, Maryland

RESULTS OF COMPARATIVE STUDIES ON REDUCTION OF SIZE PROBLEM

R. M. Mains, Department of Civil and Environmental Engineering, Washington
University, St. Louis, Missouri

STRUCTURAL DYNAMICS OF FLEXIBLE RIB DEPLOYABLE SPACECRAFT ANTENNAS

B. G. Wrenn, W. B. Haile, Jr. and J. F. Hedges, Lockheed Missiles and Space
Company, Sunnyvale, California

INFLUENCE OF ASCENT HEATING ON THE SEPARATION DYNAMICS OF A

SPACECRAFT FAIRING
C. W. Coale, T. J. Kertesz, Lockheed Missiles & Space Company, Inc.,
Sunnyvale, California
DYNAMIC WAVE PROPAGATION IN TRANSVERSE LAYERED COMPOSITES
C. A. Ross, J. E. Cunningham, and R. L. Sterakowski, Aerospace Engineering Department
University of Florida, Gainesville, Florida
R-W PLANE ANALYSIS FOR VULNERABILITY OF TARGETS TO AIR BLAST
P. S.Westine, South;, est Research Institute, San Antonio, Texas

x
 PERFORM: A COMPUTER PROGRAM TO DETERMINE THE LIMITING PERFORMANCE
OF PHYSICAL SYSTEMS SUBJECT TO TRANSIENT INPUTS
IV. D. Pilkey and Bo Ping Wang, Department of Aerospace Engineering and Engineering
Physics, University of Virginia, Charlottesville, Virginia
STRUCTURAL DYNAMIC ANALYSIS AND TESTING OF A SPACECRAFT DUAL TRACK[NG
ANTENNA
D. D. Walters, R. F. Heidenreich, A. A. Woods and B. G. Wrenn, Lockheed Missiles
and Space Company, Sunnyvale, California

Ship's Problems

DETERMINATION OF FIXED-BASE NATURAL FREQUENCIES OF A COMPOSITE

STRUCTURE OR SUBSTRUCTURES
C. Ni, R. Skop, and J. P. Layher, Naval Research Laboratory, Washington, D.C.

EQUIVALENT SPRING-MASS SYSTEM: A PHYSICAL INTERPRETATION

B. K. Wada, R. Bamford, and J. A. Garba, Jet Propulsion Laboratory,
Pasadena, California

LONGITUDINAL VIBRATION OF COM1POSITE BODIES OF VARYING AREA

D. J. Guzy, J.C.S. Yang, and W. H. Walston, Jr., Mechanical Engineering
Department, University of Maryland, College Park, Maryland

SIMPLIFIED METHOD FOR THE EVALUATION OF STRUCTUREBORNE VIBRATION

TRANSMISSION THROUGH COMPLEX SHIP STRUCTURES
M. Chcrnjawski and C. Arcidiacona, Gibbs & Cox, Inc., New York, New York

xt
 INVITED PAPERS

"SMALL SHIPS - HIGH PERFORMANCE"

RADM HARRY C. MASON

COMMANDER, NAVAL SHIP ENGINEERING CENTER
HYATTSVILLE, MARYLAND

The Navy is now into its first formance" because they can achieve
large scale Fleet renewal program much higher speeds than other ships
since World War II. And we are pro- and maintain these speeds in rough
ceeding very carefully, because we seas. Included in this high per-
realize that decisions made today formance category are hydrofoils,
will affect the Navy for many years air cushion vehicles and surface ef-
to come. Our goal is to create a fect ships.
Navy which will meet the demands of
the next generation of Americans, First, let's consider hydro-
as well as lower the age of our foils -- ships that can rise out of
ships. the water and skim along on wing-
like appendages.
In planning this new Navy, we
look, of course, to the future, The oft-made analogy between
where we see the Navy retaining re- hydrofoils and airplanes is rela-
sponsibility for covering large ar- tively accurate: when hydrofoil
eas of the world's surface. We see ships "fly," the foils function in
also that this mission will have to much the same manner as wings; and
be accomplished with fewer numbers the ship itself is weight critical.
of men and ships. Our goal, there- Also like planes, hydrofoils have
fore, is a tightly organized, effi- cockpit-type control stations,
cient force equipped with superior light weight aluminum structures and
weapons. As Admiral Zumwalt, our gas turbine engines.
Chief of Naval Operations, put
it, "a lean, mean force."

Covering large expanses of

ocean with a smaller fleet means
that our new Navy must be highly
flexible and highly mobile. We must
be able to quickly assemble units
to present a concentrated force, and SURFACE.PIERCING
to rapidly disperse ships in smaller . HYDROFOILS
groups to support friendly nations
anywhere in the world. To achieve
this mission with reasonable invest-
ments, we intend to supplement our
conventional ships with small, High LADDER HYDROFOILS FULLY SUBMEGED
Performance ships. (HYBRID) HYDROFOILS

We call these craft "High Per- Fig. 1 - Three types of hydrofoils

1

A
 There are three types of hy- Today, the U.S. Navy is the
drofoils as shown in Fig. 1 -- recognized leader in submerged foil,
surface piercing, fully submerged ocean-going hydrofoils. And, al-
and hybrids. The Navy has opted though several hundred hydrofoil
for the submerged foil system be- ferries ranging from 15-150 tons are
cause it produces far superior now in successful commercial opera-
rough weather performance and re- tion throughout the world, use of
quires less propulsion power. larger vessels in the open sea is
currently limited to military appli-
In 1957, SEA LEGS, a Chris cations where the cost of additional
Craft cabin cruiser fitted with performance is justified by tactical
fully submerged foils proved that necessity.
a hydrofoil could operate with a
fully automatic control and sta- To give you an idea of where we
bilization system. Verification presently stand, I'd like to quickly
acquaint you with the Navy hydro-
foils developed to date.

The very first U.S. Navy hydro-

foil, the HIGH POINT (PCH-!), became
operational in 1963. HIGH POINT's

best available copy.

Fig,-Z- SEA LEGS

of this system was an important

step because the submerged hydro-
foil concept depends upon auto-
matic control systems to sense the
motions caused by waves and create
forces to oppose them, thus al- Fig. 3 - HIGH POINT
lowing stable flight in the ocean.
The height at which the ship 120-ton, 116-foot hull is powered by
"flies" above the waves is deter- 2 310011P Rolls Royce Proteus gas
mined by the strut length. Dur- turbines. 1Her maximum foil-borne
ing the design process, NAVSEC speed is in excess of 40 knots, an
engineers consider the probable impyessive increase over hullborne
distribution of wave heights to ships.
be encountered and design an ap-
propriate strut length so the re- The PLAINVIEIW (AGEII-1) at 320
hicle can fly without hull impact tons is the largest hydrofoil in the
or foil broaching. world. She has had all the aches of

2
 i

waterjet propulsion. Powered by a

3300HP Rolls Royce gas turbine en-
gine, water is sucked through rear
foil struts up into the pump and
jets out at 100 tons per minute.

MAM

Fig. 4 - PLAINVIEW

a developmental program -- a ___ _ _

development in which size is one

of the technological problems. Fig. S - FLAGSTAFF

But it is really the Patrol

Gunboat Hydrofoil (PGH) program
which has demonstrated the hydro- Cpbesta:a"
foil's ability to operate in high
seas and its suitability for the
50-knot Navy. During service in :1
Vietnam, 2 hydrofoil gunboats
proved they can operate in seas
above their designed sea state;
that gas turbine-water-jet pro-
pulsion is rugged and reliable;
that the electronic/hydraulic
control system is dependable and
that a minimum of shore support is
required. All of which adds up to
the fact that hydrofoils can pro- "
vide a fast, reliable, all weather
combat system. Fig. 6 - TU.CUMCARI

These two 60-ton gunboats are Last winter, highly successful

the FLAGSTAFF and the TUCUMCARI 152mm gun trials were conducted
(PGH-l and #2), both with fully aboar the FLAGSTAFF.. These. tests
retractable foils. The FLAGSTAFF verified our prediction that the
foils are arranged in typical air- ship could "take" a recoil load
plane fashion (2 forward, 1 aft) equal to the 60-ton displacement of
with most of the lift provided by the craft. More simply, it proved
the forward foils. The TUCUMCARI that the 60-ton ship could stay
was the first hydrofoil with foilborne when the recoil force of

3
the gun equalled the weight of the
ship. (Pounds mass equal to
pounds force.)

The hydrofoil's ability to

operate at high speeds in a high
sea state is its main selling ACV SES
point. The reason the hydrofoil AIR CUSHION VEHICLE SURFACE EFFECT SHIP
can maintain high speeds in the
flying mode is because the ship's
hull is above the water's wave
action and resistance is greatly
reduced, resulting in a fast,
stable ride. Because the hydro-
foil's hull is clear of the sea- SURFACE CONTOURING SURFACE PIERCING
way, it loses very little speed SIDEWALLS SIDEWALLS
in higher sea states. For exam-
ple, sea state 5 (10' waves) Fig. 7 - ACV SES
would reduce speed about 12%
while small destroyers would have with a flexible skirt at the bow and
their speed cut in half. stern only. This configuration is
not amphibious, but should have
Now, let's consider air greater potential for higher speeds
cushion vehicles and surface ef-' and efficiency in larger sizes. Ei-
fect ships which are less ma- ther configurati;,n can have air pro-
neuverable but more adaptable. pulsion or water propulsion but, of
The concept behind these craft course, the air cushion vehicle re-
can be traced back to the early quires air propulsion to maintain
18th century (Swedenborg in its amphibious capability. Because
1716), but they are the newest water thrusters have higher effi-
in terms of development. Gen- ciency than air propellers, SES will
erically air cushion and sur- probably have either a water jet or
face effect ships are the same, a supercavitating propeller.
in that they both operate above a
surface supported primarily by a
self-generated, contained cushion
of air.

The distinctions between the

two can be seen in Pig. 7. The
term Air Cushion Vehicle has come
to mean a craft with a flexible
seal system extending completely
around the periphery of the ship. A Ll
The flexible skirt not only con-
tains the supporting air cushion -
but allows the craft to be am-
phibious, its most important
characteristic. -n contrast, the
Surface Effect Ship generally has
rigid surface piercing sidewalls, Pig. 8 - SKMR-l

4
 Early U.S. Navy efforts in this
field are usually identified with
the 30-ton test craft designated
SKMR-l shown in Fig. 8. Built in
1963, it remains to this day the
most capable ACV designed, built w-:. "
and operated in this country. ..-
Supported by her air cushion, the
ACV can travel easily from land, 36M~
where she can go over any obsta-
cle no higher than her 4 foot
skirt height, through the surf to
being completely waterborne. This
capability offers great potential
for intermodal transport situations.
It can eliminate the need for off-
loading and transfer of cargo at
the shoreline.
Fig. 9 - Bell design
The British, who call them
hovercraft, have produced a number
of successful air cushion designs,
both large and small. One 177-ton
British ACV commercially operated
as a ferry can carry 254 passengers \0,95
and 30 cars. But we now feel we
are catching up. While English
developmnent has been largely
commercially oriented, the U.S.
has pursued primarily military
applications.

The Navy is interested in air

cushion craft both because of their
amphibious capability and because
of their high speed potential,
particularly in rough water. This
interest has fostered a Navy
program to develop several config-
urations of amphibious assault
landing craft, initially 4 150-ton Fig. 10 - Aerojet design
ACVs of two different designs.
Shown in Fig. 9, the Bell design ultimate output of this program of 4
features, for the first time on a ships is the technology from which
large craft, "bow thrusters" which NAVSEC can design the operational
can be trained in any horizontal machines.
direction to improve low speed
manuevering and control. This Another program just getting
Aerojet design incorporates traiin- underway, the Arctic Surface Effects
able air propellers and a new type Vehicle Program, has as its goal the
of lift system (see Fig. 10). The development of ACVs for Arctic oper-

5|
ations. The amphibious capability
is attractiye here because the Rpodced
craft must travel over a variety bst a
of surfaces -- water, ice and
land -- and over nbstacles.

Newest of the high perforn-

ance ships in terms of develop- --. .--
ment is the Surface Effect Ship. -
It was 1967 when the Joint Surface - r . .
Effect Ships Program Office
(JSESPO) was established with a
charter to develop large (about
4000 tons), fast (80-100 knots in
smooth water) surface effect
ships. The "Joint" in JSESPO's
title referred to the fact that
initially the program was sup- -=-
ported cooperatively by the
Commerce Department and the Navy. Fig. 12 - Supercavitating propeller
It has recently become exclusively
a Navy program under a NAVMAT major differences between them is in
Program Manager. the propulsion system -- one is
powered by waterjet (Aerojet) and
As the first major step the other with a new type of super-
toward realizing their goal, the cavitating propeller (Bell). Both
program has just completed the crafts are presently being outfitted
construction of the two 100-ton, and undergoing systems check-out,
80-knot test craft shown in these and are expected to be operating
Figures (11 & 12). One of the within a few months. We certainly
hope these ships demonstrate the
potential indicated by the lengthy
analytical work and model testing on
which they are based.

This then is where we stand at

present in the high performance
- field -- at the state-of-the-art, so
" to speak. We have utilized existing
knowledge to its fullest extent and
require breakthroughs before signif-
icant advances can be made.

For example, consider the cavi-

tation problem. When a hydrofoil is
flying through the water, cavita-
tion -- the boiling of water due to
local low pressure regions -- occurs.
This cavitation can erode metal
surfaces as though a strong acid were
Fig. 11 - Waterjet propulsion applied. Fig. 13 shows cavitation

6
 iable angle of attack environment
induced by the seaway is a difficult
control problerm. To sum up it's a
long way from being reduced to prac-
tical engineering and experimental
facilities are minimal.

SUBCAVATING FOIL

Fig. 13 - Cavitation damage C

damage which had occurred on the $UPERCAVITATING FOIL

HIGH POINT's aft propeller dfter
15 hours of use under full power Fig. 14 - Cavitating foils
conditions. The cavitation
barrier occurs at about 50 knots. Another unique problem of a mili.
Until we break through this cavita- tary hydrofoil is the interfacing of
tion barrier, we are limited to weapons systems with the vehicle.
speeds below 50 knots. The motion of a hydrofoil is more
like the vibration of an aircraft
There is an answer to the di- than the rolling, pitching motion of
lemma, and that is to learn to op- a conventional ship. This requires
erate in the supercavitating re- the development of new stabilization
gime. This solution is a dramatic devices to adapt the weapons to this
indication of the technological unique motion.
problems involved. Fig. 14 shows
the two foils. If the subcavitat- Also peculiar to hydrofoils is
ing foil is pushed too fast the ef- the problem of noise interference
fects of cavitation occur directly with the proper operation of the
on the foil, causing revere damage autopilot's sonic height sensor.
and loss of control. The cavity on Sensors presently in use have been
the supercavitating foil doesn't found to give erroneous readings when
collapse until it's well aft of the they receive not only their own sig-
foil and, therefore, prevents the nals, but the noise of certain types
detrimental erosive effects from of aircraft, helicopters, and the
taking place. But the supercavi- firing of automatic weapons. These
tating foil also has its drawbacks erroneous readings have caused foil
in that it doesn't provide as much broaching.
lift as the subcavitating foil, and
the drag factor is increased. Main- Common to all high performance
taining a stable cavity in the var- craft is the need for light weight.

.7
While it has been found that rivet- light, compact, potent in the naval
ed aluminum structures have very warfare environment and suitable for
low weight, the working of the hull maintenance at sea. We should be
over time can cause loose rivets, thinking aircraft technology--but ap-
These loose connections will re- plication of that technology to a
sult in the re-radiation of elec- rugged marine environment.
tromagnetic energy, interfering
with the proper operation of elec- In discussing weapons systems,
tronic countermeasures and fire it becomes apparent that the physical
control systems. parameters of some foreign weapons
systems are better than ours. Their
All of these interface prob- weapons require lower manning and
lems point to the possibility of produce "more bang per pound." Plus,
concurrent weapons and platform de- their weapons have more R&D behind
velopment where feasible in order them.
to obtain the optimal warfare
system. We are in the early phases of
the PHM (Patrol Hydrofoil Missile)
As you can see, few engineer- which has both surveillance and at-
ing situations have clear-cut solu- tack capabilities. Basically the re-
tions. Our job at NAVSEC is to quirement is for patrol boats which
study the various alternatives and can exhibit high speed and outstand-
make the necessary tradeoffs to ing performance in a high sea envi-
achieve the BEST, cost-conscious op- ronment. The Navy's existing hydro-
erational capability, foils have demonstrated that hydro-
foils can satisfy these requirements.
In addition to state-of-the- However, we have had a bit of trauma
art restrictions in the high per- in our initial studies for we wind up
formance field, we are also nearing with a choice of foreign-made weapons
the limits of R&D technology to for these ships because, by virtue of
place aboard these craft. For ex- being smaller and lighter with com-
ample, all high performance ships parable firepower, they are more at-
are weight critical. tractive. Foreign designers have
made some concessions, such as barrel
Range is limited due to con- life of the guns, but the important
flicting requirements for light point is that foreign weapons are
weight and fuel payload. We need closer to what we needed than any-
longer range propulsion systems; thing the U.S. has produced.
that is, propulsion systems with
higher overall efficiencies because Lest I've placed too much empha-
fuel and payload are tradeable. sis on these small ships and their
fascinating problems. The Navy has
We don't have a suitable menu many missions and many types of ships
of weapons and sensors for these to accomplish these missions. We
ships. As designers, we of the en- need a balance of capability -- sub-
gineering community examine various merged, in the air and on the sur-
systems possibilities and try to face. We have a deficiency in the
bring the right system on board. kinds of ships I've been discussing
But the plain fact is, what we need today -- small, low cost, high per-
for high performance ships doesn't formance ships -- ships capable of
exist. We need systems which are patrolling and controlling large

8
portions of the ocean with effec-
tive weapons delivery or transport
capability. The technology of
materials and power packages now
permits us to consider high per-
formance with reasonable pay loads
and the technology of today's weap-
onry apparently permits us to give
the small package a massive de-
struct power.
However, even though the tech-
nology of the combat suit apparent-
ly exists, it has not yet been re-
duced to engineering practice --
and our job as ship designers and
developers -- to design and build
operational ships -- is complicated
by these gaps.
Thus, for this special class
of craft, for these new ships of
the Navy, we all have to shift our
thinking and remember that for
these applications we are space and
weight constrained. We do need the
maximum of automation. It does re-
quire a new dimension in Navy think-
ing and a new dimension in technical
support by industry -- but the re-
wards are a new dimension in naval
capability -- which we surely need
to be the strongest Navy in the
world.

9
 SPECIFICATIONS
SURVEY OF VIBRATION TEST PROCEDURES IN USE BY THE AIR FORCE

Wayne B. Yarcho
Air Force Flight Dynamics Laboratory
Wright-Patterson Air Force Base, Ohio

A survey has been ronducted of specifications and standards containing

vibration test procedures in use by the Air Force, to assess progress
toward the establishment and maintenance of uniform procurement guide-
lines. A representative sample of hardware specifications selected
from the DOD Specification Index was examined to establish the direc-
tion end extent of use of vibration tests offered in the various test
specifications and standards. A number of documents presenting
vibration test methods were reviewed to determine what procedures are
available for application to hardware items. Conclusions relative to
the current status of vibration testing methods are presented.

INTRODUCTION

The effect of the vibration environment on MIL-T-5422 (BuAer) for Naval aircraft electronic
the operational performance of military vehicles equipment, NIL-E-4970 for Air Force ground
and equipment has been recognized as a problem support equipment and HIL-STD-202 for electronic
for many years. However, prior to the early and electrical component parts. By 1962, I
1940's, little standardization existed in regard test specifications and standards were in exis-
to requirements and test procedures. Test tence, as indicated in Figure 1, and the number
philosophies and techniques varied among civi- of "standard" environmental tests had increased
lian manufacturers and government and commercial to the point that some attempt to restan?ardize
testing laboratories, for the hardware items for the general testing area was deemed necessary.
which they were responsible.

After the start of World War II, an urgent

demand for new, reliable, air and ground
vehicles and equipment brought out the need for in-LW OUNVI?
standard testing procedures in the field of -"-cuLA-Tv TOAL4
vibration. To satisfy this requirement for 4 15 4
uniform procurement guidelines, AF Specification
41065 was published on 7 December 1945, w -
containing what were apparently the first 10,

standard Air Force vibration test procedures.

Revisions to the original specification were
published at intervals during the next few years, 4-
as testing practicE developed. 2

14 4 44 5 5 5 S 5 U6 9 T 1
In August 1950, Specification 41065 was
converted to HIL-E-5272, with additional vibra- YEAR O0MIN
tion tests included. At about the same time,
other documents presenting vibration and other Figure 1. Total of Vibration Test Dccuments
environmental tests began to appear, including 1945 - 1971

Preceding page. blank

Accordingly, in June 1962 MIL-STD-810 was documents described instruments (17%), aircraft
issued, in an effort to combine at least the accessories (5.OZ),engines (2%), and miscel-
major test aspects of specifications into one laneous items (camera lenses, fire extinguishers,
document and standardize vibration and other etc. (%), as shown in Figure 2. These results
environmental test procedures to reduce con- are believed to be generally indicative of the
fusion. The "B" revision of the standard, dated types of Air Force equipment which require
15 June 1967, Notice 1 dated 28 October 1969, vibration testing. - thou- a survey in greater
and Notice 2 dated 29 September 1969 are cur- depth (of the entire specification index, for
rently in effect. Some of the older documents example) might result in minor alterations of
such as MIL-E-5272 and MIL-T-5422, have been the percentages of hardware items requiring
reclassified as "limited" (not to be used for vibration tests, it is believed that the overall
development and new procurement) in an effort result would be similar.
to gradually phase them out. The current survey
was undertaken to assess the progress toward
this objective by determining the extent to
which MIL-STD-810 is called out in hardware
procurement specifications.

70 $2
SURVEY PROCEDURE
s0
The DOD Specifications and Standards Index
contains ipproximately 43,000 document titles 50 P0AIN TOTHE 34
used in military procurement. Hany, but not ARIARtE[ $OFICATIONSI
THAT CONTAINED
all, of these documents contain descriptions * 40 VSRATIONTESTS
of quality assurance tests to be used to deter-
mine acceptability of the item concerned. The 30
vibration tests specified are usually selected
from the procedures described in test specifi-
cations or standards of the type discussed o 10%
above. These relatively few test specifications
and standards are identifiable by title from the 10 5 60%
Index, but no such indication is provided as to I
which hardware specifications contain vibration
tests. This necessitated review of many (CTf Isi ACFT (NO ISSC
documents to obtain this information. Because
of the number of hardware specifications (AJUt4 TYPS
involved, a detailed review was not considered
feasible. Instead, an examination was made of Figure 2. Types of Equipment Most Often
a representative sample of the documents listed Subjected to Vibration Test
in the DOD index, selected from those Federal
Stock Classes (FSC) pertaining to Air Force
materiel. Approximately 6,000 hardware speci-
fications were chosen for review. REVIEW OF TEST SPECIFICATIONS AND STANDARDS

In addition to the review of the hardware

REVIEW OF HARDWARE SPECIFICATIONS specifications, a number of test specifications
and standards as listed in Table I were compared
A total of 5,917 detailed hardware or to determine the number and type of vibration
equipment specifications was selected at random tests presented therein.
from 14 FSC groups pertaining to Air Force
materiel. 1,381 or approximately 23% of these
contained vibration test descriptions.

In most specifications, when a vibration

test was described, reference was also made to
the test specification from which it came. In
other cases, the orlin of the tests was not
identified. Sometimes the origin could be deter-
mined from the test details, but generally, the
modifications of the parameters of a standard
test made identification of origin impossible.

Of the hardware specifications found to

contain vibration tests, 69% pertained to elec-
trical or electronic equipment (generators,
motors, radio sets, capacitors, tubes, relays,
resistors, fuses, etc.), and the remaining

12
 Table 1
Test Specifications and Standards

Date of Last
Number Title Origin Revision

MIL-E-5272 Environmental Testing of Aeronautical 12/7/45 1/20/60

(formerly 41065) and Associated Equipment

MIL-E-5009 Turbojet and Turbofan Aircraft 6/14/46 11/13/67

Engines

MIL-T-945 Test Equipment for Electronic 7/2/47 4/11/68

Equipment

MIL-E-5400 Airborne Electronic Equipment 12/1/49 5/24/68

HIL-T-5422 Environmental Testing for Airborne 12/1/49 11/15/61

Electronic Equipment

MIL-STD-202 Test Methods for Electronic and 1/29/53 4/14/69

Electrical Component Parts

MIL-STD-167 Mechanical Vibrations of Shipboard 1/29/53 4/14/69

Equipment

MIL-C-172 Vibration for Aircraft Electronic 12/15/53 12/8/58

Equipment Cases

MIL-T-7743 Store Suspension Equipment Testing 12/5/56 3/22/62

MIL-T-4807 Ground Electronic Equipment Vibration 10/7/58

and Shock Tests

MIL-STD-750 Test Methods for Semi-Conductor 1/19/62 8/26/68

Devices

MIL-STD-810 Environmental Test Methods 6/14/62 6/15/67

MIL-STD-1311 Test Methods for Electron Tubes 4/19/68 7/10/69

MIL-STD-883 Test Methods for Microelectronics 5/1/68 5/1/58

All of the above-listed test specifications Table 1, such as MIL-STD-202, MIL-E-5272, etc.
present various types of vibration test descrip- To show the type and number of vibration test
tions as illustrated in Figure 3 or contain procedures involved, these documents are again
references to tests in other specifications in tabulated by title in Table 2.

Table 2
Types of Test Procedures in Test Specifications and Standards
Endurance Random Temp. Weight Allow Total
Spec. No. Resonance Cycling Steady Cycling Vib. Vib. Vib. Vib. Tests
MIL-E-5272 4 4 1 1 0 1 1 12
MIL-E-5009 1 0 0 0 0 0 0 1
MIL-T-945 1 2 0 0 0 0 0 3
MIL-E-5400 0 4 0 0 0 0 0 4
MIL-T-5422 2 0 0 0 0 1 0 3
MIL-STD-202 1 5 0 0 1 0 0 7
MIL-STD-167 I 1 0 0 0 0 0 2
MIL-C-172 2 0 0 0 0 2 0 4
MIL-T-7743 0 0 1 0 0 0 0 1
MIL-T-4807 0 1 0 0 0 0 0 1
MIL-STD-750 0 1 1 0 0 0 0 2
MIL-STD-810 14 6 0 0 2 1 14 37
MIL-STD-1311 0 2 3 1 0 0 0 6
MIL-STD-883 0 1 1 0 0 0 0 2
TOTAL 26 27 7 2 3 5 15 85

13

A
 (3) the location of the equipment in the vehicle
26 - (engine-mounted or mounted on vehicle structure).
Also, special procedures are applied which
:t depend upon whether the equipment is attached
directly to the structure (bard mounted) or
19 mounted on isolators. Furthermore, the selec-
14- tion of test procedures depends upon the type
it of possible malfunction which is under investi- A
10 gation, such as a possible fatigue failure or a
a" possible instrument error due to vibration.

l.oA very common test procedure indicated in

Table 2 is a resonance search followed by
-. - resonance dwell testing. This test iz normally
ctCa. EM DM PmAM WT. intended to investigate possible fatigue damage
rrmCYCL, rM retsU VWI T
Au in equipment which is to be installed in air-
v1 orAT"
TUSTMWcoEU planes and helicopters. The resonant dwell may
Figure 3. Content of Vibration Test Documents be followed by cycling tests to check for equip-
ment malfunction or for internal resonances
APLICATION which are difficult to detect. Tn other cases,
only cycling testing is required, especially
The application of the vibration test pro- for equipment in a vehicle with a short service
cedures described in the test specifications life such as an air launched missile. Table 3
and standards listed in Table 2 depends upon shows a rasige of test parameters for typical
(1) the type of hardware or equipment, e.g., a cycling type tests and for a variety of equip-
mechanical or electrical device, (2) the type ment locations. Resonance dwell and cycling
of vehicle in which the equipment is mounted tests are conducted with sinusoidal type exci-
(missiles, airplanes, helicopters, etc.), and tation.

Table 3

Cycling Test Parameters

Double Total
For Equip. Ampl. Accel. Freq. Vib.
Mtd. On (Inches) (G's) (Hz) Time (Hrs)

Recip. & Gas Turb. Eng. .036-.10 1-20 5-500 9

Turbo-Jet Eng. .036-.10 1-20 5-2000 9

Aircraft Struct. on .010 2 5-500 9

Mounts

Helicop. Struct. on .036-.10 2-5 5-500 9

Mounts

Helicop. Struct., No .10 2 5-500 9

Mounts

Air-Launched Missiles .036-.10 1-10 5-500 6

Captive Phase

Air-Launched Missiles .036",.10 1-10 5-2000 1.5

Grnd-Launched Missiles .06-.20 1-50 5-2000 1.5

Grnd. Supt. Vehicles .06-.10 1-50 5-2000 9

An additional type of test with sinusoidal some types of endurance tests. MIL-STD-8108
excitation is the endurance test, also listed does not specifically call out endurance tests
in Table 2. This type of test is a fatigue test, as such, but other specifications such as MIL-
and it is generally of longer duration than the E-5272C, do recommend this type of testing.
previously indicated resonance or cycling test.
Table 4 indicates parameters associated with

14
 Table 4

Endurance Test Parameters

Double Freq. Total Test

Equip. Ampl. Accel. Range Time
Type (Inches) (G's) (Hz) (Hours)

Hicroelect. .06 20 20-60 96

Store Susp. .03 10 50 300

Eng.-Htd. .01-.05 20 15-250 36

(on turbo-jet
or turbo-fan)

Only two of the specifications, HIL-STD- dom vibration tests are conducted, no resonance
810B and MIL-STD-202, recommend randL% vibration search or cycling is required.
testing. Table 5 indicates typical pLrameters
for random vibration tests. Mien wid.-band ran-
Table 5

Random Vibration Test Parameters

Power Spectral Overall Total

Equip. Type Density PMS "G" Time

All equipment attached .02 5.2 Three minutes to eight

to structure of air .04 7.3 hours in one or three

vehicles and missiles .06 9.0 directions as specified

powered by high thrust .10 11.6 in detail specification.

jets and rocket engines .20 16.4

.30 20.0

.40 23.1

.60 28.4

1.00 36.6

1.50 44.8

In addition to the above test procedures, found to contain the majority (74%) of the
some of the test specifications listed in Table vibration tests described, which consisted of
2 provide for combined environment type testing cycling, combined cycling-resonnnce, steady
and make special allowances for the weight of state, and random excitation.
the item being tested. Vibration tests under
conditions of high temperature are conducted to
investigate possible malfuntion of equipment
exposed to sources of heat such as electronic
gear operating in closed compartments, on
engines, or near rocket exhausts. Provision
is made for vibration testing of heavy items by
test procedures which specify a progressive re-
duction in vibratory acceleration when the
weight of the tegt item exceeds 50 lbs., de-
creasing ultimately to a minimum of 50% of the
test curve.

Oi the 14 general requirements documents

reviewed, five, as indicated in Table 6, were

15
 Table 6

Test Specifications Most Often Referenced

HIL-STD-202 Test Methods for Electronic and Electrical Component Parts

MIL-E-5400 Airborne Electronic Equipment

MIL-T-5422 Aircraft Electronic Equipment Environmental Testing

MIL-STD-810 Environmental Test Methods
MIL-E-5272 Environmental Testing of Aeronautical and Associated Equipment

Figure 4 shows these test specifications or out. Test procedures in the above five test
standards from which vibration test procedures specifications most frequently specified are
were most frequently referenced and the percent given in Table 7 below:
of all hardware specifications that called them

Table 7

Test Procedures Most Often Referenced

MIL-E-5272 Meth. V (cycl.) and XII (Resonance and cycling)

MIL-STD-202 Meth 201A and 204B, Cond. A (cycling)
HIL-STD-810 Meth. 514, Proc. I, Curve B (resonance and cycling)

HIL-E-5400 Sim. to Meth. XII, MIL-E-5272

HIL-T-5422 Sim. to Meth. XII, MIL-E-5272

DISCUSSION

HIL-STD-810 is the most detailed of the 55

General Test Specifications, has been recently W 51 %
revised, and contains most vibration tests 45 ,
presented in any of the other documents. In 40
addition, MIL-STD-810 is the only test specifi-
cation of those in this study to describe a 0
r 35- -
gunfire test (Notice 2 (USAF), 29 September o
1969) and one of two (the other is MIL-STD-202)
to present random vibration test procedures. It
is however, referenced much less often than I 14[
MIL-E-5272 which presents vibration tests formu- z 0]r
lated for the air vehicles and equipment of 15 45%145%
to 20 years ago. The reason for this infrequent 0
reference to an updated, improved specification 0 8
apparently lies In tue regulations concerning A
automatic review o! specifications at five-year E
± 8±
intervals. During these reviews, if test i j A IS
requirements have not changed, the test document
reference may be left unchanged even though it Figure 4. Vibration Test Documents Most Often
is in the "limited" category. Referenced

HIL-E-5272 is the test specification tion tests currently presented. The fact that
referenced as a source of vibration test proce- many of the hardware specifications reviewed
dures in about 51% of the hardware specifica- had not been updated for the last 15 to 20
tions. HIL-STD-202 is referenced in about 12% years might be a cause of many of the current
of the specs, MIL-STD-910 in 11%, MIL-E-5400 vibration problems. A review of outstanding
and NIL-T-5422, about 2% each. Approximately test procedures and comparison with what is
22% of the vibration tests are described in available in HIL-STD-810B is apparently needed.
detail but not identified as to origin.
Method 514.1 "Vibration", the section of
No positive indication has been obtained MIL-STD-810 pertaining to vibration test pro-
recently in regard to the adequacy of the vibra- cedures, presents a wide variety of test methods

16
and is the most complete and up-to-date of the CONCLUSIONS
test specifications (or standards) reviewed.
However, in an effort co avoid repetition of 1. Although MIL-STD-810 contains most
certain descriptive material common to a number vibration test procedures presented in HIL-E-
of test procedures, the tabular and graphical 5272, plus a number of additional tests, and
data have been highly condensed, to what some has been available for use since 1962, only 11
believe is a confusing extent. Detachment of percent of hardware specifications containing
vibration tests to form a separate vibration vibration test requirements reference MIL-STD-
test standard might be an answer, although this 810A or 810B.
suggestion has not met with acceptance in the
past. 2. Over 50 percent of the hardware specifi-
cations reviewed referenced the original
Also, results of recent vibration surveys requirements of HIL-E-5272. These requirements
have not yet influenced test specifications are based on 1950-55 technology. It appears
and standards. For example, according to data that hardware specifications should be revised
presented by Bolds and Ach (1), the test level to reflect the latest Vibration Technology.
specified in the Helicopter Vibration Test Curve
"M" in HIL-STD-8lOB is inadequate as to frequency 3. It would appear the most serious defi-
range. Inadequate design and testing of heli- ciencies of some specifications are the limited
copter equipment to realistic levels of vibra- frequency ranges and the limited use of random
tion in the 5 to 10 Hz range and the absence vibration testing requirements.
of requirements in the 500 to 3,000 Hz range
may possibly account for many current field 4. No index exists of 'tardware specifica-
equipment failures. In addition to sinusoidal tions containing vibration test requirements.
vibration testing of equipment mounted in A survey of hardware specifications to identify
helicopters, there also exists a need for random those that contain vibration tests is needed to
vibration testing in the 300 to 3,000 1Hzfre- permit compilation of an index which would be of
quency range (1). assistance in determining requirements for
development of new and improved vibration tests.

REFERENCES

(1) Phyllis G. Bolds and John T. Ach, "Inflight

Vibration and Noise Study of Three Helicopters,"
The Shock and Vibration Bulletin, December 1970.

17
 SPECIFICATIONS
A Panel Session
Moderator: Henry C. Pusey, Shock and Vibration Information Center
Co-Moderator: Clyde Phillips, 6585 Test Group, Holloman AFB

Panelists: Dave Earls, AFFDL, Wright-Patterson AFB

Robert E. Wilkus, ASD, Wright-Patterson AFB
A. R. Paladino, Naval Ship Systems Command
D. M. Lund, Naval Ship Engineering Center
Daid Askin, U. S. Army Frankford Arsenal
0. A. Biamonte, U. S. Army Electronics Command

OPENING REMARKS BY THE PANEL

Mr. Earls: I am going to talk about the costs for the whole operational system. About
effectivness in specifications and standards 20 to 40 percent of the failures that we are
from the standpoint of environmental failures having in operational aircraft are environmental,
that are occurring in service. I want to tell you and they are costing us around 16 million a
about a few environmental failures that occur year. Equipments in this particular airplane
in operational aircraft. We conduct an array of were qualified to MIL-E-5272 and MIL-E-5400.
tests using all these specifications and standards, I think it is significant that 53 percent of the
and we assume that after 2 to 5 years these air- field failures from vibration also occurred in
craft are going to be operating satisfactorily, the qualification tests. This means that when
What is the effectiveness after 2 to 5 years? they ran the test they got failures, but in many
We had a contractor make a study of this. It cases did not correct them. To solve this
was a two year field failure investigation of 175 problem we must provide good failure criteria.
aircraft scattered mostly in Vietnam with some We are putting failure criteria in MIL-STD-810
on U. S. coasts. They made a detailed analysis which we hope will be directed toward correct-
of the equipments that were failing and tried to ing this. People are not really correcting the
pin-point the environment that caused the failure, failures, they arc not retesting equipment and
They studied the failure analysis reports and they are not getting fixes into operational sys-
other documentation and made a good engineering tems. I have mentioned some environments
analysis of a selected group of these equipments other than vibration but I think most of us are
that they thought were failing environmentally, environmental people in general anyway. Pres-
It turned out that from the selected group 52 ent tests do not consider the frozen moisture
percent were iling environmentally. Twenty- situation on an airplane taking off in the tropics
one percent of the failures over the two year and going to altitude. The equipment freezes
period were because of temperature, 14 percent and will not work at altitude, but when it is
were from vibration and 10 percent from checked out back on the ground, it is alright.
moisture. The rest were from related environ- We have to start from scratch on this problem.
ments such as sand and dust, salt spray, altitude Many failures occur from forced air cooling
and shock. The 52 percent of equipments that and we expect to put a test for this into the
failed for environmental reasons represents standards. Some of our new aircraft systems
12,000 failures per year, which is quite a goof- in the Air Force already have these types of
up. What is this costing us? Fifty-five percent tests. We are working right now on a new
of all environmental failures are from temper- temperature-altitude-humidity test which we
ature, and they cost us four and one-half million want to put in MIL-STD-810. If we can take
dollars a year in one operational aircraft weapons care of some of these temperature, humidity
system. Moisture accounts for 1 1/2 million and vibration problems, we will eliminate a lot
and vibration 1.4 million. Temperature, moisture of the field failures. I think random vibration
and vibration are therefore tremendous problems may be one of the answers. With our old test
in operational aircraft. Out of these selected methods using sinusoidal dwells, one cannot
equipments, these environments cost 8.3 million really find the resonances and the ones you do
dollars per year. We hen estimated what it find are not the significant ones. I thinkwe need

Preceding page blank 19

to test all the resonances using random vibra- documents, but this did not mean anything. We
tion. We really need a functional test and an knew from some B52 experience, that in the
endurance test. People talk about how high the back end of the B58 we could expect at least
test levels are. As far as I am concerned they 15 g. This was how the Mahaffy study came
should be functioning at that level. We should about. It was a B58 contractor's specification,
even have a higher level for endurance because an estimate and prediction of what the environ-
there is a lot of fatigue in this equipment. ments were and a derivation of the test spec-
Equipment is breaking and falling apart in ifications that was reasonable. This is what
service. In general, standards should be flex- we adhere to in the Aeronautical Systems
ible. You have to have an environmental engi- Division. We will argue about what is reason-
neer or somebody who can apply engineering able, but define what the environment is and
judgement to a standard like MIL-STD-810. then proceed with the test. There are a good
You cannot expect to apply it across the board many people who are arguing that the specifi-
without somebody putting in a little knowledge. cations are too high, and they are spending more
You have to have some flexibility and adapt- time arguing than testing. This is where a real
ability. We expect to put more flexibility into problem exists. Our prime function is to pre-
MIL-STD-810 if we can. I also feel that these pare aircraft specifications, but we do get out
specifications and standards must be developed and do some testing. We do not know what we
for combat environment. Think of the mainte- are doing unless we know something about the
nance and man hours, 100,000 per year, just on environment. We probably put more instru-
the thirty equipments in this operational system mentation in the RF.4C than has ever been put
I was talking about. I think anything we can do in any airplane. We are utilizing this experi-
to improve these specifications and standards ence. We found that in the RF-4C no one even
to make them really work certainly justifies considered the vibrations in the outer wing
some expense, considering the cost of failures where levels were so high the wing was stalling.
in service. A compass transmitter was failing. We found
that the transmitter was tested at 10 g, but it
Mr.Wilkus: It is pretty hard to apply a was on a rigid block. It is never installed in
standard; they are really only guides. They the airplane this way. Nothing was measured
might be used as guides to industry as to what during the test. There was a requirement on
levels they must develop equipment. Over the accuracy but that was measured a couple of
years people have asked for the vibration envi- days later. These are just a few of our prob-
ronment of airplanes. It is like asking how lems which may be enough to generate a little
deep the ocean is, because the vibration varies discussion.
throughout the airplane. At Wright field, we
have an Aeronautical Systems Division in which Mr. Paladino: The standard that I am
we have Systems Project Offices or managers, going to talk about is MIL-STD-167. This
The managers of these systems have systems standard covers the requirements for naval
engineering groups which support all systems, equipment including machinery both for inter-
I am in an airplane directorate which is struc- nally and externally excited vibrations. There
tures-oriented. I am in the Structures Division are five types of test in this standard. Many
and have the Dynamics Branch involving vibra- people are not familiar with the five types.
tion and acoustics. There is also the Avionics Type I Ison environmental testing which in-
Directorate. There are a lot of people involved cludes testing for equipment to go on all types
in the procurement of equipment and a lot of of Navy ships. Type 2 Is on internally-excited
different attitudes. We have ours. The avionics vibration, which is better known as balance
people may well apply a specification or standard requirements. Type 3 is for torsional vibra-
that they know. We in dynamics, treating envi- tion which is usually associated with gas tur-
ronmental vibration as structural vibration, bines or high speed equipment. Type 4 is for
provide our own input to the requirements and longitudinal vibration which includes shaft and
carry It up to the test specifications. On a propulsion systems. Type 5 is lateral excita-
major system we are going to define the envi- tion associated with the shafting systems of
ronment, either by measurement or prediction. ships, which are the lateral bending modes. I
We are asking people to be reasonable. By will address myself primarily to Type 1, be-
reasonable I mean they must do it our way. If cause this really offers the most controversy.
a reasonable environment is 2 g, then that is A researcher in ship dynamics wants to under-
what it is. It may be 15 g like Mahaffy's pre- stand all the delicate forces that may act on the
diction on the B58. In the aircraft industry they ship, externally or internally, as well 9s under-
negotiate the test requirements. In the case of standing the structural response to such forces
the B58 MIL-E-5272 was in the contract and the vibratory energy transfer of the structure.

20
The ultimate concern is to bring the vibration but not one instance is recorded where equip-
to an acceptable level by having low input forces ment failed after it passed this test requirement.
if possible, and to provide design criteria for When designing equipment for shipboard use it
shipboard equipment so that it can endure its is best to avoid a natural frequency which will
dynamic environment. Shipboard equipment coincide with the natural frequency of a ship's
must be designed to enable the Navy to fulfill its hull. Such a design is possible for a selected
mission under any adverse environmental con- ship and a particular mode of vibration. The
dition. The philosophy in arriving at vibration use of one type of equipment on different ship
standards is a realistic approach that vibration classes makes the objective unattainable. As a
of shipboard structures cannot be prevented but taxpayer, gentlemen, logistics is an important
only minimized. Vibration is generated by shaft thing, to have one piece of equipment on many
and machinery imbalance, variable blade fre- types of ships. As such the provision of this
quency due to nonuniform weights, sea waves, standard to show that equipment is properly
wind, weapon firings and maneuvers. Equipment functioning even at a natural frequency falling
mounted on the hull experiences the ships greater within the testing range is considered mandatory
vibration level amplitudes by some magnification and shall not be by-passed. Finally, consider-
factor which should be kept as small as possible. ing the response of ships in calm and rough
It is necessary to consider certain functional seas, can we successfully design shipboard
threshold levels for equipment such as maximum equipment to assure the proper functioning dur-
levels tolerable to have the equipment operational, ing adverse conditions? MIL-STD-167 provides
levels at which the equipment is not operational the procedures for testing the adequacy of equip-
but functional so that checks may be made, and ment to assure that properly tested equipment
levels at which equipment is neither operable is placed on board ship. The frequency range
nor functiorai, but does not sustain structural substitution in the revision of this vibration
da.age or become a missile. Restrictions may standard has been taken down to 4 Hz and up to
also exist whereby ships may have to reduce 50 Hz. Adequate design of equipment for fre-
speed in heavy seas to minimize sliding or quencies below 4 Hz should not cause any diffi-
shipping green water. The area of ship dynam- culties when realistic calculation procedures
ics is rather complex and includes mechanical, are applied. Selection of the location on the
hydrodynamical and operational generation of ship for the installation of sensitive equipment
vibration. Response includes all kinds of vibra- must be considered from an early design stage.
tion patterns and mode couplings which are However, this equipment must also be submitted
different at various ship locations due to struc- to MIL-STD-167 testing.
tural configuration. Measurements at certain
locations may lead to erroneous conclusions or Mr. Lund: I would like to cover two basic
at best to results which cannot be generalized, points in my introductory remarks. First, for
Basically, the vibration amplitudes to which our non-Navy friends, I would like to give you
shipboard equipment should be submitted before a rough notion of just how the Navy specifies
installation on ships are given in MIL-STD-167. shock requirements. The basic Navy shock test
The philosophy in arriving at this standard is a specification is MIL-S-901C. This specification
realistic approach that vibratior of a ships' hull provides procedures for shock testing on the
and structure is considered inevitable. We must Navy's light weight and medium weight shock
accept this. The standards are kept uniform for machines and also on the floating shock platform.
all shipboard equipment. No differentiation is The specification does not require shock testing
made for either type of ship or location on the of anything. It applies only if it's involved by
ship with one exception, namely mast-mounted some other specification. The MIL-S-901C test
equipment, even though there may be areas on procedures are very generalized with the intent
the ship which may be vibrated at lower ampli- that once an item passes the test, it should be
tude levels than other areas, for example stern suitable for installation on virtually any ship and
versus midship. It must be mentioned that most in almost any location in this ship. In this
vibration standards for equipment testing for respect ii is somewhat siniilar in philosophy to
ships, even in foreign navies or foreign commer- the vibration specifications that Mr. Paladino
cial ships, consider MIL-STD-167 as a guide, was addressing a moment ago. Next step up the
The navies of NATO nations use it as a manda- ladder is the equivalent specification.usually
tory requirement. In fact there is no departure referred to as the MIL specification which may
from this standard. MIL-STD-167 was last re- invoke MIL-S-901C. It may say that you shall
vised in 1969. So far it has proven quite adequate test all valves or all turbogenerators to MIL-
in all cases where equipment has been actually S-901C, and that might be all it says. Hopefully,
tested in accordance with its requirements, the MIL specifications will support MIL-S-901C
There are many cases of failure due to vibration, by giving information, such as what the operating

21
conditions of the equipment shall be during the this test, that will not tell you If you are going
test. Which test fixture shall be used in the shock to pass the test. We have seen all orts of
tcst? What are the acceptance and rejection equipment pass the test because it is still func-
criteria? -The MIL specifications essentially tionally operable, yet it is quite a bit bent up.
support MIL-S-901C, but do not supersede it. If you had used the yield as your failure crite-
Finally, we have the shipbuildftg specifications; rion, you would have failed the test and would
these are specifications that the ship builder is have doubled the weight of the thing for nothing.
required to observe when he is building his I suggest that you consult and carefully observe
ships. First the ship building specification all the guidance given in MIL-S-901C. There
indicates which shipboard systems requires is really a lot of good information there. This
shock protection and which do not. For those information should suffice to get 95 percent of
systems which require protection MIL-S-901C you folks through the shock test with virtually
is directly involved, provided that the item in no problems at all. Finally, just to conclude
question is shock testable, which usually means with a word of reassurance, I point out that we
that it weighs less than 60,000 lbs. If the item have found that the majority of off-the-shelf
is not shock testable, a dynamic shock analysis items that are shock tested to Navy require-
procedure is invoked. I will not go into detail ments pass these tests. There might be some
on that procedure except to indicate that this minor modifications required, but not the kind
provides an analytical equivalent of the shock of thing that requires a full redesign. We end
test. It gives the Navy an indication of the up stiffening this part or that part and adding a
shock resistance of the equipment. That, in a lock washer here or there. It looks pretty
nut shell, is how the Navy specifies shock re- terrible when that ham mer comes down but,
quirements. from simple observation of the shock test re-
sults, it really is not quite that bad.
The second type is one that I am sure that
we will hear more from tonight, but I will try Mr. Askin: It is a pleasure to be here to
to stir things up a little bit. MIL-S-901C states talk about specifications and standards because
how to shock test something, what procedure to I have been in this field for many years. One
use, but it does not tell you how many g's you of the things that struck me from Wayne Yarcho's
are going to see. I get a number of phone calls talk is that MIL-STD-810B, which originated as
every year from people who ask "How many g's an Air Force specification, is only being used
am I going to see?" For the benefit of those 11 percent of the time by the Air Force. I
people, I would like to offer a few remarks to know, since the middle 60's, the rule in the
kick off what I expect to be a long discussion of Army had been to really try to use MIL-STD-
this matter. The first thing that I tell a man is 810B. I venture to say that we are using it for
that I do not know. I try to convince them that at least 50 percent of all our current new pro-
the Navy cannot tell him how many g's he will curement specifications. We are having troubles
see, because it depends on the dynamics of your with it, but I am convinced that it is the best
requirement and upon the dynamics of the test specification that exists to date. The biggest
fixture that you select. You can see from 5 g trouble with MIL-STD-810B is in Its misuse.
to 250 g, but I cannot tell you that until I know Many people that write specifications do not
exactly what your equipment looks like. If you really understand it; they do not have the
are still pressing it we will suggest you take a faintest idea of how to write or to even distin-
quick cut at dynamic analysis of the equipment. gulsh between environmental criteria and envi-
The Navy dynamic analysis requirements are ronmental test methods or procedures. Many
intended to address the same shock environment people confuse the limits of the environment
that we are trying to address by the shock test with a test specification. They will spell that
requirements. The dynamic analysis should, of out as a test requirement. So I think the big
course, include the test fixture and you will get problem is in really educating specification
some notion of how many g's you will be seeing writers and writers of procurement documents
during the shock test, I also find it is often on how to use MIL-STD-810B. There is a
necessary to point out that despite the fact that current revision of MIL-STD-810B now going
you have heard that 10,000 g's were recorded on in the three services and industry. They
during the last shock test of some equipment, have a big job on their hands in that there is a
you have no need to worry about this. This is lot of difference of opinion on the application of
high frequency garbage that has practically no this particular test specification. I think we do
damage potential. Just confine your worries to have a basic document that is worthwhile, and
the frequency range of interest, usually the first if its attacked diligently by the three services
one or two modes. Furthermore, even if I could and Industry in the present effort, we will end
tell you how many g's you are going to see on up with a much more usable document than we

22
now have. One of the things that we are now such a thing is that nobody has defined what the
trying to do is to issue along with the new stan- environments are. The reason for this is that
dard a separate document on how to use the there is an infinite number of variables re-
MIL Standard. I think this will really help a garding the induced environments. These are
lot. One of the big problems is to update the man-made environments, those which occur as
various procedures as we learn more about the the result of design of an end item such as an
actual environment. If you know what the true aircraft, a space vehicle, a ship or a ground
environment is, that is the environment you are vehicle. To try to define this is difficult al-
going to use in your test laboratory and forget though there is no doubt that with regard to a
about MIL-STD-810B. You can always get a specific application we can measure it and find
waiver on it if you can prove you know what t:he the environments. However, our basic speci-
true environment is. We, at my particular fications are designed to buy material for all
installation, have attempted to get the true applications. Therefore we need a generalized
environment on a fire control mounted in tanks specification. Such a specification does not
and helicopters because we felt that we have exist. The specifications that do exist call for
been overtesting a lot of the equipment, espe- tests based on past experience. There is a
cially the equipment meeting the existing MIL- difference between the requirements or criteria,
STD-810B. We made measurements on the and the test. We do have a three-section and a
mounting pads where our fire control equipment four-section to all military specifications. The
was to be installed and found in most cases, it three-section calls out requirements, the four-
was considerably lower than the curves that are section calls parallel or equivalent tests to
now in MIL-STD-810B. Another area in partic- prove out a particular requirement. I dare say
ular
dwellwith
test.which
ManyI find fault
p,-ople is the
spell outresonance
a resonance that industry
section. Theydoes
looknot
at care about the three-
the four-section and
dwell test on equipment for which we never find design to meet a test. This may or may not be
a resonance. Even if there was one, it would the equivalent of a real environment. With MIL-
be very hard to find it. Then, if you do find it, STD-810 we tried to arrive at a common de-
the specification says dwell for one half hour nominator for the Army, Navy and Air Force
at each resonant frequency and that is one sure for all applications. It is quite an impossible
way to kill almost any piece of equipment. It task, but we ended up in 1967 with a document
is not realistic because in the field you very that tried to do these things. There are many
seldom will find a resonance that occurs at a things wrong with MIL-STD-810, but my rec-
discrete frequency. We talked to the Air Force ommendation is to eliminate all documents
about this and we are proposing that maybe in which call out tests rather than criteria. Refer
some cases we ought to try a sweep around a to MIL-STD-810 for any test that is in a four-
resonance point instead of the resonance dwells. section of a specification. That is the only way
One other mistake that has been made is to use you are ever going to achieve any degree of
the same limits for vibration or shock testing standardization. Again remember the docu-
an item in all three directions. Very often you ment is not perfect but new people can help
know the equipment is going to be installed in make it perfect by its utilization. I realize
one orientation in which it will get its worst the importance of the degree of standardization
shock or vibration. To indiscriminately require that is required. In many cases there are
a standard test in all three directions at the occasions whereby in a specification, the end
same limits very often invokes a much more use environment is far more rugged than, for
severe test than you need. If the designer or example, the transportation environment and I
the specification writer will get together with find in reviewing MIL-STD-810 there are
the environmental people before they set up duplications of testing. If the environments in
their test program, they will have a much more which an end item is used are far more rugged
intelligent application of MIL-STD-80. than the handling and transportation environ-
ments, let us not test for handling and trans-
Mr. Btamonte: Documentation formulating portation. There has been some reference to
the basic requirements for factors both for the various procedures in MIL-STD-810. I think,
natural and induced stresses common to all if I hear right, we might end up with several
DoD military equipment is not available. We procedures for vibration. I do not understand
do have a document, MIL-STD-210 which deals why we should have more than two or three.
with climatic environments that are comnmon to Vibration is vibration. Why we need 13 or 14
Army, Navy and Air Force and Marines. There procedures is beyond me. The Army is con-
is no equivalent document covering the induced cerned, believe it or not, with aircraft and
environments for the Department of Defense. marine design. We have more helicopters than
Very obviously the reason that we do not have the Air Force and I am beginning to think we

23
have more barges and ships than the Navy, a single test document for all these
Therefore, it is high time that we end up with particular areas.

DISCUSSION
Mr. Panaro (IBM, Owego): We do environ- tell me that this is an unrealistic specification
mental analysis and testing on avionics equip- and say that our equipment cannot survive 0.2
ment, space equipment and shipboard equipment, or 0.3 g, I will not send my son on a submarine.
Presently, we are involved in shipboard equip-
ment, specifically submarine equipment. I feel Mr. Turkheimer (Wyle Laboratories): I
that some of the specifications we have designed think you have a very valid point when it comes
to were realistic. Why design for unrealistic to the survivability and the reliability of the
specifications if the environments are much equipment that your contractors are building
less than the criteria spells out? That was the and the test laboratories are testing. I think
point. As far as MIL-STD-167 is concerned, we there is a need for it. However, sometimes
are working to it. I do not think the specifications when you get down to the point of actually per-
for shipboard equipment are that unrealistic. I forming a test, you wonder how realistic the
do not think the shock specifications are unreal- specifications are. We should know what goes
istic. I do think that the specifications for on upstairs and what is really realistic. This
avionics equipment are definitely unrealistic, is one of the problems that I foresee as we
I have gotten some information from China Lake come up with more sophisticated pieces of
telling me that power spectral density levels equipment.
not exceeding 0.04 g 2/Hz were observed in
actual flight conditions and they are testing in Mr. Wilkus: I can recall a previous
specifications to 0.12 g2 /Hz, from 20 to 2,000 experience on Snark. Now this was at the
cycles. To me that is completely unrealistic, time when they had Snark infested waters down
here. There were some measurements the
Mr. Paladino: We get many statements contractor had made. The frequencies were
that the standards or specifications are unreal- limited to 500 Hz; actually they were limited
istic, but gentlemen, we have human lives in- in shake capability to something like 500 Hz.
volved. In particular, submarine type require- The instrumentation was indicating higher
ments, we have about 100 American boys on frequencies. They were measuring something
those ships. We have lost two of our ships now like 40 g which seemed impossible. The re-
in peace time and it is a sadness to all of us action of course is to start arguing about this.
that we lost them. We in the Navy have decided Actually you should just start taking a real
that we shall not lose a third one in peace time. look at the equipment. It was guidance equip-
If the skipper or a man on the control station ment and there were many problems with it.
calls for a command, and a piece of equipment The program was shut down and we set up a
does not respond immediately, that submarine program to go on a stand to fire the missile.
and 100 human lives may be lost. We cannot The question now was to U;et this stuff able to
play this kind of game. MIL-STD-167 is not meet the firing in natural environment. It was
unrealistic. Please do not look at it and go to tested to 40 g below 500 Hz. A little vibration
the Modal Basin or one of these Navy labora- testing will do a lot of good even at 500 Hz.
tories and get a specific report. These data You are principally ruggedizing the equipment.
are taken under highly controlled situations. This was done, and the first Snark that was
Everything is at a controlled speed. The maneu- fired after people really took this to heart and
vers are not there. It is humanly impossible to tested It and corrected difficulties in the equip-
set up in one document, for a situation with a ment went down range 6,000 miles the first
ship underway, what the environment is going time and hit the target. It can be done.
to be under shock or vibration. So what do we
do? We pick a set of frequencies and ampli- Mr. Johnson (Hughes Aircraft): My
tudes. Sure, you can argue that the amplitude question Is on MIL-STD-167. Much of the
is too high, but gentlemen all it is trying to do equipment we build is relatively large. In
is ferret out a resonance frequency. It is not almost every case these will be installed on
the standard that kills equipment. The standard surface craft as opposed to the submarine.
requires a maximum of 1.2 g at 25 Hz. Our Am I correct in assuming the primary reason
failures are not occurring at 33 Hz or at 25 Hz. for going to 50 Hz was to include submarine
They are occurring at 8 or 9 Hz; this is about gear? Some of the other surface vehicles
0.2 or 0.3 g. Now, if you people who are designers such as hydrofoils and others do have a much

24
higher frequency. Do the old figures of 33 Hz really matter a great deal whether the speci-
and 25 Hz cut-off still apply to the larger fication says one number or another number,
vehicles that could carry the heavier gear? somebody somewhere in the operation has got
to understand materials and components and
Mr. Paladino: The 50 Hz was added for tests and environments enough to be sure that
hydrofoils and the 33 Hz is primarily for sub- the equipment when it gets designed will func-
mersibles. The other two brackets are for tion properly. This is what we are really
surface ships. The carriers usually fall in the after. Whether the specification is right or
first bracket. The new version has included an wrong is not important and whether the con-
instruction on mass-mounted equipment, which tract is right or wrong is not important. The
is what you are working on. The mass does thing that is important is that the equipment
act as a cantilever to a soft structure which is function properly when it is installed and in
the hull which has low frequencies. They took service. I think we make too much of trying
this into account. to be explicit about everything. We can say,
for example, that the equipment has to go on
Mr. Johnson: Is there some simple way board a destroyer so the upper limit of fre-
that we can find out in advance what frequency quency is going to be about 25 Hz. So, if I
range would be applicable for a particular piece design equipment that is resonant under 25 Hz,
of gear, since they are generally designed for it is going to flunk the vibration test. So I
a particular class of ship. In other words is know that the floor of the frequency is going to
there a class listing of ships we could get? be somewhere in the 25 to 30 cycle range so I
have a shock requirement. If the shock require-
Mr. Paladino: My suggestion is that when ment is 60 inches per second, that says that
solicited for a contract for equipment, you ask you will have 1 g of equivalent static load for
the people what class of ship this is to go on. each cycle per second of frequency. You have
If they do not know the answer our office, Nay to give the designer some number to work with
Ships 037, is available and we will direct you to get this stuff sized so that the analyst can
to the proper Type desk which will give you the begin to do something with the analysis and so
answer you want. that the test men can begin to do something
with the test equipment. Somewhere you have
Mr. Mains (Washington University): I to begin. This means making some kind of
think it is possible to bring a slightly different generalization about the conditions and the
angle to this discussion which may be helpful equipment which will allow you to get started
to both sides. When I hear speakers on the in the direction that gives you a chance of
platform say they cannot answer your question finishing properly. All I can say is that it
because they are not a reliability type or be- works if you go at it right. Then, when you
cause they are an analyst and do not know about get to the shock test, you relax and everything
design, or perhaps they are a designer and do passes and there is no problem. When you get
not know about testing, this reveals to me the to the vibration test you might have a nut or a
fact that one of our principal troubles is suffer- bolt to fix here and there, but this is minor and
ing from tunnel vision. We see too much of our does not give you any trouble. When you are
own little speciality and do not take the trouble all through, the stuff functions properly on
to learn enough about the other guy's problem, board ship, which is what you are after. So I
Now about two years ago a manufacturer called would argue for trying to stand back a little
me and said that he had been asked to submit a bit and look at a somewhat wider view than is
bid on the below-decks missile handling equip- portrayed in much of what I hear at these
ment for a batch of destroyers. The prime sessions.
contractor said the equipment had to take 50 g;
what does this mean? Well, of course, it does Mr. Kilroy (Naval Ordnance Station,
not mean anything. It simply means that the Louisville): Somebody has introduced cast
prime contractor Is misinterpreting the whole iron into MIL-STD-167 and I am having a hard
business. So it takes about 6 or 8 months of time getting enough shovels to pick up the cast
cycling back and forth through the system to iron pieces. Do you know why it was intro-
get to a point where there is sufficient under- duced into that standard?
standing between the designer or manufacturer
and the guy who has a contract for ships to get Mr. Paladino: This is a type 3; it has
to the point where the designer can begin to been in since the beginning. Type 3 was not
make a reasonable proposal that will eventually changed. I think you better read the old version.
allow him to design the stuff properly. I think We have had very little trouble with Type 3.
that this is one of the big troubles. It does not This is on torsional vibrations. The major

25
changes were made to Type 1, ballast in Type 2 for adopting American standards as international
and longitudinal stresses in Type 4. We also standards. Then you do not base your standards
added mass-mounted equipment, on international standards? This is to me a
little bit surprising.
Mr. Kilroy: I still have a problem which I
will tke up with you later. In MIL-S-901, was Mr. Askin: The IEC standardization work
there a change so that we can have the report is not directly related to the MIL-STD-810 effort.
form included in the specification. We now have We have been concerned primarily with stan-
to send fifty cents to buy the forms. We cannot dardizing within the U. S., and that is-a-big
do reports, because we cannot get the forms enough job in itself. We have not yet standard-
sometimes. ized even within these three services. I sat in
on a few of the climactic standardization efforts
Mr. Lund: That is a very easy question, in IEC TC 50, and the U. S. did not have a
MIL-S-901 is currently under revision and will unified stand on the procedures, especially the
be coming out soon. We have a parting date of one on humidity that is still being thrashed out.
January and all your wishes will be fulfilled. There is a basic disagreement between the
American approach to humidity testing and the
Mr. Spang (Institutet f5r Milj~teknLk, British. I know that we in the MIL-STD-810
Sweden): This question concerns the adoption of coordination effort have not considered inter-
international standards. When you are consid- national standardization, and I think rightly so.
ering probable new revisions of MIL-STD-810 We first have to solve our own standardization
and MIL-STD-202 for component testing, do you problem.
consider the adoption of test methods given in
IEC publication 68, which are the agreed inter- Mr. Paladino: The U. S. is cooperating
national 3tandards for environmental testing with both the IEC and the ISO. In fact, in this
procedures? You would perhaps say that those country the American National Standards Insti-
testing procedures are not suitable for your tute has working groups, to one of which I belong,
purposes, but in fact many of those procedures and they tie in to the IEC. If we do not have
are very much based on military procedures complete agreement in this country between
used in different countries and on military pro- industry and the DoD community, in no way can
cedures used in the USA. You have had a very we make a united stand to an international group.
good participation in this international work and Many times the data that is involved in the DoD
you still have, both in the mechanical area and requirements is not for open publication, which
in the climatic area. Therefore I would like to makes it extremely difficult. We do participate
note that I think it would be worthwhile to find in international groups. We do have members
out your aims for the future regarding the adop- who go over to all the meetings.
tion of international standards in this field.
Mr. Root: I was chairman of the S2W60
Mr. Earls: I have had no personal experi- working group which corresponds with group
ence with the document you are talking about. TC 50. Charles Fridinger is now the chairman.
I have had some with NATO documents which We are presently working with this IEC 68 docu-
were based on MIL-STD-810. We are not con- ment, attempting to come up with our own U. S.
sidering the IEC standard as far as coming up standard on sine testing. The consensus of the
with a MIL-STD-810 is concerned. Most of our committee at this time is there are some weak-
input for NIIL-STD-810 comes from the services nesses in IEC 68 but parts of it will be incor-
and industry in the U.S. We have not taken any- porated into the standard along with documents
thing from an international standard, so far. in use in the U. S. We are not overlooking this
document, but we do not think at this time that
Mr. Sp~ng: Actually, the shock.test in MIL- it is a complete answer.
STD-810 is almost identical to the international
lEC publication shock test. I do not know which Mr. Sp~ng: It is encouraging to know that
way it has come about. I have some idea by somebody here really knows something about
asking some people in this country. There is a IEC publication 68 at least. Of course there are
lot of discussion at the moment on humidity tests weak points in publication 68; there are weak
where the Americans are asking IEC and the ISO points in MIL-STD-810 and in MIL-STD-202.
to adopt the U.S. humidity test in MIL-STD-202 Especially as an individual you will always find
for component testing. I think if you have no weak points because you have your own opinion
intention at all in this country to care about on how this standard should look. But, in fact
international standards, it coulU be rather diffi- an international standard is worked out as a
cult for the international bodies to find reasons kind of a cooperation between different countries,

26
and all the countries have their possibilities to the way It will be done and will be done no other
change the standard if they can press their argu- way. Perhaps if we standardize for one proce-
ments hard enough. In many cases this publi- dure right or wrong, we will find out whether
cation 68 is actually based on American practice, we are right or wrong and correct it. Believe
It might also be based in a certain degree on me, we are all concerned with the IEC and with
English practices, especially this humidity test. the international standards, but it is not an easy
I happen to be in the working group which pro- problem. We have nine commands within the
duced the original humidity test and I am not Army and I cannot get standardization. It is not
completely happy with that test. Actually, if an easy problem, but we are not ignoring the
you look at the other climatic test, the new dry international standards. It took so many years
heat test, you will find that it better suits your up until now and it is going to take another 10
purposes. That is a test for heated test speci- or 15 years before we get one standard.
mens and-a test for specimens which are con-
nected to an artificial cooling system. There Mr. Paladino: I am not a humidity man
is also this method on random vibration testing but I know that humidity is different on land
which is in the voting procedure. At the moment vehicles than on ships. It is different on an
as far as I know, there is no existing random airplane than on a tank in the Sahara Desert.
vibration standard. This means that you will It is not that easy. No one on this earth can
not have the same difficulties of adopting an design a piece of equipment which will survive
international standard. At least you should look the Army, Air Force, Navy and Marine Corp
at this before you start working on your own environments. So you primarily fix your own
standard; that is the whole purpose of inter- standards and design a specific piece of equip-
national standardization. ment for them. In the Navy we are concerned
with ships and therefore all equipment will be
Mr. Pusey: Part of the purpose of this designed to survive ship environments. Air
session is to solve our family problems as far Force and Army would do the same. The inter-
as specifications is concerned. Do you have play comes in for example if the Army has
this same problem in Sweden? transported some equipment to a war zone and
they use the ship as a common carrier. Then
Mr. Spang: Yes, we have. I thik the prob- we have to be careful. I do not say that you
lems are very similar, although we are fewer make them design Army equipment for the ship
people. The loudness is about the same but the environment. We had one case in particular
number of participants are fewer. We might where the Air Force was transporting jet engines
not be able to create a panel of this size. We to an island and they shipped them on a Navy
would only be about 4 or 5 people on the panel ship as cargo. When they got to the zone and
and we might get about 60 in the room, but the were checked out they fell apart. A little in-
problems are the same. vestigation showed that the ship vibration
Brinelled the bearings in the engine. We should
Mr. Biamonte: Back in 1940-41 we got always package them in mitigating capsules and
involved in a great World War, and most of our we will have no more casualties. It is fine to
forward equipment we found deteriorated be- have one document, but I think you are going to
cause of humidity. Humidity was an extreme have to have families in that document for each
problem. The Department of the Army at that service environment.
time expended an extreme amount of money,
manpower, and material to run a program for Mr. Dreher (Air Force Flight Dynamics
about four years evaluating humidity, humidity Laboratory): Will Mr. Earls say something
effects and humidity test procedures. If you about the new plans for MIL-STD-810?
remember, the Air Force was a part of the
Department of the Army in those aays. We have Mr, Earls: There is talk about two docu-
been gathering data for 30 years to come up ments. The second document is called the
with a test procedure. In MIL-STD-810, we limits document. It is more of a requirements
have five test procedures. Why should we have document on what the actual environment is and
five testing procedures? I can understand a what the test limits should be. The test methods
steady-state humidity procedure. I can under- document of MIL-STD-810 is just test methods;
stand a cycling procedure, but I cannot under- this is how you go about it. But in the second
stand why should we have five of them. Now if document there is a rationale behind it. Further,
in 30 years we cannot standardize within DoD, as Wayne Yarcho pointed out there are around
how can we standardize with other countries ? 5,000 specifications. We were asked to review
It is a difficult problem and possibly the only thirty, or forty of them and we are not going to
solution is a dictatorship which will say this is bring all of theminto line with MIL-STD-810.

27
It is the idea to eliminate as many of these requirements should be spelled'out in Section
others as we can and, I think we can eliminate 3 of your specifications. That Is where they
some of them. We will put in MIL-STD-810 should tell you what environment you will see.
what these other specifications have where MIL-STD-810 is there for the purpose of telling
passible so that it can be used in their place. you how to run a test. You have to know what
design limits you have from other documents.
Mr. Askin: Just to amplify a little on what
Mr., Earls has said, we have to keep separate Mr. Bamonte: I believe Mr. Johnson has
the environmental criteria and the test proce- a very valid question. The design requirements
dures. This is really a very difficult task that for the procedures 9 and 11 in a specification
a standardization group has taken upon itself, say that you shall meet rough handling in trans-
The effort to standardize MIL-STD-B1OB for portation environments without defining what
all three services and to eliminate thereby a they are. That is what Is in Section 3 of the
lot of the superfluous older standards that exist document and that is what he is complaining
is one of the main tasks that this committee about. However, we do not have a definition of
will have. One of the big problems is to try to these environments. So what industry does
Incorporate the Navy, Army and Air Force through trial and error is to design something
standards that are current. Wherever they are that meets the rough environment of procedures
common we must try to reduce the number of 9 and 11. Procedure 11 is nothing more than a
procedures. Right now the standard does not simulation of the environment which loose cargo
really represent all of these services adequately. would see in a vehicle. For an item being
driven as a part of a vehicle it is procedure 9.
Mr. Forkois (Naval Research Laboratory): What these environments are is very difficult
I agree with Mr. Biamonte to a certain extent, to determine. Some 20 years ago some work
For example, shipboard equipment that has was done in this area and it has been shown
been passed to MIL-STD-167 and MIL-STD-901C that vibration excitation with extreme ampli-
is pretty rugged. The Navy does not require a tudes of about 30 to 40 g and frequencies over
transportation test on top of these tests. They a range up to about 2000 Hz are excited on the
do require some protection for storage against individual item as loose cargo. With respect
humidity and they hav packaging for accidents to procedure 9, we ask that the item be designed
against marring or dents in the equipment. I to withstand repetitive shock impulses of about
feel that a lot of equipment that I have tested to 10 g. This is all the data that is available.
Navy Specifications would certainly satisfy However, to put this as a design requirement
truck or railroad transportation. A lot of the might create a situation where a specification
equipment would satisfy aircraft transportation. is not important. A specification is very
They may be a little heavy but, if you put them important, maybe not from the technical point
on Isolators they certainly would meet the re- of view, but certainly from the administrative
quirement for higher vibration. So to this ex- and contractual point of view. A specification
tent I agree with Mr. Biamonte, that in many is a legal document; it is important. We do
cases we have specifications which can embrace not know what the environment is. We there-
other specifications. I think that it is silly to fore say, and rightly so, industry has no
retest to this extent. Perhaps these committees alternative but to design to an actual test.
would do well to consider the possibility of
eliminating some tests on the basis that one Mr. Johnson: The particular part of the
test actually embraces the requirements of procedure I was referring to was necessarily
another environment, the second part, which is what you were refer-
ring to. I am referring to the first part, which
Mr. Johnson (US Army Electronic Proving is the 10 to 55 part.
Ground): We have a basic statement that the
Mil-Spec be written to require the manufacturer Mr. Biamonte: Somebody back in 1965
to use a specific design method. That is why decided to print MIL-STD-810 as a coordinated
we look at procedures 9 and 11 in MIL-STD- document. The Army Electronics Command
810 as design methods, not as a finished product had a test where we ask that the item be adapted
to meet a service environments. That is why to a plate which represented a vehicular adapter.
we do not like to test with that procedure. I The first part of this test was a survey and it
would like some comment on that. was based strictly on development specifications.
We asked the contractor to survey the equip-
Mr. Askin: You do not use MIL-STD-810 meat from 10 to 55 Hz at .0.03 inches ampli-
to design to. It is not-intended for that purpose. tudes tQ make sure that the vibration ampli-
MIL-STD-810 is a test document. The design tudes did not exceed a particular value. In past

28
experience, we found that if it did, the equip- defined environment, or you may specify a
ment would fall apart when subjected to the test procedure which is to evaluate the equip-
second part of the procedure. This was a ment once it has been designed. Now if we
diagnostic test. It was one merely to look at concede that everybodys heart is in the right
it. It did not have any criteria for acceptance place and what we really want Is for this equip-
or rejection. However, when it got inserted ment to work in the field, then we must specify
into MIL-STD-810, somebody decided that this a performance within an environment. To
was a test and applied a criterion to it. We specify design procedures or to specify a test
never, if I can help it, hold anybody to the line. begs the question. Right now, we are in the
If the item does not fall apart in procedure one, anomolous position of having to try to satisfy
we go right to procedure two and run the test, all three things simultaneously, but knowing
and use only procedure two as a criterion. I that the final criterion of acceptance will be
am sorry that has created a problem, but !t is the test. It seems to me we ought to get off
In MIL-STD-810, and you are perfectly right; the fence and stop mugwumping. Either call
it is a bad test. for performance within an environment and
shut up about tests and design procedures, or
Mr. Johnson: In MIL-STD-810 there is a frankly admit we do not know how to define the
procedure eight which is for ground equipment, environment and the performance and specify
That goes into detail as to what type of equip- the test. Shut up about the other things; set
ment for what type of vehicle. Procedure up a design procedure for them to follow and
eight seems to be a single service environment make everybody clear regardless of what
for the particular selection of equipment. Pro- happens with the test or performance in the
cedure 10 seems to be a field service environ- final environment. You cannot be all things to
ment for loose cargo or transportation. That all people at once; it will not work. From the
is what we want referenced in the specifications, standpoint of the manufacturers I have lived
instead of procedures 9 and 11. with, they would prefer and I would prefer to
have a requirement for performance within
Mr. Biamonte: I think Mr. Forkois the defined environment. We can do this. But
brought it up very well before. Procedure we know that as far as the Navy is concerned,
eight is an item in MIL-STD-810 which calls we have to pass that barge test. So we fix the
out a specific environment, it talks about a design to pass the barge test and if necessary,
wheeled vehicle, a tracked vehicle or some we dress up the other things to fulfill the re-
other kind of vehicle. There are three differ- quirements. This is not tne way it should be.
ent curves in procedure eight. Procedure 10
is what a particular group within the Army Mr. Pusey: I think you made a very good
called the Transportation environment or point. The question of whether you design to
common carrier test, which I do not want to a test or whether you design for the environ-
go into right now because that might take all ment is a very important thing. Unfortunately,
night. However, what has been happening with there are no ways to prove sometimes whether
our equipment is that when we call out proce- you have designed to the environment, unless
dures 9 and 11, a particular testing group you create a test which may be perhaps
would incorporate, regardless of our specifi- conservative, and run it.
cations, a common carrier test on the equip-
ment. This is where we run into a lot of Mr. Lund: I think the barge test, although
trouble, because a common carrier test calls you can criticize it, is about the best alter-
for resonance dwells. I dare say that there native we have. We have given design methods
is not a single piece of ground equipment that a hard try. We found that, if given the oppor-
I cannot make fail when subjected to the common tunity to choose between satisfying the Navy's
carrier test of MIL-STD-810. I do not care dynamic analysis requirement or the barge
who designed it; I could fail it. This is mainly test, most vendors really jump to the barge
because there is no procedure within MIL-STD- test. It seems to be the preferred route. One
810 that tells you how to find the resonances reason for this I think is the difficulty we had
and what resonances you shall dwell on. in applying failure criteria. If we specify
design methods, we are forced to specify
Mr. Mains: There are only, as far as I failure criteria to go along with it. If we de-
know, three ways that a design reqairement sign to yield and if in our analysis we predict
can bc specified. You may specify a design stress above yield, that must be rejected
procedure which is to be followed to produce simply because we cannot constantly predict
the given piece of equipment. You may specify what is going to happen to the equipment once
a performance to be achieved within a given it starts yielding. However, many manufacturers

29
have gotten a bit smart with the shock testing rather incredible that we have discussed shock
business, including the barge testing business, and vibration testing and design requirements
They have found that you need not design up to for some three hours now, and no one has
yield in order to pass the barge test; you can brought up the very important subject of the
get away with much less in many cases. This impedance relationship between the driven sys-
requires a lot of engineering judgement, of tern and driving system. I know that there are
course, and it is possible to make mistakes, some people who are doing some research along
but we find that they are seldom made. There this line. I am doing a little myself. We find
are not many equipments that fail the barge test that it is central to the problem of obtaining
and, when they do, no one is irticularly sur- proper test and design procedures. We heard
prised as to why they did. So I think the straight the argument earlier that test specifications
testing procedure does have certain advantages were entirely adequate or maybe not stringent
over the straight design procedure. What we enough. We also heard counter arguments
try to do here is mix these two procedures by suggesting that perhaps test specifications are
offering some guidance as to how to pass the too stringent. Both are right if you do not
test. We leave the rest up to the good engineer- experience a certain kind of frequency blindness
ig judgement of the contractor, that fails to separate high impedance products
from low input impedance areas. I think a great
Mr. Mains: This is somethin& I think needs deal of this would be cleared up if we were to
to be said. Mr, Forkois, if you wanted to make recognize that when we are dealing with shock
a piece of equipment pass a shock test you could and vibration and environmental problems, we
do so, and nobody watching over your shoulder have a situation in which the driving point im-
could tell the difference between your wanting to pedance of products can vary over perhaps two
make it pass or make it hot pass. When I have or three decades. At some frequencies we have
produced three inches of IBM output of analysis, nearly infinite impedance reflected back to the
I can make that analysis say everything is good driving system. Other frequency domains have
or everything is bad, and nobody in the world nearly zero impedance reflected to the driving
can determine which I was trying to do. So the system. I would like to hear some comments
real reason why the test is such an important from the panel on this particular aspect of the
thing in the Navy end of things is that it Is the shock and vibration problem.
one criterion that is not rigged and cannot be
rigged as long as you play it square. I do not Mr. Earls: I have very little Information
know how many know the name, H. F. Moore, on the impedance situation. We have had some
but 30 years ago he was the grand old man of limited investigations. I realize that on a
materials testing. I remember a class under shaker you use an infinite impedance input
him in which he gave out a batch of fatigue data which is not the same as on the airplane. I do
and the assignment was to take the stuff home not think we have gone far enough yet.
and come back with an argument that showed
this was a good material for the specified Mr. Paladino: In the Department of Navy
application. We brought it back th . next day he the impedance type measurements are primarily
said take it home again and prove to me it is used in acoustic work. We have not used It in
not the right material for the specified appli- the mechanical vibration test.
cation. We brought that back. He had illus-
trated the point, that given the same set of data, Mr. Blamonte: I do not think I can contri-
you can prove both sides of the question depend- bute anything with regard to impedance mea-
ing on how you present the argument. This, I surements. We have not been doing it in the
think, is a point we should all keep in mind. We Laboratory. I think the impedance type work,
can slant the analysis or slant the test or slant where we are looking at signature shock effects
the test data interpretation any old way we want and so forth, should be limited to the investi-
to. As of now we depend on the barge test or gation leading up to a specific type test. I do
the hammer test to be an impartial kind of thing not see how we could apply it to final test
if you like, adjudicated by people who are not specifications.
trying to make it go one way or another. So we
can all play it square and come out in the right Mr. Ludwig (Pratt & Whitney Aircraft):
place In the end, we hope. There are people working on Impedance. We
saw a whole session on mechanical impedance
Mr. Nankey (General Electric Company): in the Shock and Vibration Symposium here.
I think there is a very important feature of the There is a writing group, S2W58, working on
shock and vibration design and test problem such specifications. I think I will leave it to
that has not been brought up tonight. It Is them to make the specifications. Impedance is

30
a very useful tool if used in the right place; it back in the old dynamics group at Wright Field.
can evaluate development hardware and it can He came up with a simple specification for
tell you a lot about it. noise. He did not put a number on it. He had
letters ABC and there were numbers that
Mr. Dreher: It sounds to me as if every- associated with these letters, but it depended
body is trying to say that maybe we should get on the application. I appreciate what Dr. Dreher
rid of specifications. We have to admit that had to say, but I do not know whether ve are
our business, getting equipment qualified, is a going to get that many environmental engineers.
very tricky business. It is not something that In my opinion what we need is some better
some equipment project engineer can handle, guidance on testing. Engineers have to be
It sounds to me as if most of the specifications taught. We have problems such as: What is a
are called out by equipment project engineers, resonance? I do not know even whether we can
Let us assume that a radar engineer develops agree on what a sine wave is. Sometimes, it
a specification. In addition to having to be an comes to that. There is a proposal to come
expert in radar and electronics, he also has to out with a document to define procedures. I
be an expert in environment. Let us face it, think that is what we are lacking. For example,
all people that I have talked to in that situation if we improve the equipment in aircraft there
tell me they wish they had someone to help is a man back here worrying about the levels.
them out. We should take specifications off the Now he may be trapped in that, but there is an
public market. Only an environment group alternative of location. A lot of people do not
should handle them. Every development agency know that and they do not ask the question: Can
should have an environmental engineering group, it be relocated? It is amazing how much work
engineers who know what an environment is all is done on things like damping materials and
about. Equipment project engineers should get when we look at vibration in a piece of equip-
everything signed off by this group. The only or- ment, we find out that it is resonant and mag-
ganization that I have come across that really nifies the vibration quite a great deal. I would
does this is one aircraft company. This came have encouraged that it be worked on some
to pass simply because of the forcefulness of more. But maybe these things need to be
the man in charge of the dynamics group. He proven and a little bit more solid inaormation
is the kind of man that does not let anybody do put out.
anything without his knowing about it. He has
caused all the project engineers in his company Mr. Askin: I would like to agree with
to come through his office. I think that is what Dr. Dreher and Dr. Mains. I think there are
all of you are trying to say, that we need this two basic problems, One is the age old prob-
control. lem of communications. There is not enough
intelligent communication between the designer,
Mr. Paladino: What the gentleman said the tester and the environmentalist. The second
sounds very good but, for example, consider a is the parochial view taken by the designer. He
system like an aircraft carrier. It has the air- thinks he knows everything. He does not want
craft, land type vehicles, radar operating equip- to come to an environmental engineer to get
ment, missile systems, and so forth. I cannot any help. I think these environmental engineers
walk on the water. That is why we have what have to become more aware of the problems of
we call Type desks. They are specialists in the designer. It has got to be a joint effort.
particular systems. DoD is highly concerned All of this means more research to find out
about having an environmental desk, but the the actual environment that we have to design
thing is to staff it with competent engineers, for and test for. We have to educate top level
people in the government to fund more of this
Mr. Biamonte: I would like to commend research. Industry also should do more
the gentleman who suggested getting rid of research when they are designing something
specifications. I would like to get rid of speci- new. If they came to the government and
fications, also. I do not know how many industry pointed out what information they have and how
people we have here, but is there anybody here valid it is, they can jointly reach an agreement
from industry that is willing to sell to the on what specifications ought to be used. It is
government on a warranty. basically a matter of better communications.

Mr. Dreher: You missed my point. I do Mr. Lund: Starting about the mid 1950's
not want to get rid of them; I just want to take the Navy got quite serious about the shock
them out of the public chmain. problem due to installation of very shock-
sensitive systems on ships, such as electronic
Mr. Wilkus: I can remember Dr. Rogers systems and delicate weapons systems. At

31

A
that time we also had a development of the a method that hopefully will get better as we
nuclear weapons threat, and this had led to a use the specifications and can interpret them
rapid rate of change in shock requirements, better. I also reaffirm my conviction that one
This is one thing you folks put up with quite of the most important paragraphs in MIL-STD-
admirably. This goes for the contractors and 810 is in the original version as one of the
the Navy. If it is any comfort to you, I think paragraphs in the beginning. It was an intro-
we have about stabilized the situation. I really ductory remark that this standard should be
doubt that we will see major quantum jumps of used only if you did not have soniething better.
the sort you have seen in the last 10 or 15 years If you knew the operating environment better
in the shock business. Judging from the re- than stated ir, the standard, then you should
sponse tonight, the shock troops do not seem to definitely use the operating environment. In
be in very deep trouble. I do not think we will other words it is to be used as a guide. This
get you back into trouble again, at least for paragraph should certainly be in the next
some time to come. version and underlined. This would clear up
a lot of the problems. Look at the final prod-
Mr. Phillips: I have reached a conclusion uct and the final environment and get there
that we shoul &not do away with all the specifi- the best way you can. If you can use the
cations. This is a method of communicating standards use them, but if you have something
and it should be considered just as that. It is better by all means use that.

32
 SOME ADMINISTRATIVE FACTORS WHICH INFLUENCE TECHNICAL
APPROACHES TO SHIP SHOCK HARDENING

Doiald M. Lund
Naval Ship Engineering Center
Hyattsviile, Maryland

This paper briefly examines the tect.nical side of a few

administratively-oriented factoro which are of importance
to the Navy's ship shock hardening effort. It is illu-
strated that these factors will increasinaly influence the
selection of technical approaches to ship hardening.

BACKGROUND either in absolute terms or relative to other

weapon effects. When considered in conjunc-
The general objective of the Navy shock hard- tion with threat, neo for shock hardening of
ening effort is to fully optimize the shock various ship systems can be determined and
resistant qualities of Navy ships. All fac- compared uith need for hardening againnt other
tors which must be accounted :or in order to weapons effects. Such an analysis permits
optimize these qualities cannot be meaninig- development of a total hardening approach
fully discussed in a brief paper of this sort, which affords logically balanced (thus, opti-
but a review of basic considerations is needed mum) resistance against all weapons effects.
to establish a context for subsequent dis-
cussion. For purpose of this review, all fac- 3. Shock haraening State-of-the-Art Fac-
tors which influence the optimization of ship tors. These factors essentially serve to de-
hardness hase be6n loosely separated into four termine which specific technical approaches
categories; each category is defined ..rd dis- could be employed -- either now, or possibly
cussed in the following paragraphs: later after further research and development
- to reduce the vulnerabilily of ships to
1. Threat Factors. Potential shock shock. Technical confidence levels are gen-
threats are posed by any sizable weapon, erally associated with all alternatives in
nuclear or conventional, self or enemy-de- this category; these confidence levels can
livered, which can detonate underwater in the vary in accordance with shock severity, type
vicinity of the ship (primarily, influence of shipboard system under consideration, and
mines/torpedoes, near miss conventional bombs a number of other variables. Given a detailed
or missiles, and far-miss nuclear weapons). ship design baseline, knowledge of the state-
Threat considerations are strongly mission- of-the-art, and a basis for shock hardening
oriented, and include: which has been optimized from a threat/vulner-
ability standpoint, one may select technically
(a) Likelihood of exposure to enemy- optimized approaches to hardening of the ship
delivered underwater explosions, in an abso- in question. By definition, this process,
lute sense and relative to likelihood of ex- yields maximum shock hardening effectiveness.
posure to other types of enemy-delivered Were it not for the influence of administrative
weapons effects. factors (as defined below), all Navy RDT&E
programming in the shock area and all ship-
(b) Requirements to intentionally building specification shock requirements
encounter the shock environment in connection would reflect a single and outwardly desirable
with mission roles (sweeping and destruction goal -- to fully optimize the shock resistant
of mines, for instance), qualities of Navy ships through development
and application of technically optimized
(c) Benefits associated with cap- hardening procedures.
ability to withstand close-in self-delivery
of weapons which induce shock (primarily, ASW 4. Administrative Factors. For purposes
weapons), of this paper, these are defined as factors
which influence administration of the Navy
2. Vulnerability Factors. These factors shock hardening effort but which are not based
express the vulnerability of ships to shock, upon shock threats, vulnerability to shock, or

33
shock technology. The most noticeable effect DISCUSSION
of these factors is one which all of us have
probably regarded with varying degrees of dis- Since the Navy and supporting contractors are
comfort from time to time; they serve to mod- engaged in a great variety of efforts which in-
ify technically optimized approaches to ship volve selection of technical approaches to ship
hardening in response to outside influences of hardening, it would be appropriate at the onset
questionable character, such as schedules, con- of this discussion to clarify which of these
tracting regulations, availability of funds, selection processes are being referred to. Gen-
and the like. erally, this discussion is keyed to issues di-
rectly applicable to shipbuilding per se, and
Most members of the technical community which the technical approaches in question here are
supports the Navy shock hardening effort are generally tbose which are invoked as re-
naturally well informed concerning vulner- meets in shipbuilding specifications. This ap-
ability and state-of-the-art factors which preach was selected for several reasons. First,
influence ship hardening optimization. Some shock hardened ships represent, in one way or
threat factors are given less publicity within another, the end product of all of our efforts.
this community due to security restrictions, Second, it appears certain that the vast major-
but technically-oriented information in this ity of technical manhours being expended in sup-
area is nonethelesst available
aport where needed. of the ship shock hardening effort are, in
fact, being spent in direct response to ship-
Relative to the factors mentioned above, ad- building specification reqtirements. Finally, it
ministrative factors are not so widely under- is in this area that administrative influences
or appreciated by many members of the are most pronounced; ship acquisition managers,
stood owho are not card-carrying members of the Navy
technical community -- despite the strong in- Shock Program, are responsible for implementing
fluence which these factors exert upon ship ship hardening requirements.
shock hardening technical progress and the
structure of technical tasks performed in The fact that this discussion is largeply keyed
support of this effort. This phenomenom is to shipbuilding is not considered particularly
of course easily rationalized; some members of limiting. Selection of shock-oriented applied
the technical community are simply not very research goals, for instance, is strongly in-
interested in administrative affairs, others fluenced by the knowledge that most RDT&E
who might be interested may find no easy or accomplishments must eventually be reflected
sympathetic lines of communication open to all in shipbuilding specifications in order to
levels of the administrative community, and banfit the Fleets.
some administratively-oriented topics are
rater sopistiated-inter oigh and t Emphasis will be given in this discussion to
rather sophisticated in their own right and thus only three basic administrative factors: Procure-
are not easily interpreted by persons lacking ment policies, cost considerations, and schedui-
background in this area. ing considerations. This choice of dlscu-sion
topics simply reflects thE principal functions
There is little question that the above situa- of the ship acquisition manager's job--'o ac-
tion merely reflects the natural order of quire ships within the constraints imposed by
things -- an order strengthened, perhaps, by procurement policies, scheduling constraints,
purposeful organizational divisions which have and fiscal limitations.
been erected in part to protect technologists
from administrators and vice-versa. Nonethe- Navy procurement policies will be considered
less, there is equally little question that first. Some fundamental points:
the shock hardening effort will progress with I. A bedy of law has evolved over the years
maximum efficiency if all of its participating which is applicable to rovernmnt procurements.
supporters can recognize and thus account as
necessary for all of the major factors (in- This body of law reflected In the context of
Is Armed
DOD poli~ps in the ServicesProcurem-nt
cluling administrative factors) which influ- polions ithrme Navyrocuren ns
enceachevamnt
ence achiewiment off bsic ardnin objcties.
basic hardening objectives.
Regulations (ASPR). Navy procurement policies,
In turn, essentially establish how the Navy shall
go about procurin' goods ai services within
Against this background, some administrative the framework of ASF?.
factors which the writer believes are most
influential in determining selection of tech- . One of th, first sc.ps which tho, Navy tak,,s
nical approaches to ship hardening are briefly when plannin- a rhip procurment (or any other
discussed in the following section of this procur,,ment) is *to select from available alterna-
paper. The intent of the following dis- tives the procurement procedures which are best
cussion is definitely not to make the technical suitgd, legally and otherwise, tc the acquisition
community expert in the field of administra- in question. Solc-tlonr of procurement pr celures
tive affairs (it will not), but rather to con- is normally acce-mplishei wll in advanco of tho
vey a basic but perhaps improved understanding e.lr.otlon of shock rqulrements for te.. ship in
of how and why certain administrative factors question; early selrirtlon of procurement pro-
influence selection of technical approaches ce luros is abselutely n.'essary to give direction
to ship hardening, to the acquisition process.

34
 3. Central to the process of selecting two points suggest that procurement policies
procurement procedures is selection of con- will, in fact, exercise their prerogatives -
tract type. Two general categories of con- they will influence the selection of technical
tract types are worthy of brief attention approaches to ship hardening. This influence,
here: which largely reflects the firmly established
need to implement shock requirements within
(a) Fixed Price Contracts. Under any the framework of fixed price contracting con-
type of fixed price contract, the contractor straints, may often be evidenced in ways which
must assume at least some responsibility tend to counter technical considerations. Ex-
(hence, risk) for performance in accordance amination of only three issues which have a
with contract provisions at a previously bearing upon the technical quality of Navy
agreed upon price. In order to remain com- shock requirements should be sufficient to
petitive and maximize profits in an atmos- illustrate this point:
phere of fixed price contracting, the contrac-
tor must constantly strive to devise more 1. State-of-the-art. From a purely
efficient ways of doing business and must care- technical standpoint, the best approacb to
fully control his costs. The principle pre- ship hardening is the latest approach - the
requisite to fixed price contracting is that one which incorporates all new technical
the nature arn1 extent of contract effort must findings. From a procurement policy stand-
be sufficieutly well defined to permit realis- point, the best &pproach is the way we did it
tic advance estimation of contract cost. last time, because industry's experience with
the previous requirement serves to establish
(b) Cost Reimbursement Contracts. If a good basis for advance estimation of the
cost of contract performance cannot be realis- cost of fixed price contract performance.
tically estimated in idvance - as is often Some measure of shock requirement "newness"
the case with research-oriented or research- is of course acceptable. However, it is to
dependent efforts -- the government generally be expected that those who are in contention
assumes the bulk of cost risk through issuance for the contract (hereafter referred to as
of some type of cost reimbursement contract. "bidders") will compensate for their lack of
Such contracts commonly provide for payment of experience with new shock requirements by
costs borne by the contractor, plus profit factoring some "worst case" assumptions into
(fee). Contracts of this type are generally their contract bids. Cost to the Government
more difficult to administer than fixed price will thus likely be driven in an upward
contracts, and provide the contractor with direction for reasons solely associated with
little real incentive to hold down costs, the newness of the requirement, an undesir-
able effect. Of course, bidders may declare
4. Owing to the above considerations, any major alteration of shock requirements to
ASPR and derived Navy policies require appli- be unbiddable on a fixed price basis if they
catiLn of fixed price contracting procedures cannot estimate its cost implications with
whenever feasible. When a new design ship reasonable confidence. As suggested by pre-
class is being procured, some work under cost vious discussion, such a happening will more
reimbursement contracts may be required for the often force alteration of the shock require-
purpose of more fully developing new concepts ment than alteration of procurement procedures.
applicable to the ship in question; output
from these contracts serves to definitize ship 2. Complexity. The researcher right-
design and performance requirements to the fully observes that shock is a complex dynamic
point where subsequent design and construction phenomenon which cannot be rationally addressed
efforts may be accomplished under some type of by simple rules such as "design everything to
fixed price contract. Despite this potential withstand 60 G's". Consequently, as we learn
opportunity to address some shipbuilding re- more and more about shock and all its vagaries,
quirements under cost reimbursement contracts, proposed technical solutions to shock problems
it remains that procurement policies strongly tend to become more complex. Procurement
encourage development of shipbuilding specifi- policies will favor that easily priced out "60
cation requirements which can be applied G" requirement, however. If any shock require-
directly under fixed price contracts, and dis- ment is very complex, the bidder may realize
courage the specification of requirements that he simply hasn't time (within the period
which can only be addressed under some type of allotted for bid formulation) to fully eval-
cost reimbursement arrangement, uate the impact of the requirement upon costs.
His options in this case include development
The first two points cited above suggest the of conservative (protective) cost estimates,
general relationship (or pecking order) be- requesting more time to formulate a bid, or
tween shock requirements and procurement refusing to bid. None of these options are
policies. We find that the laws of physics administratively pleasing.
and the "laws of procurement" possess capa-
bility to exert a similar overall influence 3. Influence of installation variables.
upon how we harden ships, since in their own Inasmuch as equipment shock response can be
way they both ordain what can or cannot be dependent upon installation variables such as
done -- and neither will yield readily to a type of ship locetion aboard ship, nature of
desire on our part to do otherwise. The second supporting itructure, and location of adjacent

35
major masses, it would be technically desirable enterprise system (including its industrial
to express shock requirements in terms of participants) as well as the Navy's own
these installation variables. In that way, interests.
resultant equipment designs would better and
perhaps more economically reflect actual One other aspect of Navy procurement policies
needs associated with a specific shock en- should be considered in this discussion -
vironment. (As is, Navy shock requirements the issue of competition. We all enjoy the
are often predicated upon somewhat conserva- benefits of free and active industrial corn-
tive assumptions in order to address the gen- petition, which tends to keep prices down
eral case.) The procurement adminstrator while promoting quality and improvement of
would likely view the improved technical goods and services. Navy procurement policies
approach with siepticism, however . He would reflect our own feelings in this respect; they
probably point out that the contract bidders insist that competition for Navy's business
will not know, at the time of bidding, ex- be encouraged to the maximum extent practical.
actly how or exactly where most shock resis-
tant equipments will be installed. (The The influence of procurement policies appli-
successful bidder may not even have this in- -able to the competition issue is twofold.
formation at the time he must place orders for First, these regulations tend to discourage
the equipment, in fact). Therefore, the bidders implementation of technical approaches which
will understandably make conservative assump- are so technically specialized that only a
tions when estimatirn costs, which leads us few (or worse, only one or two) bidders can
right back to where we were when the Navy was realistically compete for the contract in
predicating requirements on the basis of con- question. Thus, they effectively favor rea-
servative assumptions. Or beyond where we sonably straightforward technical approaches,
were before; estimates of contract performance and favor approaches which do not demand
cost may be somewhat higher than would have possession of specialized facilities. The
been the case if the original requirement had second influence of policies in this area is
been invoked, owing to the newness and added slightly more subtle; they inveigh against
complexity of the requirement. many types of highly tailored contract re-
quirements, even though the requirements in
The impact of fixed price contracting con- question could be satisfied.by many different
straints upon selection of technical contractors. Take for example a previously
approaches to ship hardening is reasonably discussed hypothetical shock requirement,
evident from the previous discussion. Basic- which was expressed in terms of a particular
ally, we find that this particular adminis- shock environment defined by type of ship,
trative factor will always serve to promote location aboard ship, nature of supporting
simple, straightforward technical approaches structure, and location of adjacent major
to shock hardening. It tends strongly to masses. Any contractor who satisfies this re-
limit important shock design parameters to quirement during the course of a given ship-
those which can be determined in advance of building program will possess a decided com-
detailed ship design (during the bidding petitive advantage in a follow-ship program or
period), and serves to resist rapid change of in the event that some other shipbuilding pro-
technical approaches. gram finds need for the same equipment type to
be installed in a corresponding shock environ-
It should be noted that the above administra- mont. Conversely, this contractor may find
tive influences do not flow only from the Navy himself somewhat outside of the sphere of real
side of the house. Shipbuilders and their competition if his previous shock qualification
attendant sub-contractors obviously view fixed cannot be extended to the next potential appli-
price bidding issues from a frame of reference cation. In either case, desirable competitive
difference than the Navy's, but this really advantages are mitigated by the nature of the
makes little difference. When faced with lack speciallized shock requirement.
of firsthand experience with a now or very
complex shock requirement, the bidder is Again, the Navy is not the only party inter-
immediately faced with some difficult de- ested in the issues discussed above. One can
cisions: Assume the worst case (protect count upon an expression of righteous indig-
company profits by reducing cost risk), or nation from any well established Navy supplier
bid less conservatively in the interest of who suddenly finds himself effectively excluded
being the lowest bidder? Tell the Navy I from active competition by a difficult bhock
can't bid the job on a fixed price basis requirement. The usual private industry re-
(perhapn conveying the impression that I'm sponse to a highly tailored shock requirement
not as technically astute as my competitors), of the t;'pe exampled in the previous paragraph
or give it a try? No one likes to be faced is equally understandable: "But, how can I
with decisions of this sort. The fact that shock-qualify my equipment so that the shock
Navy and industrial procurement administrators qualification will be applicable to a variety
will arrive at similar conclusions (for per- of shipboard installation requirements?" In
hapa different reasons) concerning the ad- either case, the naval supplier is simply
miuistrative worthiness of shock requirements seeking an opportunity to effectively compete
is not surprising; the governing procurement for Navy business, an objective which is fun-
'poliniza re intended to protect the free damentally coincident with the intent of the

36
Navy's own procurement policies, solely related to shock will begin to in-
crease at a more rapid rate than test or
It is interesting to note that the impact of design costs as increasing level of protection
the competition issue upon selection of overtakes the individual inherent shock re-
technical approaches to shock hardening re- sistance levels of various types of shipboard
inforces the impact of fixed price contracting equipments and foundations. Principal costs
constraints. The former administrative fac- in this area are associated with the "beefing
tor, like the latter, serves to encourage im- up" of structural members, which adds labor
plementation of shock requirements which are and materials cost. These costs will begin to
technically straightforward, and discourages rise sharply at the point where protection
highly tailored "case basis" hardening levels begin to surpass the level of inherent
approaches. hardness associated with the hull and other
basic ship structure.
to the subject of
This discussion now turns
cost considerations. Little need be said con- Figure 4, which represents a composite of costs
cerning the basic impact of cost factors upon shown in Figures 1, 2 and 3, illustrates how
selection of technical approaches to ship cost considerations are likely to influence
hardening; we all know that Defense Department selection of hardness levels. If hardening
dollars are in critically short supply, and payoff is low, "hardening" within region "A"
that cost factors will always strongly favor of Figure 4 will likely be the most attractive
the least expensive hardening approach. It choice. If payoff is medium to high, hardening
may be instructive, however to examine how to some level within region 'B" is indicated.
certain shock hardening parameters and hard-
ening approaches can influence cost (and In connection with the above, we might also
vice-versa), consider how cost considerations influence the
potential for alteration of hardness levels
First, consider hardness level -- the level of from those currently specified to some other
shock severity which is to be addressed and value. Figure 4 reflects steady state con-
effectively defeated through implementation ditions, i.e,, conditions which exist after-
of ship hardening requirements. Figures 1, 2, the hardness level in question has been in-
and 3 illustrate how the major direct con- yoked for a long period of time. Figure 5
tributors to hardening cost might be generally illustrates the effect of sudden alteration of
expected to vary with hardness level. Require- hardening level from point r'A" (an assumed
ments that ships be capable of resisting sea long-established level) to any other level.
motions and shipboard vibration largely serve We find that decrease of protection level re-
as functional substitutes for shock require- sults (as expected) in a cost reduction, but
ments until we arrive at a shock severity costs do not drop immediately to steady state
level where shock loads can exceed these other levels due to the influence of design standard-
environmental loadings. In order to achieve ization. Increase of protection level will
adequate resistance to shock severities above first reduce or completely wipe out the
this "inherent protection level", we must in- applicability of previously conducted shock
voke shock design and shock test requirements tests and previous shock design analyses --
on a very broad scale. Figures 1 and 2 re- an important factor. Significant increase of
flect this consideration. Once invoked, shock protection levels would lead to at least partial
design and shock test costs can be expected to obsolescence of currently available shock testing
increase only slightly with increasing shock devices, and would naturally imply requirements
severity. Important shock testing costs which for shook hardening modificati6n of many
do not vary significantly with shock test )vel currently acceptable shipboard equipments.
are costs associated with shipment of equip- Therefore, cost considerations tend to keep
ment to and from test facilities, installation hardening levels const&nt, and particularly
of the test item on the shock testing device, inveigh against even a modest increase oT well
labor costs associated with running the test, established hardening levels.
and post-test teardown and inspection. In the
area of design, cost associated with initial In the previous paragraph, reference was made
shock design checkout of proposed designs does to a most important shock program resource:
not vary significantly with hardening level. The large reservoir of equipment designs which
The modest increase of shock testing and shock have been previously approved for installation
design costs with increasing hardening level is on shock resistant ships on the basis of satis-
largely attributable to the fact that more and faction of still-current shock test require-
more equipment areas become critical from a ments. Present Navy shock test procedures are
shock standpoint as hardness level increases, rather generalized in the sense that a conserv-
which implies increasing requirements for re- ative shock environment is usually represented
design and retesting. Design costs will during the shock test, rather than a specific
eventually increase rapidly at the point here environment associa,.ed with a specific ship and
it becomes necessary to design a significant shipboard mounting arrangement/location. This
amount of shock resistance into basic ship approach permits broad application of "shock
structure -- hull plating/framing, decks, bulk- test extension" policies, which permit accept-
heads, and the like. Ship construction costs ance of previously shock tested and approved

37

- - - -
 Figure 1
Shock Testing Cost vs. Hardening Level 4

Hardness Level-I-

Figure 2
0
Shock Design Cost vs. Hardening Level
Hardness Level *0

43
Figure3
0

Added Labor and Material Cost vs. Hardness Level

Hardness Level-*-

Figure 4

Composite Hardening Cost vs. Hardness Level

(Steady State)

Hardness Level-goo.

Figure 5 4

Composite Hardening Cost vs. Hardness Level

(Transient)

Hardness Level-*~-

38 .1

_io
equipments for subsequent installation in directly hardening equipment, as opposed to
various locations on almost any type of ship. mounting non-hardened equipment upon shock-
Furthermore, these policies permit acceptance mitigated decks or shock-mitigated platforms,
of untested equipments if they can be shown is partly predicated upon consideration of
to possess a degree of shock resistance equal costs in this category; truly effective shock
to or better than that of a similarly-designed mitigation systems are relatively inefficient
equipment which has klready satisfied the from the standpoint of space and weight added.
shock test requirement.
3. Administrative costs. These are costs
It is clear that continuing growth of our associated with administration of the shock
reservoir of previously shock tested and requirement. If the requirement is simple
approved equipments is playing a very major and straightforward, these costs will nor-
role in reducing ship hardening costs. The mally be very low. Costs in this category
majority of shock hardened equipments being can become fairly significant, however, if
Installed aboard combatant ships are now the shook requirement entails application
being accepted on the basis of shock test of stringent quality control measures or
extension, and the trend could no doubt con- leads for other reasons to the flow of a
tinue until, at some not-too-distant date, large amount of information between the Navy
nearly all equipments will be accepted on and the contractor.
this basis. It is equally clear that the cost
conscious Navy administrator will instinct- 4. Costs related to reliability. Because
ively resist adoption of technical approaches shocx requirements have a generally "rugged-
which could lessen the potential for shock izing" effect upon equipment design, a shock
test extensions, i.e., approaches involving hardened equipment is generally a more reli-
increase of hardening level or tailoring of able equipment. Nonetheless, reliability
shock test requirements to reflect specific issues will be considered when alternate
shock environments in specific ships. approaches to shock hardening are being
evaluated - particularly when the shock
Similar considerations apply to our other on- hardening process will involve addition of
board shock program resources. It would be any sort of moving parts or addition of
quite undesirable, from a cost standpoint, materials which are subject to gradual deter-
to adopt new technical approaches which would ioration.
significantly reduce the potential for utili-
zation of presently available shock test faci- 5. Maintenance costs. Anything added to
lities, shock design resources (computer pro- an installation for shock purposes will
grams, able personnel, etc.) and related in- likely require some form of periodic main-
dustrial capabilities/facilities. These re- tenance during the life of the ship; for in-
sources have not been easily won; any technical stance, the added items may require painting,
rationale which leads to their dismissal must lubrication, or may wear out or deteriorate.
therefore be extremely compelling. Related maintenance costs can become quite
significant when considered ovor the 30 year
Before closing this portion of the discussion, lifetime of a class of ships, so careful
it might be worthwhile to briefly summarize attention will be given to this potential
all principal cost considerations which would source of shook cost. The previously dis-
normally be given administrative attention cussed question of direct hardening vs.
during the development of shock specifications. shock mitigation also serves as a good ex-
These are as follows: ample of how maintenance considerations can
influence technical approaches to ship
1. Direct labor and material costs. In- hardening. The essential question is usually
cludes only those costs directly and immediately simple: Will the cost to develop, manufacture,
associated with satisfying the requirement, such provide space for, install, and maintain
as shock design costs, shock testing costs, cost shock mitigating devices and associated
for "beefing up", and the like. structure on a given number of ships exceed,
or be less than, the cost to develop a
2. Indirect labor and material costs. If directly hardened design?
satisfaction of the requirement entails
addition of weight, it is possible that the 6. Maintainability-related costs. As the
ship will have to grow (increase in length, name implies, costs in this category reflect
beam, or both) in order to accomodate the the impact of shock requirements upon ability
weiir. increase. (Weight which is added low to maintain shipboard equipments and struct-
3n the ship may serve only to decrease re- ures. To date, the maintainability issue has
quirements for ballast, but weight added high been raised seriously only with respect to
in the ship can be p6nalizing). Shock re- foundation shock requirements ("bulky shook
quirements which entail added space require- resistant foundations restricting access to
ments can also exert a direct effect upon surfaces which must be painted and to equip-
ship growth in some cases. Ship growth, of ment which must be maintained"). The writer
course, implies added cost; more hull and is firmly convinced that shock requirements
structure, longsr piping and cabling runs, need not (inherently) cause such problems,
and so forth. The Navy's preferonce for but the issue is nonetheless sensitive and

39
 I
and will be given close attention during re- goal of maximizing compatibility in this
view of proposed shock specifications. area, in the interest of keeping hardening
costs low.
7. Operating costs. Any shock requirement
which causes addition of ship weight elso Specifically, these scheduling considerations
causes increased fuel consumption. Soae sort influence selection of technical approaches
of penalty (possibly a cost penalty) will also to hardening in two principal ways. First,
be borne if shock requirements cause any ship- and in direct reflection of the fact that
board operation to become more diffliciUt or "time is money", they strongly favor the least
time consuming. tim-consuming approach. This influence is
.~ est, of course, if there is any possi-
Shock hardening costs in many of the above bility that introduction of the shock require-
categories are currently very low or near- ment will cause stretch.out of time between
zero, and costs in al categories of signi- major shipbuilding events or cause overall
ficance are dropping sharply in response to lengthening of the ship design and construction
the shipbuilding industry's rapid adaption to period. Naturally, this scheduling factor
presently-specified shock requirements and due fundamentally influences the question of how
to increased shock test extension opportunities. precise a technical approach shall be taken.
Needless to say, the administrative community This, in turn, influences the degree of shock
has a strong interest in preserving present design conservation associated with any
cost-reduction trends and in holding costs hardening approach, since we must necessarily
down in areas where shock requirements adopt a conservative approach in cases where
currently exert little cost impact, time does not permit development of techni-
cally precise solutions.
It can be seen that discussion of cost con-
siderations amounts to the central theme of The second principal influence of scheduling
this paper. Procurement policies were found consJderations involves sequence of ship-
to be predicated in many ways upon cost con- building events. Sequence of shipbuilding
slderations, and the immediately preceding events can vary to some extent, depending
discussion illustrated a number of relation- primarily upon which sequence of events best
ships between cost and selection of shock suits the individual shipbuilder and second-
hardening approaches. The following dis- arily upon other factors such as ship size.
cussion of scheduling considerations will Therefore, it is undesirable to key general
again illustrate the predominance of the cost shock requirements to any assumed sequence
factor. These scheduling considerations are of shipbuilding events, unless it is implicit
discussed separately only for convenience; in that the 3equence In question would always he
effect, scheduling considerations are strictly
cost-oriented administrative factors for all
followed. Many examples could be given of how
shipbuilding event sequence and the variability
)
practical purposes. of same influence shock hardening approaches.
Perhaps most important, we find that con-
Assume for a moment that a shipbuilder is sideration of these factors is partially re-
bidding on a contract to build a class of com- sponsible for the Navy's "piecemeal" approach
batant Navy ships for which no shock require- to the shock hardening of ships. A systemg
ment has been invoked. During the bidding approach to achievement of shock resistance
period, the shipbuilder will develop schedules may be preferable from a purely technical
which define when various shipbuilding eventE standpoint, but the success of such an approach
(design, procurement, and production events) is dependent upon "everything coming together
would be performed during the term of the at once" -- implying an assumed sequence of
contract. This schedule will reflect the events which could easily disturb the usual
shipbuilder's lowest cost, most efficient way routine of shipbuilding events. The piecemeal
of building non-hardened ships. Now, assume hardening approach, which allows shipboard
that the Navy changes its mind at the last equipments and other items to be "shock quali-
moment and imposes a shock requirement upor. the fied" on an individual basis ani with minimum
ships Jn question. The shipbuilder, in re- regard for the design of supporting decks and
sponse to this change, will likely alter his other "system" characteristics, is thus
formerly-optimum schedule of events at least highly attractive from the standpoint of
slightly to account for the impact of shock scheduling considerations,
requirements upon design, procurement: pro-
duction, and other shipbuilding activities. Having considered the pr:mary influences of 4
some administrative factors upon selection of
The extent to which such otherwise-optimum ship hardening approaches, it seems appropriate 4
shipbuilding event schedules must be altered to briefly reconsider the basic objectiv6 of the
to permit satisfaction of the shock require- Navy ship hardening effort: To fully optimize
ment has a direct and significant bearing upon tho shock resistant qualities of Navy ships.
hardening cost, and provides a direct measure What is really being sought here is maximum
of the compatibility between the shock re- shock hardening cost-effectiveness; no other
quirement and the shipbuilding process in parameter can adequately define "what's opti-
general. Selection of technical approaches mum" in a meaningful way. It directly follows
to ship hardening is strongly keyed to the that administrative factors (cost factors,

40
essentially) and effectiveness factors must normally be needed (in the writer's opinion)
be given equal consideration as we go about by technical personnel who are not reqUred
the business of optimizing the shock resistant for other job-related reasons to be know-
qualities of Navy ships. ledgeable in these fields. Note, however,
that this paper does not tell all that could
CONCLUDING RBMADS AND RBD0ENDATIONS be told concerning shipbuilding event seq-
quences, shipboard maintenance requirements/
We might first consider some trenis which problems, and other specific cost-related
have an important bearing upon the subject matters. The general importance of these
of this paper. The need for improved ship factors relative to the selection of ship
shock resistance took a sharp swing upward hardening approaches has iJready been
during the 1950's, in response to a marked stressed; the reader may judge for himself
increase in the inherent shock sensitivity whether or not further study in such areas
of many new shipboard systems (primarily, new might prove beneficial.
and sophisticated electronics and weapons
suites) and due to development of nuclear Improve efficiency at the technical/admini-
weapon threats (which added a new dimension strative interface. A more deliberate effort
to the shock problem). This situation led to could be made to address administrative fac-
Navy sponsorship of a considerable amount of tors in technical proposals and in technical
shock-oriented RDT&ME, which was understand- reports which will receive administrative
ably directed almost entirely at effective- attention. A possible criterion for deter-
ness-related goals. Recent full scale ship mining how much administratively-oriented
shock tests attest to the success of these information might be included in such
efforts; we now have the technical capability documents: Provide all readily available
to deliver almost any reasonable amount of information in this area which might be un-
ship shock resistance. known to the cognizant administrator. The
writer of the technical proposal or technical
The administrative influence did not lay report may understandably be in a poor position
dormant while the events described above to furnish explicit information in this area.
were taking place, however. Owing at least However, his superior acquaintance with the
in part to DOD's recent difficulties with cost technical concepts/practices in question ani
overruns and associated contractor claims, his usually intimate acquaintance with private
considerably increased attention is now being industry's technical capabilities and limi-
given to DOD procurement policies -- and to tations may easily place him in the best
the requirements which we attach to things position to ascertain the basic admi.lstrative
being procured. Attention to cost consider- implications of his proposal or recommendations.
ations has also greatly increased, for reasons Some example "administrative implications"
we all know about. which could often be addressed (with a mini-
mum of homework) at the technical level are:
The overall influence of administrative factors Whether or not implementation might force
is thus seen to be rapidly increasing at a alteration of shipbuilding event sequences.
time when the requirement for further up- Extent to which private industry is prepared
grading of ship hardness characteristics is to respond to the technical requirements.
declining. The implications of these trends What sort of shipboard maintenance might be
are clear, and some of you may have already required. Whether or not added weight will
felt their impact. Those technical personnel result. (An example administrative impli-
who to date may have seen no real need to cation could be cited for every such factor
seriously account for the administrative side mentioned in this paper).
of the shock hardening effort might reconsider
their position in the light of present trends; Structure technical tasks to yield information
things are changing, of administrative interest. Naturally, it
cannot be expected that meaningful administra-
So much for trends; what might we do about tively inclined information of the sort referred
them? Above and elsewhure in this paper, it to above will be initially available in all
has been suggested that the technical com- cases -- particularly when basic technical
munity can contribute significantly to the concepts, rather than technical practices,
ship hardening effort by "accounting for" are the subject of the technical proposal
or "considering" administrative factors which or report. However, it is generally possible
are becoming increasingly important to the to structure follow-on developmental efforts
ship hardening effort. These are nice words, in a manner which will permit early identi-
but they lead to an obviols question: Ex- fication of important administrative impli-
actly how might the technical community cations.
"account" for needs in this area? The follow-
ing paragraphs summarize some possibilities. Finnlly (and inevitably): Directly address
administrative factors at the technical level;
If necessary, learn more about administrative take a head-on approach to solution of pro-
factors. This paper provides about as much blems related to procurement policies, costs,
shock-orlunzted insight into procurement pol- and schedules. Opportunities to promote
icies and basic cost considerations as might further optimization of ship hardness qualities

41
are perhaps nowhere more numerous than in this
area. A few of many technical objectives which
directly reflect administratively-oriented
needs: Develop ways of expanding the scope and
aplicability of shock test extension policies.
More fully develop a basis for extension of
dynamic analysis qualifications. Develop
short-cut methods for optimized design of shock
resistant foundations. Pursue investigation
bf maintenance-free shock mitigating devices.
Further standardize shock design/shock testing
procedures. Develop additional shock test and
shock design guidance material. (This list
could go on and on. The general theme should
be clear, however).

We've seen, by example in this paper and else-

where, that administrative influences can run
counter at times to rapid technical progress
and other technical interests. Hopefully, it
has also been conveyed that the administrative
influence must be accounted for in a construct-
ive fashion at the technical level if we are to
realize fully optimized ship shock resistance.
The question of how our technical resources
might best be directed at administratively-
oriented needs is at once difficult and im-
portant. Think about it.

42

42
 MEASUREMENT AND APPLICATION
OF MECHANICAL IMPEDANCE
FORCE TRANSDUCER CALIBRATIONS RELATED TO
MECHANICAL IMPEDANCE MEASUREMENTS
E. F. Ludwig
Assistant Project Engineer
N. D. Taylor
Senior Engineer
Pratt & Whitney Aircraft
Florida Research & Development Center
West Palm Beach, Florida
(U) Mechanical impedance measurements have become a popular time
and cost-saving, nondestructive industrial test tool. To obtain these
measurements, only smoll, nondestructive dynamic forces are
required at the test item. A common problem in measuring any
parameter is the suitability of the transducers and associated instru-
mentation required. In studying mechanical impedance parameters,
cognizance of the hase strain amplitude and sensitivity linearity of the
force transducer is particularly important. Special emphasis is given
to these two topics and to a survey of several time and cost-saving,
mechanical impedance test programs.

INTRODUCTION measured by force transducers, allow the-

test engineer to use portable exciters, which
The Pratt & Whitney Aircraft Florida can be maneuvered into almost any test loca-
Research & Development Center has used tion. Some transducer characteristics are not
mechanical impedance measurements exten- defined clearly in specifications. Base strain
sively in the past few years on jet and rocket sensitivity and amplitude linearity are discussed
engine development programs. Much state-of- in this paper.
the-art experience has been acquired in imped-
ance testing techniques. * These techniques . To evaluate base-strain sensitivity, the
save time and cost in nondestructive industrial force transducer is attached to a simple, fixed-
testing. To obtain these measurements, only free beam. The structural characteristics of
small, nondestructive dynamic forces are this beam, such as spring rate, effeciive mass,
required at the test item. These small forces damping, strain, and natural frequency will be
presented analytically. The force transducer
base-strain sensitivity test procedure and data
*The author E. F. Ludwig, is presently a mem- obtained from the described test beam will be
ber of the American National Standard Writing discussed. The beam calculations will be com-
Group S2-58 that is currently preparing a pared with impedance and strain test data taker
standard on the experimental measurement of during the force transducer base-strain test
mechanical impedance. The document to be program. A math model of the test beam's
produced by this committee will serve as a bending inodes also will be presented and com-
general guide for the user in the selection of pared with impedance test data.
calibration techniques and will recommend
types of evaluation tests necessary for deter- The force transducer calibration test data
mining the suitability of transducers and described in this paper illustrates the need for
associated instrumentation. A section of the standard procedure to describe amplitude lin-
S2-58 document will discuss force transducer earity and sensitivity deviation. A proposed
base strain sensitivity and amplitude linearity, procedure is defined and discussed. The paper
This paper, which will be presented at the concludes with a description of several imped-
42nd Shock and Vibration Symposium in Key ance test programs that yielded considerable
West, Florida, will contain a more compre- savings at Pratt & Whitney Aircraft's Florida
hensive discussion. Research & Development Ccater.

43
SYMBOLS acceleration levels (a) with a known mass
attached (m)and measuring its output signal
A Beam Cross-Sectional Area in2 (Ex). A standard 200 Hz frequency should be
a Acceleration - g's used if it iswithin the rating of the transducer.
c Damping - lbf/in/sec Calibration data points will be taken at least at
CC Critical Damping - lbf/in/sec octave band intervals between one and 200 lb
d Displacement - in "nat 100 lb intervals over the rest of the
E Modulus of Elasticity - Ibf/in2 traf-cer span. For transducers rated at less
Ek Kinetic Energy - lbm/in 2/sec 2 than one p,,nid, data should be taken in the
Ex Output Signal - my. or pc same manner to cover the span of the transducer.
F Force - lbf
f Frequency - Hz The above calibration data will permit cal-
g Gravitational Constant - lbf-in/Ibm-sec 2 culations of force transducer sensitivity as a
I Moment of Inertia - in 4 function of applied force.
J 7 Applied Force F = ma
K Spring Constant - lbf/in Force Transducer Sensitivity Sx = Ex/F
L Beam Length - in
M Moment - lbf-in Since the force sensitivity is now known over
In Mass -Ibm the span of the transducer, its deation can be
me Effective Mass - Ibm calculated in plus or minus percent from the
Sx Sensitivity - mv/lbf or pc/lbf mean.
SD Sensitivity Deviation
t Time - sec Sensitivity Deviation
V Velocity - In/see Sx-S v
W Weight - lbf SD = S X1 00
We Effective Weight - lbf avg
x Distance Along the Beam - in This seems to be one of the better statistical
y Deflection of the Beam - in methods for stating deviation. A calibration
ym Maximum Deflection - in curve supplied with the transducer specification,
Z Mechanical Impedance - lbf/in/sec however, would be most desirable. This infor-
a Logarithmic Decrement mation would allow the user 'o establish sensi-
E Strain - uin/in tivity deviation over his span of concern. For
P Density - lbf/in3 example, Figure 1 illustrates a transducer
u Stress - lbf/in 2 torqued to 25 footpounds with a deviation of
Wo Circular Frequency - rad/sec approximately *9% when defined at 1800 pounds.
The same transducer at 400 pounds is a 3%
AMPLITUDE SENSITIVITY LINEARITY transducer. Mony specifications state sensitiv-
ity deviation at * percent of full scale. This
The study of dynamic structures using the percentage loses significance at relatively low
mechanical impedance approach entails the force levels which are normally applied during a
measurement of the applied force. This force mechanical impedance test (i. e., a transducer
may vary over a wide dynamic range due to the with a 500 *2% pound full scale force rating
responsiveness of the structure under test. would allow for a i10 pound deviation. If a test
For this reason the amplitude linearity charac- requiring 10 pounds of applied force were
teristics of the measuring transducer must be required, the user of this transducer could only
established. Published specifications of force guarantee that the applied force was withla 0 to
transducers do not always provide sufficient 20 pounds). The calibration specification should
data for the mechanical impedance user. Some also include information about preload versus
users may not be aware of the insufficient data torque for the type of attachment bolt recom -
that is important to make a valid measurement. mended and mass and accelerometer NBS
The following may affect amplitude sensitivity traceability.
and its deviation: preload mounting torque;
different type attaching bolts; lack of standard To test the practicality of this procedure,
methods of stating linearity and methods of quartz and plezite annular transducers were
obtaining calibration data. A standard calibra- calibrated using these guidelines. Figure 2 is
tion procedure must be established, a photo of one transducer mounted on a dynamic
exciter with the calibration mass attached.
The following test procedure will allow this Figures 1 and 3 illustrate empirically derived
standardization to be accomplished: The amp!i- force transducer sensitivities versus applied
tude linearity calibration is performed by force for both transducers. These figures Illus-
vibrating the force transducer at various trate that increases in compressive preload
caused increased sensitivity and improved
44
 TEELDOLT SxAVO. SDAT IOLB AT400LB a

OI . II 2.43 I 7A1 I 2 -..

-5 BERYLI MCOR BLT

1&0 8PRELOAD 100 IN..LB DYNAMICDATA
0 Pr.LOAO 100 I.L)STATIC DATA
-2.4 X PRELOAD60 IN..L DYNAMICDATA

S16 -STEEL BODLT

100 IN..LB STATICDATA
2. PRELOAD

DL -- 00 t0 160 -240 32
1000 1200 1400 1600 1800 APPLIEDFORCE- LB
0 200 400 600 SW
APPLIEDFORCE. LB

Fig. 3 - Annular Quartz Force Transducer

Fig. 1 - Annular Piezite Force Transducer Calibration Data
Calibration Data
sensitivity tended to increase at low force
levels at a different rate than static sensitivity.

FORCE TRANSDUCER BASE STRAIN

SENSITIVITY
Determination of force transducer base-
strain sensitivity involves mounting the trans-
ducer on a surface which could be stressed to
measure accurately the strain at the base of the
- transducer. The transducer used was an
annular piezite element that requires a center
-, torque bolt for attachment.
A cantilever beam was selected as the best
surface for determining strain. If all proper-
ties of the beam were known, strain could be
- determined both theoretically and with gages.
The theoretical spring rate and effective mass
" ": for a 16-Inch cantilever beam at first bending
." ' were computed as well as the first five reso-
nant frequencies. The data was then compared
with actual mechanical impedance data taken at
various points along the beam. Spring rate was
checked against the impedance data. The effec-
tive mass also was checked against the data.
Fig. 2 - Amplitude Linearity Calibration Resonant frequencies computed theroetically
Setup were very close to those appearing in the imped-
ance data.
amplitude linearity. Figure 3 illustrates the
effect of bolt type on sensitivity. The steel bolt Holograms were made of the beam during
decreased sensitivity but showed better linear- vibration at first bending. Displacements were
ity compared to the berylium rooper bolt. taken from the reconstructions of these holo-
grams and compared with theoretical deflection
Both static and dynamic test data were values calculated using the standard cantilever
taken on the quartz force transducer. The beam deflection equation and sinusoidal
same trend in data waE. consistent for both con- approximation.
ditions; sensitivity increased with torque, .iad
linearity improved. The main difference noted A mathematical model of the beam was
was that static calibrations consistently showed made to further substantiate the mechanical
a lower sensitivity. Also, the dynamic impedance at the tip of the beam. The model

45
was a se-ies of mass-spring-damper systems. The spring constant K at d.
The values of spring rate, effective mass and
damping were based on data taken from static K n 3EI/d 3
loading tests, logarithmic decrement or read
directly from the impedance data. For example, the theoretical spring con-
stant at the tip of a sixteen-Inch steel beam with
After determining values of spring rate a cross-sectional area 3.0 x 0. 5 in. Is S86
and effective mass, the transducer base strain lbf/in. Testing determined the static sprLn.g
was calculated for various tip displacement and constant for this beam in a clamping device to
acceleration levels. These values were corn- be 562 lbf/in. Deflection inside the clamping
pared with test data taken at these levels, device is the most likely cause of the discrep-
With strain determined for different loads, the ancy between these constantly. The theoretical
sensitivity of the transducer to base strain was letwul thene 17. 1 inche
recorded and analyzed. length would then be 17. 1 inches.
The effective mass of a cantilever beam at
BEAM ANALYSIS FOR IMPEDANCE first bending can be calculated by energy
CORRELATION methods. The beam deflection mode shape can
The equations used in the majority of this be described as
section stem from beam theory. The basic 4 1
differential equation is for the beam's elastic 3
Y Ym/3 (x/L) - 4 (X/L) + 31
curve.
The kinetic energy of the cantilever is
d y/dx 2 /EI
W Ek = 1/2 fL (dy/dt)2 dm

I wvhere dy/dt =wy and dm = NV/gL dx

Integrating we get

y - X Ek 0.1283 Ww 2
M /g

AThis is equal to the kinetic energy of the

SP_ effective mass:
2 E = 1/2WW2ym 2/g
k
me = 0.2566m
Using the boundary conditions of a cantilever
beam under uniform loading, this equation can That is, a steel beam sixteen inches long
be solved and used for theoretical checks on with a cross-sectional area of 1. 5 in 2 would
actual cantilever beam data: have an effective mass at first bending equal
4
-pA/24E (0~-419,x# 31, ) to 0.00483 Ibm. (We = 1. 86 lbf.)
y
A second method of determining the effec-
Ym- -pAL 4 /8EI at x 0 tive mass was suggested in Reference 1. The
mode shape at first bending was approximated
These equations are used to determine the by using
theoretical spring rate at first bending and the
3
effective mass. They also provide a ch:c¢k on Y m (I - cosrn/2L)
natural frequencies.
Figure 4 compares this mode shape with the
The predicted spring 4,nstant for an ideal elastic ctirve. Solving the integral for kinetic
cantilever beam can be calculated as a function energy yields a slightly different value for effec-
of length from the clanip
1 d end. The beam tive-mass.
deflection produced by a force at a distance
d from the clamp point equals: me 0.227m

3 2 2 me 0.00427 Ibm
y -F/6I (-d , 3d L - 3d x)

46
 Tf

Sproduced from
best ava:lable copy.

CURVE
"EI.ATc
100 icOhINE F.OXIMATION

* HOLOGRAPIHYDATA
so (SHIFTED
TO ZERO)

a40 8 _ _

20-

0 20 40 60 so 100 ,.
LENGTH• With Transducer-Mounted Without Tansdumw Mounted

Fig. 4 - Deflection of a Cantilever Beam Fig. 5 - Holograms of the Cantilever Beam

To determine which deflection curve where with appropriate boundary conditions,
approximated more closely the mode shape at the nontrivial solution yields:
first bending, a hologram of the sixteen-inch
beam was made during vibration. This study cosz -1/coshz
of the data from the holography curve indicates rooL
that perhaps the beam was not secure in its where z =
fixture during the mode shape holography. In w
general, the trend of the data is good, but the
shift at the zero point indicates slight motion s z4
at the clamp. For the hologram chosen, tip
deflection was 64 microinches single amplitude. z z2.
Data taken near the clamp indicates approxi-
mately 1. 5 microinches double amplitude with
the true zero deflection point approximately two
inches beyond the clamp. Shifting the zero
point makes the mode shape agree closely with The natural frequencies are
the cosine curve.
n IElg/pA (Zn/L)2
As shown in --igure 5, the holograms
revealed no difference in the deflection with the
transducer on or off the beam. This means that The first five resonant frequencies are:
the preceding calculations are valid for calcu-
4
lating transducer base strain. z1 1.875 EIg/PAL - 114.

The resonant frequencies of a cantilever wI z2 EIVAL 4 - 401. f 63.8

beam are obtained by solving the partial differ-
ential equation of motion. This will be done for This frequency would be 55. 9 Hz If the beam
a steel beam with dimensions 16 x 3 x 0. 5 inches.
Beamequaion:were 17. 1 inches long.
Beam equation:
z 4.694 f2 /2i(114.) 400.
2L2\2 2 2 z1121
._.x / x , 7.855 f
x4 g 4 - 10.996 f 2200.
4 4
Since I and A are not functions of x, and assum- Z5 14.137 f5 3620.
ing no external loading, Tw3 other methods of finding the resonant
frequencies w, Investigated. The Rayleigh
method used the inusoidal approximation deflec-
a2 4
y/Jx 4 --
2
y/3t 2 a2 Elg/pA tion curve. The other is the energy method, for
which the beam deflection equation was used.
Beam equation results agreed very well with the
results of the previous section.

47
CORRELATION WITH IMPEDANCE DATA
An actual steel beam measuring 16 x 3 x
0.5 inches was tested for impedance with the
drive point located on the centerline at differ-
ent lengths from the clamped end (Figure 6).
TheoretiL.al values of spring constant, effective
mass and exponential decay damping data will
be compared with impedance test data. At
resonance, the impedance due to the spring
term and mass term cancel each other leaving
only damping.
= +Twm -TK/W

At resonance,
wi= K/1
IZI =c

Since the spring constant at the tip of the

cantilever is known, and the frequency at first
bending is known, the effective mass can be
calculated by this method also. Tip spring
constant
K = 686 lbf/in Fig. 6 - Strain Sensitivity Calibration Setup

Effective mass Logarithmic decrement

6 (1/n-i) inx/ n
me = K/(2rf)
2 where f - 63.8 1z
From actual decay data:
me = 0.00427 Ibm Xl Xn

Effective weight
e = 1.65 1bf 2.175

This agrees well with the effective mass x5 1.94

(weight) calculated by the kinetic energy 46 in (2.175/1.94)
methods. Mass and spring constants can next
be used to calculate the critical damping coef- 6 0.0294
ficient.
cc 2 1Kme 3.42 lbf/in/sec To calculate actual damping

With spring constant and effective mass deter- c/c c -6

mined analytically, damping must be derived
from the logarithmic decrement. A decay c 0.016 lbf/lin/scc
curve of tip acceleration was made so that the
damping could be derived. The frequency of Figures 7-10 show data taken at different
this free vibration was at 57. 9 Hz. points along the beam. (16, 12, 8, 4 inches
from the clamped end.) In Figure 7 where the
beam was being excited at the tip, the theoret-
ical spring rate, actual damping and effective

48
 STIFFNESS MASS. LB (WEIGHT)
loS 02 103 10 102 IN. f 10

104

FROKO H)FREQUENCY- HZ
Fig. 10- Impedance of the Cantilever Beam
Fig. 7 - Impedance of the Cantilever Drive Pt. 12 in from 'lip
Beam Drive Pt - Tip weight lines have been drawn. Tl'e inserts are
mode shapes of the beam at the particular reso-
LB nant or antiresonant point. Theoretical spring
STrIFFNESS -
IN.
MASS, LB (WEIGHT) rates
From are
theseshown
data onit istheevident
three other figures. ratio
that damping
3 5
.010. 101
3 0 10 /1 10 is fairly5 constant
10.001001 along the beam.

°4 ' " " .o COMPUTER SIMULATION

zo o.1To further correlate the actual and theoret-

102ical impedance of the cantilever beam, a com

into a series of mass-spring-damper systems.

Each system matched one resonant frequency
1.o0 on the impedance data. The impedances of all
=[ systems were added to calculate the total
10 5
1"
10
100 0 _oN
.1
impedance.
Fo hs aai seietta apn ai
FREQUENCY •HZ
The impedance of a single mass-spring-
Fig. 8 - Impedance of the Cantilever Beam damper parallel system can be described by
Drive Pt. 4 in from Tip using a combination of real (damping) and imag-
inary (mass and spring) vectors.
STIFFNESS MASS, LB (WEIGHT)

i_- 0 . .0Imaginary '

FRGUNC -
HZ'( 5-K °
. Real-

5 10 10 1000 00s Since systems added in series must be

FREQUENCY1 HZ summed as the inverse, it is easier mathe-
Fig. 9 - Impedance of the Cantilever Beam matccally to sum mobility, the inverse of
Drive Pt. 8 in from Tip impedance.

1 wn
am-lt} 4

49J
I
 The real and imaginary terms can be =
1
2I
first antiresonance
summed separately. Then, the mobility will be
the square root of the sum of the squares of the me= IC2s d o
real and imaginary terms. A comparison of 2 second resonance
the computer simulation with impedance data The simulation's spring rate asymptote at
is shown in Figure 11. It Is obvious that the low frequencies is lower than the spring con-
antiresonances are not being modeled properly. stant used tn model the first resonant mode.
Investigation into this phenomenon using differ- This is due to the additive effects of springs in
ent models will continue, series. Most test specimens at FRDC, however,
have shown the asymptote of the impedance data
to be very close to the spring constant value for
the fundamental mode. Spring constants of
-L.J4J4J JL4.JJ secondary modes have not lowered the asymptote
Iappreciably. Even though the above approach
seems basic for a simple cantilever system, the
theoretical calculation, impedance testing and
00000 904850 computer simulation method can be broadened
l ol o3 106 102 1o to encompass more complex systems. A

BEAM AS A STRAIN INPUT

- With the deflection of a cantilever beam
0 thoroughly investigated, this beam will be con-

force transducer mounted on the beam. Output

- V of the transducer as a function of strain for
- various torque loadings and rotational positions
were determined. Strain was predicted as a
001 function of effective mass times acceleration
and also as a function of spring rate times tip
__ _ _ deflection. These values were compared with
FREOUENCY (Hir actual strain readings.

Fig. 11 - Computer Simulation - Cantilever CALCULATION OF STRAIN

Beam Impedance
A bar bends when subjected to a moment
The spring
model tested wasrate at first
chosen bending
as 562 for the
lbf/in., the M. For a small
acodn beam deflection, the strain
oHoeslwi
static spring rate. This value was chosen according to Hooke's law is
rather than 686 lbf/in. because of the move- zx = 1/E(ax-Vay) -My/El
ment inside the clamping device discussed pre- Where Cy 0
viously. The length of the cantilever was con-
sidered as 17. 1 inches. The first resonant For the calibration of a force transducer, we
frequency was 57. 9 Hz, the free vibration value are interested in the strain at its mounting
measured on the decay data. The calculated point at the tVansducer base. Data taken from
effective mass used for first bending was strain gages located near the transducer mount-
0. 00427 Ibm. ing pohit were compared to calculated values
of strain based on spring rate and effective
Damping is equal to the value of impedance mass in Figure 12. The beam experienced
at each resonant point. The values of spring local stiffening when the transducer was mounted.
rate and effective mass for each of the other Figure 12 shows the effect on strain of mounting
modes were based on the frequencies at the the transducer as well as the range due to torque
resonant and antiresonant points obtained from and rotation with the transducer mounted.
test data. In formula,
Strain in the beam is dependent on the
2 force applied at the cantilever tip. Tip accel-
me 1 Ri renaceeration and displacement at first bending
together with effective mass and spring constant
yield two methods of determining force: one

50

~- . .
 CALCULATED ROTATION 300 TORQUE
STRAIN. 6 0 25FT.LE

2f120 WITIH3 T

2 i-

40- ,, TORQUE
-REDUCED ;5
RANG0E
WITH

MOUNTED INERTIAL
.0 - -I-- . . . FORCE
2 4 a 10 0 40 60 80 100 120
TIP ACCELERATION
•G STRAIN. • NN.

Fig. 13 - Force Transducer Output -

Fig. 12 - Transducer Base Strain Effect of Torque at 300

potential and one kinetic. Peak values are

obtained. The added mass of the drive rod at
the tip of the beam shifts the resonant fre- _0_ROTTION_0

quency lower (49 Hz) than noted for free beaRATON 0

vibration. This extra mass must be added to 50 0 26 FT.LB
the effective mass of the beam to obtain true C> X0
strain data. Electronic mass cancellation is 4o014 3

used to correct mechanical impedance dat-, to El 4s

RESPONSE OF A TRANSDUCER TO BASE
STRAIN 2 20

Mounted on a sixteen-inch cantilever to

beam, a force transducer was subjected to base
strain induced by vibration at first bending. 0 20 40 60 80 100 120
The outpv.t of the transducer was recorded at STRAIN. ,-INAN.
different strain levels for different torque
loadings and as a function of rotation around
the transducer's axis. The transducer's output
due to base strain was 200 times its normal Fig. 14 - Force Transducer Output -
inertial force output. Findings indicate a sig- Effect of Torque at 00
nificantly higher output as torque is increased
(Figure 13, 14). They also show that output is
related to rotational position (Figure 15). One -0? _ T_
1
characteristic of output during these checks TOR
OUE
was that initial loading sometimes caused a - 02S FT.L8
wa- -R45 FT.8L
peak transient reading. This reading then STRAIN - ,&3PININ.
decayed to some steady-state value after a few 25L
INERTIAL FORCE 2-
seconds. At present, no explanation for this
can be offered although the trait did depend
somewhat on rotation. Further studies will be
made.
10
Actual inertia force, shown in Figure 13, .

illustrates the magnitude of error due to base B

strain sensitivity. Most applications should 60 12 0
avoid base strain if possible. If not, then pre- ROTATION. DEG
cautions should be taken in using force data.
Fig. 15 - Force Transducer Output - Effect
of Rotation

51
SUMMARY
A -B~
Thus far this paper has presented two dif- STIFFNESS
L8 104 MIGHo" ST FN2 ,IF
T to)~cr
1 WAS L1
.- o 10. 0 "SS, 12CT
ferent transducer characteristics for users to IN. ° -<I -
consider when evaluating force transducer a -- I: I
:1410 . a ... .10
calibrations. The first section suggests a
0
standardizing procedure for amplitude linearityI " tJI n '
and sensitivity deviation. The second, base 00i
strain sensitivity, suggests a procedure for -:10 0
_i.00j 10 . i' 0.001

determining output as a function of base strain. lo 0 1 10

Cognizance of these areas will help a user ,RIOUNCY Z 4Z
100FREOUENCY
determine the validity of his mechanical imped-
ance measurements.
B -to STIFNESS 7-N 10 102 MA3 LS
1 7
Studies of the cantilever beam system used STIFFNESS 0 STIFIHTS 10 -.
10 103, to~
for base strain sensitivity revealed a means for . "'>." o
i- -.
comparing theoretical and test results and _0 10 ! " ..1.0....
served as an aid to simulate the mechanical ,0 01 .
impedance of this system with a math model. . 3,

This same procedure could be extended to more ; - - I....

complex systems. 111 ~ 5--. .j- .0
1000 1 100 10 (HZI
000
FRDC IMPEDANCE TEST PROGRAMS 11FREQUIAIXOZ FROEUENCY

Fig. 16 - Louver Section Damping Study

Mechanical impedance measurements have
become a popular cost and time saving tool. frequency range that could cause clamp wear.
This section discusses several representative This frequency also correlated with a critical
impedance test programs conducted at the frequency source generated by the development
Pratt & Whitney Aircraft Company, Florida engine. With the test proven effective in diag-
Research and Development Center. During nosing clamp wear, this procedure could be
these programs, considerable time and cost extended to determine if a burner section requires
were saved by implementing this nondestructive overhaul during prescribed maintenance
test technology, schedules.

COMPARISON STUDY OF DAMPING SCHEMES MECHANICAL IMPEDANCE DEFINITION OF

CRITICAL PLUMBING BENDING MODES
impei-
A determination of the mechanical
ance characteristics of several different louver Point and transfer impedance data taken on
section designs permitted selection of a practical the pump plumbing (Figure 18) defined critical
and optimum design. Of all seven louvers resonant frequencies. Laser (time average)
tested, the "handlebar mustache" design (D)Pos holography was used to signature plumbing
sesses the best damping characteristics. From deformation at each minimum I.ipedance drive
a practical design standpoint, however, the point location so that particular component
short-hooded louver (c) possesses sufficient damp- spring rates and mode shapes could be defined.
ing properties below 750 Hz. Figure 16 com-
pares four louvers tested. VERIFICATION OF PREDICTED ANALYTICAL

BURNER CAN CLAMP WEAR DIAGNOSED PARAMETERS

THROUGH IMPEDANCE MEASUREMENTS Impedance meaiuremente were used to
AND LASER HOLOGRAPHY verify analytically-predicted dynamic spring
rates of a rocket engine fuel pump bearing sup-
Excessive clamp wear during an engine port. The bearing support and housing were
development program prompted an investigation mount6d for test as shown in Figure 19. Trans-
of burner can damping characteristics. The fer impedance data was takei on the critical
data from the impedance test program described members of the bearing support system. Analy-
critical frequencies at which mode shape holo- sis of this data showed good correlation with
grams were taken. These holograms (Fig- calculated spring rates. The test results indi-
ure 17) depicted a bending mode in the 100 Hz cated that a critical speed analysis program was
within design tolerances.

52
 Fig. 19 - Bearing Support Test Setup
ROCKET ENGINE SIDE LOAD DETERMINED
BY DYNAMIC SPRING RATE TEST
Substantial cost savings are realized by
the effective use of mechanical Impedance
measurements. During a recent rocket engine
test, for example, a standing shock wave was
ingested into the nozzle of a staged-combustion
r.g and became attached during the run. The
point of attachment could be determined an well
as the side displacement when attachment 4
12 NZ 11 ioccurred. To estimate the amount of side load-
ing which caused this displacement, the spring
Fig. 19 - Buraer Ca. %Clamp
Wear Test Setup constant of the nozzle system had to be deter-
mined. Since the engine was in the test stand
and time was a very critical factor,impedance
was measured at the test site. The exciter was
positioned at several locatiors along the nozzle.
Figure 20 illustrates the exciter at the tip.
From impedance data taken in the 5-5000 Hz
range the first bending made of the extendible
nozzle assembly was found to occur at approxi-
mately 5 Hz, the first ring mode occurred at
42 Hz. A spring rate of the nozzle system was
determined by anayzing the impedance plots and
constructing mathematical models fron the low
order modes.

Fig. 18 - Pump Plumbing Mode Test Setup

53j
 DISCUSSION

Mr. Bouche (Endevco): As chairman of

American National Standards Institute commit-
tee S2-58 I would like to compliment Mr. Ludwig
on doing a very excellent job in making these
A. ,calibrations. Early in the work of this commit-
tee we recognized that these two characteristics
were very import-nt for this standards document
and that additional calibration work wou!d have
to be done in order to have enough information to
include in the standard. So we asked Mr. Ludwig
to make these measurements, and the results
WI ~ are good. The amplitude linearity measure-
ments turned out better than expected. The
base-strain sensitivity turned out to be quite

I11
serious, because if there is a strain present in
actual experimental measurement of the order
of, say 100 microinches per inch, then large
force errors could result. It would be neces-
sary then to select locations on the test speci-
men where it is known that the strain environ-
ment is small.

Fig. 20 - Nozzle Spring Rate Test Setup

REFERENCES

Bonesho, J. A. and Bollinger, J. G., "Self-

Optimizing Vibration Damper, "Machine
Design, Feb. 29, 1968, pp. 123-127.

Freberg, C.R. and Kemler, E.N., Elements

of Mechanical Vibration, 2nd Ed. John Wiley &
Sons, Inc., 1966.

Harris, C. 0., Introduction to Stress Anslysis,

McMillam Co., N.Y., 1959, pp 93-94.

Wylie, C. R., Jr., Advanced Engineering

Mathematics, 3rd Ed., McGraw Hill Book Co.,
N.Y., 1966, pp 323-326.

54

- ------
 THE MEASURNP0 OF MBHANICAL DMPEtANCE

A D ITS USE IN VIBHAT7O TzsTING

N. F. Hunter, Jr., and J. V. Otte

Sandia Corporation
Albuquerque, New Mexico

Mechanical impedance has, in tne pa~t, been used almost exclusively as an

analytical tool in the vibration field. This paper considers both the
laboratory test procedures and instrunentation used to measure the imped-
ance of structures. This is followed by both an interpretation of typical
data and the utilization of this data for the following:
1) derivation of laboratory test specifications,
2) comparison of a system's impedance characteristics before and
after tests to determine structural failure, and
3) electronic simulation of a structure during force controlled
vibration tests.
,Each of the above applications is supported by specific examples.

MECHANICAL D4PEDANCE AND APPARENT WEIGHT on the system is termed transfer impedance (ZI).
Since
Mechanical impedance (Z) is defined as the
complex ratio of the driving force acting on a
system to the velocity response of the system. H(2) =

where

where H(w) = transfer function

Z(w) n mecnanical impedance** vI = velocity at point of force input

F(w) - driving force v2 A velocity at some other point on

system,
response
v(w) = velocity
then
€(w) = phase angle between F and v.
L_ F . 1 Z
An alternate expression for Z is in terms ZT" v2 TRW-, HFw) p
of its real and imaginary components:

Z a Ecos 0 + JEsin where

v v transfer impedance

If the velocity response is measured at Z a driving point impedance.

the point of force input, the Z is referred to P
as driving point imredance (Zp). Conversely, Apparent weigit is defined as the complex
the use of velocity response at another point ratio of the driving force nnd the acceleration

*Tlis work was supported by the United States Atomic Energy '7ommission.
*0Mechanical impedance and all related terms are a function of frequency.
Hereafter, the (W) will
be omitted and hence implied.

55
 response of the system, Acceleration is in where
terms of G, where one G is equal to 386 a/sec.
W mapparent weight (pounds)
F
11 - Fz.0 g a acceleration of gravity (386 in/see)2
where k a spring constant (f/in)
mhr =88 (1le-. 2 )
weight
-apparent
I
F • driving force v static weight
0 - acceleration response (0 ) c coefficient sc
adamping
in
- circular frequency (2nf).
-0= phase angle between F and 0.
The apparent weight characteristics
Using the relationhip ex-
tion and velocity; v a Gg/djw,between
then
accelera- pressed above are plotted in Figure 1. Knwl-
edge of the above characteristics will often
prove valuable in the interpretation of labora-
V- z tory data obtained during system analysis.

where
Wa apparent weight
MASS Wm =mg
Z * mechn-nical impedance
g = gravitatiorl acceleration Q
(386 in/sec2)
w a circular frequency (2nf).
L
LOG f
The point and trensfer concept applies to
apparent weight the same as for mechanical im-
pedance. Namely,

PT1(w DASHPOT Wd =
waw Z-
where Qt

wT a transfer apparent weight -6 dB/octave

V a point apparent weight
P LOG f
HW()- transfer function.

INPERTATION OF APPAREW WEIGMr DATA

The apparent eight characteristics
three ideal elements (mass, dashpot, and of the
spring) SPRING W -

are given by the following equations. 9 -

Has4 W a mg - w -12 dB/octave

dLOG f
Spring w 2 Fig. 1 - Apparent Weights of Ideal Elements

56
 A typical apparent weight (W) plot, ob- is very non-resistant to motion (an
taied in th3 vibration laboratory, is shown in apparent weight less than 5#).
Figure 2. d) Weight of Decoupled
Subsystem
130" If one is fortunate enough to be working
go with a simple system, it is sometimes
possible to determio the weight of the
decoupled subsystem -+-r resonance.

Consider the siml~lified system and its

apparent weight cirve depicted in Figure
100 3-

500

SYSTEM A-l
r-RESONANCE
(SUBSYSTEMRESONANACE) Wk C
100

22

Inti ae h ltd ofW wela 1 '(v + fW

2t 50 10 500 1000 2M0

Fig. 2 Typical Apparent Weight Plot

NSYSTEN RESONANCE

In this case, the amplitudie of W, an wefl as '2f 0 f4

the phase between force (F) and acceleration wO FREQUENCY
(HZ)
(0), are plotted on log-log paper as a func-
tion of frequency. This data is discussed Fig. 3 - Apparent Weight Plot
below. Showing Decoupled Mass
a) Lo Frequencies - Rigid Body
The system is rigid over the low fre- Over the frequency range f 3 to f4 Hz, after w2
quency range fror 20 Hz to 50 Hz. The has decoupled, the weight of w2 is found as
magnitude of the system apparent weight follows:
'W) is equal to the static weight (145#);
i, is constant; and the phase is zero
dagrees. This is the frequency range w2 W1 W2 '
vkere pre-test measurement of the static
weit hv facilitates calibration and pre-
test confirmation. It should be noted that as a general rule the
peaks and notches of apparent weight respec-
b) System Anti-Resonance tively correspond to subsystem and system
resonances.
The peak in V, associated with a phase of
900, at 400 Hz signifies an anti-
resonance with respect to the point of APPARNT WEIGlHT MECHANICAL NPEWICE
measuremant. In other tords, a subsystem,
somewhere above the control yoint, is Although mechanical Impedance is a familiar
resonating. The peak of Wsignifies vibration analysis parameter, apparent weight
resistauce to motion at the driving point is the more efficient and convenient environ-
(an apparent weight of 7000#). mental test tool. The reasons are outlined
below.
c) System Resonance
The ntch in W associated with a phase aN Accelerometers are the most common trans-
shift from 180 to 900, at 780 Hz sign.- ducers used in the laboratory.
fies system resonance where the system

57
 b) Apparent weight has dimensions of pounds. A fixture Is usmally required between the
The fact that a system normally acts as a force transducer and the system being analyzed.
rigid body at low frequencies (10 to 20 Several fixture requirements are listed below:
Hz) means that the apparent weight is
equal to the static weight within this a) The fixture must be rigid over the fre-
frequency range. This provides both an quency range of interest. In addition
easy and accurate pre-test calibrate to axial rigidity, there must be no
scheme. resonances between the transducers in a
muti- ge array.
reported in terms
c) Field data is normally
of acceleration. As will be demonstrated b) The attachment of the transducers to the
later, apparent weight can be used in fixture must be rigid.
direct calculations involving field data
since both involve acceleration. c) The attachment of the fixture to the
system must be rigid.
d) Many vibration tests, particularly the
surveys designed to determine the dynamic d) The fixture should adapt to the test item
characteristics of a eystem, use a con- in the same way as the field mount.
stant acceleration input control. If
the control accelerometer is located at e) Commercial force transducers are moment
the point of force measurement, a ratio sensitive. The fixture must be designed
is not required. The force output is to minimize moments acting on the trans-
directly proportional to the apparent ducer.
weight.
f) Motion should as nearly as possible be
e) Apparent weight (along with blocked force) confined to the axis of force input.
may be used to define the foundation's
characteristics by a test
an seenweight item To force
sumarize a, b and c above, the motion
[2]. In fact apparent may be in- at the transducer face must equal that
serted in any equations involving at the base of the system (both amplitude and
mechanical impedance by using phase). When a fixture is between the force
transducer and the system, the force required
aZ to drive the fixture must be subtracted from
Z =.W. that indicated by the force transducer(s) as
expressed below.

1EASURO1I o
OF APPARR;T WEIGHTP Fss. "PFT -_F
F
The basic requirement is measurement of
the driving force and the resultant accelera- where
tion. These measurements, as well as the
fixturing, are outlined below prior to con- Fs - force into system
sidering the excitation source and the instru-
mentation for ratio and phase measurements. FT a force indicated by transducer

1) Force Measurement FF - force to drive the fixture.

Commercial force transducers, as well as Since the fixture (by design) is rigid,
one developed at Sandia ill, are all acceptable. the force required to drive it can be exprebsed
All known force transducers are moment sensi- as:
tive. This must be considered in test design
and data analysis. The transducer(s) must be F -
inserted intermediate to the path of force F
transfer. The input force must be controlled
and/or measured at the point or plane where where
apparent weight is deaired. If the interface
are& is small, one force transducer Is used. w = fixture weight
However, ekmulti-gge array will be required
for systems with a large interface. The total 0 = acceleration of fixture ( )
force from a multi-gage array is obtained by
summiag the instantaneous output from each Therefore,
transducer. Note that a phase sensitive sum
must be used since apparent weight is the F a F -
complex ratio of force to acceleration. A DC s T
summation (i.e., peak averaging) cannot be
used since the resultant is a DC, phase insen- This subtraction can be done electron-
sitive, signal. ically [2]. The electronic subtraction

58
technique will be included in the discussion of
apparent weight simulation below.

For a single force transducer, calibration

is accomplished (1) through use of the voltage
4) Instrumentation
The instrumentation used to obtain the
complex ratio of F to G is constantly being
improved. The system depicted in Fig. 4 is
I
or charge sensitivity, or (2) by measuring the fairly representative of the required setup.
acceleration of the rigid system above the force
input (at low frequencies) and using F - ma a wG.
Multi-gage array calibration is normally by
method 2 above. 1 6

2) Acceleration Measurement S.MA

Commercial accelerometers are acceptable.

Measurement of point apparent weight requires
that the accelerometer(s) be mounted on the
test item or fixture immediately above the force
transducer(s) so as to represent the accelera- "M
tion at the point of force input. Transfer
apparent weight requires that an accelerometer ,
be mounted at the point of interest above the
force transducer(s). Multi-accelerometer
arrays are normally used to monitor accelera- .
tion of systems with a large interface. The A. rADA
general rule is to locate an accelerometer
directly adjacent to (above) each force trans- Fig. 4 - Apparent Weight Measurement System
ducer. In order to maintain a phase sensitive
acceleration signal, the multi-gage arrays are
instantaneously (AC) averaged during the test. Signal conditioning amplifiers are required for
It should be re-emphasized that for point the piezoelectric force transducers and accel-
apparent weight, the acceleration required is erometers. The F and 0 signals are filtered
that on the test item at the point of force (extract fundamental) and fed to log converters
input. Therefore, any fixturing must be rigid whose outputs are DC signals proportional to
(acceleration equal in phase and amplitude to log F and log G. These DC signals are sub-
that at the point of interest on the test sys- tracted to yield log FIG which may be recorded
ten). Accelerometers are calibrated according or plotted.
to voltage or charge sensitivity.
Note that the log converters used are
3) Force Generators designed for apparent weight measurement. They
provide constant amplitude outputs phase co-
Several factors influence the size (power) Lerent with the input F and G signals for phase
of the vibration source. These are discussed detection. In the newer and more sophisticated
below. apparent weight systems the trucking filtering
and log converting may be done in one operation
A small amount of force can be used to to increase system dynamic range.
measure apparent weight of heavy items. For
example, r.50# generator was used to analyze The phase meter provides a meter reading
a 3000-lb system. However, it should be noted and a DC output proportional to the phase dif-
that the acceleration level will be extremely ference between F and G. This DO proportional
low for even moderate apparent weight values. to phase may then be recorded or plotted.
A highly sensitive accelerometer will thus be
required. 5) Inpuu Control
The size of the vibration machine required Primarily, apparent weight is computed
is often dictated b. the size of t' system during sinusoidal tests. In practice, experi-
interface to be analyzed. For example, multi- mental determinations using random vibration
point force input to a test system of large inputs have been made [3]. Using the equations
diameter raquires a large vibratfon machine.

In cases where the apparent weight is - IuI 2'

measured fr the purpose of comparison (dis-
cussed later) low force input is acceptable. Gaf -W
However, in cases where the apparent weight is
to be used in conjunction with fiell data or where
environmental tests, the force and/or acceler-
ation test levels should approximate those W apparent weight
anticipated in field and laboratory usage.
This requirement is primarily due to the non- 7- force spectral density (lb 2 /Hz)
linear characteristics of most systems.

59
 I

* acceleration speatra1 density (G2/Hz) Note that this relationship is valid only
for a rigid connection at the interface. In
Gaf- acceleration force cross spectral other words, validity holds only over the fre-
density, quency range where the acceleration of the two
mounting surfaces is equal.
the apparent weight could be determined.
Thus far, experimental techniques using the 2) Vibration Response of Foundation With and
equations above have been tried at Sandia, but Without Payload
no test results are available. The experimen-
tal results are considered by Hunter and Otts
[3].
WTWF WP UF + WT - U7

PFLUMDZAL MUATIONS INIVLVING A1,TARM? WMCMT

The fundamental equations used to derive Gr or, F oon0
random and sinusoidal test parameters will be hr
considered prior to a discussion of specific I'
laboratory procedures. Note, again, that
apparent weight and vibration amplitude are
complex functions of frequency. Therefore,
the equations below must be applied at dis-
crete frequencies over the froquancy band of
interest. Also note that point apparent weight
will be Implied unless otherwise specified.

1) Interface Apparent Weight When Two Struc-

tures Are Joined F or or I
t
F or

Fig. 6 - Resly~se of Foundation With and Without

wPayload and With Different Payloads
T F WP Refrring to Figure 6,a foundation with
apparent weight VF (at Payload mounting point)
is subjected to a sinusoidal force F.
The acceleration response (a%) at the mounting
point results from the input F.
WP If a $lwyload (Wp) ismounted to the foun-
dation (W ), the interface response ((1) to
the same rorce spectrum F is as shown below.

Fig, 5 -Apparent Weight Interface)

of Joined "F M 4.
Structures (At
where
Referring to Figure 5, a payload andW
foundation with apparent weights and F, G, a acceleration at interface with
respectively, are to be joined. TIe total payload mounted (sinusoidal)
apparent weight (Wr) of the interface is the
vector sun of the apparent weights of the two (' " acceleration at interface without
structures. payload (sinusoidal)

W" a apparent weight of foundation as

WT~ ~ ~ P"_Wseen by payload
where W - apparent weight of paylocd as seen
we by foundation
p a r w interface apparent weglht with
foundation and payload joined
W .n foundation apparent weight (NP 4*- Wp)
= total apparcnt weight.

60
 Similarly, if the force excitation were 1) Driving Force at Inteiface
random (), and the random acceleration re-
sponses were 4 and t with and without the If the payload-foundation interface re-
payload respectively, the following relation- sponse (G1 ) Is known and the foundation and
ships would apply. payload apparent weights are measured to be
4p and W.n respectively, the following infor-
jWP 12/WF\2mition c&n be calculated.

I 41
-I a) Force Driving Payload
Fp a G, W (sinusoidal)

whe --OiW 12 (random)

p I
4 PSD response at interface with
payload Fp u sinusoidal force driving payload at
interface
PSD response without payload
F 7 - random force spectrum driving pay-
W = apparent weight of foundation as P load at interface.
seen by payload
b) Force Driving Foundation
W - apparent weight of payload as seen
p' bfoundation FF - OiW
interface apparent weight (We +- w). -W
F pF JI W1
3) Vibration Response at Interface with
Different Payloads
Continuing with example 2 and Figure 6, FFASF M sine and random force respec-
tively driving fowidation from
interface.
above, the response of a different payload
(wl) would be related to the response of Wi
as shown below. Both sinusoidal (Gi) and c) Total Interface Driving Force

random Oi) interface resporses are considered. F G'W41W' aF F

The same sine and random force excitations
(F and 5) as in example 2 are assued. 2

w "F" sine and random force respec-

GIl G + pl) tively driving foundation and
payload at interface.

(WF + 5) Transfer Function

1I5 "1 I\lwF + W- 2 / Acceleration response ratio and transfer
apparent weight are the two forms of transter
function which will be considered. A transfer
where function is the complex ratio of an input to
the response at a different location and can
= apperent weight of different therefore take many forms.
S payloads
a) Acceleration Response Ratio Transfer
G11 a sinusoidal response of payload P1 Function H(w)
on foundation
Consider the teat item in Figure 7. The
4I, = PSD response of payload P1 on response ratio between input at point 1
foundation and response at point 2 is

0 = same as in example 2 02

same as In example 2 H(w) (sinusoidal)

W.
F same as in example 2 11(w) 12 2
- (random)
same as in example 2

61
 where vibration environments, requires that the
laboratory test be as accurate and realistic
H(w) - transfer function as possible. These requirements dictate that
field data be used in establishing test cri-
2 "
-G sinusoidal acceleration teria whenever applicable field data is avail-
able. Unfortunately, many times field data
L1, 2 0 acceleration spectral density. cannot be measured in the form required. For
example, the input force spectrum at the base
of a field system is difficult to obtain since
G2 or 42 force transducers must be inserted intermediate
to the path of force transmission. This Is
norm lly prohibitive due to tolerance changes
and/or strength changes resulting from the
transducer(s). Also, apparent weight measure-
F1 C2 ments on field foundations are hard to obtain
w1 - 2 H() due to their remoteness, size and availability.
1I) rObviously,
2 one is forced to compromise with
1w2 2 1-H() 2 L the ideal situation. Typical compromises, In
1 2lieu of more realistic techniques, are outlined
below.
It should be noted that the purpose of
this paper is to demonstrate the use of ap-
t
G, or 41 parent weight and not to Justify the techniques
to which it is applied. The arguments for the
F1 or :t various tests are thoroughly covered in cited
references associated with each case discussed
Fig. 7 - Transfer Functions
Case #1 - Derivation of Sinusoidal Force Input
b) Transfer Apparent Weight Spectrum [H
Field vibration responcep of z aystem were
Again consider the test item in Figure 7. measured during a large number of field tests.
The ratio of input force at point 1 to This data (OF) obtained from field accelerom-
the acceleration response at point 2 is eters located at the laboratory input control
interface 6.s analyzed in the form shown in
F1 Figure 8.. This plot represents the maximum
acceleration within selected fr,.quency bands
02 and is obtained from Sandia's ,.MBAN [5] data
system.
From (a) above we also can write this as
follows

F1
W12 1

Py previous definition, FI/G1 is defined

as point apparent weight. Therefore

The basic relationships presented above

will now be applied to the derivation of labor-
atory test specifications and procedures.

VIBRATION TEST SPECIFICATIONS AND TEST PROCE-

DURES AS A FNOCfION OF APPARET WEIG
in the
The fundamental equations outlined
previous section will now be applied to various '' ""
test programs. Evaluation of a system's func- Fig. 8 - Measured Field Vibration
tional and structural integrity, under field (!.Wxim Response) (F

62
Using example 4a from the previous section, the x = maximum apparent weight in each
Wma
maximum force driving the system is band.

Minimum apparent weight corresponds to maximum

F = W acceleration response. Therefore, the apparent
weight is biased to minimum values since it is
where to be used with maximum field acceleration.

F = force input to test system Finaly, application of F = F

in the force spectrum depicted in Figure 10.
C = field acceleration at input interface (Note that the force spectrum could be faired
over the frequency test range to provide a
W= apparent weight of test system. non-stepped control.) This force spectrum was
used as the input control. The test item is
In this case, the field acceleration (Gs) now driven with an approximation of field force.
at the base of the test system is known from In addition, interface acceleration was limited
Figure 8 for each pre-selected frequency band. ro as not to exceed the maximum field accelera-
It is desired to calculate a constant force tion () from Figure 8.
input F, for each of these frequency ranges.
The system apparent weight Wwas measured
in the laboratory and is presented in Figure 9.

SOLID LINE: HAXDAVERAGE

DASHEDLIKE: FAIRED VALVES

DASHEDLINE: BAND AVERAGE P.

APPARENT WEIGHT (Q)

IO0V

I ~ 4 00 00

FREQUIfCY (Hz)

00 IFig. 10 - Computed Field Forces

.~.' J Case #2 - Derivation

Spectrum [6]of Random Force Input

As in Case #1, field vibration measure-

ments were made during field operations with
10- several units. A random PSD analysis of each
field test was plotted. The PSD plot in
FRFQr1,H;(Y
(.) Figure iU was obtained by enveloping the maxi-
Fig. 9 - Measured Apparent Weight mum from the composite spectra.
of Test System Again, using example 4a
(random) the force
spectrum is calculated as follows
As shown in Figure 9, the apparent veight over
each bandwidth was biased to a min.'mut constant
value as follows. 7I S'

W wuin * 0,1 (wm, - W in The random acceleration spectrum 4 was the

enveloped composite spectrum in Figure 11. The

where system apparent weight was measured and band

averaged as described previously.
-= averaged apparent weight in each
band x mn
Wmi n minimum apparent weight in each
band The band averaged apparent weight is
depicted in Figure 12.

63 j
 xZ1
"
.,, : L~ xz1s ,.l*I*. ' ." a "a' a'ag a

,N - E£NVELOPE0
COMPOSITE
SPECTRIM. Vi)
FIELD SPECIMEN
TEST
APPARENT
WEIGHT.
, Iwa,,

~ ixio3 I 190MSINUSOID Lx0

COMP3SIEFIELD
S
SPECTRUM - AII

V ',
La0 l APPARENT
WEIGHT, 4
I1 , LxO BANDAVERAGED
I x I0D1,S I 1 11111 10 x10 -llIl
1110 1xleI 1041 LOx 10 LOx 11? Lax 10 4.0: 10
FREQUENCY FREQUENCY
Ik)

Fig. 3. - Fiel Acceleration Fig. 12 - Test Specimen Apparent

Spectral Density Weight Characteristics

I x105: ' ' ' ' I ' '"''1

ICALCULATED FORCE
04e SPECTRUM
' 9 R-
z SINUSOID

Ii LABORATORY
INPUI
13 FORCE
SPECTRUM

FI

Ix 10 XI XI
(H)
FREQUENCY

Fig. 13 - Calculated Laboratory

Force Spectral Density

64
 The 71 spectrum resulting from use of the Py comparison, whenever the limited force
above equation is shown in Figure 13. spectrum YL is lets than the derived spectru
[, the compensation must be properly equalized.
Case #3 - Derivation of a Limit on Random Force 3v.will be used as the input spectrum over what-
Spectrum from Case #2 [6] ejer frequencies it is le s than 7'.
Limiting the acceleration response spec- Note that one could just as well determine
trum at a location above the control interface transfer apparent weight between input force
was also required. Unlike sinusoidal testing, and acceleration response at the limit point.
there is no way to limit random spectrums with
commercial or in-house electronics. It thus
becomes necessary to limit the random response Iw 1,i2 1
through proper equalization of the input force 2
spectrum. The procedure is illustrated below.

Assume a point above the control interface where

is to be limited so as not to exceed a random
spectrum A. during the force controlled test w = transfer appurent weight
derived earlier. The transfer function H(w)
=
between the input (G_) and limit point (02) can 71 input force spectrum
be measured in the laboratory.
• -response PSI.
12
1(w) 2 (sinusoidal) Then, replacing 42 with AL one can find 7L.

a_L .12
IH(w)12_ (random). IW12JL
1 4
where
The required acceleration spectrum ('01)
at 7 a input limit force spectral density
the input is therefore L
W12 a transfer apparent weight

A w, limit PsD.
11(w) I'
As discussed previously,
where
a- input PSD P. a

L - limit PSD
where
H(w) a transfer function between input
and limit points. W12 = transfer apparent weight

Using the definition of apparent weight, W a point apparent weight

the required force spectrum is
H(w) - transfer function (c l).
2
i a HIWI) Substituting into tne above yields

where 9;L _ I W 12~

=
a; input limit force spectral density
Wa test item apparent weight which shows that both approaches yield the same
results.

I a input P3D Case #4 - Simulation of a Structure's Apparent

Al - limit PSD Weight
I1()- transfer function as described above. Many situations arise where it is desired
to insert the apparent weight of a structure
into the test configuration [2]. It isn't

65

!.I
normally feasible to physically Insert the s
structure, so a technique [2] has been developed 40
whereby the apparent weight is electronically 30
simulated. This technique simulates the blocked 20 0
force and apparent weight of the test foundation 10
according to Norton's theorem. Sk

mlectronic simulation techniques have been 3

developed and applied at Sandia [2,3]. Basi-
2
cally., the technique consists of utilizing an
analog computer that determines what the re-
sponse of the test item would be to a force
controlled input if the simulated apparent
.4
weight were physically present. It then main-
tains the determined acceleration response at .2
the test item shaker interface. Full coverage
of the technique and the electronics required .1
is included in the cited references. f2
10 200 300'W500 1000 2000
FP.E=C (H42)
Fig. 14 - Similarity Between Apparent Weight of
ASB2T WEIGHT AS NONDSTRUCTIVE TEST TOOL Two Systems (Same Design)
Apparent weight has numerous uses in too
defining the mechanical characteristics of so
system. These include system modeling, system C
comparison, and post-test diagnosis.

1) System Modeling

As discussed previously, the mechanical 0 10

characteristics of a system can be interpreted X / TI

from the apparent weight data. Of particular
advantage are the overall system apparent - - ,
weight characteristics which can be determined. - - '
The dynamic analyst uses this apparent weight
data to confirm and/or modify the system model. 1
In addition, this data can influence system
design changes. .5

2) System Comparison ,,

a) Comparison of Different Units ',/

Many times the question arises as to the 2o

similarity of two units. The dynamic FREQU''CY (11)
characteristics can be effectively com- Fig. 15 - Differences Between Apparent Weight
pared through their apparent weights. of Two Systems (Same Design)
Figure 14 shows the similarity between
two units analyzed in the laboratory.
Although there are minor differences, the
systems are within expected tolerances. 20 , - StRYSTI 94 A
77 '
As another example, two distinct patterns /i
were observed as shown in Figure 15.
Approximately 50 percent fell into each
category. It was necessary to know which a
type system was used in field tests since
the apparent weight characteristics were -
electronically simulated in laboratory
tests [7,8]. /

b) Comparison of Same Test Unit- Different a L

Configuration I a /

Figure 16 is an apparent weight plot of _ _ _ _ _ _

the test unit before and after addition 10 20) too 200 3W $0V
of a subsystem. The anti-resonant peak rk1QI17CY (lit)
at about 40 IU is unaltered, but the
higher frequency (A0 Hz) characteristics Fig. 16 - Unit Apparent Weight with Subsystems
have changed. A and B

66
 Kiowledge of this type characteristic may not be possible. Variable force or accel-
change is important when field data and eration input and force-acceleration product
apparent veight plots are used to derive control are solutions, but should be applied
laboratory test specifications as de- with caution since system nonlinearities can
scribed previously, affect the test results. Problems resulting
from nonlinearity are discussed below.
3) Post-test Diagnosis - Detecting Unit Damage
by Apparent Weight Analyses [9] 2) Phase Measurement

Pre-test and post-test apparent weight Since apparent weight is a complex ratio,
analyses were made on a unit which had been care must be taken to preserve the proper phase
subjected to a vibration test program. The between force and acceleration. Tracking fil-
results are shown in Figure 17. ters and log
maintain phaseconverters
coherency.do not
Also,necessarily
tape machines
20 may pose problems for phase coherent recording.
10 PETEST"WA 3) System Nonlinearity
-POST-TST WA A nonlinear system will exhibit apparent
weight characteristics which are different,
10 - -PoS--r"sT --- L WA nni e naor systemin exhbi aepparen the
, differences normally being a function of input
amplitule and/or direction of sweep.
3 '' a) Input Amplitude Effects
ii 'Figures 18 and 19 aeplots
ferent a-Alyses (upsweep vc. ofdownsweep)
two dif-

of the same test system. Bth plots were

from /2 g input control. All conditions
(fixturing, input, etc.) were unaltered
for the two tests. The results show that
10 26 50 lob 200 ;00- the data is quite repeatable. (Figures 18
(11z)
REQ11.NCY and 19 on following page.)
Fig. 17 - Pre-Test and Post-Test Appnrent Weight Figure 20 depicts the results of a 5 g
acceleration Input control (vs. 1/2 g in
Figs. 18 and 19). The consequent differ-
Damage during testing is evident. In this par- ences in the system characteristics are
ticular case the data revealed that structural obvious.
failure of a support flange had occurred. The
support flange was redesigned. ~ .

Pre-test and post-test comparison is be- ......

coming standard procedure on units subjected t
to complex test series (radiant heat, shock,
vibration, etc.). Many times this allows the
project consultant to evaluate test results
without having to disassemble the unit for ..-
inspection.

Apparent weight is a powerful tool in

vibration testing. However, it Is not simple .
to measure this characteristic in the labora-
tory. Some of the problems and associated
precautions were discussed previously in the
I,
'
section entitled "easurement of Apparent
Weight." Further consideration to this problem .
is given below. H

1) Dynamic Pange
Tracking filters, log converters, voltage "
amplifiers and phase meters normally have a r t
maximum of 70 dB dynamic range. On the other
hand, apparent weight plots many times exceed .........
this range limit.
Fig. 18 - Apparent Weight of System Measured
As a result, control of constant force or During Upsweep (1/2 g Constant Input)
acceleration input across the frequency range Compare to Figure 19

67
 -J-

- -

FUQLTKiY(HU) MRQLVIC (1)3Z

Fig. 19 -Aparert Weight of yintgm Measured Fig. 21 -Apparent Weight of Syste

During Donve 12gConstant Upeveep at 5.0 g
Input) to Figure IS
-Compare Compare to Figure 22

-I- -T -- -

111
T.**-- 30

IL 4

Fig. 20 -Apparent Weight of System (5g Constant Fig. 22 -Apparent Weight or system
Input) -Compare to Pp.1an19Downsveep at 5.0 g
Compare to Figure 21

68
 b) Sweep Effects 8. C. 3. Nuckollo and J. V. Otts, "A Progress
Report on Force Controlled Vibration Test-
Figures 21 and 22 compare a unit analyzed Ing," The Shock and Vibration Bulletin 35,
during an upevaep and downsweep, respec- January 1968.
tivey. Note the difference in the anti-
resonant amplitude at 50 Hz (900 vs. 9. R. B. Tatge, "Failure Detection By Mechan-
1400 1b). It should be noted that such ical Impedance Techniques," Acoustic
a dratic change is an exception, not a Society Journal, 41:196-20O, May 1967.
rule. iowever, one must be aware that
this type discrepancy can occur.

CONCON

Apparent weight is a valuable tool for the

test engineer and dynamic analyst. The field
of application covers system aralysis, test
specifications, test technio-aes and trouble
shooting.

The intent of this paper has been to

acquaint test and design personnel with typical
applications, techniques and problems.

In particular, it should be noted that

apparent weight is a relatively new tool in
the vibration laboratory. Known limitations
and inherent inaccuracies should be ful-
understood prior to application.

REFERENCES

1. M. W. Sterk and Z. A. Ellison, "Development

of Low Cost Force Transducer," The Shock
and Vibration Bulletin, Bulletin 36, Part
6, February 1967.
2. N. F. Hunter and J. V. Otts, "Electronic
Simulation of Apparent Weight in Force
Controlled Vibration Tests," 22nd ISA
Conference, Preprint No. P15-1-PMI-67,
September 1967.
3. N. F. Hunter and J. V. Otts, "Random Force
Vibration Testing," The Shock and Vibra-
tioL Bulletin, Bulletin 37, February 1965.

4. W. B. Murfin, "Dual Specifications in

Vibration Testing," The Shock and Vibra-
tion Bulletin, Bulletin 38, Part I, Aug.
1968.
5. J. T. Foley, "An Environmental Research
Study," Institute of Environmental Sciences,
April 1967 Proceedings.

6. A. F. Witte eAnd R. Rodeman, "Dual Speci-

fications in Random Vibration Testing -
An Application of Mechanical Impedance,"
The Shock and Vibration Bulletin, Bulletin
41, October 1970.
7. J. V. Otts, "A Study of the XW-25 Vibration
Response to Structure-Borne Input from the
Genie Rocket," Sandia Report SC-DR-66-186,
May 1966.

69
 DISCUSSION

Voice: The amplitude did not matter under that response to get a clean sine wave, the phase
testing, because you had lower accelerations of the clean sine wave output can be measured
than in the field, In your tests did you have any pretty well. We have had good results with sev-
problem with nonlinearities, in the system that eral different types of phase meters. However,
you later discussed, where the accelerations if you want to argue the question, "What phase
were greater? Did that affect your apparent are you trying to measure with this noise on the
weight calculations? signal?" that would be a different question.

Mr. Hunter: Tiere are a couple of dif- Mr. Bouche (Endevco): I assume from your
ferent systems here. On the system that I discussion that you are measuring point apparent
showed for our unit diagnosis, all of the accel- weight or point acceleration impedance. Is that
erations were low. We did not use this apparent correct in-'it of theee practical applications?
weight in deriving test specifications for any
unit. In the earlier part of the paper, where I
talked about deriving
eivn the
takdabu h test
ts specifications
peiiaiosfr for Mr.and
failure Hunter: Theofpractical
that sort diagnoses
application of unit
were really
a unit, the forces and accelerations in the mea- f
surement of the apparent weight did, indeed, point-apparent-weight measurements using a
approximate the field measurements. If we had small exciter and a single force gauge. For
tried to use those later measurements just for most of the test specification derivation work,
unit diagnosis, I am sure we would have had
problems, we have been using a ring of force gauges con-
because it looked like a pretty non- taining perhaps 10 to 15 force gauges in a sand-
linear unit. wich fixture arrangement and measuring result-
Vie:r
unit. hant acceleration on a ring just above the gauges
Voice: I have not had very good luck with o the
on h unit.
nt
phase meters and I cannot seem to get better
than 10- or 15-degree resolution in the angle. Mr. Bouche: Do all accererometers ex-
Maybe it is because the signals I measure are perience the same acceleration motion?
not pure sinusoids. They are sinusoidal inputs,
but the outputs are not sinusoids. Did you have Mr. Hunter: Not necessarily. That is a
any better luck than that? For me 15 degrees pretty strong limitation that I should have men-
was a real problem. With the kind of resolution tioned in the discussion. We assume that all
I get, it hurts if I have that much scatter. points on the test item have the same motion. If
they do not, of course, an error is introduced.
Mr. Hunter: I think normally we have got- The amount of error depends on how much mo-
ten better results than that, probably in the tion gradient you have across the base of the
neighborhood of 1 5 degrees, but a lot depends item. For most of our applications, we have
on what you define as the resolution. If you are tried to hold our frequency range down, e.g,
testing an item that is severely non-linear so below 500 Hertz, in order to get some sort of
that there are many second and third harmonics reasonable correlation between the acceler-
in the acceleration response, and if you filter ometers.

70

*
 TRANSIENT TEST TECHNIQUES FOR MECHANICAL
IMPEDANCE AND MODAL SURVEY TESTING

John D. Favour
Malcolm C. Mitchell
Norman L. Olson

The Boeing Company

Seattle, Washington

The historical development of transient test techniques util-

izing the modern digital computer to analyze transient data to
define mechanical impedance and modal survey information is
discussed. Complex frequency functions, such as apparent mass
and transmissibility transfer functions, are developed through
ratios of Fourier transforms and digitally lotted both in rec-
tilinear (magnitude and phase vs. fre uency and polar (Nyquist)
forms. Validation of early software ?prior to the advent of
Fast Fourier Transforms), through mechanical impedance measure-
ments on the Bouche aluminum beam demonstrated dynamic range and
frequency accuracy superior to existing analog techniques. The
operating logic of present digital software routines is dis-
cussed. Attention is drawn to the digital plotting of transfer
function data in the polar (Nyquist) form, and in particular,
the advantages of the annotation method developed.
'he developed transient test techniques have been utilized in
the analysis of spacecraft, missiles and airplane flutter model
testing.
The major bending and torsional modes, and natural frequencies,
of a large spacecraft structure were clearly identified via the
Nyquist plotting routine. The structure was excited with a
force step function and the transient response at various loca-
tions analyzed. This data is presented as a relatively
straight-forward use of the technique.
A cantilevered supersonic wing flutter model was tested using
a transient fast sine sweep. The structural transfer function
obtained is compared to a transfer function obtained by an
analog steady state vector analyzer. Because of the close
spacing of modes, typical in airplane wi g structures, the Ny-
quist presentation of the transfer function is used. This
data demonstrates the ability of the transient excitation
techniques ind the Nyquist presentation to obtain the best
estimates of normal mode response without multiple exciters
when the actual response is far from being orthogonal.

BACKGROUND to excite the test article. Narrow

bandwidth tracking filters were gener-
During the early and mid-sixties, ally the "heart" of the analysis equip-
the aerospace industry had a love affair ment. Those practitioners of the "art"
with the concept of "Mechanical Imped- soon found out, amongst other things,
ance". The Boeing Company, like all that the accurate measurement and plot-
other aerospace contractors, developed a ting of the mechanical impedance of a
mechanical impedance measurement capa- High Q, mechanical system required very
bility. The equipment suppliers offered slow sinusoidal frequency sweep rates
a variety of equipment, all of which and could consume from 20 to 40 minutes
relied upon sinusoidal forcing functions time to complete one plot. Furthermore,

71
non-linearities and drift were a prob- APPARENT MASS MEASUREMENT ON
lem with individual components, such as ALUMINUM BEAM
frequency converters, oscillators,
tracking filters, phase detectors, etc., io.ooo
such that the final plot could not be
expected to exhibit a high degree of
accuracy. To compound these problems,
the data plot, to be of any further use __000

analytically, had to be reduced, gener- '.o

ally by some manual means. This added
to the errors and the overall time con-
sumption. Because of this, a parallel
research effort was initiated to devel- 100
op an accurate and stable technique of /2
measuring mechanical impedance, trans- a
missibilities, transfer functions and
other forms of input-output relation-
ships generally associated with linear 1o
systems analysis techniques.
w
Because of some prior experience
on a linear system research [1] problem, _._

the digital computer was chosen to be

the nucleus of the new mechanical im-
pedance technique. The reasons for this
were the following: a) it was stable,
repeatable and accurate, b) the output o.1 _=

data could be in both plotted and listed

form, c) it permitted the utilization of
non-sinusoidal excitation, thereby sav-
ing time, and d) it simplified the on-
site data acquisition requirements. o01
This research led to the development of o 1o0 1.oo
a package of computer software entitled FREQUENCY, Hz
"IRES" (Impulse Response). The "IRES"
program Ts designed to analyze transient Figure 1
or impulsive forcing functions and the
resultant response data. It does so by THEORETICAL [4]
computing the Fourier Integral Trans- RESONANT 74.5 118.6 474.4 640
forms of the transient forcing functioa FREQUENCY, Hz
and response data and subsequently MEASURED
ratioing the two transforms to yield the RESONANT 74.3 118 459 625
system transfer function. This is a FREQUENCY, Hz
straight-forward linear systems approach.
This software program was developed TABLE 1, THEORETICAL & MEASURED RESONANT
prior to the development of the Fast FREQUENCIES OF STANDARD BEAM
Fourier Transform (FFT) [2] and relied
upon a Fourier transform algorithm en-
titled "FXFORM". FXFORM permitted the These results were superior to any-
solution of the Fourier Integral equa- thing produced by our commercial analog
tion at ony desirable frequency, not equipment. The only disadvantage to the
just integer multiples of the funda- digital method was the data turnaround
mental frequency (inverse of record time, 24 to 48 hours after test. The
length, T). test data was recorded directly in digi-
tal form by a portable analog to digital
The FXFORM program provided very converter and tape storage unit. Three
high resolution in the frequency domain channels of data could be recordeo to
and the ability to define very sharp accommodate the necessary mass cancella-
peaks and notches. Figure I illustrates tion computation. Upon completion of
the high resolution and sharpness made testing, the digital magnetic tape was
possible with FXFORM. Figure 1 is an sent to . e Central Computer Processing
.apparent mass" measurement on the Center and processed on the IBM 7094
"Bouche" (3] aluminum beam measuring computer. This technique, developed as
3/4 in. x 3 in. x 36 in. This figure a result of "in-house" research, found
illustrates the quality of resonant fre- scattered usage in many tests but failed
quency correspondence between theoret- to receive general acceptance due to the
ical and measured. Refer to Table 1. long turnaround time required end some
customer reluctance to test with a "new"
method.

72
DYNAMIC DATA ANALYSIS SYSTEM AND IRES The individual transforms of the
forcing function and responses to the
In 1968, a new data acquisition forcing function may be plotted if
and analysis system designed around an desired.
XDS Sigma II digital computer was ac-
quired. This system, the Dynamic Data The frequency domain transfer
Analysis System (DDAS) [5) was developed function is calculated as
for test support and analysis of trans-
ient and random data. The system pro- T(f) = Rf
vides "on line" time and frequency
domain analyses for both local and re-
mote test operations, thereby elimin- where T(f) = Complex transfer function,
ating the long turnaround time problem. R(f) = Fourier transform of
The IRES program was rewritten (using response function,
the same general methodology) to be com-
patible with the DDAS. FM function, of
forcing transform
Fourier
T
The source of data inputs to the
DDAS is analog FM tape containing call- and may be plotted, listed, and or
bratlon data, the forcing function, the written on digital magnetic tape, for
data and an and
response Calibration B time
IRIG data code reformatting and plotting ';n Nyquist
signal. signals form as described below.
are digitized by a dual channel, 10-
bit-plus-sign, synchronous-sampling NYQUIST PLOTTING ROUTINE
Analoa to Digital Converter (ADC).
Digitizing is initiated from the IRIG With the reduced turnaround time of
time code signal at a time and for a the IRES program on DDAS, the utiliza-
duration specified by program control tion of IRES increased quite dramati-
parameters. cally. Those structural dynamicists and
engineers, charged with the responsibil-
The digitized dual channel sinu- ity foi laboratory verification of the
soidal calibration signals, in con- structural and dynamc aralysis of a
junction witn the known transducer complex structure, became increasingly
sensitivity, calibrates the 'raw' data interested in the possibilities offered
into engineering units and provides the by IRES. The only problem was that they
cross-channel time delay or skew infor- were not happy with the rectilinear
mation, which is used to remove un- plotting of transfer function magnitude
wanted skew from the time domain data. versus frequency on one page and the
plotting of phase on another (Fig. 2A).
Any D.C. offset, or tare value, Specifically, the methods developed by
immediately prior to initiation of the Kennedy and Pancu [7) were being used
transient test data is subsequently wth analog equipment and the Nyquist
subtracted from test data be;ore trans- or polar plot of a transfer function
lation to engineering units, was most useful. The Nyquist plot per-
mits the direct plotting of magnitude,
The resultant transient test data phase and frequency (see Figure 2B).
is r'eTormatted for tiime history plots The major problem involved with Nyquist
(if desired) and Fourier transform com- plotting, both with analog equipment as
puta'ion. well as digital equipment, has been the
proper and accurate annotation of the
The D AS Fast Fourier Transform is running parameter, frequency. A Nyquist
a modified Cooley-Tukey algorighm [6] plotting routine was added to the IRES
calc jlated in a Digital Spectrum Analy- software and the solution to the cnno-
zer !'SA). Thp DSA is a sma,. spucial tation problem was developed.
purpo>e computer, controlled from the
Sigma I Central Processing Unit (CPU). Transfer function d3ta is input to
It ib comprised of two 4096 word mem- the tyquist plot routine via digital
ories, one of which typically contains magnetic tape generated by IRES.
time domain data, and the other, a sine
or cosine funct~on. The multiply-add The frequency rdnge for plotting is
cycle time of this device is approx- selectable by program control, allowing
imately 900 nanoseconds. The DDAS FFT any desired isolated modal response to
algorithm requires that the time domain be plotted, thereby eliminating the
data to be transformed be composed of visual confusion of multiple-overlaid
an integral power-of-two data points, modes, which would result from display-
Zero data values are added onto the end ing the transfer function over its
of the digitized time series to yield entire frequency range on a single
the required po er-of-two values for plotted page. Modal responses to be
transformation. pl.tted are typically selected by quick

13
look examination of transfer function
magnitude and phase plots on a storage
and display scope.

The Nyquist polar plane represen-

tation of the system transfer function
is constructed from 3 variables. Fre- TRANSFER FUNCTION OF SECOND
quency is the independent variable and ORDER SINGLE DEGREE OF
the real and imaginary pasts of the
transfer function are dependent vari- FREEDOM SYSTEM PLOTTED IN
ables. RECTILINEAR AND NYQUIST FORMS
At each d4 screte frequency, fi,
the real and imaginary parts of the
transfer function, Ri and 1 , are
plotted, respectively, on the real and
imaginary axes as illustrated in Fig-
uie 2B.
Discrete frequency points on the - - - - - - - ..o
locus of the Nyquist plot are denoted MAGNITUDE
by a small square (a) and consecutively
connected by line segments. The set of %-
points is plotted within a polar circle % .. goo
of 3 inch radius. The engineering unit
equivalent of this radius is annotated PHASE
on the plot. Io

The problem of clearly and accur-

ately indicating running frequency was LOGFREOUENCY n
solved by numerically annotating the L
frequency of every 'kth' p;jlnt. "K" is
a value supplied to the p-ogram by the 2A. COMBINED RECTILINEARTRANSFERFUNCTIONPLOTS
DDAS operator.
Of particular interest to the
structural dynamicist is the notation
cf maximum rate of change of arc length
with respect to frequency along the
locus of the Nyquist plot. This point IA
approximated by locating the maximum
distance between discrete frequencies
within the frequency range plotted.

Clearly, this approximation appro-

aches true rate of change of arc length
in the limit as Af-.-0, hence, suffi-
ciently fine frequency resolution is of
considerable importance. Plus (+) sym-
bols on the locus denote the two fre-
quency points between which the maximum
distance occurs.

areOperator/program communications
are performed at the DDAS teletype con-
sole in a conversation-l mode. Required
control parameters a ew in number and ______-f-o

operator intervention is thereby mini- REAL

mized.
After plotting has been completed, 2B NY UIST OLARPLOT
the operator may elect to a) replot (on
the same, or on a different plotter); Fiure2
b) terminate program execution; c)
change any combination of proqram con-
trol parameters and r -plot from the
stored data or d) reaC' in another
transfer fuinction from the digilal da~a
tape

74
UTILIZATION Al(f)
-_71 a point iner~ance
7*he remainder of this paper is de- f
voted to illustrating two of the many
ways by which this transient test tech- A2 (f)
nique has been utilized. The first ex- a transfer inertance
ample illustrates verification of
structural dynamics of an ultra-light-
weight, spacecraft solar array struc- Both transfer functiovs vere plot-
ture, thru a simple "twang" test and ted in Nyquist polar form, Figures 4
analysis of the structural response. and 5, over e limited bandwidth of 10 to
The second example is a more complex 40 Hz. Both plots indicaie cwo separate
illustration of how transient excita- and distinct resonant modes at 12.8 and
tion and analysis techniques have been 29.0 Hz. Clues to the identification of
utilized to analyze the modal response these two modes are :learly provided
of a cantilevered supersonic wing flut- within the two polar plots. in the
ter model. point measurement, Al(f)/F(f), kFigure 4)
the magnitudes of the two resonaint modes
SOLAR ARRAY STRUCTURE are quite different with the 29 Hz mode
having the larger vector magnitude. In
In the spring of 1970, a test pro- the transfer measurement, A2 (f)/F(f,
gram was in the process of verifying (Figure 5) which is more sensitive to
the dynamic structural analysis of an torsional motion, the two modes are
ultra-lightweight spacecraft solar nearly equal, but in thiL case the 12.8
array structure. Steady state sinusoi- Hz mode has the lcrger vector magnitude.
dal response or resonant searches, res- This, along with the prediction from
onant dwells and resonant delay phenom-
ena, were beir.g used to verify the
analysis. In conjunction with this pro-
the dynamic analysis, clearly estab-
lished the identity of the 12.8 Hz mode
as first torsion and the 29.0 Hz mode
I
cedure, and because of its simplicity, as first bending.
an additional series of "twang" tests
was also conducted. In each case the TWANG TEST SETUP FOR SOLAR ARRAY
test configuration, instrumentation STRUCTURE
(partial), location and method of apply-
ing the step function excitation, or
twang, are shown in Figure 3. On one
side of the structural frame, a woven
fiberglas skin is attached. The solar
cells are mounted upon this skin. For
this test, a lightweight string was
attached to the back side of the fiber-
glas skin, in the center of one of the
structural subdivisions, and run over a
pulley. A 1.5 pound weight was attached
thereby preloading the structure. About
a dozen response accelerometers were
located on the structure. Two will be
discussed; accelerometer Al, located at
the point of string attachment, and
another accelerometer A 2 , located on
the frame of the structure, outboard of
A1 . The preload caused both bending
and torsion in the structure.

The test was conducted by simply

cutting the string and thereby applying
a 1.5 pound step function (release of
preload) to the structure. The response
aceleration signals at A1 and A 2 were
recorded on FM magnetic tape. For pur-
pose of analysis, the force step func-
tion was synthesized on FM magnetic tape
at a later time. The data was then
analyzed via the IRES program, on the
DDAS, and two transfer functions were
computed:

Figure 3

75
75
POINT INERTANCE MEASUREMENT, TRANSFER INERTANCE MEASUREMENT,
A1i (f)IF(f) A2 (f)/F(f)

31.7

15.0

. 1 1 90070O 8.

I ' " x °SO_ _ -o "R-.o9o

° I
270 27&)

FREO RANGE 0.9766E + 01 TO 0.4028E + 02 CPS

FREO RANGE 0.1iG +02 0.4,028E 402 CPS F- 0.3052E + 00 CPS
AF=.03052E + 00 CPS 0.122E + 02 > A F MAX

0.238E + 02 > 0.1312E + 02

0.2859E 02> As/A F MAX Figure 5
Figure 4

By strategic location of the res- at a time until the locus of points des-
ponse transducers and the point of in- cribing a mode is plotted. This system
put forcing function, all of the dynamic produces excellent complex plane plots.
modes of interest can be clearly iden- Its main disadvantage, however, lies in
tified through the analysis of a few long test times since it requires steady
simple "twang" tests. state response of the test specimen.

SUPERSONIC WING FLUTTER MODEL The excitation system consists of

a voice coil weighing 20 grams placed in
For preliminary 'valuation of the a constant magnetic field with no mech-
techniques of the Fourier transform and anical coupling between the voice coil
transient excitdtions applied to the and the field-producing structure. The
testing of complex airplane structJres, voice coil is attached to and supported
a cantilevered wing flutter rmodel of a by he structure under test. Provided
supersonic airplane was chosen is a rep- the field intensity is constant over the
resentative structure. The morl was range of displacement of the voice coil,
constructed primarily from fiberglas and the derived force from this coil is pro-
balsa wood, with aluminum being used for portional to its current. Using a con-
the nacellc strut springs. A picture stant current source to drive the voice
of d structure similar to the one tested coil, the coil back EMF is working into
is shown in Figure 6. a relatively high impedance and theye-
fore.adds insignificant damping forces
As a basis for judging the results o the model. The force 'ignal consists
of the transient testing, the model was of a voltage proportional to the voice
first tested using a sinusoidal steady coil current and phase-coherens with it.
state vector ana~yzer (Aeroeldstic
Modal Analysis System - AMAS)[81 (9). The response measuring system con-
By setting a frequency on the frequency sists of an Endevco 2264 piezo-re~istive
synthesizer the resulting complex plane accelerometer vith associated power and
vector is plotted after the rystem under balance units folloved by a broadband
test has reached a steady state response. instrumentation amplif,cr.
This process is continued one frequency

76
 CANTILEVERED WING FLUTTER MODEL

F _°_~ L ~ ~
...
{I
~ 1M- !.. _7
I

Figure 6

Figures 7 and 8 present the resul- This relationship is valid for low
ting AMAS plo ts for the direct inertance damping typical of airplane struct;,re.
measurement for the first 3 modes. The For optimum results the angle 8 should
method used to extract modal data follows be kept between 10 and 30 degrees. Fig-
that outlined by Kennedy & Pancu [Ref. ure 7 details the circle fitting and
6] in that resonance is determined by calculations for the first mode and
the maximum rate of change of arc with Table 2 tabulates the results from the
frequency, dS/df. With the resonant AMAS steady state response tests. This
frequency determined, a circle is fitted data shows significant non-orthogonal
to the data around resonance from which response in modes 2 and 3.
the angle e is determined between two
vectors. This angle is used to measure Af = 0.01
the damping value from the relationship f
MODE f 2C
Af
fntan 9 1 7.00 190 0.017
2 10.12 320 0.0072
where damping ratio 3 10.28 16.50 0.014
f = frequency
fn a natural frequency TABLE 2: STEADY STATE TEST RESULTS
0 = phase angle

77
 AMAS ANALYSIS, FIRST MODE AMAS ANALYSIS, SECOND
AND THIRD MODES

7.200 6.8 1.6% 0

06.85
10.500
06.9 0 10.18 0 0.0
7.100
10.180 .0
0 1°'4° 0 0 0 010.2 G
7.0 0
706.94 06.40010.150 010.22010.06
7.060 10.38 0 10.230 010.07
(6.96
10.2400O10.08
10.130010.09
6 190 e10.25
4
0 Af -. 01 Ho.Z 06.97 ( 10.12) 010.10
0.1 10.0
7.020 0 0 10.320 10.11 010.26

0 010.27
7.00
10.2
fn 7.00 Hz
2 =2_Af_
S2-
e2/ o0.017
~f, TAN 8.0 G/LB

8GIFigure
8.0 G/L8 8

Figure 7

For the transient test a fast sine 105 seconds (real time). Taking the fast
sweep was chosen as the forcing function, sine sweep to the limit for its duration
since it can input about two orders of results in about a 90 second sweep time
magnitude more impulse, or momentum (lb- leaving some cushion for the response to
sec), into a structure as compared to die out at the completion of the sweep,
using a bandwidth limited delta function thus avoiding truncation errors.
(assuming the same resulting bandwidth
and peak force). This is obtained be- The results of the transient fast
cause the duration of the sine sweep sine sweep are shown in Figures 9 and 10
approaches the data sample length, T, covering the first three modes. Table 3
whereas the impulsive excitation usually lists the frequencies and damping ob-
lasts less than 10% of T. For the fast tained from the first three modes using
sine sweep tesL, the initial or starting the transient excitation.
frequency of the sweep was chosen just
below the first mode frequency and the Af 0.0095
ending frequency was picked to be some-f
what above the highest mode of interest MODE _ n 8

1
(mode 7). The resulting swept bandwidth
was 5 to 50 Hz. A log sweep rate was 1 7.01 230 0.0135
chosen. 2 10.12 340 0.0063
°
Since the required Af is in the 10.28 230 0.0094
neighborhood of 0.01 Hertz, the DDAS TABLE 3: TRANSIENT FAST SINE SWEEP"
equivalent to this is 0.00953 Hz and
therefore, the required data length T is TEST RESULTS

78

3a
 TRANSIENT ANALYSIS, SECOND AND
THI RD MODES
TRANSIENT ANALYSIS, FIRST MODE

leo 18o ° 0

NYQUISTPOLAR PLOT
FREO RANGE 0.6500E + 01 TO 0.7250E + 01 CPS FREG RANGE 0.9750E .01 TO 0.1075E + 02 C'
fF-O .9E5E-02 CPS f0.128E
F * 0.9535E - 02 CPS 002
0.7010E+ 01 > 11/hF MAX 0.1029E +02 > SMF MAX
0.7020E+001

Figure 9 Figure 10

The results of both methods compare -l X

favorably even though the analog system '' =
takes more test time. where K =spring constant

The following is a discussion of M = mass

additional data which can be obtained X= displacement
from the Nyquist plots where it can be
shown that the diameter of the fitted
(or constructed) circle is equal tof=fr
1/2 K, if the plot is the ratio of dis-f
placement divided by force (see Ref. 6),
assuming orthogonal response. Now, for
a single degree of freedom system,
(2f K2 X 1- 1
27& 27j X

fn
n
F
=

=foc
=
acceleration

oc
uec
frqec
natural frequency

damping coefficient
(2 =fn == =With this relationship, the modal
2~M(nf~)point mass can be measured from the plot.
measuring acceleration per unit force This method is subject first of all, to
inta fdisplacement errors in the damping measurement,
Intado errors resulting from a high degree of
X.-X = 1 coupling and also to errors in the ab-I
T \ 2~t~nn\Z
/rZ~n\ isolute sensitivities of both the accel-
(2nf
2M(2Ifn)erometer
F and force transducers.
and at resonance, f fn, The alternate method of added mass
is applicable to the fast sine sweeps

791
and the Fourier transform. Using the The method of estimating the ortho-
fyquist form of data presentation, the gonal mode shape proposed here, consists
resonant frequency measurement is sub- of using a single shaker, the transient
ject to the least percentage error with fast sine sweep and the Fourier trans-
respect to the other measured param- form. That is, if a series of accel-
eters; therefore, this method of added erometer responses to a transient fast
mass seems appropriate under the cir- sine sweep excitation were recorded on
cumstances. It follows: magnetic tape, an estimate of the
Morthogonal mode shape could be obtained.
n
a unknown modal mass First, a Nyquist plot of inertance would
be calculated for each accelerometer
Ma =small added mass (known) location, using the transient fast sine
fn a resonant frequency without sweep excitation as a common denominator.
added mass M For each mode on each. plot, a circle
f = resonant frequency with would be constructed to fit the data
a added mats M around resonance. A plot of the con-
structed circle diameters versus each
2
(2 in) K and a2 K location would result in the estimated
n n and = + orthogonal mode shape. The resonant
frequency and damping for each mode
substituting for K, would be calculated directly from The
one of
the Nyquist plots of inertance.
(2f )2M measurement of modal parameters (mass)
n Mn = (2ffa) (Mn+Ma ) would also be performed based on the
most orthogonal-looking response. The
2
Mn [(2fn ) 2 _ (2nfa) 21 = Ma(2wfn) generalized parameters can then be cal-
culated by normalizing the point of
measurement to the point of maximum res-
(2nrf a22 ponse using the estimated measured mode
Mn =Ma (2 fn) 2 - a= shape. Better yet, if an analytical
n2a1?fZ- analysis exists, the analytical gener-
alized parameters can be normalized to
This method is valid only for small the accelerometer location where the
added masses since it assumes no change modal parameters were measured. A com-
in the mode shape. For the least amount parison between measured and cnalytical
of error, this calculation should be quantities could then be performed.
performed at a shaker location producing An interesting footnote to this
the best orthogonal point inertance
response. If large couplings exist, the investigation ib the non-linear char-
calculation is probably
pro not
Waai rsecttoth n worth
bley making.
wof mlevel
acteristic of mode 5, where a response
above a reference threshold was
With respect to the problem of obtained. The relatively high force
measuring the orthogonal mode shapes of level produced a rattle in a nacelle
an airplane, the method of multiple bearing. This rattle produced the un-
shakers is the best, considering the usual AMAS steady state response of
accuracy of the measured estimate, flow- Figure 11. The fast sine sweep using a
ever, the method requires a considerable force level where the bearing rattled
expenditure of effort during and in prep- produced the plot of Figure 12. A
arations for the test lower force level produced the more
usual plot shown in Figure 13.
It has been ig
shown
and
o later by CraCa
andlaterby [11] bythat
Pendered [10T
the method These experiments on the flutter
(11]othatuhe meutod model wing are part of a continuing
of Kennedy-Pan~u can produce inaccurate development program directed toward
mode shape data when the degree of developing practical methods and tech-
coupling is very high or when there is niques of conducting airplane ground
damping coupling. However, the Ken- reso f anductig f l ae tests
nedy-Pancu method does provide data resonance and flight flutter tests
which, if used in conjunction with the using the Fourier transform and trans-
determinate method of Asher (12], can lent excitations.
set up an effective multiple shaker
test. The minimum test time consumed CONCLUSIONS
during transient single shaker tests
would allow those modes to be identi- The techniques of mechanical im-
fled which require multiple shaker tests. pedance and structural dynamic testing
The data has also been obtained in a thru transient excitation and digital
short time period, after which the mul- computer analysis have been transformed
tiple shaker test can be designed and from the realm of academic curiosity to
set up. existance as a realisitc, accurate and

80
 AMAS ANALYSIS, HIGH LEVEL TRANSIENT ANALYSIS, LOW
NONLINEAR RESPONSE LEVEL NONLINEAR RESPONSE

28.0
00
0
30.6 30.8 002.

3D.40 Z0.0 e29.2

L O 29.4 .73

-\ 30.2 29.6 R 6.0X109

29. 2700 AF -0.0035

66.0 G/LD Figure 13

Figure 11 economical engineering tool. The work
reported in this paper represents a
small segment of a continuing effort to
TRANSIENT ANALYSIS, HIGH LEVEL provide more economical and meaningful
NONLINEAR RESPONSE information from the test laboratory.
o REFERENCES
9
(1 J. D. Favour, "Transient Data Dis-
tortion compensation", 35th Shock
and Vibration Symposium, February
1966.
[2] J. W. Cooley, P.A.W. Lewis and
P. A. Welch, "The Fast Fourier
Transform Algorithm and Its Appli-
cations", IBM Research Report
RC 1746, 1967.
.. 0o 3 R. R. Bouche, "Instruments and
Methods for Measuring Mechanical
Impedance", 30th Symposium on
Shock, Vibration & Associated
.5 Environments, October 1961.
[4] C. M. Harris and C. F. Crede,
"Shock & Vibration Handbook",
Volume
1.
[5] M. 0. Michellich,
"Dynamic Data
Analysis System", 40th Shock and
2700 Vibration Bulletin, December 1969.
[6] J. W. Cooley and J. W. Tukey, "An
FREORANGE 0.2925E+02TO0.3055E+02 CPS Algorithm for the Machine Calcul-
AF-0.O535E-f02CPS atlon of Complex Fourier Series",
0 304 5E+O2>A.S/FMAX Math. of Comput., Vol. 19, pp 297-
03046E +02 301, April 1965.
Figure 12

81
[7] C. C. Kennedy and C. D. P. Pancu, Close Natural Frequencies in Res-
"Use of Vectors in Vibration Meas- onance Testing", J. Mech. Engr.
urement and Analysis", Journal of Sci., V7, n 372-379, 1965.
the Aeronautical Sciences, Iovem- [11] R. R. Craig, "A Study of Method
ber 1947. for Determining Pure Modes and
[8] Instruction Manual SD1004-18B, Frequencies of Complex Struc-
Aeroelastic Modal Analysis System, tures", Boeing Document No. T6-
Vi and V2, Spectral Dynamics Corp- 5558, Informa1.
oration of San Diego, 1968.
oraionnofSn Dig euenc
1 . (12) G. W. Asher, "A Method of Normal
[9) Monsanto Digital Frequency Syn- Mode Excitation Utilizing Admit-
thesizer, Model 3100A, Monsanto tance Measurements", Proceedings -
Electronic Instruments, West National Specialists Meeting on
Caldwell, New Jersey, 1969. Dynamics & Aeroelasticity, Fort
[10) .1.W. Pendered, "Theoretical In- Worth Institute of Aeronautical
vestigation Into the Effects of Science, 69-76, November 1958.

DISCUSSION

Mr. Schrantz (Comsat Laboratories): What Mr. Schrantz: What was the sweep rate?
Is your definition of a fast sweep-rate in terms
of amplitude? Mr. Favour: On that particular example it
went from 5 cycles to 50 cycles in 90 seconds.
Mr. Favour: It was a constant amplitude. Recognizing that, in order to get an analysis of
0.01 cycles delta F, we had to have 100 seconds
Mr. Schrantz: What level? of data. If you want to call that a transient, it
really is. The sweep went from 5 to 50 cycles
Mr. Favour: I do not know the level on that in 90 seconds and left about 10 seconds of resid-
wing. It depends upon the structure of course, ual transient to die out, so we had all the data
but I do not know what that level was. without truncation.

82
 PREDICTION OF FORCE SPECTRA BY MECHANICAL IMPEDANCE
AND ACOUSTIC MOBILITY MEASUREMENT TECHNIQUES

R. W. Schock
NASA/Marshall Space Flight Center
Huntsville, Alabama

and

G. C. Kao
Wyle Laboratories
Huntsville, Alaba m

Structural impedance, or its reciprocal mobility, has seen limited use in recent years to
tailor dynamic tests of highly critical and expensive components to ensure a highly accurate
control of response loads. These efforts have been generally confined to sinusoidal test
simulations of rocket vehicle longitudinal and lateral modes in frequency ranges at or near
the component response frequencies.
A method, described in this paper, has been developed to calculate broad frequency range
vibration criteria which account for both primary and component load impedance for
structures subjected to random acoustic excitation. These criteria rather than being defined
in traditional motion parameters am defined in force parameters and are, therefore, termed
force spectra. The force spectra were predicted by a one-dimensional equation which
utilizes four types of data measured at equipment mounting locations. These data consist of:
Input impedance of support structure; Acoustic mobility of support structure; Input imped-
ance of component package; and Blocked pressure spectrum.
An experimental program was conducted to validate the prediction equation. A stiffened
aluminum cylinder with the dimensions of 3 ft (diameter) x 3 ft (height) x .02 in. (skin
thickness) was used in acquiring input impedances and acoustic mobilities. An 8 in. x 8 in.
x 1/2 in. aluminum plate was used as a simulated component. The plate was supported by
four sets of leaf springs with four loadwashers attached to the bottom of each spring for
measuring loads. The blocked sound pressure spectra were obtained from microphone
measurements on a rigid dummy concrete cylinder. Two equipment mounting pozitions were
used in the tests.
All test data were acquired on-line to analog/digital acquisition systems. Computer
programs were written to reduce and analyze the acquired data, and also to predict the inter-
action force spectra. Good agreements between the predicted and measured force spectra
were obtained.

INTRODUCTION would probably be overtested or under~ested as com-

pared to actual inflight environments. A more
Component packages of rocket vehiclr are tradition- realistic approach would be first to determine the
ally qualified for acoustic environments by motion- interaction forces butiween a component and its
control (or response-control) testing. Vibration envi- support structure, and then to test the component undei
ronments used in this type or testing are obtained by the force-control environments. This approach
enveloping peak amplitudes of measured or predicted requires a prediction technique to define the dynamic
response data, but ignoring effects of component - forcing criteria as indicated above. The primary
primary structure coupling. Consequently, compo- objective of this paper is to preseitt a method for pre-
nents qualified under the motion-control criteria dicting interacrion forces between components and

83
corresponding support structures subjected to acoustic The driving force FL(u) can be shown [ III to have the
excitations. Such force environments which are following expression:
determined in spectral forms are referred to as "Force
Spectra" in this paper. Research in seeking practical ZZ L
techniques for defini a force-control criteria has F (0)= V M ys )
gained considerable attention during recent years. L 0
Available results which are relevant to force-control
studies are presented In References I through 10. Based on Equation (1), the component-structure inter-

FORCE-SPECTRA EQUATION action force spectra of vehicle structures subjected to

acoustic excitations could be expressed by, 11 :
It is assumed that dynamic responses of a structural
system subjected to excitation by external forces are
2
predominantly one-dimensional. Thus the dynamic 0L(0) = (t)• 2
characteristics of the structure can be represented by a L() (u) • T2)
one-dimensional impedance model as shown in Figure 1.
In this figure, the basic unloaded structure is replaced
by an equivalent structural "black box"; external loads Where
are applied at terminals 1, 2, and component packages
which are treated as load impedances, ZL(w), are L O(N)
p Power spectral density (PSD) of
reference sound
pressure which is
attached to terminals 3, 4. The corresponding veloci- reference onr erwh suI
ties and interaction forces at the attachment points are assumed to be constant over the sur-
Indicated by VL(L)and FL(u), L respectively.
I The face over the component attachment
locations.
structural impedance model, as shown in Figure 1, can
be represented by the equivalent constant-force model
(Thevenin's model) and the constant-velocity model (
2
(Norton's model) as shown by Figures 2 and 3, respec- = Acoustic mobility at component
tively. The dynamic characteristics cf the attachment mounting locations, and is defined by
points (terminals 3, 4) are represented by Zs(w) which the ratio of the rms velocity
response of a support structure and
is defined as the support-structure impedance, or source its corresponding rms sound
impedance, i.e., the impedance looking back to the pressure.
left of terminals 3, 4 without any loads attached.
Due to structural complexities of rocket vehicles,
l.ood precise anaiytical approaches to obtain the parameters
Sructurol
EteaForBock
3
Attachment
Point
Component
Pck.
defined in Equation (2) are not practical. Therefore,
in order to validate Equation (2), experimental
3 L(u) techniques were used to acquire measurements

Fo(4)
2u StUnloa e d4 -noae FL 9 E q im e n t of these parameters on the selected test specimens,
S-kCtur (ZL and to compute the force spectra quantitatively.

the force spectra computed in the

Figure One-Dimensional
u. impedance Model of a The manner of
aboveaccuracy could then be checked by comparison
Structural System with force responses neasured from a simulated com-
ponent mounted on a support structure which was sub-
_ v_1 ) jected to random acoustic excitations. The approaches
ZS LC F_,1_7u)
( used to implement the measurement techniques outlined
F,(4) J above are depicted in Figure 4.

Figure 2. Equivalent-Constant Force Model TEST SPECIMENS

The support structure used in the experiment was a
VQ stiffened aluminum cylinder, as shown in Figure 5.
30 -The cylinder's dimensions were 36 in. (diameter) x
36 in. (length) x 0.02 in. (thick). The cylinder con-
v{L F,(U) sisted of five aluminum rings and twenty-four longitu-
dinal stringers equally spaced along the longitudinal
and circumferential directions, respectively. All
Figure 3. Equivalent-Constant Velocity Model stiffeners were mounted to the cylinder wall by rivets.

84
 TYPWIN E APOTYPISO AND MLCTION
DATAANALYSIS
SEJW~WN TIORESPONSE SIGNALS

IECNANICAI. TEST CYUINDR ANALYSIS

SINEMWEP
----------------- ---------------

MECHANEA S"w...P SIMUTE

IMPEANCECOMPONENT PACKAGE

L------~ - --------- ~ ----------

ACOUSTICALcelrIGinDRPDAAYI
RANDOM

___ L --------------- --------------- ---------------------------

AI -cSecr o m. IIa SW . I SAcoustic

Vcity

I es I

* II

[1111111](9
TEST

CYLINDER *Ta
ca i Spectvo Accelmifo Acce.Iueflen

PSD
LP.
Z.
SIMULATED
COMPONENT
*PACKAGE ON TEST
CYLINDEt

COMPONENT Aoi pcr

k[SPONSEaspt3-Tf

4 ---------- I I ------

00
O-DNAMIC COMPONENT
SMULATED
NRAKE PCKAGE

CONTROL RN0 S
TESTING

Sh&
le~t Foce Spectreto

Figure 4. Riock Diogram of Force-Spectra Program

85
 The dimensions of the individud~ curved eanels formed
by the stiffeners were 6 in. x 4.75 in. Two steel rings
of I in. x 1 in. x 1/8 in. angle cection were riveted at
both ends of the cylinder, and two circular sandwich
plate bulkheads were bolted to these end rings by 1/4
in. diameter hex bolts and nuts. Each circular bulk-
head was constructed from two steel Plates of 1/8 in.
thickness, septirated by a 1/2 in. thick plywood
I1 section. Two component mounting oo-Itions will be
discussed in this paper, and they are designated as

a Positions RI, R2, R3, R4: Centers of ring

0 frame segments
U .4 Positions SI, S2, S3, S4: Centers of lorgitu-
WIG dinal stiffener
> segments

The dummy rigid c/linder consisted of reinforced con-

~ , ~ *crete having dimensions of 48 in. (length) x 36 in.U
(diameter) x 4 in. (thick). A photograph showing
the rigid cylinder is presented in Figure 12.

The simulated component package, as shown in Figure

6, consisted of a 1/2 in. aluminum plate with lateral
dimensions of 8 in. x 8 in. The plate was supported by
Figure 5. Stiffened Cylinder and Input Impedance four sets of leaf springs at its corners. The bottom of
Measurement Locations -each spring was fitted with a loadwasher assembly.

NOTE- All path, except [tern No. 2, we* mods of Aluminum

AllY ype 66-6

so - G-
(D4- 0.008 SteelLeaf Springs
0.031 x 2.00.x 4.193 Strip
- -- (~) 0.063 x 0.50~ . 2.00 Strip
0.063 x 0.550.x 2.00 Strip
%2.00 Strip
rii$Fori'
* -----0.063,x0.550

1
Q No. 10432 1000 Flat Head Screwx W/6 Lo.
t 00~ (1 No. 10-32?HexNut

o ~®
I~
~® ' ~09
D1O- No. 10-3? 1000Flat Head Screwx 3V 14.
toodwasherAssembly; Details Shrown
inFigure

Figure 6. Dimensions of the Simulated

so®000Compnent Package (All
- - IDimensional Units are in Inches)

86
Each assembly, as shown in Figure 7, consisted of a DATA ACQUISITION SYSTEM
Kistler 901A loadwasher which was sandwiched between
an anti-friction washer on the top and an aluminum The data acquisition systems employed in the tests are
mounting stud at the bottom. These elements were shown in Figure 8. Basically, the systems consisted of
held together with the top clamping strips by a center four major subsystems described as follows:
bolt as shown in Figure 7. Each loadwasher was pre-
compressed to approximately 1000 pounds level, so * Analog data system
that the tensile and compressive forces induced during
testing could be measured. The total weight of the 0 Central control system
component package was 3.81 pounds; the resonances S A/D convrsion system
of the package were measured at 110 Hz and 1200 Hz,
respectively. The fundamental resonant frequency of * Recording and displaying s..t
the 1/2 in. plate was found to be 1200 Hz. The key functions of each subsystem are described as
AI4... topCt Stip
,ping follows:

\p';Q \sAnalog Data System

This system was comprised of transducers, such as accel-
erometers, microphones, strain gages, signal ccndition-
S,
.,I-FrIW,,.w. ringamplifiers and patch panels. Electrical signals
Ci. sp
t,;p _, S .. produced by these transducers were conditioned by
-0,,appropriate
SW _- amplifiers to achieve desirable signal levels
.V16- prior to the link-up with the A/D systems. A maximltm
Al,;- Moly of 128 data channels are available through the current
system.
"P - Central Control System
Fare. Axi The central control system was the brain of the entire
data acquisition system; it consisted of the following
Figure 7. Details of Loadwasher Assembly units:

A-"
. N
c k S, D.,-1 C-tZ
8.. ... .. -----------

A,.

* * 1-1. W T .' _ThL

L1

1 4)4 325.W4C

Figure 8. Block Diagram of the Analog/Digital Data Acquisition System

87
 - XDS - SIGMA 5 Central Processor * One 64-channel multiplexer
* Cere Memory Unit * One sample-and-hold amplifier
* A/D Controller * One analog/digital converter (15 bit,
* Multiplex Input/Output Processor 100 K sps)

* Frequency Controller The multiplexer's function was to transmit incoming

analog signals sequentially to the somole-and-hold
The peripheral equipment used in the tests is shown in amplifier at a predetermined rate. The amplifier sam-
Figure 8. The main function of the central processor pled the signal and applied a gain of 1, 2, 4 or 8 (as
was to coordinate work performed by various units selected by the digital program) to convert the signal
according to instructions as given through input devices to a desired range. The digitized signals were stored
(card reader or magnetic tapes). The input instructions temporarily in the core unit, and then transferred to
established the requirements for: other recording or display devices upon receiving
commands from the central processor.
a Type of analog signals to be acquired
Recording and Displaying Systems

I
(sine or random)

I
* Sensitivity factors The recording system consisted of two magnetic tapes
• Number of data channels and one random access disc for storing digital data
output
Sampling rate received from the core memory unit. Other
formats could also be obtained by plotting or punched
* Frequency range cards, as desired.
* Full-scale ranges MEASUREMENT APPROACHES
* Data recording or displaying formats
Impedance data for the support structure were acquired
Each unit would then act accordingly to specific in- by the impedance measurement systems as shown sche-
structions during the length of a data acquisition matically in Figure 9. Three types of signal were
process. acquired for each measurement point. These signals
consisted of:
A/D Conversion System * Force signal

There were two A/D conversion units which formed * Acceleration signal
part of the data acquisition system. Each unit consisted
of the following equipment: 9 Constant Amplitude Reference Frequency
Signal (COLA).
SWlIMEN

......

WI. A/L
0.. Ac q.,'0 I
Splct,*I Dr0,.... LoglVolI.g.

I.. .,OG sp..

Mdl 1012 Mdll11.1-2
#M.llot?.
. SD112- IW

0-V
Apd ~ A, .
__ _cv2 _Add. _ -

My-~ ~ ~ ,101 d _ 001-

k lw.w AlIdl ,d --

0411lot
Co. $.

Figure 9. Instrumentation Block Diagram of thc Mechanical Impedance Measurement System

A.p,.
88
 The force and the acceleration signals were obtained by aluminum cylinder. The two cylinders were placed
employing an impedance head which was bonded to near the center of the reverberation room and were
the aluminum cylinder by Eastman 910 or commercial spaced 10 feet apart from each other. A control
dental cement. All measurements were made In the microphone, used to acquire the reference sound pres-
frequency range from 40 Hz to 2000 Hz with sinesweep sures, was located near the center of the room and 10
forcing techniques. The sweep rate was set at one feet above the floor level. Figure 12 shows the rela-
octave per minute for all test runs. During each test, tive positions of the microphones and tha iest cylinder
controls were exercised to maintain minimum force and in the reverberation room. Blocked sound pressure
acceleration levels by the amplitude servo monitor in
order to achieve high signal levels throughout the
frequency range of interest. Wilcoxon impedance
heads Z-602 and Z-13 were used to measure imped-
ances at Locations Rand S, respectively. "

The measurements of the component package imped-

ances utilized the instrumentation set-up as shown in
Figure10. The component package was mounted on aW
shaker (Ling Model No. 286), as shown in Figure 11,
which was used toResponse
the component. provide signals
input forces
were for drivingfromL
obtained - - -'
two accelerometers, Al and A2, mounted on the
center of the 1/2 in. plate and the base plate, respec-
tively. The force signals were generated by the out-

iiii,
puts of individual Ioodwshers. The input impedance
signals consisted of COLA, force and A2 responses;
whereas, the transfer impedance signals consisted of J
COLA, force and Al responses.

The measurements of vibro-acoustic data were conduc-

ted in Wyle Laboratories' 100,000 cu ft reverberation
room. The acoustic mobility measurements consisted
of measuring sound pressure near the surface of a con-
crete cylinder, and simultaneously acquiring the res- Figure 10. Measurement of Component Package
ponse data at the component mounting positions of the Impedance

A.fmnic. Sig~...
Oicilloh.
Analog Loo,.jh. 1,2,3,4
Il0911WW Summd Feedback
M"curW4nt
A ~System.
Acceleomet.Sigol
Fedock 104feAmp Summi.gJ 6cio
Choge A-p 50 _,/lb*So't1-1 12 5 mv/lh-GC..0 25 I
50 -v/lb te A-p

Plot" ..... Sl
fre..nce

TOCOMPUtR

Figure 11. Instrumentation Block Diagram for Impedance Measurement of the Component Package

89
 sigrxls were acquired by four microphones which were
placed about 1/4 in. above the surface of the concrete
cylinder. The ,'nter dist :,s between microphones
were set identical to that of the component mounting
positions (6 in. x 4.75 in.).

-1

..

ysC . Figure 13. Measurement of Component Package

.. jResponses

Figure 12. Relative Positions of Test Cylinders and

Microphones in the Reverberation Room ...........
.M A

To obtain the responre data of the aluminum cylinder

'
4 •
for computing acoustic mobilities, the average accel-
eration responses at the Rand S locations were requir- _ ______

ed. The average responses at these locations were _

obtained by placing four Endevco Type 2226 accelero- !i: A

meters at positions indicated below: ____ _______

R1, R2, R3, R4 ___ A

SI, S2, S3, S4 ]___ C A

Following the acoustic mobility measurements, the

component package was mounted on tre aluminum ,m A.--

cylinder at locations R and S,respectively, for ___--_

acoustic testing. A typical test position of the com-

ponent is shown in Figure 13. The blocked pressures ..
were acquired by four microphones located near the
-
surface of the concrete cylinder. The interaction
forces between the component and the aluminum cylin-
a...-
C AC A-&',....
I--,i*-" .I
der were sensed by the Kistler loadwasiers; and the
4.0 A-044*

responses of the plate were acquired by an Endevco

Type 2226 accelerometer. Instrumentation employed in Figure 14. Instrumentation Block Diagran, for Vibro-
the vibro-acoustic tests is shown in Figure 14. Acoustic Expeiiments

90
REDUCTION OF MEASURED DATA C

The acquired data may be classified into two general 8

categories: the sinesweep data; and the random data
obtained from the vibro-acoustic tests. These data
were subsequently analyzed to obtain: the component
package impedance, support structure impedances,
acoustic mobilities, blocked sound pressure spectra and 0.
the measured force spectra. The reduced data were
stored on magnetic tapes and were used as "inputs" to
compute the predicted force spectra.

The component impedance was obtained by dividing the Fq.......' cy..... ". ...
total force acting on the component by the velocity Frequency, Hz
response of the base plate (A2). The total force was Figure 16. Summed Impedance at Location R
obtained by summing the loodwasher responses. The
component impedance plot is shown in Figure 15. The
resonant frequencies of the component as seen from the
shaker are located at 110 Hz and 1200 Hz. The latter 10'r
frequency is the fundamental frequency of the /2 in. , ".
plate. .0

00

0o. Frequency, Hz

z. Figure 17. Summed Impedance at Location S

Frequency, Hz
The analyzed results from the random data acquired
Figure 15. Impedance of Component Package during the vibro-acoustic tests are expressed in terms of
the acoustic mobility data as shown in Figure 18 and
The resultant impedances at the Rand S locations were the blocked pressure data as shown in Figure 19.
obtained by summing the individual impedances mea-
sured at these locations. The summed impedances are
shown in Figures 16 and 17. The impedance at the R
location is controlled predominately by the stiffness of ,o .. -r-rr- -r-r-," ,, ,- -

the ring frame from approximately 100 Hz to 1000 Hz.

The characteristics of ring impedance are not obvious. - R Locations
This phenomenon might be attributed to the weakening 0 S Locations
of the ring stiffness due to the deep cut-outs in the ring Lcaton
frames, and the stiffening effects contributed by closely
spaced longitudinal stringers. The dynamic stiffness is ,0
approximately 35,000 lb/in. The impedance data at the U
S location for frequencies between 40 Hz 140 Hz , "-
cnd

appear to be low in magnitude; this was caused by the

low acceleration outputs of the Z-13 head. The low- -8

frequency stiffness (between 140 Hz and 500 Hz) is > .. 4---,, .

estimated as 10,000 lb/in. Major resonant and anti- Frequency, Hz
resonant frequencies occur in the frequency range from
600 Hz to 1500 Hz. In general, data above 140 Hz Figure 18. One-Third Octave Band Velocity Acoustic
appear to be valid. Mobility Levels at Locations Rand S

91
 !I
; I I pI.
I
I ,,.., i I
• 6h f*
i iiiii , , ,Ill i l i ... i p
I
! i

,q~ /'\ - Predicted

W. --- Measured
> Acoustic Test
-h
Level at
A / \Modified
Acoustic Test Level at-f R Location yLocalized
S Location S "
'a W.

5r r5t

Frequency, Hz Frequency, Hz
Figure 19. One-Third Octave Band Blocked Pressure Figure 21. Comparison of Predicted and Measured
Levels at Rand S Locations Force Spectra (1/3 Octave PSD) at
Location S: OASPL - 136 dB
In general the frequency intervals at which results have
been obtained from the mechanical impedance, blocked
pressure, and acoustic mobility spectra are not the TABLE I. Comparison of RMS Force Magnitudes
same. It was necessary, therefore, to convert the three
spectra to the same frequency intervals by averaging
and interpolating in each spectrum such that they were Equipment Refereace Acoustic RMS Force, Lb
all reduced to identical frequency intervals determined Mounting Test Level (50-2000 Hz)
by the widest interval of the three spectra at any paint Location dB * Predicted Measured
in the frequency range. Since the random data analy-
sis was at constant bandwidth and the sinesweep data R 136 4.43 4.41
were collected on a variable frequency interval basis, 2.50
the final frequency interval was determined by the S 136 (4.40)t
random analysis at low frequencies and by the sine-
sweep analysis in the upper part of the range. dB Re 2 x 10 W'
DISCUSSION OF RESULTS t Predicted force based on modified upport-

Based on the input data as described in the previous structure stiffness

section, the force spectra were computed for the R
and S locations. The computed results are presented
in Figures 20 and 21, respectively. The measured
force responses are also presented in the same figures Generally speaking, the force comparison at location
for comparison. The rms values of these factors have R is considered quite satisfactory both in rms values
been computed and are tabulated in Table 1. and spectral characteristics. The comparison at the S
location shows poor agreement for frequencies below
140 Hz. Such discrepancies are attributed to the
Predicted errors incurred in the measurement of impedances of
N /- .. Measured the support structure at the S locations.

E"l Sr'
"
- . The problems associated with impedance measurements
originated from two main sources; namely, in the
CD " electronics of the impedance measurement equipment
and the localized structural effects at measurement
points. Based on the experience of the present pro-
gram, the acceleration outputs from the Z-13 head
appeared to be low at low frequencies. Consequently,
Frequency, Hz it was found that impedance data measured at S loca-
tions are not valid below 140 Hz. The rocking and
Figure 20. Comparison of Predicted and Measured pitching motions generated by impedance heads during
Force Spectra (1/3 Octave PSD) at sinusoidal excitations would also affect the accuracy
Location R: OASPL - 136 dB of the measured data.

92
The localized structural effects are attribuled to the The value of f is estimated as 100 Hz from Figure 21.
application of the dental cement in bondigr, imp..ance Thus the localized stiffness of the support structure is:
heads, or the mounting :tuds used to connect the com-
ponent package to the support structure. Cemeniputtles, 4760
depending on their sizcs and the mixture ratios of the k - = 2.27 x 104 IVin.
cement powder and its solvent, would tend to generate - O"T
higher stiffnesses and increased modlsl masses at these
points. Such effects would create higher stiffnesses. The above value indicates an increase of stiffness which
This is particularly true for very flexible structure, is 2.27 times greater than the originally measured value
such as measurement locations at S. (104 IbVn.).

At low frequencies, the increase in stiffnesses attribu- The predicted force spectrum, which is c,,nputed based
ted to dental cement putties could be estimated by the on the modified stiffness, is plotted against the measur-
filowing equation: ed force in the frequency range of 50 Hz to 400 Hz.
F/f.. 2 i Significant improvements in amplitude accuracies have
k = k [( f -1 (3) been achieved.
fI
FORCE-CONTROL SHAKER TESTING EXPERIMENTS

where The objective of the shaker testing was to determine the

accuracy of the predicted forces as applied to the test-
k localized stiffness ing of component packages. The assessment of accur-
acies was based on the comparisons of component res-
k2 stiffness of the component package ponses obtained in thetesting
vibro-acoustic tests with
(4760 Wbin.) obtained from shaker by controlling inputthose
forces
through loadwashers. Instrumentation and testing
f = resonant frequency of the component- equipment employed in the tests are illustrated by the
support structure system block diagram as shown in Figure 22. The Ling Elec-
tronics Automatic Spectral Density Equalizer/Ai.a;yzei
f = resonant frequency of the component Model ASDE-80 was used to control the test system and
2 package (110 Hz) to perform the following functions:

Log Convert.

-8 ; " .cn. r I.Mate.-1 ' -1

I
I,,
Fp -u .....
t-

-
-
I_
" ----_ A?2A
- -- J*
::: F A
:::::::::::::::::
A 2 Ac ,,,-o--.--i-t--

C oFo l i Sh ak e P oiptsr A m l f e ArIoi Curment

/L/ I -- mme oc I g. outtII Ckrg.mp, I

_ _ ____ Cr ectro-O
rxi Shu

Figure 22. Schematic Diagram of the Shaker Testing Program

93
 * To generate inpt force spectra for .*... .. "
vibration tests;
* To measure the PS[3 of feedback signals N TR
from loadwashers on a continuous basis; and, TRPF
MR
To provide automatic control of the servos.
automatic feed- ,/ \TRMF
back PSD's by the use of

A control spectrum was shaped in the frequency range of V :Base Plate I

10 Hz to 2025 Hz by 85 separate servo control channels 'Resonan '
according to the specified bandwidth; of the ASDE-80. Effects
The PSD's of the control forces were based on:
a) The predicted force spectra; and, Frequenc), Hz
Figure 23. Comparison of Component Responses at R
b) Loodwasher outputs obtained from vibro- Location (1/3-Octave Frequency Band)
acoustic tests.
The comparisons of the component package responses IV F T--7

at locations
24, P. and Sare
respectively. thepresented in Figures 23 and
characteristics, theInfollowing
discussion of the response
terminologies are
/, r----TRMF
defined: :z: ' t ' MR
-TRPF

* Measured Response (MR) - Response A.

measured from a vibro-acoustic test o -

* Test Response by Predicted Force (TRPF) - iBase Plate'

Response generated by predicted force spectra ' / Resonant
Effects
" Test Response by Measured Force (TRMF) - 1Ec ,.

Response generated by measured loadwasher Frequency, Hz

outputs during a vibro-acoustic test.
Figure 24. Comparison of Component Responses at S
In Figure 23, the MR correlates satisfactorily with the Location (1/3-Octave Frequency Bond)
TRMF for frequencies below 300 Hz and between 800
Hz and 1200 Hz. The comparison between MR and
TRPF show reasonably good the forcing
input cha-acteristcs, thereby
20agreements
H, b
lrgedisrep~cis
bel~ for frequencies
ocurbeteenthe AEocn there
(25aHz)tcod
below 200 Hz, bur large discrepancies occur between creating different component responses.
200 Hz and 1200 Hz. The increase in response
amplitudes behveen 1200 and 2000 Hz was attributed Dynamic Range of the Control System - The
to resonant vibrations of the base plates. maximum dynamic range of the control system
was estimated as 40 dB which, in many
In Figure 24, the MR and TRMF are in satisfactory instances, was considered inadequate for
agreement for frequencies below 200 Hz and between controlling responses of high Q systems.
800 and 1200 Hz. The TRPF is in poor agrenment
with the MR below 800 Hz. This was certainly, in Control Correction Time - The analog con-
part, caused by the errors incurred in the predicted for trol system requires a minimum of 5 seconds
force spectrum below 140 Hz. But, it correlated well to
tioncorrect a 40-50
time may have dB change.
been too slowTheforcorrec-
the
for frequencies between 800 Hz and 1200 Hz. Large
amplitude discrepancies between 1200 Hz and 2000 Xz structural systems that were tested.
were caused by resonant vibrations of the base plates.
CONCLUSIONS
The results of the force-control testing are not quite as
encouraging as compared to that of the force spectra The following conclusions may be drawn from the
prediction. The possible causes for the large ampli- results of the experimental program:
tude discrepancies in the component responses are
presented as follows. 0 The Force-Spectra equation can be used to
predict interaction forces between support
0 Input Forces - The difference in the structures and component packages with
frequency-bandwidth resolutions between reasonable accuracy provided that sound
the narrow band PSD analysis (approximately estimates can be made of:

94
 - Input Impedance of Support Structure REFERENCES
- Input Impedance of Component Package 1. Schock, R., "Utilization of Force-Spectra
- Acoustic Mobility at Mounting Location Technique for Dynamic Environmental Prediction
- Effective Acoustic Pressure on Support and Test," To be published in Shock and
Structure Vibration Digest.
2. Kao, G., and Sutherland, L.C., "Development
component mounting locations dueat the of Equivalent One-Dimensional Acoustic Force
application of dental cement will signifi- Spectra by Impedance Measurement Techniques,"
cantly alter the characteristics of meosured Wyle Laboratories Research Report WR 69-11,
impedances. Therefore, it is important that May 1969.
input impedances of support structures should 3. Jones, G., and On, F., "Prediction of Inter-
be determined as close to actual mounting face Random and Transient Vibratory Environ-
conditions as possible. ments through the Use of Mechanical Impedance
e When corrections were made to account for Concepts," The Shock and Vibration Bulletin
stiffening effects due to localized supports, No. 40, Part 3, pp. 79-88, December 1969.
the predicted rms forces were found to be
higher than the measured forces. 4. Kana, D., "Response of a Cylindrical Shell to
o Since the coupling effects between a com- Random Acoustic Excitation," AIAA Journal
ponent package and its supporting structure Vol. 9, No. 3, pp. 425-431, March 1971.
in the finalofovetetln
are included theeffct vibration
an criteria,
uderestng5. Ballard, W.C., Casey, S. L., and Clausen, J.
the effects of overtesting and undertesting C., "Vibration Testing with Mechanical Imped-
on component packages can be avoided.
once Methods," Sound and Vibration, pp. 10-
* Digital approaches to measure mechanical 21, January 1969.
impedances and acoustic mobilities hove
been proved to be practical and feasible in 6. On, F.J., "A Verification of the Practicality
providing dynamic information for the com- of Predicting Interface Dynamical Environments
putation of dynamic environments, by Use of the Impedance Concept," The Shock
* Improvements in mechanical impedance and Vibration Bulletin No. 38, Part 2, pp. 249-
measurements are required to achieve de. 260, August 1968.
sired measurement accuracies.
7. Nuckolls, C.E., and Otts, J.V., "A Progress
9 The analog control system used in performing Report on Force Controlled Vibration Testing,"
shaker testing lacked the needed dynamic The Shock and Vibration Bulletin No. 35, Part
ranges and correction time. This has caused 2, pp. 117-130, January 1966.
poor agreement between component responses
obtained from the shaker testing, and the 8. Murfin, W.,., "Dual Specifications in Vibra-
vibro-acoustic testing. tion Testing," The Shock and Vibration Bulletin
However, it is felt that response agreements No. 38, Part 1, pp. 109-113, August 1968.
could be improved if a digital control system
were used. Current digital control systems 9. Otts, J.V., "Force Controlled Vibration Tests:
can provide 60 dB dynamic range with a A Step Toward Practical Application of Mechan-
correction time of less than one second. ical Impedance, " The Shock and Vibration
Frequency resolutions from 4 Hz to 8 Hz are Bulletin No. 34, Part 5, pp. 45-52, February
obtainable. 1965.

10. Noiseux, D.J., and Meyer, E.B., "Applica-

bility of Mechanical Admittance Techniques,"
ACKNOWLEDGMENTS The Shock and Vibration Bulletin No. 38, Part
2, pp. 231-238, August-1968.
This work was supported by National Aeronautics and 11. Kao, G.C., "Prediction of Force Spectra by
Space Administratin, Marshall Space Flight Center, Mechanical Impedance and Acoustic Mobility
under Contract No. NAS8-2581 1. Measurements Techniques," Wyle Laboratories
Research Report WR 71-16, September 1971.

95
 DISCUSSION

Mr. Shoulberg (General Electric): Are you Mr. Schock: Yes, the force spectra are
proposing to average your force at the four averaged, but the total impedance is the sum-
Input points? mation of the Impedances at thefour inputpoints.

96
 DYXAWIC DESIGN ANALYSIS
'PIA THE BUILDING BLOCK APPROACH

Albert L. Klosterman, Ph.D. and

Jason Re Lemon, Ph.D.
Structural Dynamics Research Corporation
Cincinnati, Ohio 45227

The availability of mechanical impedance testing

equipment for determining the dynamic characteris-

IJtics of structures has caused considerable interest

in describing a complex component from test results.
However, the direct use of digitized response data
for highly resonant, multi-input/output components
yields unsatisfactory results when used for further
analytical total system Anvestigations. This paper
describes how the response data can be used to describe
the component under test and the mathematical formu-
lation necessary to represent the equations of motion
of the total system. The procedures are then applied
to large complex mechanical machinery to perform a
total system dynamic design analysis*

INTRODUCTION
Several authors have attempted to system dynamic analysis.
use experimentally measured impedance
data to represent various componento in Reliable representation of com-
a total system dynamic analysis. 11, 2, ponents from dynamic test results
3, 4]. However, in cases where the allows for the complete implementation
component under test is highly rasonant of a dynamic design analysis procedure
and has multiple connection points# the which parallels the design process,
direct use of this response data is where major structural components or
unsatisfactory L13J. Small errors in substructures, are often designed or
the measured data are unavoidable with analyzed by different engineering
any of the present or proposed test groups or at different times. There-
equipment. When these multi-input/out- fore the design or analysis of each
put subsystems are included in the total component could proceed as independently
system dynamic analysis, the errors will as possible with due consideration being
be greatly magnified whenever diffrenceoe given to th6 final coupling of sub-
of nearly equal large quantities cccur structures to form the complete struc-
in the calculations. This work illus- ture. Groups which are designing the
trates techniques which can be used to extremely complex components could rely
determine an analytical representation on impadance test results while groups
of the dynamic characteristics of a designing structurally simpler compon-
system from the response data so that ents could use analytical finite element
the above mentioned errors are avoided, investigations. The results of the
In particular, response data is used to dynamic analysis on each component can
set up a modified real or complex "modal" then be evaluated by the department with
representation of the system under test. system responsibility to evaluate total
Although the modified modal representa- system dynamic performance before the
tion is not completely general, it is system is completely assembled. There-
convenientto manipulate and has been fore full implementation implies mean-
used satisfactorily to represent highly ingful ipecimcations on the dynamic
resonant components. These representa- performance of each component.
tions can then be combined with
appropriate mathematical representations This procedure is also attractive
of other components to perform a toal because each component is represented in

97

IL
 I
terms of a reduced number of modal modes as follows
degrees of freedom. Therefore the de-
grees of freedom of the final matha-
matical model are significantl reduced n [rJfrPT
F ]
with respect to the total ycal do- ] (
grees of freedom. This allows the rul I . + i D- X r]
development of a mathematical model
.hich is flexible enough to predict the or
necessary phenomena under investigation,
but small enough to be easily manipulated. n [ r]Trp]rr
DETERMINATION OF DYNAMIC CHARACTERISTICS IQ] =5 (8)
FROM RESPONSE DATA r=l mr( 0 r2 "i 0 / g r" 02 )
In order to determine a modal re-
presentation of the dynamic characteris- where E and mr are complex modal
tic consider the equations of motion wh er T hd ae oml modal
for
ing steady state harmonic
the presence motion
of hysteretic assum-
damping. parameters. The above
Canl be determined modal parameters
by various testing
procedures and data analysis techniques.
[M-jV' + i [D ] + [K)[q]-[FPei~t (1) Some of the common methods are the
followings
Normally the assumption of proportional 1) Use of multiple shakers
damping is made where damping is assumed 2) Least square curve fitting
proportioned to stiffness and/or mass, 3) Graphical curve fitting
in order to uncouple the equation of ) Eigenvector search techniques
motion with modal coordinates. However, Reference El] lists a fairly complete
this assumption is not necessary if com-
plex modes are used in the solution of overview of the available techniques.
the equations of motion [1]. To deter- Space does not permit a complete descrip-
mine the steady state response due to tion of the above techniques here, how-
sinusoidal excitation [f]=[F] ei t ever for completeness, one of the basic
seek a solution in the form 1q]-[Q]ei0 t techniques will be reviewed.
to obtains The procedure is basicly similar to
r[K+iD 0 2 [M]] (2) the method of Kennedy and Pancu [5) but
K - E =F] (2) extended to include complex modes 1l).
Let 02 . X and consider the homogeneous A typical polar plot of the response of
equations a
inhighly
Figure resonant component is
i. shown
[K + iD] - X [M]] (Ql = [o (3)
This set of equations has a non-trival IMAG.(Q)
solution if
X [M] =0 (4) REAL (Q)
det [[K+iD]-
There is a set of "n" complex eigenvalues dO 0
X and
E"Tr]associated complexthis
which satisfies eigenvectors
home- N,5
R
geneous equations

[K ]
+ iD][1r] -Xr)C4]ETr] -[0] (
These complex modes depend on [K], [M]
and [D] and differ from the real normal
modes which depend only on [K] and [M].
Orthogonality of the above complex modes Figure 1 - Typical polar plot
isfollwingequations:
easily shown and results in the of a highly resonant system
It can be snown that each
0_ is a fre-
C rr]TM)C IT5 ) Co0 quency where the rate of chane of the
and for r a are length with respect to frequency is
a maximum. Once each of these frequen-
+ i D] C =[K0 (6 cies is located, a circular arc can be
It can be easily shown [1] that fit to of
value this dportion
2 of the curve and the
the solution to the equations of motion v o
can be written in terms of the complex d at Ozr

98
can be determined. It can be shown that equations of motion in terms of modal
the value of gr can then be determined coordinates was pioneered by Hurty [6),
from the following equationa but further advances have resulted in
the practical implementation of the
1 d p2 method for experimental data [1,7). The
g= 2 (d -- at P= r (9) method will be illustrated for zero
damping, but is easily extended to damped
system. Let the degrees of freedom at
By constructing a normal to the curve at the points of connection to other com-
r and recording the anglea , the ponents be designated by the vector [Q).
components in the complex eigenvalue can Then the motion of these points is
be tionsi
determined from the following equa- related to the modal coordinates Y r of
the component by the equations

[Q] [IT ] (13)

r2[Ur + i Vr] (10) The equation of motion for the general-
ized modal coordinate, Yr is
where2r
U
ere gr2R cos a. (-p 2mr +kr) Yr= [I r]T[F] (14)

,r = g 2R sin IF] is the vector of forces applied to

r the substructure and [Trj is the eigen-
R = the radius of curvature at vector abbreviated to intlude only the
coordinates at the connection points,
=0 r and those where additional external forces
are applied.
to any
(Notes mr can be normalized
value desired) In order to set up the mass and
stiffness matrices for the entire assem-
By a close examination of Equation bly, Equations (13) and (14) can be
(8)it is recognized that for a parti- implemented in the following mannere
h
The
cular frequency range, the component can equations of motion for the st com-
be adequately represented by the modes ponent can be described in terms of modal
which are resonant in this range and the coordinates ass
complex flexibility of the modes -.
bove
the range of interest. Therefore, r[ ks 2
2 C
me
S [y s]=[ 5s][F s
]
represent the dynamic flexibility equa-
tion IQ] = [G(i )0 i?] in the follow-
ing forms (15)
where [FO] is the external forces
applied to the sth component. The
Q] +[GM(i )J[F]+[Z][F) (11) uncoupled equations of motion for the
total systems aret
Where [G1 4 (iP ) is determined from the
resonant modes and [Z] is the complex kl mI Y
flexibility of the higher modes of vib- k2 2 m2 Y2
ration. The complex flexibility matrix 2
[Z] can be obtained by subtracting the . .
modal response representation, deter-
mined by polar plots, from the total
response data. e.g. n n
[Z][F] = [Q] - CGM(i )][F] (12) .. ..
The best accuracy for this calculation is
usually obtained at a frequency where [lJIT[F1
G~j Fj - 0 (i.e. near anti-resonances [I2)T[p2)
in the modal response . (16)
representation)

[ Fn]
DETERMINING THE EQUATIONS OF MOTION OF I In]T
THE TOTAL SYSTEM
The technique of setting up the

99
where Therefore,
EQ'1 ["E V]E [YD] rEcD0zl Ecz E vz (19)
[Q2] - 2E2]
IT ....... [] C
[-G,-
2
EQ ) (17) J
* Write Equation (19) as,
[Qn Rn]
When the components are connected# con-
straint equations are necessary to re-
late the various physical coordinates
[Q] at the connection points. Let these
[ D
I
I [i
EY
I
constraint equatiuns be represented ass where

Q2 E 3] 3 ...[I
[H] -[0] (18) L1
Therefore, Equation (16) becomess
J
Qnp [C 2,m] r][ Y] [ f Tp
Substitute
Equation (18) Equation (17) into
to obtains or

[01 L[ IT[
] a a 11 E

Vn J 0 ynJ TC
][TF ]

or This set of equations can then be solved

1 to determine the natural frequencies,
mode shapes, frequency response, etc.
[C)I u [0J The
previous basic approach taken with vhe
technique is to approximate a
* continuous system which has a infinite
number of modes by a system which has a
* finite number of modes, Use of a
n smaller number of modes to represent a
•Y dynamical system always results in a lose
When the components are connected, the of mass, a lots of flexibility, or both.
total degrees of freedom are reduced by Through practical experience, it has been
the number of equations of constraint, observed that frequently the flexibility
Therefore, break the total de rees of of component modes outside the frequency
freedom into independent [ ? ] and range of interest are important in the
dependent yD] coordinates. overall simulation, whereas the inertial
properties are not. Therefore the
r , following technique has been devised to
[DJ _ £0) include
ithout the flexibilit
including of a higher
the inertI modes
propertie

The alnusoidal response of a parti-

,CD .
-D oular Oomponent can be represented in
modified dynamic flexibility form ass

[ YD1 - - ECD]'I [CI ] I TI) IQ 0) [ t [ Y I + rZ)[F0 ) (21)

100
 where 2) A fewer number of modes are
needed to obtain an adequate
EZ] is
thethe residual
modes flexibility
of vibration of
not in- representation
is if the component
tested in a manner similar
eluled in the modal coordinates to the state in whioh it will be
Emounted in the entire assembly.
° E J s o e ,tth seectonowever.to determnall the necessar
:
is forces
pontswhen
at the connection parameters from the experimental test,
the component is restrained, a
points.
Also following
the forces relationship exists sohisticated mounting system is required
between on modal coordinates and whch will monitor the forces at the
forces applied at the connection points, connection pointa which are restrained.
pIn the ease whre constrained modes re
EFy I- - C ITC O] (22) used to represent a particular component
it can be shown that the residual mass
Multiply Equation (21) through by EZ) "l of the constrained coordinates also has
to obtains a rather significant effect on the system
E
r]" l r'['[ ]E (23) simulation.
APPLICATION - LARGE MOTOR SIMULATION
Substitute Equation (23) into Equation
(22) to obtains In order to evaluate the use of this
approach to studying the dynamic char-
E]3-.] f 731j[Q ]E+[IT 3TCZJ'i ])E "y aoteristics of complex mechanical mach-
(24) inery, SDRC and U.S. Steel Marketing
undertook an anlysis of a 1750 HP
Hanoe, Equations (23) and (24) becomes electric motor. An overall view of the
motor is shown in Figure 2.

• (25)

Equation (25) represents the stiffness

matrix form of an additional element
which can be included in the mathematical
model in order to represent the residual
flaxibility of a particular component.
The inclusion of this elument into the
mathematical model has significantly
improved the simulation for numerous
practical cases. In some of the more
critical cases the inclusion of residual
flexibility has meant the difference
between excellent correlation and no
correlation. Therefore, either a
large number of modes should be included
to account for all important flexibil-
ities, or residual flexibility should be Fig. 2 - 1750 HP Electric Motor
added to approximate the effects of the Foundation System Which
omitted higher modes. Was Analyzed.
The above technique was illustrated
for the case where the modes were deter-
mined from a free-free excitation test The system was divided into the
(i.e. the connection points were unro- following components which are shown in
strained). Procedures have also been Figure 3.
developed and will be published at a 1 Motor Bass
later time for using the modal properties 2 Rotor
generated from excitation tests where Fluid Film Bearings
some or all of *he connection points are
restrained durin, testing. In some cases Str
this technique i& attractive for the 5 Foundation
S following reasonss The motor base and rotor were modeled
with finite olemont techniques and their
1) The free-free zero frequency modes individual modal characteristics deter-
which are sometimes difficult to mined. The other components were tested
determine, are eliminated, to determine their respective character-

101
 a Fig* 5 -Typical Free-Free Rotor Pre-
Lip? quency Response*
The fluid filmtca~rire were tested
and represented by a simpla Spring-
daohpot model@ The comparison of a
typical analytical model to test results
Fig. 3 -Components Selected for the aesoni iue6
Dynamic Design Analysis.#_______________

The motor bass was represented with --

6 rigid body zero frequency modes flex- i l ?~

of all other modes* Correlation bet-
ween the computer model and the test
i :4.i
results on the base alone are shown in A.t (K S),
C.7
Figure 4. 106~[*V~KWV~I~

j I t o
10

IFig* 6 -Typical Bearing Frequency

Thestator wstested and repro-

- --1sented with 6rgdbody zero frequency
~ ziijjj
\jfj~jjL.modes and the residual flexibility of
all other modes. T1he comparison of the
teat data to the analytical representa-
tion for a typical plot is shown in
Fig. 4 -Typical Free-Free Base Fre- Figure 7a
quency Response.
The rotor was represented with 6
rigid body zero frequency modes and 3
flexural modee. The cor~relation bet-
ween tho computer model and teat results
for the rotor are shown in Figure 5.

102
 AA

_ _0' __
* ilo 2~~ -: FYI
~10 100
-P .U 1000 F 2 V

Figs-7Typical Free-Free Stator Fre- I I 't

quency Responses F

The foundation was tested and re-

presented by 6 rigid body zero frequenc
modes and 6 flexural modes. The compri-
eon of the tout data to the analytical Fig. 9 -Stator and Base Free Body Dia-
representation is shown in Figure 8 * gram showing forces at the
connection points.

_J:TI!~i1

jg~i
.A

IRK__________
4.4. -. 4,A

-A-4,~

the
used4for i o- 1ttri
0hwi

combine this componwnt with the motor

base to see if meaningful results can be - dUoA.~

obtained with Jusot these two components.

The forces assumed at the connoction*I
points for the motor base and stator 1

model are shown in Figure 9. The toot

result. of the motor bso and stator ,,
combination
ntmdoe
are compared
nFgures to10 the
and compon-to
11., !:!/~~

The total system was then assembled 10

mathematically and the forces assumed at.
the connection points are shown in Fig- :2:
ure 12. The total system response isI:F
compared to the computer predicted
response in Figures 13 and 14. Fig. 11 Predicted and Measured Fre-
quency Response of Base-Stator
Assembly.

103
 CONCLUSIONS
The use of a modified modal repro-
sentation hau been shown to be an ads-
- -~ quate reproesentation of the dynamic
- characteristics of each component for
perfcrming a total system dynamic
analysis. The modified modal represent-
ation. anm be determained from either t~3t
data or an analytical investigation*
Reliable representations determined from
tout results allows for the complete
* '~ - *,,implementation of a "building block
approach to dynamic analysis" which is
/'"[.,extremely attractive since it is practi-l
cally feasible for large complex systems
and parallels the substructurirg design
process.
REFERENCES
[1) Klosterman, A.L. - "On the Experi-
mental Determination and Use of Modal
Representations of Dynamic Character-
istics", Ph.D. Dissertations Univer-
sity of Cincinnati 1971.
Fig. 12 -Base, Stator, Rotor and [2] Klosterman, A.L. & Lemon, J.R.
Foundation Free Body Diagram "Building b3lock Approach to Struct-
___________ural Dynamics", AS=E Publication,
1969 VIBR - :30.
opMo
V r""M-Odd ~F. BM Rq0d@
[3] Wright, D.V. - "Sound Radiation and
the Force Ratios of Foundaktion
Structures", Paper presented at Ship
~L ~ Silencing Symposium, Groton$ Comne
;:.~-[4) Ballard, W.C., Casey, S.L. a Clausen,
44, J.D. - "Vibration Testing with Mach-
'iI,.Lanical Impedance Methods": Sound and
~41IIL
4 .Vibrations January 1969.
[5) Kennedy & Panou - "Use of Vectors in
Vibration Measurement and Analysis",
Journal of Aeronautical Scienee,
Fig. 13 -Predicted and Measured Fre- Vol. 14, No. 11 November 1947.
quency Response of Bases,6 utWle
Stator, Rotor and Foundation [)Hry atr-"yai "yai nlsso
nlsso
System. Structural Systems Using Component
4 s Modest" AIAA Journal, Volume 3, No.4
~I1 J~T~[7) MacNeal, R.H.(ed) - "The Nastran
- Theoretical Manual"t, NASA SP-2210
~ September, 1970.
fi l

tot

Fig. 14 -Predicted and Measured Fre-

quency Response of Base,
Stators Rotor and Foundation
System.

104
 MOBILITY MEtSUREMENTS FOR THE VIBRATION ANALYSIS
OF CONNECTED STRUCTURES

D.J. Ewins and M.G. Sainsbury

Imperial College of Science and Technology,
London, England

The mobility or impedance coupling technique is widely used

for the vibration analysis of structures which comprise an
assembly of connected components. Its application is
straightforward when the components are amenable to theore-
tical analysis, but if certain components are too complex
to be analysed their mobilities must be obtained experimen-
tally. Standard 'impedance' testing methods are generally
inadequate for measuring the required multidirectional
mobility data, and the work described in this paper is an
attempt o develop techniques for obtaining such data.
Measuremerts have been made on a freely supported beam and
on a resiliently mounted block, and these data have been
used to predict the response of the system formed by bolt-
ing the beam and the block together. The results illus-
trate the importance of obtaining sufficiently complete and
accurate data if mobility measurements are to be used for
the vibration analysis of connected structures.

INTRODUCTION ordinates at each connection point as

are necessary to realistically des-
ViP.ration analysis of complex cribe the conditions at that junction.
structures which comprise an assembly The motion of a point on a structure is
of connected components is often made completely defined by six co-ordinates,
using the mobility or impedance coup- three translational and three rotation-
ling technique. This approach per- al, but in the analysis of specific
mits analysis of each component indi- cases it is often possible to ignore
vidually and then couples them some of these by virtue of the symmetry
together by matching forces and velo- of the structure. In the simplest
cities at each connection point, which case where motion is known to occur in
is considerably more convenient than a single direction, such as a mass
attempting to analyse the complete moving in a straight line, then only
structure at once. It is in fact a one co-ordinata is required. In prac-
standard technique in dynamic analysis. tice, this degree of symmetry is seldom
However, it often happens that one (or encountered, especially with the corn-
more) of the components is itself too plex engineering structures that we are
complex to be analysed directly and considering, so vibration analysis
for such a case, recourse may be made using a single co-ordinate is usually
to an experimental approach in order unrealistic. However, in many practi-
to obtain the mobility data which is cal cases, motion is confined to a
required for the analysis of the com- single plane involving vibration in
plete assembly. This paper is con- three directions - two translation and
corned with the development of experi- one rotation - and for those it is
mental techniques suitable for measur- necessary and sufficient to include
ing this data. three of tie six co-ordinates in the
analysis of vibration.
When the mobility coupling tech-
nique is applied analytically, it is Such considerations are made regu-
customary to consider as many co- larly in theoretical vibration analysis

105

*2
 -- i

Fig. I BLOCK AND BEAM ASSEMBLY

BEAM (CL)I

BLOCK (b)

7Fx

Fig. 2 CO-ORDINATLS FOR COUPLING PFEAM AND SLOCK

106
but are seldom included in those oxer- bly formed by these two components, use
cises which make use of experimental is made of the mobility coupling tech-
data for one of the components. Stan- nique. By considering each of the
dard 'impedance' testing techniques are components individually, we may derive
confined to measuring mobility in a mobility expressions for each of these
single direction and are generallyinade- which relate the velocities at the
quate for the acquisition of such com- point of connection to forces and
plete data as are required for a real- couples applied at that point. This
istic analysis of a complex structure, data may be presented generally by the
This limitation has been identified matrix equation:
already C13, [2] and previous work has
demonstrated the difficulty and extent
of the task of measuring the complete lr} .rI{ { r {*,
6 x 6 mobility matrix for a structure [y]Lil' or 1F L [J J1)
[2]. However, as mentioned above,
many practical structures do have a
certain degree of symmetry and this
permits us to confine our attention to where fY] is the mobility matrix and
pemotion in on [y- is the impedance matrix for
only three
motion co-ordinates
in one plane and at
to a time.
consider ta
ta on ntesrcue sa.
vector of velocities in the co-ordinate
The mobility data required in this directions included and fF1 is the vec-
case constitutes a 3 x 3 matrix which tor of forces (or couples)-in those
is
the considerably
complete 6 x easier to handle than
6 matrix. directions.
ietos These
hs matrices
mtie and
n vec-
e-.
tors, and all those which follow, have

This paper describes a case study complex elements in order to describe

both the amplitude and the phase of the
made to assess the feasibility ofvaiu qantesrpsned
using ex)erimental data to analyse various quantities represented.
connected components and employs a Equations for two components (a
particular structure
components as composed
an xample. Theof first
two and b) connected together may be rela- .
compnensxampe.
asan he frstted by virtue of the fact that at the
stage was to specify those nobility cectio o the ective ve
data which are required in order to connection point their respective velo-
analyse the vibrations of the assembled cities must be identical, so that
structure. Next, an experimental
technique was developed for obtaining (,,} { { (2)
these data to the required accuracy. (2)X
Finally, predictions of the dynamic
characteristics of the assembled Furthermore, by considering equilibrium
structure based on measurements of the at the point of connection,
individual components were compared
with measurements made on the structure
itself. {F.] + {Fb} = {PI (3)

CASE STUDY
where fPI is an externally applied
The structure treated in this force,
study comprised two simple components. Combination of equations (2) and (3)
st c rp leads to an expression for the mobility
The first was a solid steel block 9of the combined structure at the con-
911 x 12" resting on four identical nection point tye
rubber pads, one under each corner,
while the second was a uniform rectan-
gular steel beam 11" x 2" x 72". The r:i (rv,-Irv "I
assembled structure was formed by L[cJ = LGI+[ybJ,
attaching the beam to the block in such
a way as to obtain symmetry in one ver- As mentioned earlier, as many as six
tical plane but not in the othcr, as co-ordinates may be necessary to do-
shown in Figure 1. Interest was con- fine the motion of a point on a struc-
fined to vibrations in the vertical ture (in which case the order of these
plane containing the longitudinal axis matrices and vectors would be 6), but
of the beam, although a similar analy- in this case, as in many others, we may
in the iti ae si ayohrw a
sis
siser could
cldaes. mlimitfor motion
be made
our interests to a limited number
of these. Figure 2 shows the con-
noction point for each component and
THEORETICAL APPROACH indicates those co-ordinates and forces
which should be included. Because of
In order to predict theoretically the symmetry of the structure. the
the vibration properties of the assem- other three co-ordinates (which include

107
 P,

P2.

FIG. 3(a) EXCITING BLOCK MK I (Not to Sc~e) 1

P21
ru

Y 3 9V .,V
vc
r3/

FIG. 3(b) Y 2 (Not to 3(- 4e)

NKirN;1'.)-

108
motion out of the plane of the paper) three co-ordinates in the theoretical
may be ignored in this analysis or, if approach, so it is necessary to include
required, treated separately. Thus all three in an approach using experi-
the equations defining the mobility mental data. The requirement for such
matrix above (1) are in this case: comprehensive data may not be met using
conventional methods of measuring mobil-
(*)
X
'~
Y..Yxy
Ye
iii t
y (collectively
F.'impedance
referred to as
tests') as these are insuf-
- y. yvy Yy* FY (5) ficiently developed to provide either

)Ye Yee Mo. I the completeness or the accuracy

demanded for this application. In all
but very exceptional cases, mobility
The object of this analysis is to measurements are confined to those ex-
predict the vibration characteristics pressions which relate response to a
of the complete assembly formed by the translational force in a direction nor-
two components, and this information is mal to the surface of the structure
contained in the assembly mobility,lY., under test. The response to applied
It will be seen from equation (4) that couples or in-plane forces is not gen-
this technique involves a number of erally measured so that the mobility
matrix inversions and these can be the matrices required here may not be ob-
source of some difficulty in numerical tained directly from current experimen-
applications. In the first instance, tal methods.
the component mobility matrices [Y.] and
[Yb] become ill-conditioned and are One previous worker has attempted
difficult to invert at frequencies to measure the complete mobility matrix
close to one of their respective natur- and has developed a twin shaker unit to
al frequencies. A similar limitation apply either a direct force (with the
applies to the combined expression two shakers in phase) or a couple (with
(EY-1"+ lyi") which also has to them in antiphase) to the structure
be inverted. Secondly, it is known [2]. An alternative solution to the
that operations involving the mobility problem of rotational excitation is
matrices can be sensitive to small being explored in the design of a ter-
errors in the individual mobility ex- sional electromagnetic shaker, but this
pressions, and this feature must be is at an early stage of development.
taken into account when experimental A third approach is provided by the
data is to be incorporated, technique described in this paper which
is based upon tho use of a sing!-
Theoretical expressions for the shaker and other standard impedanc,2
component mobility matrices were do- testing equipment with the minimum nf
rived by standard methods 131, specially designed attachments. The
although it was necessary to use exper- aim of this measurement technique is to
imentally derived data for the dynamic determine the 3 x 3 structural mobility
stiffness and damping capacity of the matrix described above in relation to
rubber pads which support the steel the case study which is used here as an
block. Calculations were made using example, and is achieved in the follow-
these expressions and the results are ing way.
presented later in the paper. A fur-
ther series of calculations were made There are nine elements of the
to predict the mobility of the assem- mobility matrix to be determined and
bled structure, again using the mobil- these are obtained by measuring the
ity coupling technique, and these will three responses (i, and 6) to each of
also be presented and discussed later, three different excitation conditions.
together with corresponding results Thus three tests over the chosen fro-
from the experimental approach. quency range are required and in each
of these the shaker is attached so that
,1 different combination of the three
EXPERIMENTAL APPROACH reference forces (F., F and Me) is
applied to the structure. Knowing
We shall now consider a situation these specific combinations, it is then
in which mobility data for the indi- possible to extract the responses to
vidual components are to be determined each of these forces individually and
solely from experimental measurements, thus derive the required mobility
as opposed to theoretical analysis. matrix. Details of the specific
This is a matter of necessity when shaker attachment arrangements used are
dealing with particularly complex discussed in the next section.
structures.
It is clear that whatever '.ech-
In exactly the same way that it nique is employed to measure this mobil-
was considered necessary to include ity deta, there is a vast quantity of

109

qI
information to be handled. Essen- assuming the block to be rigid, those
tially, all that is required from the could be related to the response of the
exercise is a prediction of the vibra- structure itself in the three reference
tion properties of the assembled struc- directions by means of a simple geo-
ture and all the mobility measurements metric transformation:
made on the separate components are of
no inherent interest beyond that of
enabling the calculation of the assem-
bly properties. The mobility coupling
process (i.o. application of equation
[] 9 ~
LTJV (7)

(M)) is applied at each of a number of V

Y
discrete frequencies using the component
data measured at those frequencies. For each excitation position, we may
Thus it is convenient if such data are write
in a digital format rather than
analogue (i.e. in the form of graphs). T= = [YVF.I
Since there is considerably more of L J(3"
this intermediate data than is even- "-ilrYlD
=
tually required to describe the assem- L ir'l
bly properties, it is most conveniently or
acquired and stored in digital form,
either on punched tape or in the memory
of a digital computer which is subse- M ' I=E

rvMr (9)
quently employed to carry out the matrix
manipulations for extracting the re-
quired information. This requirement
""
"T11Ypi

where [yL
=

is a vector of apparent
I
indicates the suitability of the digital mobilities and [r}h is the force trans-
transfer function analysers currently formation vector for the ith excitation
available which may be readily inter- position. Combining the three equa-
faced with a small digital computer to tiens for i = 1, 2, 3 gives
provide an automatic testing facility.

MEASUREMENT TECHNIQUE or

Based upon the concept of a single [Y] [T][Y ](10)

shaker technique for measuring multi-
directional mobility data, a number of relating the measured quantities with
specific configurations for attaching'' the required properties.
the shaker were investigated. It was
necessary in all of these to attach an The first stage in assessing the
'exciting block' to the point of inter- suitability of the proposed exciting
est on the structure and to connect the block design was a check on the sensi-
shaker to this block in different ways tivity of the numerical manipulation
so as to exert the chosen excitation described by equation (10) to realistic
conditions to the structure. The errors in the measured data. This was
first design for this block is illus- performed numerically by taking a speci-
trated in Figure 3(a). The shaker was fic mobility matrix for the structure,
attached to the block in each of the computing the apparent mobilities that
directions indicated by P, , P2 and P3 would be observed in the three runs
in turn, and the correspiiding compo- with the exciting block, polluting this
nents of the excitations in the three data with random or systematic errors
directions (Fx, Fy and Me) are given typical of' those which might be expect-
by the matrix equation: ed from the experimental equipment used
and then recomputing the structure
Sm oobility with this polluted data. it
FY -- .or f~ ={}F1
Fl.or IoP1= Ptwas
Pi (6) the soon foundblock
exciting from shown
this exercise
in Figure that
3(a)
Me [would L not be suitable for obtaining the
i|L,2,3 required mobility data because errors
in measured quantities of the order to
For each shaker position, measurements 0.01% in amplitude and O.O ° in phase
were made at specific frequencies of were sufficient to generate large
the input force and the three res- errors in the computed mobiliies (in
ponses, and from the collected results excess of 5 dB). Accordingly, alter-
of all three runs the required mobility native designs for the exciting block
expressions could be derived as follows, were investigated and the one finally
The response was measured at three con- adopted for all tno tests reported here
venient stations on the block and by is shown in Figure 3(b). This has a

110
simpler transformation to the reference the steel block using a range of de-
co-ordinates and the checks on error coupling rods and clearly demonstrates
sensitivity described above indicated the limitations of a direct connection
that a measurement accuracy of 1% in between the shaker and the structure.
amplitude and 10 Ln phase was usually
s,,fficient to maintain an accuracy of In order to measure and process
',etter than I dB in the mobility of the the data required in this exercise, a
structure computed from equation (10). digital transfer function analyser
(DTFA) was used in conjunction with a
In addition to the basic force small digital computer to form an inte-
transformation from the shaker posi- grated system, described in reference
tions to the reference co-ordinates, it [5]. Measurements were made at dis-
was also considered necessary to allow crete frequencies in a chosen range (as
for the presence of the exciting block opposed to a continuous sweep through
and the transducers mounted on it that range) and the computer was pro-
between the force measuring transducer grammed to select these frequencies,
and the structure. By assuming the control the DTFA while it measured each
block to behave as a rigid mass, it is of the transducer outputs in turn, and
possible to incorporate this correction then store or output these results for
in the transformations described above subsequent processing. The amount of
and the complete equation relating information which ha to be acquired in
measured and required data then becomes: order to derive the required mobility
i is considerably in excess of tlhe actual
[y] = "T11Y , (i)
5 ILD>rj-iMJLTJL¥["-.
- data to be determined for the assembled
structure. In order to compute the
where [Md] is an inertia matrix for the amplitude and phase of one mobility ex-
specific exciting block. pression for the complete assembly, it
is necessary to measure a total of 24x
Consideration of mobilities in more quantities on the components. Thus,
than one direction leads to an appre- while the information sought may be
ciation of some of the limitations of handled or presented in the form of a
conventional impedance testing. In graph, the same is not true of tile
particular, it is noted that if a intermediate data and because of this
structure is not symmetrical, or is not it was decided that the automatic data
excited through its mass centre, there handling facility was an essential Cea-
will often be a significant response in ture of' this measuring technique.
directions other than that in which the 4
forcing is being applied. If this
effect is not properly allowed for, RESULTS
erroneous results can easily Le obtain-
ed. For example, in measuring tle The measurements which were made
mobility at the end of a simple canti- on the individual components - tle
lover beam it is found that there is a steel block on rubber pads and the
significant rotational response to an freely supported beam - were carefully
applied transverse force. In most checked against theoretical predictions
cases, the shaker and its connection of their mobil-ties. Care was taken
are not completely free to accommodate to ensure a high degree of accuracy
this rotation and as a result, they throughout the experiment. The vo lme
impose a restraint on it. This is of the complete set of results (a total
equivalent to applying a couple to the of 18 graphs for tie two components)
end of' the beam iii addition to tie proh ibits their inclusion here. flow-
intended force. This couple is not ever, two of' the nine mobility expres-
measured, but the response which is sions for each component are iIlustra-
measured is the sum of' that generated ted in Figures 5 antd 6, for the block
by the force and the couple and is not and beam respectively. In each case,
the information which is solught. In the two examples shown represent the
order to overcome this difficulty, it best and( the worst correlation between
is generally sufficient to incorporate theory and experiment, and apart from
a 'decoupling rod' between the shaker some di fficulties encountered with si g-
and thre structure which has a low sttff nal noise at the low frequencies
ness in all directions other than that (below 50 liv), tie results are consid-
in which it is desired to apply excita- ered to be sat ], ctory.
tion [I] and accordingly, tiis has been
adopted as standard practice in alI The complete mobility data for Ilh(h
tests of the current study. A series two component s were then used Iii con-
of' measurements were made to indicate junction with equation (1') to predict
tile importance of' using such a device the mobl lity of the as sembled structuire
The results are given in Figure 4i which formed by connecting them together as
shows measurements of ill( response of shown ill Figure 1. Tli s pro(e s

111

I
derives the 3 x 3 mobility matrix for plete and accurate and this require-
the assembled structure and from these ment is not usually satisfied by con-
data the specific information that is ventional impedance testing techniques.
required may be extracAted. In many
cases the properties of greatest inter- At the outset of a theoretical
est are the natural frequencies of the vibration analysis it is necessary to
structure, and these may be convenient- consider which co-ordinates must be in-
ly described by a single mobility cluded in order to satisfactorily de-
expression. fine the motion of the components con-
cerned. Only exceptional cases demand
For the case being studied here, all six co-ordinates at each point of
the direct mobility of the assembly in connection between components, but on
the x direction was computed from the the other hand, very few may be treated
component measurements and .. shown in with only a single co-ordinate. Exact-
Figure 7 alongside the corresponding ly the same considerations must be made
data obtained by measurements made for an analysis which employs experi-
direct on the assembled structure. mentally obtained data. There are
There is an unexplained discrepancy be- many applications in which one or more
tween the two results in the magnitude components may not be analysed directly
of the third natural frequenzy (near because of their complexity and for
90 Hz) but otherwise the agreement is which experimental measurement of mobi-
quite close. Perhaps the most signi- lity data is undertaken. The majority
ficant result, however, is that illus- of mobility measurements are made with
trated in Figure 8 which again shows standard 'impedance testing' techniques
the mobility of the assembled structure and these are often totally inadequate
predicted from the measured component for the application of such data to a
mobilities, but in this case we have mobility coupling exercise. Standard
used the limited data which might be testing is usually confined to a single
obtained from a conventional impedance direction - that normal to the struc-
test. This result is obtained by ture in most cases - while mobility
application of the mobility coupling data in other directions can be of equal
technique but with only a single co- or even greater importance. This fact
ordinate included in the analysis (in is illustrated by the case studied hero
this case, the A co-ordinate). The where exclusion of the rotation co-
errors in the natural frequencies pre- ordinate, , results in a marked deter-
dicted by this over simplified approach ioration in the quality of the pre-
are quite striking, and they clearly dictions (cf. Figures 7 and 8).
Lndicate the need for the more complete
data which has been measured by the While it is essential to include
method described above. as many co-ordinates as are necessary,
it is also important to ensure that the
A final result given in Figure 9 unimportant ones are omitted. Tie
shows the measured mobility of the reason for this is that as tie order of
assembled structure compared with that the matrices increases (by the inclu-
predicted entirely from theoretical sion of more co-ordinates in the analy-
analysis. Such an approach was only sis), so does the sensitivity of tie
possible in this case because of the numerical operations to small errors.
simplicity of the two components. The Thus it is possible to gain no benefit
result illustrates the important point from the inclusion of an extra co-
that even for such simple structures as ordinate (providing this is a relative-
were used in this study, agreement ly unimportant one) without simultan-
between theoretically predicted and ex- cously improving the accuracy of the
porimenta l ly measured characteristics measured quantities.
is not remarkable. In fact, tile data
predicted from the component mobility In the majority of cases, it will
measulements shows better agreement be necessary to measure mobility data
than that predicted from theory. in more than one directlon. The two
main problems encountered in making
these measurements are (i) the applica-
DISCUSSION tion of a couple, and (ii) the applica-
tion of "in-plane folces, to the struc-
We have shown that it is both ture under test. The first of these
possible( and practical to predict the may .be overcome by using a torsional
vibration charactoristic.s of an assem- shaker or by using a twin shaker sy.-
bly of connected components using ex- tem but the secon.l is very difficult to
porimentally measured mobility data. overcome. In some cases, it is
Hlowever, it must be emphasised that possible to excite with a force parallel
such ant approach is only reliable when to the surface but raised above it b>
tile measured dat a , Suff l lent ly coM- an amoilt suffic ient to accommodate the

112
 IA
shaker. This gives rise to an applied satisfactory and that the additional
couple in addition to the desired force complexity involved in multi-directional
and although this may be small, the measurement is essential to a realistic
response due to it could be of the same analysis of typical engineering struc-
order of magnitude (even greater) as tures.
that due to the force. Thus unless
the response to a couple (i.e. the ro-
tational mobility data) is known, this ACKNOWLEDGEMENT
technique must be used with a degree of
caution. The authors wish to thank the Min-
istry of Defence (Admiralty Engineering
One general solution to these dif- Laboratory) for the sponsorship of the
ficulties is the experimental technique contract under which this work was con-
described in this paper, where it is ducted.
accepted that the required mobility
data may not be measured directly but Thanks are also due to Solartron
must be extracted from a series of Ltd. who supplied the computer-
carefully controlled and designed ex- controlled transfer function analyser
periments. By this approach. it is for making the measuremerts.
possible to determine the mobility
matrix foi a component using a single
shaker in conjunction with an 'exciting REFERENCES
block' which is attached to the test
structure, and other standard equipment. 1. D.U. Noiseux and E.B. Meyer,
However experience has shown that care "Applicability of mechanical ad-
must be exercised in the design of the mittance techniq(1ue," Shock and
exciting block in order to maintain an Vibration Bulletin, 38, 1968.
acceptable degree of accuracy in the 2. J.E. Smith, "Measurement of the
final results. It is also an accepted total structural mobility matrix,"
feature of the technique that a large Shock and Vibration Bulletin, I0,
amount of data has to be measured, 1970.
acquired and processed in order to oh- 3. R.E.D. Bishop and D.C. Johnson,
tain relatively little desired informa- The Mechanics of Vibration,
tion. Since it is likely that this Cambridge University Press, 1960.
data is itself required for detailed le. Kerlin and Snowdon, "Driving
analysis (as for the mobility coupling Point Impedance of Cantilever
in this example), then it is conVen- Beams," JASA, V 1.47 N .1(P ?) ivm.
ient to present it il a digital format, 5. A. Martin and C. Ashley, "A rcm-
readily stored on punched tape and puter controlled digital tr%,'- ,r
avallable for further computation. Thus function anal yser and its appiica-
it was decided that an automatic in- tion in automobile testing,"
strumentation system based on a digital Society of Environmental Engineers
computer and digital transfer function Symposium on I)ynamic Testing, 1971.
analyser was neceisary for effic nit
application of the technique.

CONCLUSIONS

It has been shown that vibrat ion

analysis of an assembly of coninected
structures may be made using exporit-
mentally measured mobility data fox the
components. However, th is approach is
only practicable when the measurod
data is sufficiently accurate and com.
plete for the structure in question.
This requirement demnands the meas.ure-
ment of' mobility data in more than osiv
direction for most practical struc-
tures.

An experimental techn:que using a

single shaker has been described for
measuring mobility in up to three
direct ions and this has been tested In
a case study of two Connected compo-
nents. The results of' the exercise
confirm that the exporzment.l method is

113
 > aw

7. CD

LLCL

Mi

42

Ti ~ p A ~ V.~3I .NO

11
 7 V "-7 ' I-7r •
-T • T

IYj
II I

, - ' , *.
. ..;: .
... . . .".
-/ - i .. i.. ..
-

_- r -
" " 4
1 ..
!...
.- 4.

--

-I
SP

.... .. '. .o..11

:(I
. ..... .. ...
.
: ++- -- == :* :
:
I
 .. .. . r 17 ~ l -,

-2

"UNI aSp AL J _OO

~ ~*---~ I i{f116
 VF 7 TpT~vT'iz~zfzzJ~]_

I I

I ____ .~ 1
4
- u

I
:..~

~.~1

~+bI 11ji1>..i
.- *- -

z *-~. .-. . ,. -1-

-1-

i4~ j~ .

I-.-.-

~ .L~
'U -- - - - ,.~---,--- .. - -

I. 777:7

dd
- . 77K ~J*.tI..I.
* ___ 7T *

110
,.-

p
-

a
'1* 0

- -t~z~-- - - - - -- ~------- ----

 coR

II
Sia
U ,.

Cal -17

u• :j
a:: .. .... I __-_-

.. . . . ..... ............ .

V
p.. ..
I:
CLL
-

i-
- - .f. ,ji
i 1 !
Iif'
- r7i 7

'310

,.2i . t. t. , .. "

o o o 0o o

119
........
. ..........-. . ... ... .

... ... .... .....

.....
..... .

I.

I. J

T T.

- . 0. ~: i12
.
 ::7 i. i
IY:: I i

. ... . . .......

I I ... ....

- IE= F-

I '8
z "- '- i , -S-.--
. - - -
: . . ~: .. . .

""1..
..
,4
..
4.
1...I
* ....
,-o .
I

. . . t. . ....... ....... .... ..,... • +.. -..

E
E

IT to t I, 717 TtT

:,,
pT

~~121
 DISCUSSION
Voice: Could you say something about the Mr. Bouche (Endevco): Is there some prac-
time required to compute the measurements tical solution to your noise problems at low fre-
from the raw data? quencies such as using accelerometers with
higher sensitivity or narrow band filtering?
Mr. Ewins: The accelerometers that we
Mr. Ewins: On the full 3 x 3 matrix, we used had a sensitivity of 300 ptcacoulombs per
have not used the PDP-8 for that calculation. g. I do not think we could use more sensitive
The system that you saw in the photograph was ones. The digital transfer function analyzer
the preproduction prototype. There is only one does the filtering in a quasi-digital mauier. It
in existence, and the store that is attached o it handles a very high noise level very well. The
is not sufficient to do the matrix calculation, problem was that the mobility levels we were mea-
We have been using that for running the test, suring were so low at these low frequencies that
for storing the data and putting the data onto they just could not be handled. The main difficulty
paper tape which we feed into our CDC com- on the beam example, where we had rather poor
putcr. The actual computation time in a CDC agreement between the theoretical curve and the
is a matter of seconds - less than a minute. If measured one was that It was in the region of
we were to use the PDP-8, it is not sufficient to the first antiresonance where you have effec-
store all the data. We need an extra magnetic tively zero mobility. It was in getting this anti-
tape or extra disc storage. It would presumably resonance through that we ran into the noise
take somewhat longer, but, using the PDP-8 to problems at about 15 Hz. I think it was proba-
interface with a larger computer, it is a very bly an extreme condition for the equipment to
short time -- on the order of a minute. be tried on.

122
 LIQUID-STRUCTURE COUPLING IN CURVED PIPES - II

L. C. Davidson and D. R. Samsury

Machinery Dynamics Division
Naval Ship Research and Development Center
Annapolis, Maryland

The coupled vibrational characteristics of a pipe assembly

comprised of straight sections and uniform bends arranged in
a nonplane configuration have been analyzed. The results in-
dicate a significant level of coupling between the plane com-
pressional wave in the contained liquid and beam responses of
the pipe. Experiments confirm the general level of coupling
but indicate some difficultyin predicting the fine detail of
frequency response.

BACKGROUND Reference [Ia described a technique for

deriving and a matrix iteration tech-
Liquid transfer systems continue nique for solving the differential equa-
as a major class of machinery recog- tions describing the uniformly curved
nized as a serious noise problem. A liquid filled pipe in the plane of the
principal cause of this noise is the bend (8 degrees of freedom). Experi-
produces a dynamic pressure pulsation. level of liquid-structure coupling.

Because of the relatively low frequen-

cy, this pulsation is transmitted This work has now been extended to
through the liquid as a plane compres- nonplane pipe configurations and, addi-
sional wave. To the pipe, and any tionally, a closed form approximation
other structure encountered, this pres- to the differential equations has been
sure is an exciting force. In a obtained.
straight pipe the structural response
is a radial extension or breathing of The structure shown in Fig. 1 was
tho pipe wall. In the frequency range analyzed and its frequency response, in
of interest, which is well below the the form of mechanical mobility, com-
breathing resonance, this mechanism puted. Of particular interest were the
does not transmit a significant amount transfer mobilities from an acoustic in-
of energy. It must be considered, how- put at the "free" end to structural re-
ever, for its apparent effect on the sponses at the flanged end. Several of
compressibility of the liquid and at- these were measured and the results gen-
tendant change in the sonic velocity. erally confirm the analysis. The experi-
mental procedure necessarily employs
In a curved pipe the change in di- novel measurement techniques.
rection of the plane wave front pro-
duces a transverse force on the pipe MATHEMATICAL ANALYSIS
wall which has been shown to be an ef-
fective coupling mechanism between the The computations performed in this
longitudinal wave in the liquid and work rest on the ability to character-
beam responses of the pipe. ize each component of the structure by

123
I~_
S manipulated to yield the mobility param-
eters of the structure 2,_J

In Fig. 2 weld joints are indicate

by heavy lines. The numbers shown at
the ends of the structure and at each
weldment facilitate identification of
each component and serve as indices for
their transmission matrices.

Ap 4

Fig. 2 -Piping Arrangement

Coordinate System
Fig. 1 - Liquid Filled
Pipe Configuration A right hand coordinate system is
chosen and oriented so that the x-axis
its transmission matrix. The choice is always directed along the center-
of components follows naturally from th line of straight sections or tangent
fabrication of the actual structure, to the centerline of elbows. The z-axis
The structure consists of four straight is always directed toward the origin of
pipes and three 900 elbows as shown the radius of curvature of any
in Figs. 1 and 2. The object of the "straight-elbow-straight" section of the
analysis is to obtain transmission mat- structure and thus always lies in the
rices for these components and with them plane containing a section of this type.
construct the system mathematically. In this structure those sections are
The advantage of formulating the ptob- (3) - (0), (5)- (2)and (7)- (4) . There.-
lem in terms of transmission matrices fore, the right hand triad rotates
is that the complete structure (includ- about the x-axis as one travels from
ing physical boundary restraints) may (0) to (7) and hence the necessity of
be synthesized by multiplication of the rotation matrices as indicated above.
component transmission matrices.in the
appropriate order. Due to the complex- A transmission matrix relates the
ity of the structure's geometry, coor- forces, moments, displacements and ro-
dinate system rotations are necessary tations at one point in a component
to obtain a result describing the do- (or structure) to the same quantities
sired structure. This is done by in- at another point. Stated mathematically
cluding rotation matrices at proper in-
tervals during the synthesis. Finally,
the overall transmission matrix which ZM UZ|g1
results from the synthesis can be

124
or yields a 12 x 12 transmission matrix
which does not include the presence of
the liquid. However, by considering the
t
d dliquid
;:
2 to behave as an elastic beam with
3 (2) no bending or torsional stiffness, a
f tf2 x 2 transmission matrix relating I.
W and P at the ends of the pipe is ob-
tained by analogy to the catalogued
where U is the transmission matrix and matrix for an axially loaded beam.
Z is a state vector. Eq. (2) shows the This is done by replacing E, Young's
relation in partitioned form where modulus, by B, the effective bulk modu-
lus of the fluid [4] . These matrices
U N can then be combined to form a 14 y 14
transmission matrix for the liquid-
V VY filled pipe if the mass of the pipe,
VZ and the liquid, p , are added in
d= -f T (3) those elements of U describing bend-
MYl ing behavior.
1 zj Unfortunately, the 14 x 14 trans-
MP mission matrix for the liquid-filled
elbow in closed form is not as readily
obtained. However, the literature does
The quantities in Eq. (3) are defined provide guidelines. Reference [31 gives
as follows: the transmission matrices for rigid el-
bows with distributed mass and for mass-
u,v,w - displacements along x,y and less, elastic elbows. In the case of
z axes respectively, the massless elastic elbow, the effects
of shear deformation are neglected.
--, *,i- rotations about the x,y and (These effects are accounted for in
z axes. this work.) Neither of these models
is sufficient but they do serve as a
.- displacement of a fluid par- starting point to obtain the desired
ticle along x axis. matrix. The massless model is rejected
for obvious reasons and, although pre-
N,VyVz - normal and shear forces along vious unreported mathematical analysis
x,y and z axes. indicated that large increases in stiff-
ness do not significantly change the
TMyMz - torsion and bending moments behavior of short radius elbows, the
about x,y and z axes. conclusion that infinite stitfness may
therefore be assumed is erroneous. As-
P - force on fluid along x axis suming the elbow to be rigid causes the
(pressure not used to avoid submatrix t 2 in Eq. (2) to be null. Al-
unit inconsistencies) though the actual values of the elements
in t2 are small in comparison to other
It follows that U is a 14 x 14 square elements of U (on the order of 10-6),
matrix and each of the submatrices in their contribution in synthesizing the
Eq. (2) is 7 x 7. For this problem complete system is significant.

i =1,2,....7 Therefore, rather than trying to

amend and append existing forms of U
and U is a rearward transmission mat- for the elbows, a derivation from first
rix 21 principles is used.

Obtaining the transmission matrix Consider the free-body diagram of a

for a straight pipe section filled with differential element of a liquid-filled
liquid is facilitated by [3] , where elbow in Fig. 3. By writing the equi-
transmission matrices for a straight librium equations and the appropriate
beam under axial, bending and torsional elasticity equations that relate forces
loads are catalogued. Combining them and moments to displacements and

125
 Z is defined in Eq. (3), .' is the
14 x 14 matrix of constant coefficients
My~v and s = Ra, R being the radius of curv-
II1Myature
?it~ of the elbow and e- the angle the
elbow turns. The matrix A is shown

Equation (4) is analogous to the

linear, first order different equation

j ft!II-,- ,vf dz =
0 --a
'
S- "whose solution, given Z(O) = ZO, is
P Ud
~ Ii zze Szo
Similarly, the solution to Eq. (4) is

Fig. 3 - Free Body Diagram, Z eAS Z0 (5)

Differential Element
Where Zo is the state vector at S = 0 and
rotations through the elastic proper-
ties of the elbow and which reflect the As (-
presence of the liquid, a set of first e U (6)
order, linear differential equations
with constant coefficients results [3]. is the desired transmission matrix.
This set of equations, 14 in number, Further, since
is shown in Eq. (4).

(4) eas =1 +(as) +

dZ -
ds =A

r0 0 0 00 R0o A'00o oo o v'

Ur -1 0 0 0 -R 0 010 ORW0 0 0 0 1
S00 ooQ 0 0 0 0 /0 0 0 0
o0 0
040 00 00 010 0 0 00%~0 0 3
1 0 0 0 o 00 0
oi0'0 0 0oR 0 t'
0 oouooo- ooo o o- o
d V, -0 oo
0 -t 0o 0-00o 0 o, 0 0_o~
0 o 0o
0 Vs (4a)
N
€y 0 o#W0 0 0 0 010 0 0 0 o 0 0 V
T 0 00 -R'0 00 1 0 0 0 0 0T
My 0O0O 0 IRat0OO0O0RO0OO0O 9,
NM 00 0 0 0 .Pt010 -R 0 4 0 0
0 0 0 0 0 0 o1W
0 0 0 0 0 0 0 a
o-i l oooooo v

126
 it can be shown that the matrix equa- represents based on what the available
tion terms imply. A problem arises here
since each elemental series is of such
((A A 2 complexity as to defy complete identi-
As S) ( s) fication in terms of known functions.
IT FTo ease the situation, only those terms
in each elemental series containing a
single stiffness parameter (E, G or B)
I being a unit matrix, is a valid rep- to the first power in the denominator
resentation of the transmission matrix. (or not at all) are retained. Under
Eq. (7) is used to obtain U for the these conditions, only those terms of
elbow. each elemental series which were readily
identifiable with known functions at
Two approaches to finding U are the outset still remain.
available at this point and both are
used. In the first, the elements of A Futher simplification is made by
can be computed numerically, the powers examining with this second method the
of A found and Eq. (7) formed using cases of a rigid liquid-filled elbow
enough terms in the series to insuze with distributed mass (E,G, and B in-
accuracy. Although A is complex (E,G finite) and a massless elastic elbow.
and B, Young's, shear and bulk modulus In the case of the rigid elbow with
respectively, are given small phase mass, comparison of numerical results
angles to represent damping), Eq. (7) with results obtained by the firjt tech-
converges absolutely [5] and hence it nique outlined above reveals that in
is only necessary to choose a criteria the submatrix t 3 of Eq. (2) the ele-
for accuracy that determines the num- ments correspond exactly with the ex-
ber of terms necessary to be included. ception of the element at row 11, col-
In this work, the criteria chosen is umn 3 of U . This element vanishes
that the elements of the nth partial in the rigid elbow with mass and is
sum in Eq. (7) exceed the correspond- small (in comparison to all other ele-
ing elements of the (n+ 1) term by 106 ments in t 3 ) but nevertheless nonzero
in ratio. For the elbow considered, using the numerical method first pre-
10 terms of the series are necessary sented. By including in this element
to satisfy this requirement. stiffness terms according to the cri-
teria above and leaving all other ele-
This technique has been used to ments in t3 unaltered, this element
find U for an elbow where only de- also compares well (but not exactly).
grees of freedom in the plane of curva- It is concluded, then, that in t3 , with
ture are considered [1], yielding U the single exception noted, the terms
as an 8 x 8 matrix. The method has used to represent U in closed form
been extended in this investigation to correspond to those for a rigid elebow
include all degrees of freedom and with distributed mass. Thdt is, stiff-
yield U as a 14 x 14 matrix. The ness parameters need not appear in t3 ,
difficulty that arises in using this at least in thp frequency range of in-
approach is not accuracy or complexity terest (20-550 Hz) for the elements in-
but economy of time. dicated.

The second method which results For the oiassless elastic elbow
directly from the difficulty cited ( / " and .
I', U.' + . " vanish),
above is to find U in closed form. comparison with the numerical method
The approach is similar to that above first presented reveals that all ele-
in that Eq. (7) is used. Here, however, ments of submatrix t2 compare exactly.
the elements of A are written symboli- Further, in both techniques the elements
cally and the powers of A up to the are constants over the frequency range
tenth term are obtained by performing of interest.. It is concluded that no
the necessary matrix multiplications frequency dependent terms need be re-
by hand. The partial sum is formed tained in t 2 in approximating U in
and each element contained in the re- closed form and those which do appear
sulting matrix is examined to deter- correspond to the elements in t2 for a
mine what function or combination of massless elastic elbow.
functions the elemental series

127

-4
 Finally the two limiting cases since in general V = jWd. In Eq. (10)
indicate to a limited extent the par- the upper and lower left hand submat-
tial form of some of the elements in rices have been multiplied by (-l) since
tI and t4 . Their final form and the fo in transmission notation is the nega-
form of those elements which vanish in tive of fo in robility notation.
these cases are obtained as described
above. The computed results presented in
this report were facilitated by the use
Having obtained the required trans- of a general purpose computer program
mission matrices, the synthesis process (6J which allows convenient matrix mani-
is a simple one. Consider for example pulation and includes a graphing sub-
the interface of straight pipe and el- routine. In all figures shown herein
bow at point 6 in Fig. 2. Continuity the subscript 7 has been interposed
requires that the displacements and ro- to 1 for consistency with available
tations of the elbow at this point be literature (3]. Finally, it is noted
equal to those of the straight pipe. that mobility parameters obtained using
Having adopted the force and moment sign the numerical technique to find U for
convention that positive forces on posi- the elbow compared almost exactly to
tive faces are positive as well as nega- those obtained using the U matrix of
tive forces on negative faces and again the elbow in closed form.
invoking continuity it is seen that
the state vectors of the elbow and EXPERIMENTAL MEASUREMENTS
straight section are equal. Thus, if
Z7S = U76Z6S and Z6E = U65Z5E and since The pipe configuration of Fig. 1
Z6S = Z6E it is correct to write Z7 = was fabricated from standard 4-inch I.D.
U76U65Z5 where S and E have been dropped. seamless steel pipe and 6-inch radius
elbows (900). The structure was filled
This process can be continued until with lubricating oil and tested by meas-
point (0) is reached yielding uring four transfer mobilities. In each
case excitation of the structure was
u
3..I U2a 0 (8)7produced by an airborne sound generator
Z? JU7D6 5 Ur U (8) suspended above the unflanged end of the
pipe and acoustically coupled through
or a rigid cone adaptor. During tests an
air gap of approximately 1/64-inch was
d7 t tmaintained between the cone and pipe.
f4~tt~ffo]
1(9)

3pressure
Two wall mounted microphones in the cone
rovided for detection of the excitation
and also indicated the fre-
quency range over which a plane wave
was present, as shown in Fig. 4.
where the matrices R3 and R4 are rota-
tion matrices ensuring proper orienta- Structural details of the flanged
tion of the coordinate triad as outlined end are shown in Fig. 5. Structural
above and the ford of the state vectors responses at this end were detected with
is defined in Eqs. (2) and (3). The the pair of accelerometers which can be
matrix UL is the transmission matrix of seen on the plate structure in Fig. 1.
the flange and plate assembly shown in As arranged, their sum detects transla-
Fig. (5). tion along the x-axis; their difference,
rotation about the y-axis (1). These
Equation (9) may be transformed to were moved to positions corresponding
a mobility matrix to 9 and 3 o'clock on the flange to
detect rotation about the z-axis ( W).

.j'
t f The fluid response at the flanged
3 4 end was detected with a wall mounted hy-
Ii .0 drophone which can be seen on the under-
V7 t I3 7 sure detector, being very near the "free"
surface effected by the rubber membrane

128
 Li

-- 20

LJJ-40

I I
1K

FREQUENCY, 1-Z

Fig. 4 - Pressure Distribution in Adaptor Cone

(Fig. 5), is loaded by inertial charac- presented in Figs. 7 and 8. From these.
teristics of a short liquid column and, appropriate spring rates were obtained
therefore, serves as an acceleration for use in the computations. These
detector. measurements were subject to a poor sig-
nal to noise ratio which accounts for
For computational purposes the the departure from spring-like behavior
blocking structure was assumed to be at low frequencies.
characterized by a diagonal mobility
matrix. Mobilities in the plane of the Figures 9 and 10 present a compari-
plate structure were assumed tobe ideal son of computed and measured transfer
stiffnesses, dependent only on struc- mobilities for the coupled structure of
tural geometry and material properties. Fig i. The complete 14 x 14 mobility
The mobilities out of the place (trans- matrix was computed but only those param-
lation along the x-axis, rotation about eters involving liquid-structure cou-
the y-and z-axes) were measured with pling and considered significant from a
the apparatus of Smith [7] . This noise transmission viewpoint were meas-
technique involves the use of a spe- ured. In all cases the mean levels of
cially designed exciter, Fig. 6, and computed and measured mobility spectra
required only acceleration response agreed quite well, indicating that the
measurements. For low mobility struc- coupling mechanism was accurately repre-
tures, such as encountered here, no sented. Agreement in detail of the
correction to the measured data is re- rotational mobilities is considered rea-
quired. Two of these measurements are sonable for a structure of this

129
 I

iY 0
:I
TX
#48SWBR4/4 L

Fig. 6 - Mobility Measurement

Apparatus
if rotational response was the problem
since the rotations were not the same
RIS 5r/PPEE1WD in the two planes.

The final frequency response in

this group is the transfer through the
liquid column. This also indicates some
difficulty in establishing the exact de-
Fig. 5 - Physical Restraint and tail of response, but it is not likely
Coordinate System at that the response transducer (hydrophone)
Flanged End is at fault since this measurement tech-
nique has been employed successfully in
complexity. The translational response, a wide range of applications. Rather,
however, shows some rather large dis- it is thought that structural details
crepancies. A possible explanation for such as the bolted flange and welded
the peaks which occur in the 100 to joints, which are not considered in the
200 Hz range of the measured data and analysis, produce these errors.
also for the generally poor agreement
at low frequencies, is that accelerom- The significance of liquid struc-
eter response due to rotation was large ture coupling as a noise transmission
in these regions. However, transla- mechanism can be appreciated by compar-
tional measurements were made with ac- ing the responses at the plate structure
celerometers in both 12-6 and 9-3 due to the liquid input (pressure x
o'clock positions and they were nearly area) and due to a force applied direct-
identical. This would not be the case ly to the free end of the pipe
i 130

/
 _____

- 4 t

T-- -T - ~ -- T
ISOr--1 - ~
-- 12o.-

J 2050 0 6010

loBlckin - Structure

~
(Figure~~~~
isse ~ ~~
htter-i ~ 11) It los hsdpnso hi

spct. Snc all copig 7cur

Fig
74
Meh:re
Roainl oiiy -esue

Roai
be n Mobalit aton

131 ECY z50 o

 I4~-- --

I 10I

0 so 10 FREQUENCY, Hr 10

Fig. 8 -Measured Translational Mobility,

Blocking Structure

132
 +I
-6oC

SM+
LU/ UILMO
4+ 4. 4

<:- - +" -
+ +

i .,oo +
+ ..
4++

C " +,-,t.4.. 4.

. 0
zm
WI L
, I I I i Ii

I+ Fig. 9 -
F,?EQL'AICY, AlE
Transfei Mobilities, Liquid Filled Pipe

13
 MXILI
Q
LIu
V)

A++

z ++
F~4
_ __ ___ __ __

1+

+05 o o 0
F/wUWCH

Fig 10 +Tase oiiieLqi ildPp

E- MU134
 U

CO-40 + +.
+
.I + +* +4

2+
+ +

L -60 +0,,- + Q • + + O O0

us +
Lu 4.4.

+ 4-0
G + +

440

-J
-/0000
0o
%b Z+ 0 + MXIL
ILu0 o 0 00;Zo 0
LL)

Repns u t iui n
zos 410 200500

Fig. 11 - Transfer Mobilities, Relative

Response Due to Liquid and
Structural Excitation

REFERENCES 4. E. J. Waller and L. E. Hove,

"Liquidborne Noise Reduction, Final
1. L. C. Davidson and J. E. Smith, Report of Contract NObs 86437," School
"Liquid-Structure Coupling in Curved of Civil Engineering Research Publica-
Pipes," The Shock and Vibration Bullo- tion No. 8, Oklahoma State University,
tin, No. 40, Part 4, pp. 197-207. February 1963.
2. Sheldon Rubin, "Review of Mechanical 5. Erwin Kreyszig, Advanced Engineer-
Immitance and Transmission Matrix Con- ing Mathematics, pp. 604-606, John
cepts," (Presented at 71st Meeting of Wiley and Sons, Inc., Hew York, 1962.
Acoustic Society of America, Boston,
Mass., 1966). 6. D. R. Jordon, A General Digital
Computer Program, CODAT, for Solving
3. E. C. Pestel and F. A. Leckie, Noise Transmission Problems," MEL R&D
Matrix Methods in Elastomechanics, Rept 548/66, March 1967.
pp. 138-152, McGraw-Hill BookCo., Inc.,
New York, 1963. 7. J. E. Smith, Measurements of the
Total Structural Mobility Matrix,"
The Shock and Vibration Bulletin,
Bulletin No. 40, Part 7, pp. 51-84.

135

____________
 DISCUSSION

Voice: Did the resonance of the fldd in the having a much stiffer system than we should. It
column correspond to what normally would took a little refining of the model to make it
occur? work.
Mr. Samsury: The prediction is on the mark Voice: I think you misunderstood the
as compared to the measurement. What this question.
means is that the model is a good model.
Mr. Samsury: Yes, I did misunderstand
I just wondered if you saw two corn- what you meant. In the coefficient matrix of this
14 x 14 set of equations, the bulk modulus is in-
pliances that
lower than dropped
your the frequencies
calculated points,.ovdia little bit volved in one
n offtelqudtrs
the liquid terms. The
h bulk
ukmdmod-
ulus is determined on the basis of effective bulk
modulus which reflects tMe material of which the
Mr. Samsury: If anything, in generating pipe is made. In other words, yes the compli-
transmission matrices to perform the theoret- ance of the system containing the fluid is in-
ical analysis we had problems with respect to cluded in the sound speed in the fluid.

136
 TRANSPORTATION AND PACKAGING
A SURVEY OF THE TRANSPORTATION SHOCK AND VIBRATION
INPUT TO CARGO

F. E. Ostrem
GENAL AMERICAN RESEARCH DIVISION
GENERAL A1MERICAN TRANSPORTATION CORPORATION
NILES, ILLINOIS

The shock and vibration environment encountered by cargo during

transportation is reviewed. Available data describing the environ-
ment on trucks, railcars, ships and aircraft is summarized. The
vibration environment is described in terms of probability of
occurrence of peak accelerations (within selected frequency bands)
as a function of frequency. Peak acceleration levels, 99.5%, 99%,
98%, and 90% probability levels are presented for particular vehi-
ales covering a wide range of operating conditions. Curves are
presented to show the effect of direction, load, location and speed
on the environment. The shock environment is described in terms of
shock spectra. Shock spectra are presented for typical events en-
countered during transportation, such as trucks crossing railroad
tracks and backing into loading docks, railcars crossing railroad
switches and railcar coupling events.

INTRODUCTION tered by cargo during transportation(l). The

information will be used to develop performance
In order to design an efficient and econom- tests for evaluating containers used in the
ical package or container to safely transport shipment of hazardous materials. This paper
goods and material, detailed information on the summarizes the data on two of the transporta-
transportation environment must be available. tion environmental conditions, namely shock and
Such information is also useful for developing vibration. The data covers shipments by the
tests to assess the ability of new or existing major transportation modes (rail, air, road and
packages and containers to protect or contain water).
the contents during transportation. Designs or
tests based on valid quantitative data can be Although a wealth of data exists describ-
cause for debate only as a result of a change ing the shock and vibration environment for the
in the transportation system or condition. The various modes, only selected data which presents
design or tests based on estimates or judgement the latest in measurement and analysis tech-
are always subject to question. niques is presented in this paper. More detailed
information on the transportation environment is
Although extensive data is available on the presented in Ref. (1).
transportation environment, the information is
widely scattered and fragmented. This is a re- VIBRATION
sult of the approach to the transportation
environment which has generally been to measure The vibration environment, in most studies,
the response of cargo with little concern for is defined as the motion of the surface support-
the actual inputs. As such, only occasionally ing the cargo, i.e., the motion of the cargo
is there relevant data on the environment and floor of the vehicle directly adjacent or be-
this must be sought out in each report. Few neath the cargo. For a container restrained or
programs have been conducted with a view towaro in contact with the floor, this is assumed to
providing basic descriptions of the transporta- be the input to the container. However, even
tion environment. this description of the environment is open to
question. Measurements taken next to the con-
A survey study recently completed for the tainer or cargo can differ from actual load in-
Office of Hazardous Materials of the Department puts. It has been demonstrated in controlled
of Transportation has compiled, reviewed and laboratory tests employing a road simulator
assessed currently available information de- (tethered truck)(2) that although the accelera-
scribing the environmental conditions encoun- tions measured directly on rigid weights

137
(representing the cargo load) were low when com- 10 - 15 Hz 12-1/2 Hz
pared to the unloaded condition at the same 15 - 23 19
location, the accelerations at locations re- 23 - 30 26-1/2
moved from the neighborhood of the load were not 30 - 44 37
affected much by the presence of the load. There- L4 - 63 53-1/2
fore, the cargo load and the proximity of the 63 - 88 75-1/2
transducers will influence the measurements, and 88 - 125 106-1/2
the currently employed method of describing con- 125 - 175 150
tainer input may be very conservative. A more 175 - 238 206-1/2
accurate determination of load inputs would re- 238 - 313 Hz 275-1/2 Hz
quire measurements (with load cells) at the
container/cargo floor interface. Other tech- Results of the analyses are summarized in
niques are required in the event the cargo leaves Fig. I for the loaded trailer. The data have
the cargo floor, e.g., loose cargo, been plotted at the center frequencies of the
filters used in the analyses and present the
In many of the vibration studies conducted, probabilities of peak accelerations being less
only measurements in the vertical direction are than a given value for that frequency band. The
reported, since this direction is assumed or has figure stummarizes measurements made at various
been determined to be the severest. This is locations on the cargo floor in the vertical
justified for general cargo which is loaded onto direction only, since these were determined to
vehicles regardless of directions or labels be governing. It was concluded from this study
(e.g., this side up). The vibrations are assumed that the environment over most roads consists of
to be capable of being applied to any surface of low-level complex vibration upon which are super-
the container and thus only the direction of imposed a great number of repetitive shock pul-
severest vibration environment is measured. ses.
Road Vehicles A question frequently raised in the descrip-
tion of the transportation environment is the
Typical sources of vibration to road vehi- duration of time that the various frequencies are
cles include road surface roughness, engine, excited. This has been determined by evaluating
transmission and drive assembly, wheel unualance, data reported for a typical road condition and
wheel shimmy, and wind gusts. Except for the speed. In the study described above, aaditional
road inputs and occasionally wind gusts, the vi- information concerning the number of peaks coun-
bration due to the other sources are kept low as ted in each bandwidth is given. These peaks are
a result of proper design and maintenance. For presented below for the unloaded tractor-trailer
poorly maintained vehicles operating on the road, travelling on level concrete at a speed of 50
the other zources can exist and present a major miles per hour. The data have been normalized
source of vibration. Unfortunately, in most first to the lowest frequency and then to the
studies of the vibration environment on road frequency having the hi6hest peak count ratio to
vehicles, the vehicles are carefully selected frequency ratio.
and in excellent condition, thereby eliminating P/P P/P1/f/f
many of the potentially severe sources of vibra- Center Peaks
tion. Frequency Counted 1 'P/P
1 /f/1l max

Semi-Trailers - An extensive measurement 1.25 207 1 .64

program of the vibration environment on a flat- 3.75 965 1.56 1.00
bed tractor-trailer is reported in Ref. (3). 7.5 1350 1.09 .70
Measurements were made at various locations on 12.5 2470 1.2 .77
the cargo floor of an unloaded tractor-trailer 19.0 2948 .938 .60
cabination and with cargo consisting of a 26.5 6169 1.41 .90
radioactive materials cask weighing 15 tons. 37 7561 1.23 .79
53.5 10952 1.24 .70
Sixteen different road conditions were en- 75.5
io6.5 992-1
22851 .80
1.28 .51
.8
countered and identified in the study and these
were traversed at different speeds. Representa- 20.5 13041 .3 .24
tive data are presented for each of these con- 27.5 36 .99 .24
ditions in the report. Probability factors were 275.5 24736 .99 .0
developed in the study to account for the var- If it is assumed that a single frequency is ex-
ous speed and be
road type combinations
nnconteed
acros-cunty likely to
tip.cited all the time (corresp~onding to a normal-
ized value of 1.0), the other frequencies are

The data were analyzed with a series of seen to be excited in time ra..ging from 24% to
filters having the following bandwidths: 9Ad of the trip time. Only peaks greater than
0.1 g were usel in these comparisons. The fre-
quency assumed to occur cqntinuously Js near the
B.ndwidth Center Frequency natural frequency of the suspension system, as
0 - 2-1/2 11z 1-1/4 liz would be expected.
2-112 - 5 -,
5 - 10 7-1/2 Tractor-Trailer (Renewed) - A study similar
to the one described above was conducted on a

138
renewed tractor-trailer combination (4). Dta Event Factor
obtained in these tests were to show the effects 1. Backing up to dock 1
of rebuilding and reinforcing the trailer. In 2. Crossing railroad tracks 8
addition to monitoring accelerations, load cells 3. Dip 2
were used to monitor input loads to the caxrgo. 4. Low level t;) high level 2
However, i'strumentation problems rendered the 5. Overpass 1
load cell data unusable. 6. Asphalt road at 50 mph 10
7. Access road 1.5
The neasured data are summarized in Figs. 8. Four-lane highway 1.5
2 to 4. Fig. 2 summarizes all of the data meas- 9. Construction zone 1
ured in the vertical direction for the various lO.Hlacktop at 60 mph 10
locations on the cargo floor, for all of the
road types and vehicle speeds. Weighting fac- It was conc uded from the study that the
tors were employed to account for the probability severest vibratiors occurred in the vertical
of occurrence of the various road speeds and road direction and resulted from driving over pot-
type combinations. Fig, 3 summarizes the same holes and bumps. The location of the cargo on
data for the lateral direction and Fig. 4, the the truck bed has an e.Tect o: the severity of
longitudinal direction. The effect of rebuild- vertical inputs witb cargo located over the
ing a tractor is seen to be a significant reduc- rear wheels getting the roughest ride.
tion in the vibration levels at the high fre-
quencies. This is to be expected, since any Rail Vehicles
looseness in the system would be removed.
Vibrations in railroad cars emanate from
Because of the extent of data available for a variety of sources. Vertic1 -,',rations re-
this particular vehicle, additional plots are sult from the unevenness or rougnness of the
presented to show the effect of various opera- rail, discontinuities at the rail Joints, flat
ting conditions. Fig. 5 shows the effect of spots on the wheels and wheel unbalance. Later-
vehicle speed, Fig. 6 the effect of load, and al vibrations are caused by the tapered wheel
Fig. 7 the effect of location on the environment, treads and the wheel flanges. (The purpose of
Only the peak and ms values of the reported the tapered wheel treads is to keep the car
data are plotted. The effect of direction of trucks centered between the rails while the
medsurement is shown in Fig. 8 by a plot of the flanges of the wheels limit the lateral excur-
90 probability curves of Figs. 2, 3, and 4. sion of the car trucks.) Longitudinal vibre-
The curves show that over most of the frequency tions result from starts, stops, slack run-outs
range investigated: (1) higher speeds result in and run-ins. These latter effects result from
higher levels of accelerations, (2) ar unloaded the inherent slack in each coupler which can
vehicle experiences higher levels of accelera- build up to large values for long trains.
tions than a loaded vehicle, and (3) the aft Characteristic frequencies associated with rail-
location is less severe than the forward loca- car vibrations are described in Refs. (6) and
tion in the vicinity of the fifth wheel. Further, (7).
the peak levels are approximately an order of
magnitude larger than the rs value. Flat Car - Extensive measurements of the
vibration environment on a railroad flat car are
Flatbed Truck (L-1/2 ton) - Extensive meas- reported in Ref. (8). Data are presented for
urements have been made of the vibration envir- measurements in the vertical direction, the
onment on a 2-1/2 ton flatbed truck (5). In lateral dircction and the longitudinal direction.
this study, the data have been separated to show The data has be2en replotted to the same format
the vibration levels under what is considered as the previous data and are presented in Figs.
normal operating conditions and those for ab- 10, ii, and 12. Events included in the data are:
normal condItions. The abnormal conditions train leaving svitching yards, stopping, cross-
included: (1) driving with two wheels on the ing intersecting tracks, climbing a hill, going
shoulder of the road, (2) driving completely on downhill with braking, on level runs at 4U aph,
the shoulder, (3) driving off the road in desert crossing switches, crossing bridges, on rough
brush, (4) driving on the median of a 4-lane track, on curves, and in tunnels. Weighting
highway, and (5) driving on a dirt road. f.umry factors were used to account for the probability
results of this study are presented in Fig. 9 of occurrence of these events when developing the
for the normal conditions. Comparisons indica- summarized data. The test car was part of three
ted little difference between the normal and ab- different train lengths varying in size from 65
normal conditions. Fig. 9 reflects factors to to 120 cars.
account for the frequency of occurrence of the
various conditions investigated. The specific The measurements indicate that there are
weighting factors for normal operating condi- two frequency bands in which the highest ampli-
tions were as follows. tudes occur, the 0-3 Hz and the 5-10 Hz bands.
It is reported that the amplitude distributions
Factors applied to road types for deriving in these bands showed little resemblance to vi-
composite descriptions of environment: bration type distributions. Most of the peaks
in these bands were a result of transient im-
1pulses rather than steady state vibration. Above
10 Hz, the vibration levels were below .72 g In

139
all the requency bands analyzed, the peaks in the water. The hydrodynamic forces include those
the vertical direction were highest. As a re- resulting from slamming, pounding and the wave-
suit of this study,it was concluded that the induced motion of the ship. Slamming is defined
rail environment consists of low level random as the impacting of the ship with the water after
vibration with a number of repetitive transi- the bow has left the water. Pounding is the im-
unts superimposed in the low-frequency ranges. pacting of the waves on the ship when all por-
tiors of the botton are submerged. Wave-irduced
A comparison of the frequency spectra for motion is the motion of the ship in response to
various operating conditions (4) is shown in the waves, excluding those resulting from alam-
Figs. 13 and 14. Fig. 13 shows the effect of ming and pounding. A complete description of
speed on the vibration spectra. Fig. 14 shows vibration sources in ships is given in Ref. (11).
the effect of direction of meastrement for a
particular event. Only the peak and ms values A characteristic vibration fre-ucn.ci as
are plotted on these curves. ated with ships is the blade passage frequency
which results from the non-uniform presaure field
Ai.rcraft acting on the hull as each propeller blade passes
close to the hull.
Vibrations in aireraft result from runway
roughness, propulsion or power plant dynamics, As with other modec of transportation,
unbalance in propellers or rotors, aerodynamic proper design and maintenance reduces the sever-
forces and acoustical pressure fluctuations. ity of many nf the sources of vibration. In
In addition, the surrounding air will also induce addition, operational restrictions may be im-
vibrations due to its turbulent nature. The air posed to reduce the levels of vibration. Fcr
has vertical components of velocity which impart example, ships may be requested to reduce speed
vertical accelerations to the aircraft, in rough water in order to prevent slammi.ig, or
ships may be routed around rough seas.
Characteristic frequencies associated with
the propeller-driven aircraft are the propeller Much of the ship vibration studies have
blade passage frequencies. bee conducted
at various in quiet
propeller water on
speeds. straightandruns
Maneuvers
Helicopter - The results of a recent study crashbacks (sudden reversal of direction) whicl.
of the vibration errvironment on helicopters is generate higher levels of vibration than straight
presented in Ref. (9). The helicopter, an HH43B, runs are also conducted.
has a pair of contrarotating rotors with blades
47 feet long and is powered by a turoo-Jet en- In the past, the vibration levels for rough
gine. The events included in the study were: seas were established by using estimated magnifi-
motor starts, rotor engagement, take-off, hover, cation factors. The factors were based on experi.
climb, cruise 90 knots, straight flight and des- ence and some scattered data and were akplied to
cent. A summary plot of this data is presented vibration levels measured in calm seas (Ref.(1l)).
in Fig. 15. The curve includes measurements
recorded at various locations for the various Dry Cargo Vessel - An extensive measurement
events. Weighting factors are applied to the program of the vibration environment on cargo
data for particular events to account for the vessels is reported in Ref. (12). Data on ex-
frequency of occurrence of the events. Some treme values of load conditions to which cargo
conclusions from this study were: (1) hovering mii-.t be subjected is described. Seven acceler-
produces the severest environment while rotor ometers were installed at various locations
start and engagement produces the least, (2) the aboard a 520-foot dry cargo ship operating in
longitudinal
vironment, anddirection produces the severest
(3) straight or level cruise en- regular North Atlantic service. Data were re-
corded intermittently over a 15-month pericd.

results in insignificant levels when compared to

hover, climb and high speed events. The vibrations recorded aboaed the ship in-
cluded only the seaway-induced motions charac-
to et - Statistical data describing the terized as slamming, whipping and wave-induced
vibration environment on an NC-135, a version of acceleration. It is reported that the wave-
the commercial 707 jet, is reported in Ref. (10). induced frequencies ranged fran 0.030 to 0.205
The severest environment was measured in the Hz, while slaming-induced transients having two
vertical direction and occurred during take-off. basic components of .1.4 and L.6 Hz. The whip-
The data are presented in Fig. 16 in terms of the ping component of the ship induced by slamming
peak values and the probability of peaks being occurred at 1.5 Hz.
less than indicated levels. The data, however,
only applies to one location on the floor of the During the test period, the wave-induced
aircraft. acceleration reached a maximum of 0.88 g (zero
to peak) in the vertical direction at the baw.
Ships Slamming, or the impacting of the ship after it
has left the water, produced higher frequency
Typical sources of vibration in ships it.- accelerations (approximately 10 Hz) in ejcess of
elude the propellers, the propeller shafting, I.> 8,C (.ero to peak). Statistical analyses
the power plants, a~xillary machinery, and the were only conducted on the wave-induced accelera-
hydrodynamic forces as th-. ship passes through tions, since it is reported that sufficient uata

140
were not recorded for slasming events. The suspension van are reported in Ref. (13) and are
results are presented in Fig. 17. The data are shown in Fig. 22. The curves envelope the maxi-
presented in terms of Gri s veruus probability mum shock spectra in each direction for the vari-
of occurrence. It can be seen that une bow ous events encountere& during a cross-country
vertical accelerations were most severe,followed shipment.
closely by the transverse direction and the
stern vertical direction. Fore and aft or longi- Rail - Shocks to cargo in railroad cars result
tudinal accelerations were least severe (apprexi- from: car switching or humping operations, stops,
mately 40% of the bow vertical). The bow (verti- starts, slack take-up in the coupling r,stems,
cal) was also most severe for slam. A technique and transient inputs from the track.
for extrapolating the data to extreme values is
also presented. Based upon the analysis, it is Shock spectra for inputs resulting from
reported that the most probable maximum bow ac- slack run-ins end run-outs are shown in Fig. 23.
celeration on the vessel, operating on the same (4) The curves envelope spectra for a number of
route over a seven year span,would be 2.97 g's events and are separated into longitudinal, ver-
(peak to peak). tical and lateral directions. Run-in and ran-out
SHOCK ahocks result from the inhei'-nt slack which
Shock is defined as a sudden and severe e:;.sts in each co--ler. While "he cleick is small
non-periodic excitation of an object or system. for each coupler, the total slack fo. a long
In most studies, it is defined as the motion of train is large. The accumulated slack produces
the cargo floor or platform on which the cargo a loritudinal whip action when the train moves
or container is supported. Since there is no up or down hills. These shocks are most severe
precise distinction between vibration and shock, for cars near the end of the train.
the data in some instances has not been reported
separately.Shock spectra envelopes for inputs rsulting
from road crossings and switch crossings are
In this paper, shocks are described in shown in Fig. 24. The envelope curves have been
terms of shock spectra when available. (A separated to indicate he levels for the three
shock spectra is the response of a series of directions (longitudiual, vertical and lateral).
single degree-of-freedom systems to the excita-
tion.) This is considered the most descriptive The severest shock loading for railcars
format for the complex transient inputs. It occurs during coupling or humping operation. The
provides a convenient means for comparing a automatic feature of railcar couplers requires
large amount of data describing very complex that they be impacted together to actuate the
motions. It also gives information on the ener- couplers. Due to uncontrollable factors (equip-
gy levels as a function of frequency of the ment and operators) affecting railcar accelera-
shock excitation, tions during make-up of trains at a railroad
yard, there is a wide variation in coupling
Truck - Shock inputs occurring during speeds.
truck transportation include bumping into load-
ing docks, crossing railroad tracks, cattle The shock levels resulting from coupling
guards, and other transient road inputs, impacts are dependent primarily on the car
weighit, impact velocity, and the type of shock
Shock spectra for typical shock events en- absorbing system (draft gear) on the coupler.
countered by a rebuilt tractor-trailer combina- A standard draft gear allows a travel of approxi-
tion (4) are siown in Figs. 18 and 19. Fig. 18 mately four and one-half inches before bottoming.
is for the loaded tractor-trailer crossing rail- Cushioned underframes or cushioning draft gear
road tracks a' 40 mph. Fig. 19 is a spectra have a travel of as much as 30 inches. Thus the
for the semi.trailer traversing a dip in the car can be brought to a stop in a much longer
road at 40 aph. time and therefore with lower deceleration.
Damping factors (C/Co) used in ccmputing Envelopes of shock spectra for common im-
the shock spectra generally range f.tom zero to pact speeds of 2 to 5 mph are shown in Fig. 25.
one. A commonly used ratio is .03 to .05 which (4) urves are presented for the longitudinal,
is the damping in most structurus. Zero damp- vertical and lateral directions. Although 2 to
ing provides an upper limit on the shock spec- 5 mph is a common coupling speed, a frequently
tram while .10 is characteristic of shock iso- used upper limit is 10 or 11 mph. This impact
lation systems. speed includes a high percentage (approximately
98%) of all impacts.
Shock spectra for typical events encoun-
tered by a two and one-half ton flatbed truck Air - Shock excitations occur in aircraft
(4) are shown in Figs. 20 and 21. Fig. 20 as a result of landing impacts, braking, and
applies to the input resulti . from crossing gust loading. Recent studies of the shocks en-
railroad tracks at a slow speed and at 45 mph. countered during landing operations with modern
Fig. 21 is a bhock spectra computed from lori- aircraft have shown that this event produces
tudinal inputs resulting from the truck backing relatively low levels of shock when compared to
into a loading dock. other events. Extensive measurements of aircraft
response to landing impacts have been made with
Shock spectra envelopes for an air ride the NAZA velocity-acceleration-height (VGH)

141
recorder. The measurements are read directly able. Data on the vibration environment defines
from the records and reported in terms of peak the maximum acceleration as a function of fre-
acceleration. Spectral analyses are not per- quency and also the probability that accelera-
formed. Typical results (14) showing the proba- tion peaks are below specified levels. The data
bility of exceeding given accelerations during indicates that only occasionally do the acceler-
landing impacts are presented in Fig. 26. The ation peaks reach the maximum values. Cmpari-
data shows that only once in 1000 landings did sons of maximum peak values with root mean
the normal acceleration reach 1.1 g's. square values indicates an order of magnitude
difference in some instances. Thus extrapola-
Extensive measurement programs of the gust tion from rms to peak value or the reverse can
loading of aircraft have also been conducted, yield erroneous values when the commonly used
However, as with the landing shocks, most of the factors of three or four (sigma) are used.
data has been measured with the VGH recorder and
has therefore been reported in terms of peak Shock data is available in terms of shock
accelerations of the aircraft center of gravity, spectra which facilitates the comparisons of the
Typical data (14) showing the occurrence of gust very complex transient events. Fig. 28 compnares
accelerations per nautical mile are shown in the shock spectra for typical events on trucr
Fig. 27 for a four-engine turbo-jet transport. and railcar. The severest of the shock events
is seen to be that which occurs during coupling
Ship - Shock loadings occurring on ships events. The least sevi-e shocks were those
result from slammings and impacts with piers measured on an air ride suspension van.
during docking operations. In general, both of
these events are controlled to a large degree, The available data on shock and vibrations
resulting in very low shock levels. Data has is generally applicable to transport vehicles
not been reported describing the shocks encoun- which are well maintained. Thus the descrip-
tered during docking. Data on ship slamming tions may not apply to systems which are poorly
are generally included in the description of the maintained and could be used in the shipment of
vibration environment. Ref. (12), for example, general cargo. Further, the data is only appli-
describes a slam event as having a peak of 1.5 cable to cargo which is restrained or always in
g's (zero to peak) with frequency components of contact with the cargo floor. There is a total
ll.4 and 4.6 Hz. lack of data describing the shock and vibration
DISCUSSION environment encountered by unrestrained or
loose cargo. Techniques are required to meas-
Comprehensive descriptions of the trans- ure the shock and vibration environment on loose
portation shock and vibration environment for cargo and to translate the data to laboratory
typical transport vehicles are currently avail- tests.

REFERENCES

1. Ostrem, F. E.; Libovicz, B. A.: A Survey of 8. Gens, M. B.: The Rail Transport Environment.
Environmental Conditions Incident to the The Journal of Environmental Sciences,
Transportation of Materials. GARD Report July/August, 1970, pp. 14-20.
1512-1, Contract DOT-OS-00038, May, 1971. 9. Gens, M. B.: A Preliminary Observation of
2. Clements, E. W.: Measurement and Analysis of the Dynamic Environment of Helicopters.
Acceleration Environments Generated by NRL Institute of Enviroamental Sciences, 1968
Rough Road Simulator. Naval Research Labor- Proceedings, pp. 423-432.
atory, NRL Report 2097, Feb., 1970. 10. Harley, R. A.: Impromptu Vibration Data
3. Foley, J. T.: The Environment Experienced by Acquisition with EL 1-31 Recorder. Insti-
Cargo on a Flatbed Tractor-Trailer Combina- tute of Environmental Sciences, 13th Annual
tion. Sandia Corporation Research Report Technical Meeting Proceedings, Vol. 1,
SC-RR-66-677, Dec., 1966. April 10-12, 1967, pp. 83-93.

4. Anon.: Data Package of 182 Documents from 11. Buchmann, E.: Environmental Vibration on
AEC/DOD Environmental Data Bank. Sandia Naval Ships. Technical Note AVL-244-962,
Laboratory, Albuquerque, Ncw Mexico, Feb., Naval Ship Research and Development Center,
1971. April, 1969.
5. Foley, J. T.: Normal and Abnormal Environ- 12. Bailey, F. C.; Fritch, D. J.; Wise, N. S.:
ments Experienced by Cargo on a Flatbed Acquisition and Analysis of Acceleration
Truck. Sandia Laboratory Development Report Data. Report SSC 159, Ship Structure Com-
SC-DR-67-3003, Feb., 1968. mittee.
6. ilenkovic, V.: Feasibility Study for a 13. Schlue, J. W.; Phelps, W. D.: A New Look
Wheel-Rail Dynamics Research Facility. at Transportation Vibration Statistics.
General American Research Pivision, Dec., Bulletin 37, Part 7 (of 7 parts), Jan.,
1968, DB 182472. 1968, The Shock and Vibration Bulletin.
7. Luebke, R. W.: Investigation of Boxcar Vi- 14. Hunter, P. A.; Fetner, M. W.: An Analysis
of VGH Data Collected from One Type of Four-
brations. DOT, Federal Railroad Adminis- Engine Turbojet Transport Airplane. NASA
traction Report No. FRA-RT-70-26, Aug.,1970. TN D-5601, Jan., 1970.

142
 43 0

to 0

.3 -H1
4. 43

43 43

I'V -0 1. XU O .0 klv33 INI**

@343

9);
_ __ _ _ _ _ _ _ _ _ _ _ _ _.49

I - to - clV333 0 - 6ll~

143'
 141

____________ __________4

3g
4

a- 4-

/1 14

0 .0

4 H4
V V

Ia
r~r-4
to - .LVI0VIVdC k- 2'LV3103

1444
All"

-P -P

9 8 go

-4I

(I -)29 NIV332

2 ..

43

a --.

4)
9 I 9 .

145

'L" x,
 I IJ>
00

.043
I~~~4 S

cc 14

2 CO

'IA

9 10 10

14
0 5. 0 . 0

lynd-1
ixYvDM-OI
to "011 .9 - 10I±V313My

146.
 WiRTICAL.
- - -..LV4TM.
-- -- LATIVAL

L010

II *0

os

000

2 No
FREQUENCY )Iz
Fig. 17 OKO
Ship-acceleration vs probability Fig, 18 - Shock spectra, tractor-semitrailer

1
of occurrence ,0 (rebuilt), crossing railroad tracks
at 4o mph
WI / -

/,../a.I t',"" " I

LAIIAIA
Fig. 5w
17 1- Ship-acceleratinV
-: pracob-etabil r Fig. 18 - Shock spectra, .Utrac2tons fle <
w 10/7/-
oocurne(rebuilt), tk p rcriroarssingsra
ci track

06 .
0 mph
C
141.

1071
0 01 .. II/ 0 1

W 10 10 1/ WO 01 O OI

F. 19-Sokseta rco-eiriei. 20 Soksecr,212tonfate

(rbitdi t4 p trck railodcosW

1471
$I

P- -
. AFTiUWII

( 1.0 '. 1Q

ow
a. a

I 0 1I0 0 I 4

'00 *IIIoI!I 0 ,,Ilr

FREQUENCY - Ht. FREQUENCY " 4

Fig. 21 - Shock spectra, 2-1/2 ton flatbed Fig. 22 - Shock spectra, tractor-trailer
truck backing into loading dock air-ride van

- --- v.sar1Ti1

. . TCRI,
AL 1

-- -- -- -- ------ A--

148 L
S.... w
IV 0

10 to IOD KOD t0 ID KO
FREQUECY '- Ht FREQUENICY Hr

Fig. 23 - Shock spectra, railroad, slack Fig. 24 - Shock spectra, railroad, road cross-
runs ins/outs ings, intersecting track, and switches
148
 0 0 Ow
104

I\\

FROUNC -s. TNi

Fig. 25 -Shock spectra, railroad coupling Fig, 26 - Landing-impact accelerations,

(2-5 mph) turbojet aircraft

0-) - -O

0-- - CONM

0- - mvmv

0--0-

0a -

Fig. 27 -Gust accelerations, turbojet Fig. 28 -Summaryoftpclraprain

aircraft shock events
149
 DISCUSSION

Voice: What kind of instrumentation did Mr. Westine (Southwest Research Institute):
you useto'get this information? When I looked at your curves for different ve-
locities, I wondered if you could Improve results
Mr. Ostrem: This was a review, and we and reduce the data to a single curve by non-
did not make any measurements. Most of the dimensionalizing the accelerations. For example,
data has been developed by the Sandia Corpor- you might multiply acceleration by a character-
ation. Maybe Mark Gens from Sandia would istic length and divide by the velocity squared.
like to comment on the particular type of in- I notice that your axis numbers are nondimen-
strumentation that is used. sional, in essence, and the curves are all simi-
lar in shape. It seems to me that you may be
Mr. Gens: We will put you off until the able to telescope all these different velocity
next paper. The same instrumentation was curves into a single curve. Have you tried this?
Involved.
Mr. Ostrem: No, I have not.
Laboratories):
Mr. Gaynes (Gaynes Testing
First of all, you showed the rebuilt tractor. Mr. Clevenson (NASA Langley Research
Were you able to get comparisons with a new Center): First let me compliment you on a fine
tractor that had not been rebuilt? compilation of quite a bit of data. It is very in-
teresting. I do have a question, however. The
Mr. Ostrem: Not on a new tractor; but the first half of your slides all started at 20 cycles-
first slide presented data on a non-rebulit trac- per-second. Does this imply that there are no

data inorthe of us down

think mostrange
lowI frequency to half a
tor. The firsttraier.
scon wasrebultcycle
one was he a typical tractor-
just so? realize the low

Mr. Gaynes: What about freight cars? Is frequencies are also vitally important.
the difference in car age and rebuilt cars ap-
parent in freight cars too? Mr. Ostrem: Yes, I think that is a good
point. Some of the data did go down to about 2
Mr. Ostrem: Yes, this is one of the biggest Hertz. What we are really lacking is the low
problems. Most of the vehicles are carefully frequency, the so called rolling or rocking fre-
selected and this is the one drawback. It is not quency information. Workers in the field are
the information that would be applicable to gen- just starting to collect data in this area. The
eral cargo where there is no option on the vehi- feeling is that this environment is doing a lot of
cle to be used. damage, particularly for loose cargo. When car-
go Is stacked, many problems can occur due to
Mr. Gaynes: Are you showing higher g for- this low frequency rocking phenomenon. No, I
ces at the higher frequency range as shown above am not Implying that there is no problem below
about 10 cycles per second as compared to where 20 cycles. We are waiting for environmental
we have been working in the past? Investigators to generate information on the low
frequency phenomena.
Mr. Ostrem: Yes, I think that is generally
true. Mr. Monroe (Babcock & Wilcox): Do you
have any data on barge shipments?
Mr. Gaynes: In the past we have worked
with much lower levels. Mr. Ostrem: No, all I can do is refer you
to the one study that was recorded in one of the
Mr. Ostrem: If you are talking about pack- Shock & Vibration Bulletins* and that was in the
aging, I think you are trying to point out that shipment of a Saturn rocket on a tote barge.
most of this is damped out before it reaches the
package. I think this is one of the big problems. Mr. Deitrick (Hughes Aircraft Company):
We do not have any information that would re- Why is not more work done in presenting this
late to the actual cargo. The response of so- data in the form of PSD's. John Schlue of JPL,
called loose cargo Is what we really need to some years ago, put out the only work I have
know and we have no data on that. seen in this respect. By using peaks, you very
*R.W. Fruedell and K. E. Elliott, The Dynamic Environment of the S-IV Stage During
Transportation, Shock and Vibration Bulletin 33, Part IV, p. 111, Naval Research
Laboratory, Washington, D. C., 1964.

150
often wind up with a very conservative evaluation vibration equipment necessary for testing. They
of the transportation environment, are trying to stick with some presentation relat-
Mr. Ostrem: I think the main reason is that ing to peak value and a sinusoidal frequency.
Maybo, for very sophisticated cargo, one would
most testing laboratories, at least for package go to other methods. I think some of the later
evaluation, do not have the sophisticated random papers will discuss that.

151
1*<
 ISE DYNAMIC ENVIROMIDNT OP SELECTED MILITARY IHELICOPTERS*

Hark B. Gens
Sandia Laboratories
Albuquerque, New Mexico 87115

The purpose of the study was to determine the

dynamic input to cargo from the floor of the cargo
space in the OH-6, UH-i, CH-46, and CH-47 helicopters.
The instrumentation, test procedures, data reduction
processes, and results are discussed. The vibration
regime for helicopters is shown to consist of a base
of Gaussian random excitation with superimposed
decaying sinusoids which are associated with rotor
activity.7

INTRODUCTION C. Data should be obtained during lUnd-

ing.
The increased use of helicopters for the
transport of military supplies has caused With these points in mind, a program was under-
interest in the dynamic input of this type of taken to define the dynamic environment of
aircraft to cargo. Since the Environmental helicopter cargo spaces.
Criteria Group of Sandic Laboratories is
charged with the task of supplying descriptions The helicopter differs from other aircraft
of the environmental levels which will be en- in that it is supported by a rotary rather than
countered during the life of AEC/DOD systems, a fixed wing. While airborne, the rotary
it became essential that the dynamic environ- motion tends to exert torque on the fuselage of
ment of cargo during helicopter transport be a the craft with the result that, failing correc-
part of those descriptions. A literature tion, the fuselage has a tendency to rotate
search revealed little information on dynamic also. Two general methods have been employed
input measurements in the cargo space and that to control this unwanted, if not dangerous,
which was available was fragmentary because of motion. One is to provide in the tail area a
specific areas of interest. In view of the vertical propeller which exerts a countertorque
Environmental Criteria aroup's mission and the equivalent to the force exerted by the rotor.
availability of instrumentation, it was decided The second control method is to provide counter-
to measure the environment and produce a torque by means of a contrarotating rotor. The
spectrum of vibration in helicopter cargo latter has the advantage of providing lift and
compartments. drive capability as well as rotary control.
Helicopters employing each of these %.ypes of
Since no firm data were available, the control needed to be considered.
first step taken was to measure the dynamic
input of an aircraft of convenience, even THE HELICOPTERS
though not a likely cargo carrier, to attempt
to define parameters for further work. This In addition to the plane used in the pro-
effort, reportea elsewhere 1], suggested liminery work (the HH43) two examples of each
several areas of inquiry: type were employed to derive data.

A. A specific effort should be made to A. Contrarotating types:

measure low (0-10 Hz) frequency
excitation. 1. CH46

B. The presence or absence of a series 2. CH47

of decaying sinusoids needed to be
investigated. *This work was supported by the U. S. Atomic
Energy Comission.

Preceding page blank

15
 B. Vertical propeller types: E. Climb

1. OC6 F. High cruise

2. UHi G. Docent

The CH47, OH6, and UHI were made available by H. Io flight

the U. S. Army Aviation Board at Ft. Rucker,
Alabama, while the CH46 was made available by 1. Turn
the U. S. Marine Corps. at the Patuxent Naval
Air Station, Maryland. J. Hover

Both the CH46 and CH47 aircraft are In cddition to these, data points of opportuni-
primarily cargo-carriers. They are very ty, such &s maxim=m power take-off, hover at
similar in appearance, both built by the Vertol three feet, at tventy-five feet, and clear air
Division of Boeing Aircraft Company, both turbulence were sampled.
powered by two turboshaft engines. They are
slightly different in size, the CH46 being 84 While helicopters are capable of direL
feet long and 16 feet high with 51 foot diameter take-off in a vertical direction, two factors
rotors while the CH47 .s 99 feet long, 19 feet tend to limit the frequency of such action.
high and has 60 foot rotors. The rotors are First, when operating from a standard airfield
mounted fore and aft and rotate counter to with its attendant traffic, the helicopter must
each other. For the purposes of this work, the move from its Fad into the regular traffic
principal difference between them was that the pattern in order to take-off. Secondly, for
CH47 has an isolated floor in the cargo compart- aircraft with skids, this must be accomplished
ment while the CR46 does not. The CR46 cruises while out of contact with the ground. Often
at 110-120 knots while the faster CH47 cruises the wheeled helicopters do the samo. This the
at 140 knots. The rotor speed of the latter is sequence of events includes liftolf to a height
230 rpm (3.83 rps). of two to three feet, thin moving to the edge
of the runway and finally, .:,--;ccipt of towar
The OH6 is primarily an observation plane clearance, take-off which usually combines lift
but is adaptable to carriage of small cargo 6y and forward notion. For these reasons, the
removal of cabin seats. Manufactured by Hughes, test sequence shown was uried.
it is 30 feet long and 8.5 feet high. The rotor
is 26 feet in diameter. It has a cruise speea INSTRUMENTATION
of 110-120 knots. Rotor speed is 465-514 rpm
(7.75-8.57 rps). It is powered by a single The instruesntation consisted of two
turboshaft engine. clusters of accelerometers. One array was a
triaxial set of piezoelectric accelerometers;
The UH1, built by the Bell Helicopter Co., the other a triaxLal set of piezoresistive
is considered a utility aircraft. It may be accelerometers. Each group was mounted on an
used to carry up to seven passengers or opproxi- aluminum block which in turn was mounted on the
mately 300 pounds of cargo. It is 57 feet long, cargo floor at a structural member. With the
14 feet high, and has a 48-foot rotor. Powered exception of the C1147, the airframe structure
by a single turboshaft engine, it cruises supported the floor directly. On the CH47, a
85-120 knots and has a rotor speed of 294 to 324 structural mambar of the isolated floor was
rpm (4.9-5.4 rps). (The specifications for all chosen. The accelerometers were positioned to
helicopters were taken from the respective be normal to the major axes of the aircraft.
pilot's manuals and Ref. (21.) (See Fig. 1.)

THE TEST PLAN The two types of accelerometers were used

to permit accurate measurement throughout the
The test plan was developed partially as a entire range of interest from DC to 2000 Hz.
result of the preliminary work and partially The piezoelectrics provided information on the
during conversations with the pilots. The higher frequencies while the piezoresistivea
standard data points which were established wsre valuable for lower frequency response.
were: The particular area of interest for the latter
was below 10 Hz.
A. Lift ofi
The recording equipment and powe: supply
B. Low hover iere located in the vicinity of the transducers.
This rovided a loading of approximately 200
C. Take-off pounds on the cargo floor. The recorder used
was the ELI-31 developed by Sandia Laboratories
D. Low cruise (3]. It consists of a 14-channel tape recorder
vith t:,eelectronics necessary to provide

154

iniiix niem glal anl '.'i'~i +iI

i M nm,1i pb . ' . . .
 I
signal modulation and transducer calibration. CONDUCT OF THE TEST
The power supply is a 28 volt DC supplied by
a NiCad battery. (See Fig. 2.) When possible, two operators accompanied
each flight. One acted as a talker, comauni-
The instrumentation was placed as nearly cating with the pilot and to a portable tape
as possible in the same position in each type recorder. The other operated the test recorder
of aircraft. On the OH6, the equipment and made a voice record on the test tape. This
occupied most of the floor of the aft compart- procedure permitted complete documentation.
ment with the seats removed. The floor of the Calibrates were placed on the test tape prior'
aft compartments of the UHi accoodated all to the flight. The pilot had a list of the
of the apparatus with the seats in place. In desired events. He would inform the test
both the CH46 and CH47 there was, of course, a engineers when he was about to execute a
surfeit of room. specific event and upon receiving assurance
that the recorder was ready, would proceed with
the maneuver.

The data were recorded in 20- to 30-second

somples for two reasons: First, the magnetic
tape on the recorder was limited to 7.5 minutes
of record and, secondly, mosat aircraft events
Ire of relatively short duration. For longer
:vents, such as cruise, the dynamic excitation
ft- is essentially stationary, so samples are
typical of the entire ride.
Upon completion of the flight on one
* ,J.. helicopter, the equipment was installed in
another until all had been sampled. The same
.,instruments were used for each flight.

DATA REDUCTION

Data reduction was accomplished in the

following steps:

A. Real-time analog oscillograph record

Fig. I - Typical Accelerometer Installation
B. Bandpass oscillograph record

C. VIBRAN analysis

D VAIL antlysis

The analog oscillograph record waa utilit-

ed to determine:

A. That the data had been recorded.

B. The portions of the data to receive

:JV further consideration.

Selected portions of the data were then

ell "subjected to the bandpass analyela technique,
6 which provide further refinement of the
portions to ba reducer.. In addition, it made it
- possible to begin to see the types of motion in
the lower frequen'y ranges,. We found, for
instance, a very clear trace in the lowest
Op frequency band ,ofan excitation at about 0.3 Hz.

Fig. 2 - Four.een Channel Recorder and Power Sections of the record were selected for
Supply - Typical Inatallation more detailed analysis. After careful examinn-
tion, it was noted that the levels of excitation
were generally higher for the CH46 and CH47
aircraft. It was decided to reduce samples of
each event 'for these, but to select only a few
events froeethe records of the 111 and OH6

155
 4

helicopters. Those selected Included low for presentation because it not only is
flight, climb, high crnuie, and flight through composed largely of steady state vibration,
clear turbule=e. Theee records were reduced with few interdittents, but it tends to respond
by program VUbRA in three anas. The selected to the dictribittion within each frequency band
sazples ware from 5 to 10 seconds In length. more closely than do the peak values. The
single rotor helicopters have low amplitudes
Program VIBPAII [3,4] prod-2css an in the Lover frequencies. None of the ampli-
amplitude distributioa table wiin frequency tuder is =uch above 0.5 g until the frequency
bands ea shmn In FLS. 3. Thic muthod parmits bands beginning with 250 Hz are reached. The
a detailed took at the dietribution of peaks reotest amplitudes are found bet een 500 Hlz
by p2rzent within a frequency band or within a and 700 Hz, where there appears to be vigorous
record. An additional progras VAl [5]. was response in the longitudinal axis. The prdomi-
employed to cabine e VIAMAN record1a into a nance of the longitudinal axis also appears
single record for gach axis of each helicopter beween 180 1z and 350 Hz as well as between
and, finally, into a single record rer anio 700 Hz and 1000 Hz. The transverse axis gains
for all of the helicopters. Program VAII, ascendency beyond 1000 Hz. In general, however,
opertes by daterznln& the rumbar of peaks at the vertical axis shows the highest peak
each miplitude on each record, cozbining them, amplitudes.
presenting the percentage of the total peaks
at each level, and calculating the root Map An analysis of peak distributions within
square for each frequency band and overall for each frequency band shows that 50 percent or
each resultitt record. The report is In the ore of the peaks lie below 0.14 g in the
same form as a VIBRANI record. lower frequencies in the longitudinal and
transverse axes. Thiz condition obtains up to
Further analysi , by frequency bsnds, was 120 Ha and again from 1000 Hz to 1900 Hz. It
undertaken using the method proposed by Curtis is true up to 240 Hz in the transverse axis.
(6]. He has shown the use of Rayleigh probe- In the vertical axis, however, one must go to
bility paper as a means of estiating uhether the 0.27 level before such a statement may be
a peak amplitude distribution is random or made. Nven at that level, in three bands
sinusoidal. Each frequeacy band in the (0-20 Hz, 700-1000 Hz, and 1000-1400 Hz) the
suvoaries for each helicopter and for eli percentage of peaks at or below 0.27 g lies
helicopters was plotted to provide a curve somewhat under 50 percent. The vertical axis
which would indicate the characteristics of shows an absolute peak at 1.9 g from 500 Hz to
the vibration. 140U Hz; the longitudinal one of 2.7 g from
500 Hz to 700 hz. All of the transverse axis
RESULTS lies belw these levels.

The vibration records for the two t)es The presence of excitation caused by
of helicopters were sufficiently different in decaying simnsoids appear to be a characteris-
characteristics as ro require separate discus- tic of the helicopter enviroumant. In the
sion. In general, both showed the presence of tingle rotor types, the distribution of peaks
decaying sausoids, but in widely varying is largely Gaussian random, but at certain
dogree. The overall amplitudes offexcitation frequencies, particularly in the longitudinal
in tht aL3le rotor aircraft were mrkedly and vertical axes, the character of the die-
lowmr than those in the dual rotor planes. The tribution is modified by the sinusoidal
vertical sis generally had higher amplitudv influence. Figure 5 is an illustration of this
in both types, but in at least one fyequency phenomenon. In the longitudinal axis, the
band, the lcngLtudinal axis predominated in sinusoidal presence appears between 120 Hz and
each t)pe. 180 Hz and between 500 Hz and 1000 Hz. In the
vertical axes, It is apparent below 20 Hz,
Figure 4 is a representation of the 99 btwesen 40 Rz and 80 Hrz, betvnen 180 Hz and
percent level of vibration for the aingle rotor 240 Hz, and beLveen 1400 Hz and 1900 Hz.
helicopters. The 99 percent level was chosen

156
 . 0 N9 .4 0AC
N N N .4an .
r 0% 10

N .8 P) P) a .4 0 .9 W%
4
It. . . 1: . .4 It.
Y ey V) t N .0 0) Ca
.4 N NY .4 .4 .0 C1

N A
L4 a .oa N
.4~ ~ N NCC

W~~. % NlU
I
N 8N N 9
P
VN CU -r 0 .9 a
U 4 Ni N .4 .4 CC a,

14 CP IA N .4 . .4 .4 P

4-m . 0 A% .4 a 10 .4C N
I Ny N N 4MI

- 0 $0

N. IA N N N D 4 0 0

00 N. .9 C I N IAI4C
.Z W" N .4 P N 0 0.4 0 . N M
.4i 1 .4
N 0~.4 % 4 . i CS
440 tW P .4N Nf
U -%
2W W4 W .4 P

>W 42 u W .4 N N .4C

02V
I CW .4 N N% on 0 N N
V) 0 C u~

C,
U 0w0 N .9 . IA IA 94 .8
0 .4

4S .4 N

14
CY

o4 e m

or 01.1 P4 .8 P4 .
CL 94w A N P 8 N.0 .

C :,::a C N
Ne N
IA N
94 NN . . 4.a C
C o N N00 A 94 N . 4 . C C C C C 1

00~~~~~~
. 4 . Cz A 9
af 2 -

* 4157
 representative of the CR46. An exception is
NOTS, LONGIU0 ___ the highest band (1400 Hz to 1900 Hz) which is
ECOD
PNiLONCIUDINAL TRAs45V1RS - CH47 data.
21. RANSVERSE
AND VERTICAL
)LtIUD~ ICR. VERTICL
HAVEQUAVAUS Peak distribution analysis reveals that
50 percent or more of the peaks are at or
below 0.14 g in the longitudinal and transverse
axes, except from 2.40 Hz to 700 HE where the
Will -level rises to 0.52 g. In the vertical axis
the 50 percent peak &mplitude level is found
orI at or below 0.37 g except in the 240 He to
E2350 Ht band where it rises to 0.52 g. The
distributional shape is very similar in the
two horizontal axes, but differs in the
0.1 1 ,' 11lvertical.
... . Absolute peaks are in the 350 Hz
IR10tEhCY
Wto 500 Ht band at 5.2 g for all axes.

Fig. 4 - Helicopter Vibration - Single

Rotor 99% Level

Z.0"-' --- "

1.0 APPROXIATft EQUAL

HAVE
LO L I

1.0 v sv1.0

GAuSSIAN
RANDOM -

I!q toIO4Uz)
I1Z I
IHI I.Fig. 6 -Helicopter Vibration - Dual
3 ,AtIT%
<t[RIANDOM[ Rotors 99% LevelIs
210 / I ! The presence of the decaying sinusoidal
excitation is much more noticeable in the dual

, ]rotor type of helicopter. In contrast with

S4 the random vibration with superimposed decaying
_ __I sinusoids discussed above, the dual rotor
_ aircraft seems to have decaying sinusoids with
)C 9 superimposed Gaussian random characteristics
98 g Figure 7 is an example. This trait is particu-
A( ,'01AT1 VIAS ,- larly evident in the vertical axis where it
appears to some extent in each frequency band.
Fig. 5 - Typical Pure and Hixed Gaucsian In the longitudinal and transverse axes it is
Random Peak Dist.ributions found only in the 240 Hizto 700 Hz and the
1400 Hz to 1900 Hz regions. Several other
Figure 6 shows the 99 percent level of frequency bands exhibit traces of the decaying
peak amplitudes for dual rotor helicopters, sinusoidal excitation, but are predominantly
In this type, the lower frequencies show a Gaussian rindom in nature.
much higher level of excitation. All of the
frequency bands show values in excess of I g It would appear on the basis of these
up to 500 Hz. The hLghest levels are found data, in comparison with those of conventional
in the longitudinal and transverse axes aircraft, that the reason for the appearance
between 350 Hz and 500 Hz. It is interesting of the decaying sinusoidal characteristic is
to note that this is the only frequency band the action of the rotors in the rotary wing
in which the vertical axis does not dominate, vehicles. Additionally, it seems that the
The high levels up to 120 Hz are contributed presence of two rotors intensifies the einuo-
by the CH47, while those above that are oidal action to the extent that it predominates

158
in solme frequencies and axes. While it is not amplitudes were under 0.037 g in the vertical
the function of this study to determine causes, and transverse axes, but rose to 0.10 g in the
it might be conjectured that the sweeping of longitudinal axes. Beyond 10 Hs the data
the rotor blade above the fuselage imparts a becomes more regular and assumes a Gaussian
momentary additional lift. Another conjecture random distribution. There is evidence of the
is that the entrance of the blade into the Influence of the decaying sinitsoid here again.
turbulent wash of the preceding one causes this This effect is particularly evident in the
action. At any rate, regardless of reason, vertical axis. Peak amplitude values range
the decaying sinusoid is a major component of from 0.14 g to 0.19 g in the 10-20 Hz bands
the dynamic environment of the helicopter, and from 0.072 g to 0.52 g in the 20-30 Hz
frequency bands. The high values are in the
vertical axis.

:I
ft~CVIN xCA.'.G
1111 SURMOI

- g g Fig. 8 - Stylized Low Frequency Bandpasa

A( tIl
ATI (Y " Oscillograph Record

Fig. 7 - Typical Pure and Mixed Decaying

Sinusoid Peak Distribution

Data on low frequency vibration were

derived from the piezoresistive accelerometers.
They were reduved in the snme manner as that
obtained from the piezoelectric transducers,
but in narrower frequency bands. The bands of
interest to this discussion are 0-5 Hz, 5-10
Hz, 10-20 Hz, and 30-45 Hz.

The bandpass oscillograph records were

particularly valuable in considering the band
of lowest frequency, They revealed the S
presence of very low frequency excitation on
the order of 0.25 to 0.33 H:. Figure 8 is a
representation of the appearance of these data. 4
The low frequency of these peaks permits very [
few of them during a sampling period of
approximately 5 seconds. Indeed, the narrow ow ,,A
band analysis of one record shown in Fig. 9 10, 10o0 20 nOD
reveals that the peaks are discrete and that '(0v

gaps exist between them. It is these discrete

sinusoidel-like vibrations that are so notice- Fig. 9 - Narrow Band Analysis
able to the passenger in the helicopter. Peak
levels In the 0-5 Ht band were as high as 1.4 g
in the vertical snd transverse axes, but only
0.27 g in the longitudinal direction.
Converaely, in the 5-10 Hz bands, the peak

159

0'j
SUIMARY AND CONCLUSIONS

This study consisted of the ieasuremnt [4] J. T. Foley, "ProlL/Ainary Analysis of

of the dynamic environent of the cargo floor Data Obtained in ta Joint Army/AEC/
of two types of military helicopters. The Sandia Test of Truck Transport Envlron-
single rotor type was represented by the OH6 ment," Shock and Vibration Bulletin,
and UHI aircraft. The dual rotor type was go. 35, Part 5, The fihock and Vibration
represented by the CH46 and CR47 helicopters. Center, U. S. Naval Research Laboratory,
Reduction of the data for the complete flights Washington, D. C,, pp. 57-70, February,
of the cargo carriers (dual rotor) and for 1966.
selected events for the single rotor carriers
was accomplished. [5] L. A. Faw, "Program VAIL, User's Manual,"
unpublished manuscript, Sandia Labora-
Analysis of these data revealed several tories, Albuquerque, New Mexico,
a.pects which appear to ba unique to the heli- August, 1970.
copter environment. Among them are:
[6) C. K. Harris and C. E. Credo, Edo.
1. The presence of decaying sinusoids Shock and Vibration Handbook, Vol. 2,
distributed in several frequencies. Chap. 22, McGraw-Hill, New York, 1961.
This type of excitation is often
dominant In the dual rotor craft, but [71 M. B. Gan, "Dynamic Envitcnm nt of
its presence in the eingle rotor Relicopters--Complete Data,"
environment causes a distortion cf SC-M-71-0604, Sandia Laboratories,
the otherwise randomly distributed Albuquerque, Ne Mexico, November, 1971.
vibration. (To be published.)

2. Discrete very law frequency excite-

tion of moderate amplitudes are
present.

3. Vibration amplitudes for the single

rotor helicopters are lower (1.4 g
maximum) than for the dual rotor
types (2.9 g maximum).
4. The vertical axis generally ha& the
greater amplitudes. In the higher
frequencies, however, the longitudi-
nal and transverse azes have
amplitudes which approach those of
the vertical.

While, in general, the dynamic inputs to

cargo by the helicopters considered herein are
not excessively high in relation to other
modes of transport, the presence of relatively
high amplitude, low frequency excitation is
vorthy of consideration during design of
packages and tiedowns for helicopter transport.

BIBLIOGRAPHY

(11 M. B. Gens, "A Preliminary Observation of

the Dynamic Environment of Helicopters,"
Proceedings, 14th Annual Technical Mst-
ing, Institute of Environmental Sciences,
April, 1968, St. Louis, Missouri.

[21 "U. S. Rotary-Wing Aircraft," Aviation

Week and Space Technology, Vol. 94,
No. 10, p. 76, March 8, 1971.

(31 J. T. Foley, "An Environmental Research

Study," Proceedings. 13th Annual Techni-
cal Monting, Institute of Environmental
Sciences, pp. 363-373, April, 1267.

160
 DISCUSSION
Mr.Hughes (Navtal Weapons Evaluation Mr. Earls (Wright-Patterson AFB): I un-
Facility): Did you try to correlate the frequency derstood you did your analysis on an octave
a-e- occurrences of the decaying sinusoid basis and the peaks, for example, in the 500 and
with the beat frequency of the rotors? 700 Hertz range, were around 5 g's. What is
the implication on.specs for cargo? Should we
MY, u h rtest to 5 g's, or should it be done on an average
basis? What about the peaks that you would
quency lies somewhere between 4 and 8 Hertz have if you analyzed on a narrow band basis?
which has to be multiplied by 4 because four What is the implication on specifications for
blades are involved. We seem to find the har- cargo?
monics rather than the basic frequency. The
decaying sinusoid is not quite as prevalent at Mr. Gens: Expecting this question was one
that frequency, but often at a multiple of it. reason that I mentioned we are purists, and

Mr. Hughes: But it does come from the therefore, not very good at the testing angle.
r rdoes it not? That is one reason we like to use a display like
rotor, o no the VAIL VIBRAN display which I showed, or
something like Mr. Ostrem showed. With these
iVI-. Gens: It appears to come from the data presentations, the design and test people
rotors. We have not attempted to find the can use their judgement in choosing the level.
cause, but we guessed one of two things - Obviously a level of 5.2 g's peak would be ex-
either as the rotor passes over the fuselage it tremely conservative when we are looking at a
adds some additional lift, or perhaps it is en- little oi,,r one g at the 99 per cent level. This
countering the wake of the preceding rotor is my thought about it. Is this the sort of an-
which causes a perturbation of some type. Yes, swer you were looking for, sir? I hope I did
we may be all wrongI Helicopter manufac- not tell you anything, because I do not want to
turers can leap down my throat. commit myself.

161

Lj
 HIGHWAY SHOCK INDEX (U)

Robert Kennedy
U. S. Army Transportation Engineering Agency
Military Traffic Management and Terminal Service
Newport News, Virginia

The Army, Navy, Air Force, and Marine Corps have jointly
sponsored and participated in the development of a Shock
Index (SI) for highway transportation. A numerical SI,
associated with a particular vehicle-load combination,
can now be determined at a low cost by application of
simple static field measurements. The SI provides clas-
sification for vehicle-load combination as regards prob-
ability of shocks transmitted to the cargo during highway
shipments.

It has long been recognized that reasonab- are unimportant vehicle differences in shock at-
ly accurate estimates of shock transmitted to tenuation. Several years back at the initiation
the cargo during highway ransit can be made by of this work, it was decided to try to obtain
a combination of experience and intuition. One physical measurements of static vehicle charac-
experienced in conducting test runs and observ- teristics contribution to cargo shocks to facil-
ing accelerometer readings can usually predict itate more organized and reproducible shock es-
either a rough ride oz a smooth ride for a par- timation. The concept of shock index, then as
ticular highway vehicle with a known loading now, is basically to measure static vehicle
configuration. Basically, the four most impor- characteristics, run the measured vehicles over
tant cues used either intentionally or acciden- controlled courses, and mathematically fit the
tally by estimators are the capacity of the static measurements to the test shock measure-
truck tractor, the relative amount of cargo, the ments. The resulting formula for shock index
usable suspension and tire deflections, and the is then tested with instrumented highway tests
load position. Other factors as trailer dynamic to determine the degree of accuracy of the em-
beam deflection, speed, and condition of the pirically developed formula or perfvrmance pre-
highway pavement contribute to cargo shocks, but diction process. The results of these efforts
to date have produced the following formula:

4
5+ KLKS + KL 2- KS2
si -,. max rated
I [45max grosswtw0]
rated net 0-' 4 2 0.53] [log pct rated load- 2.25]
KLKS +4K j

+ (M+N) (P) + (S+T) (U) + 4.92

(F+C) (P) + (I+J) (U)

The maximum rated net weight is (A+B) - (C+D) and the maximum rated gross weight is (A+B).
The symbols in the equation are defined as follows:

Preceding page blank

163

___ _ _ _
 A - Combined front weight - rated load at any position.

B - Combined rear weight - rated load at same position as "A".

C - Combined front weight - no load.

D - Combined rear weight - no load.

K - Greatest combined suspension spring rate - front or rear.

L
KS - Least combined suspension spring rate front or rear.
F - Combined front suspension deflection - rated load located forward.

G - Combined front tire deflection - rated load location forward.

I - Combined rear suspension deflection - rated load located rear.

J - Combined rear tire deflection - rated load located rear.

M - Combined front spring deflection - rated load at test position.

N - Combined front tire deflection - rated load at test position.

P - Combined front weight - rated load at test position.

Combined
- rear suspension deflection - rated load at test position.I

T - Combined rear tire deflection - rated load at test position.

U - Combined rear weight - rated load at test position.

At first look the SI formula appesrs some- 5.0-

where between awesome and gross. Further curve
fitting should produce a more simple formula
with accuracy consistent with data. The formula
assigns numbers and coefficients to what is
known about factors that determine truck-trailer
ride performance. If a truck is loaded to near
capacity, if the truck has a usable long-travel 4
suspension, and if the load is positioned to
work all springs, the best ride for the cargo
will result. The other direction on any of the
above factors will produce a rougher ride. The
other end of the scale is a truck loaded with .
10 percent or less capacity, with only a small
portion of stiff suspension springs working,
with high pressure tires, and with the load lo-
cated directly over the rear axle. This ar-
rangement produces the roughest ride and conse- TEST
CONDITIONS
quent high loss and damage to the cargo.
1. TANDEM AXLE TRAILER
The most significant factor affecting the
2. GROSS L11 WEIGHT
70,000 VEHICLE
SI for a particular vehicle is the percent of 3 CARGO CENTER GRAVITY
rated load. Figure I is a plot showing how the ATMIOSPAN
shock index changes with percent rated load. It 4. RATEDTIREPRESSURE
is apparent that most highway vehicles perform I11 11
well as regard shock and vibration when loaded 1.0
to near capacity. Most standard highway vehi- 10 20 30 40 55 0 70 so 100
go
cles should range from SI 2.0 to SI 5.0 with PERCENT
RATED LOAD
changes in the weight of cargo. The SI should
aid in communicating to those responsible for
establishing the percent rated load in practice Fig. I - Vehicle performance for
the great sensitivity of the relative weight percent rated load
on the cargo bed to the shocks transmitted to
the cargo.

164
 Highway vehicle mechanics have resulted in Not to belabor the gasoline analogy, but
cumbersome and complicated practical applica- with an indicated octane rating for a gasoline,
tions. The essential elements as masses, all from automotive engineer to hot-rod hobby-
springs, capacities, and load positions are ist benefit from the classification. Octane
quite understandable. The approach used to the rating has done considerable to improve communi-
problem has been to match up the understandable cations. The empirical, simplified approach
elements to the results sought. The factors in has aided in pursuit of more precise, techno-
the formula have a mechanical basis. However, logical characteristics for gasoline perform-
the goal was to obtain a formula of static ance. When one starts to investigate the need
measurements that works without too much con- for a shock index, one finds out that just about
cern for the theory involved. The formula is everyone in the field of transportation has the
intended to give a shock index rating which need for a term that will roughly indicate the
would be similar to and as useful as octane rat- degree of shock and vibration severity to the
ings for gasoline or tire ratings for automotive cargo.
tires.
Briefly, the origin, generation, and devel-
In much of the work today the question opment of the highway shock index came about
"who needs it?" requires immediate and prompt over the last 4 years. The Army, Navy, Air
attention. Traffic managers and shippers today Force, and Marine Corps entered into an inter-
rent or purchase highway equipment with the in- departmental agreement to pursue and sponsor
tent of providing minimum shocks and vibrations this type of work Jointly for all modes of
to the cargo. At present, the only criteria the transportation. The steering group, consisting
traffic manager can use are the manufacturers' of Mr. J. Pellant, Navy; Mr. H. Leonard, Air
advertisements and the cost of the equipment. Force; Mr. T. Quast, Marine Corps; and Mr. R.
Other than for particular and fragile cargo, the Kennedy, Army, formulated the basic concepts for
traffic manager must choose transportation shock index during series of work sessions.
equipment ignorant as to its ride-cushioning General Testing, Incorporated, was awarded a
ability. There is no documented correlation be- contract to perform the necessary test runs and
tween the cost for rental of the equipment and mathematically fit the data to the shock index
the quality of the ride given to the cargo. For equation. J. A. Johnson, Incorporated, was
many years, automotive tires have been purchased awarded a contract both to verify and determine
by the consumer on the same basis of trust or the degree of accuracy and the range of appli-
ignorance. The only criteria available to the cability for the shock index formula.
small consumer for purchase of gasoline have
been the cost. Recent developments have made it It was realized early in the work that in
posslble for the consumer to purchase gasoline order to obtain reasonable acceptance the en-
or tires based on a general empirical rating. tire scientific community need participate and
This probably does little more than indicate approve, from concept to development to check-
comparative worth between the other items avail- out. An advisory group was set up which com-
able; however, it is of great value to the con- prised representatives from the National Bureau
sumer. Traffic management people have indicated of Standards; National Academy of Sciences; De-
a need for a factor as the shock index that will partmept of Transportation; Aerospace Indus-
give them a comparative value between the vari- tries, Incorporated; and NASA. This group has
ous trucks available, especially for rental type been instrumental in the development of the
transportation. mathematical concepts, engineering concepts, and
all scientific considerations.
Designers also have a need for a quantity
term shock index that will enable them to set The project proceeded generally in the fol-
goals for shock attenuation. Also, when a par- lowing manner: mathematical equations for shock
tic'ilarly good shock attenuating vehicle design index were developed; sensitivity analyses were
has been ochieved by highway vehicle design or performed on the various equations to establish
development, a single factor term to describe the number of pertinent variables. All of the
this improvement is needed for proper promotion. work was set up to limit tie shocks of interest
to frequencies below 60 cycles per second; to a
The third big need exists in communication threshold of Ig, or 10 percent of the maximum
in this area. Of the many tests and monitored rating, whichever is the greatest; and to the
instrumented movements, it has been difficult to 95-percentile shock amplitude.
communicate in general terms the relitive merits
of various highway systems. Shock and vibration It is realized that many cargoes will re-
people know the relative work by studying varn- spond to frequencies higher than 60 cycles per
ous curves, spectra, and tables. However, these second; also that a threshold of Ig may not be
data are Indeed hard to communicate because of adequate to cover shocks for extremely fragile
their great depth and the specialized knowledge Largo; and lastly, that eliminating the top 5-
required for understanding. percent high shocks may be insufficient for

165
those cargoes that experience the damaging out during the verification phase. All the ia-
shocks that the formula ignores. The basic ap- dications to date are that the general approach
proach, however, was to loosen up the require- has considerable merit and that more factors
ments as far as possible to see if the shock should be brought into the shock index consider-
index still had a justifiable value. It turned ations to increase its range of applicability.
out that even with the loose tolerances the Most of the test runs were made with the pave-
shock index is of great value to the field to- ment indexes recorded for future correlation
day. Further work will be accomplished to make with the resulting shocks and vibrations.
the shock index applicable for higher frequen-
cies, lower threshold, and a higher range of The vehicle characteristics are altered
extreme values. for each test, and, in most tests, the charac-
teristics are the extremes for rapid curve de-
A series of trucks covering as broad a velopment. The most pressing need for progress-
range as possible for the trucks that are in- Ing in this work is to establish mathematical
tended basically for transporting cargo were and engineering relationships between the vehi-
selected and static measurements made for each cle shock index and the pavement index.
truck and load configuration. The trucks were
then run over a standard short road course that This work was only performed because a
had well-defined impediments or shock genera- need existed. It is planned to implement this
tors. A relationship between the 9J-percentile shock index work by using the shock index con-
maximum shock measured from the test runs and cept and formula in operations. It is antici-
the shock index factor was developed. The pated that traffic managers will specify shock
shock index was set to rate from 1.0 for a very indexes for particular cargoes that need
poor riding truck,producing high shock to the better-than-average shock attenuation during
cargo, to 5.0 for the best available truck, highway movement. This use of the shock index
This range of numbers was selected to match the will consider such factors as cost of packaging,
performance serviceability index which indi- price of the item, and military importance of
cates general condition of highway pavements; the item. It is intended to start with the
this also runs from 1.0 for the worst, to 5.0 most important items first for specification of
for the best. the shock index and broaden out to cover the
majority of cargo within the next few years.
After the formula was fitted to agree with Shock index as part of the specification for
the results of the shocks measured on the pre- procured vehicles will be introduced during the
calibrated road course, it was evaluated by sep- planning stage. For evaluation of various vehi-
arate contract over public roads using actual cles that are still in the design and prototype
commercial equipment. The shock index valida- stages, shock index will be calculated and used
tion test included trucks with SI ranging from as a basis of worth. Manufacturers will be en-
approximately 2.0 to SI approximately 4.5. The couraged to use the shock index as part of their
amounts of load or cargo were varied, the load specifications where applicable.
placement was varied, and various size trucks
were used. During the validation contract all The highway shock index is intended prima-
test runs were conducted at maximum legal speed rily for the transportation users, and it is
and at one-third and two-t'irds maximum legal predicted that in this area the shock index will
highway speeds. Tests were also run over both find its greatest acceptance. Small users and
flexible and rigid pavements. illwork to date consumers that do not have access to shock and
on the shock index has been entirely successful. vibration specialists or to en engineering de-
The basic concepts checked out during the devel- partment will have a guide that will indicate to
opment phase, and the development work checked some degree what they are getting in shock and
vibration control for their money.

166

I4
 DISCUSSION

Mr. Schell (Shock and Vibration Information manufacturers want to know what their snock in-
Center-: Are you aware that there is another SI dex is, because if it is low, they will go about
Index in the transportation field which is an in- raising it. It does not take much work on a truck
dex of shock severity ? This is n the Department to improve its springs and damping to the point
of Transportation Motor Vehicle Safety Stan- that it has a good Shock Index.
dards. It is the time integral of a weighted ac-
celeration. It is called the "Severity Index", Mr. Bacile (Martin Marietta): I guess I
and the symbol SI is used. It is an Index of need to be led by the hand. Let us assume I
shock severity on the human being or occupant have some cargo that I want shipped. Let us
of an automobile. r also assume that I have already defined the fra-
SI =J an dt gility envelope for this cargo. I want to ship it
from here to there. How do I tie in this fragil-
Mr. Kennedy: We had a DOT representa- ity envelope to the shock index? How do I use
tive on the advisory committee, and with the this protective tool?
crowd you are talking about and the crowd we
are talking about we do not think they are going
to get mixed up, because most of our people are Mr. Kennedy: The traffic manager would
not going to be worried about the first integral specify, he tells the shipping company that he
or the second. They would not really know what wants a truck with a shock index of 4.5 or bet-
it is. ter for this particular cargo.

Mr. Gaynes (Gaynes Testing Laboratories): Mr. Bacile: How do I get from the fragility
Are you not getting involved slightly with a envelope to the shock index?
legal problem in relation to a gentlemen picking
out a certain shock level truck and then finding Mr. Kennedy: If you refer to a sophisticated
out that he had bad damage? Is there a liability missile component or something like that, shock
involved here? index would not come into play at all. But if you
mean troop support cargo such as rations, small
Mr. Kennedy: 1 do not know about a liabil- ammunition, and that sort of thing, we have quite
ity - of course there are legal implications, a record of experience in shipping such items.
but in our business there is a big legal depart- As shippers, we at least know the things that
ment to handle those problems. You are correct, break easily and those that never break. We
but this method will involve rate-making, and know the things that are expensive and those that
there will be interstate commerce problems, are not expensive. We know the things that are
and we have lawyers that work on this sort of well packaged and those that are not. Traffic
thing all the time. managers know all these facts. The only thing
he does not know is which truck Is good and
Mr. Gaynes: But you are setting up a defi- which Is bad. If he has fragile, expensive cargo,
nite or finite number in relation to a shock level poorly packaged, he will ask for a good truck.
which, if it is exceeded, supposedly you could be You have to pay more to get a highly rated truck,
held responsible. one with a high Shock Index. Most of our very
fragile items, such as warheads, are closely
Mr. Kennedy: As I mentioned, on household studied to determine accurate shock and vibra-
goods containers, we did almost the same thing. tion fragility levels. If we put them on a light
We had several thousand manufacturers to whom truck, the accelerations could be a factor of three
we sent a performance standard. Anybody that to five times higher than they would get on a
could equal or better the standard requirements heavy truck. We are trying to get away from this
would get preferred treatment by the traffic man- kind of thing because we who have the technical
agers. We got a little rumble at first. Put they ability do not get an opportunity to order trucks.
are not being put out of business by such a re- We have to instruct the traffic manager, the trans-
quirement. Their services are simply not se- portation officers and that type of personnel so
lected, so they quickly come up to the standard that they can order a truck. That Is why the
requirements. As a matter of fact, the truck Shock Index was developed.

167
 DEVELOPMENT OF A ROUGH ROAD SIMULATOR AND SPECIFICATION FOR
TESTING OF EQUIPMENT TRANSPORTED IN WHEELED VEHICLES
I
Harold M. Forkois and Edward W. Clements
Naval Research Laboratory
Washington, D.C.

A Rough-Road Simulator machine has been constructed and placed in

successful operation. The simulator employs rotating drums with detach-
able road profiles, or bumps, which are interfaced with a wheeled vehicle
to provide a random shock and vibration environment. This machine was
the experimental device employed to simulate field conditions in the labo-
ratory for the acquisition of shock and vibration data. The analysis of ac-
celeration records of the response of the wheeled vehicle platform, both
unloaded and with incremental loads, provided a description of the simulated
environment, or what alternatively may be described as a calibration of
the performance of the rough-road simulator machine. A proposed draft
of a Wheeled-Vehicle Rough-Road Transportation Test Specification is
presented, which is related to the use of this machine.

INTRODUCTION DESCRIPTION OF THE ROUGH

ROAD SIMULATOR
A definite need has existed for many years
for the development of an adequate and realistic This machine consists of a truck, trailer,
method for laboratory shock and vibration test- or other similar vehicle, which has its rear
ing of Navy and Marine Corps electronic equip- wheels supported by steel drums onto which
ment and missiles transported in conventional suitable bumps can be attached (Fig. 1). The
wheeled vehicles. Current specifications for upper surfaces of the drums are at floor level,
ground eqp..ent require separate and distinct and all associated equipment is mounted in a
vibration and shock tests using conventional pit with a depth of 5 ft below floor level. The
testing machines which do not satisfy adequately pit is covered with steel gratings to maintain
the requirements of this ordinary but complex floor continuity. The steel di urns, which have
environment [1-2]. It is agreed, generally, that a diameter of 33-1/2 in., are driven by V-belt
the most sophisticated method of simulating drives powered by two hydraulic motors. At
this shock-excited environment within a labo- 1500 psi operating pressure the maximum aver-
ratory building would be to employ an appro- age speed of the drums is 380 rpm correspond-
priate number of hydraulic actuators which ing to hydraulic motor speeds of 1265 rpm, es-
would be programmed to provide the required timated horsepower of 28.5 per motor, and
six-degree-of-freedom motions to a suitable speedometer reading of approximately 38 mph
structure or platform. The cost of such corn- for a 2-1/2-ton 6 X 6 military truck. The speed
plex test facilities for general use is considered controls of each set of drums on each side are
prohibitive when compared with a substantial independently adjustable. Enough slippage oc-
investment decrease effected by utilization of a curs in the belt drive between the fore and aft
recently constructed type of testing machine drums so that a random and variable phase
which formed the basis for the development of relation exists between them. The vehicle is
the proposed test specification. This machine constrained laterally by one large turnbuckle
has been described previously in recent publi- hooked to the frame of the truck at the center
cations under the descriptive title "Rough Road location of the rear spring, and longitudinally by
Simulator for Wheeled Vehicles," prior to pub- a hook arrangement attached to the front
lication of the experimental results 13-61. bumper. Adjustable steel brackets, bolted to
These results together with the proposed rails anchored in the concrete floor, provide
specification appear in Ref. [7-81. the necessary end restraints. When a smaller
Preceding page blank 169

__ _ i
 -,- -'

* =, •. .

I'. .4"'i"; i '...1 -"

Fig. 1- Rough-road simulator for wheeled vehicles (general arrangement)

vehicle is utilized, such as a jeep, its rear tension varies and when the tension dccreaseb
wheels are located on top of the forward drums, slipping of the belts on the pulleys o,-curs which,
and manual disengagement of the clutches pro- in turn, causes the phase changes n bump ex-
vided on the shafts of the aft drums renders citation. Building disturbances have been small
them inoperative, excitation to the vehicle then even though the simulator is in close proximity
being provided by the front drums only. The to wall and column footings.
forward pair of drums are adjustable in the
longitudinal direction to accommodate vehicles,
of different wheel bases. The same lateral and Description of Hydraulic Drive
longitudinal restraint arrangement as described
for the truck is utilized. Heavy vehicles can be Figure 3 is a schematic drawing of the ele-
driven by the simulator at speeds up tc 40 mph. ments utilized in the drive and control system.
Greater speeds are possible for powered ve- A 100-hp constant speed electric motor drives
hicles if the engines are used to boost the driv- two constant-volume fluid pumps which, in turn,
ing power of the hydraulic motors. Figure 2 is energize the two fluid motors. Speed control is
an overall view of the installation, achieved by the "on-off" operation of two pilot-
operated relief valves (one for each fluid mo-
tor), with maximum pressure adjustment of
Ground Shock Isolation 2000 psi. The actuating pilot valves are sole-
noid operated, and in turn the solenoids are
To mitigate ground shock effects, the steel actuated by the "on-off" operation of two inde-
drum-foundation brackets are bolted to a sepa- pendent centrifugal switches located on the con-
rate reinforced concrete block which rests on trol panel. The centrifugal switches are driven
the soil directly. The block Is approximately by two independent flexible cables attached to
10 x 8 ft on its base and 3 ft deep, extending the motor pulleys. Two tachometer generators,
below the bottom of the pit floor level. It driven by the rear drums, provide signals for
weighs about 20 tons. A 1-in. thick preniolded two panel meters for rpm speed indication of
joint filler separates this foundation from the the drums. Also. two panel gages indicate in-
concrete floor of the pit. In addition, resilient stantaneous pressures in each drive system.
pads are Interposed between the bearing pedes- Accumulators, pressurized to 500 psi initially,
tal and pedestal brackets. The resilience of are provided to smooth out the speed control.
these pads under dynamic loading causes rela- Speed regulation is in the order of 5-10 percenlt
tive motions between the drum shaft pulleys, depending on the operating speed and vehicle
Because of these relative motions, the V-belt loading.

170
 -Ig. 2 - Overall view of simulator installation

fication simulator machine was generated by the truck is used to supplement the hydraulic

equipping each of the four drums with two

bumps with maximum eccentricities of 3/14 in.
and 1-1/2 in. They are spaced 180 degrees
apart on the drums. The contacting surfaces of
motors. The use of these road profiles was not
considered feasible especially with the small
diameter drums provided.
4
the bumps are circular in form, and for the
chord length used, are nearly the same shape
as a half sinusoid. One bump has an external
radius of 10-3/4 in. and a chord length of 12 in.
The other has an external radius of 8-5/32 in.
and a chord length of 12 in. The bumps are
aluminum castings made of alloy 356-T6.
WIIEELED-VEHICLE SHOCK
ENVIRONMENT

In determining the accelerations trans-

mitted to equipment itemrs transported in a
wheeled vehicle, there are three separate phe-
nomena which must be considered 1101. The
I
21
Three other road profiles were provided, first is the forced vibration of the vehicle body
These profiles are similar to three vehicle on its springs usually associated with low speed
courses located at the Munson Test Area, (5-15 mph). This phenomenon corresponds to a
Aberdeen Proving Ground, Md. and are identi- forced low frequency vibration with displace- 2
fled as the Six-Inch Washboard Couroe, Radial ment amplitudes dependent on the spacing and
Washboard Course, and Three-Inch Variable heights of the road irregularities and on the
Washboard Course [91. The use of these pro- spe, of the vehicle. The input acceleiation
files on the drums poses a practical difficulty levels are substantially below I g. The second
because of the high starting torque required to phenomenon is associated with the higher speeds

4
 FLUID MOTOR M A

ACC ACC
CHECK VALVE NO.N2

PILOT VALVE-
SOLENOID OPERATED _

RELIEF VALVE-"'

Fig. 3 - Schematic drawing of hydraulic control system

(15-40 mph) over the same irregularities as for acceleration data were taken in the cSkrgo
the virst Phenon::non and results in a severe spaces of the vehicles operating over mi.,vh
w-1.bounce producing a random vibration. -oads at speeds Whel. cacsed discomfort to the
The third phenometion occurs when a singte driver. Some of the bigh-g level points were
ropi discontinuity of large magnitude is tra- recorded on body appendages and rot in cargo
versed at sufficient speed, causing the body of spaces. The frequency and amplittode distribu-
the vehicle to strike the rubber pads (snubbers) tion indicates the complexity and randomness
on the axles. 1Xring the time the axle pad and of the motions involved. The values of the con-
the body are in contact. the suspension system stant "gg"levels presented by the sloping straight
is changed by the effective removal of the main lines leads to the conclusion that Weo data does
spr-ings from the system, thus giving the body not include incidents Of "hard" spring bottom-
of thp vehicle a large acceleration or decelera- ing. The predominant frequencies of vibration
tion. This impact can prcviuce a medium-.level occur in the ranges of 2-5 i~z, 7..15 Hz, and
shock pulse In excess of 10 g maximum for a 60-200 Hz. These Ireauency ran~ges are th'eI
duration of seve-al milliseconds. natural frequencies respectively of the ain
s,)rigs. the tires, and the body structures. it
ir nr ted that the avera -. accvleration ,.niplitudeA
Statistical Data is constant at about 0J) g with most pointu being
contained under a line representing 3 g ampli-
Figure 4 predents data frin tests on nine tude. As would be expected, the low frequen-
trucks constituting 1848 test 's. The cies up to 5 IN are associated with iarge

172
 o 00(oe7

000
0
010 =2

~ ~ ina
data
pont 1

011
icetel qae tr~rt
displacement amplitudes of 1-10 in., and the Typical Records
higher frequencies of 50-200 Hz are associated
with small displacement amplitudes generally A typical reproduction of acceleration rec-
less than 0.01 in. The intermediate range of ords (taken originally on magnetic tape) for a
frequencies 5-50 Hz have displacement ampli- location at the center of the cargo platform of
tudes in the range of 0.01-1.00 in. Since pre- an unloaded 2-1/2 ton 6 x 6 military truck, for
dominant frequencies are absent in the range the three mutually perpendicu'r directions
of 15-60 Hz, the optimum natural frequencies (vertical, longitudinal, and transverse), is
for equipment shock isolators is in the range shown in Fig. 5. The corresponding truck speed
of 25-35 Hz for conventional wheeled vehicles. is 20 mph. The acceleration signals have been

z
0

TIME (SEC)
(a) Vertical ReprocM

z
0

W
0

0 i2 3 4 1
TIM~E (SEC)
(1.) 'rran ve rs*

F A.C-/elerdtt,)n'-tlflle$gn~turei of the- principal components of

Input acceIc r~tion ot the cc nter 'f the c, rgj platfovn. of an unloaded

2-1/2 ton, 6 b mniltary truck. recorde'd ait a speed of 2(0 miph3

174
TM (
 - .. . ..- -

I-

0 3 4
TIME (SEC)
(c) Longitudinal

Fig. 5 - Acceleration-time signatures of the principal components

of input acceleration at the center of the cargo platform of an
unloaded 2-1/2 ton, 6 >, 6 military truck, recorded at a speed of
20 mph-Continued

passed through a 50 Hz low-pass filter, and edge of the cargo space and the left main frame
clipped by the recording system at ± 2 g. The member, Column 2 at the center line of the
dc offset is due to slight unbalance in the re- cargo space and Column 3 over the right main
cording system, and should be disregarded. frame member. This arrangement provided a
reasonably complete sampling of the varied
Even though the excitations of each of the structural details of the cargo space (Fig. 6).
driven wheels are relatively periodic, the initial
states of the truck at the times of transit over The acce!--":-meter signals were low-pass
the bumps are always different. This causes filtered and recorded on magnetic tape. For
more randomness in the magnitude of the ex- analysis, the .'ms acceleration levels were read
citation than might be expected. Thus little off during playback and selected recordings
periodicity is noticeable in the truck motions, were transcribed onto loops for extracting
Of course, dominant frequencies exist which power spectral density and amplitude distribu-
are caused by particular modes of vibration of tion. The latter analyses were performed with
the truck structure. the loop recirculating at eight times original
speed to reduce the processing time and trans-
late the signal's component frequencies to a
4Procednre for Measurement and more convenient range.
Analysis of Rough Road Simulator
Outpuia Motions
Instrumentation Characteristics
The test vehicle was an M35 6 x 6 truck
carrying dead weight loads of 0, 2316 and 4522 The accelerometers used ww're piezoelec-
lb. With each load. triaxial accel-ratiun meas- tric units with a natural frequency of 12 kHz,and
urements were made at each of twelve selected a linear range of 0.001 to 20 g. Thpy were e'n-
locations of the cargo space at speeds of 5, 10, nected to wide-band charge amplifier3, the fre-
15, 20, 25. 30, and 35 mph. The locations were quency response of the combination beiig vs-
arranged in a matrix of four rows and three sentially flat from 0.05 to 4800 Hz. The outputs
columns: Row 1 midway between the front of from the charge amplhiieri were low.pass fil-
the cargo space and the forward rear axle, tered at 2500 Hz and recorded in high-density
Row 2 over the forwaa'd rear axle. Row 3 mid- FM mode at 7.5 ips. Each was recorded on two
way between !he rear axles and Row 4 midway tape channels, at high gain having nominal dy-
between the after rear axle and the tail-gate. namic rrnge of 0.03 to 10 g and at low gain with
Column 1 was chosen midway between the left dynamic range 0.06 to 20 g. The lower limit

175

A
 Cob

II'i , ,

II I

iI
,[I * , I
,Row

jL V -t^ ' Row 2

.. I '.
i r I

I, I

Column Column Column

2 3
Fig. 6 - Arrangement of measurement
locations on the truck

in each case corresponds to the rms noise level plotted by a swept-frequency heterodyne wave
of the recorder. analyzer. Even with the factor of eight speedup
the time required for a spectral analysis was
The rms acceleration levels were moni- long, so the frequency range covered was re-
tored directly from the master tapes at normal stricted to 2 to 300 Hz (real time). The anal-
speed with a tru2 rms voltmeter. The selected ysis filter bandwidths employed (in real time
recordings transcribed ento tape loops were equivalents) were 0.25 Hz from 2 to 12.5 Hz,
played back at 60 ips and contained a (real time) 1.25 Hz from 12.5 to 62.5 Hz and 6.25 Hz from
forty second sample of the iceorded signal, 62.5 to 300 Hz. For a completely random sig-
The amplitude distribution of each of these sig- nal the wort case error estimate would be
nal samples was measured with a cumulative * 25% at the 80% confidence level. The Instru-
distribution function anlyzer, and at the same mentation setups are outlined in the block dia-
time the power spectral density function was grams of Figs. 7 and 8.

176 I
II
 r--- -- -
VOLTMETE OSCILLATOR

,S-I' CAIBRATOR

L-- CAIBRTOR--~BAND
SIN PASS
FILTER
NOISE
GENERAO
I

ACCELEROMETERW PASS

TACHOMETER

Fig. 7 - Block diagram of measurement system. The acceleration channel shown

in one of three, the drum speed channel one of two. Acceleration channel calibra-
tions were inserted behind the charge amplifiers, the charge-to-voltage transfer
coefficient of the charge amplifiers being check independently.

RESULTS Figures 12 and 13 are photographs showing

the steel-weight arrangements for the nominal
Figure 9 displays the rms accelerations at 2300-lb and 4500-lb truck loads. These loading
each speed for each of the three load conditions assemblies were bolted to the transverse chan-
averaged over the ensemble of measurement nels supported by the truck's main frame mem-
positions; the bar labelled L shows the average bers.
rms longitudinal component, T the average rms
transverse component, V the average rms ver-
tical component, and R the rms resultants of Determination of Severity Level of
the individual L, T, V sets averaged over the Random Vibration Inputs
ensemble of measurement positions. Figure 10
summarizes the rms resultants at each position The basic data which formed a guide for
averaged over all speeds for each of the three the selection of the eccentricities of the circular
load coditions. The bars are plotted in roughly contoured bumps described in a previous para-
the geometrical relationship of the actual meas- graph, was taken from Fig. 4 which showed that
urement locations, and their heights represent practically all g levels were contained under a
the rms accelerations (in g) whose values are line representing a 3-g peak amplitude for the
indicated at the tops of the bars. The average frequency range 1-200 Hz. An estimated aver-
values for each row and each column are indi- age acceleration amplitude is shown at approxi-
cated by vertical lines at the ends of the row mately 0.5 g. Referring to Fig. 9, the maxi-
and column markers. Figure 11 shows a typical mum vertical rms acceleration is indicated
group of spectral density plots obtained for about 1.4 g at 10 mph for the zero-lb load con-
platform position row 2 column 3, for zero-lb dition. It was assumed that this average value
load and speed of 20 mph. of 1.4 g rms at 10 mph for the empty or light

177
 OSCILLOSCOPE

FDISTRIBUTION
LFUNCTION
MODE) ANALYZER

VOLTMETER

0 VIBRATION

OSCILLATOR T

VOLTMETER

Fig. 8 - Block diagram of analysis system

30

320

10

5,.0t,2
(a) Zero-lb load

Fig. 9 - Average rms accelerations of all the

measurement locations

___ __ __ ___ __
___ _______ __
178

*
 30 I a

Z--20

01

Ii

T VR VRLT VR L 7V R L TIVR L TIVR L

0 5 10 15 20 25 30 34
SPEEDW/H

(b) 2300-lb load

20 --

0O 5 10 . 0 25 ?.0 54

(c) 4500-1b load

220 Fig. 9 - Average rr accelerations of all the

measurement locations -- Continued
1r1 -4

load condition represented the approximate viewpoint, where the maximum level may be
maximum severity level tolerated by the drivers arbitrarily set at 3 g as suggested by Fig. 4.
of the nine trucks for the tests which produced The zero or light-load condition represented by
the data presented in Fig. 4. Also, It has been Fig. 9(a) and 10(a) in terms of g rms, repre-
shown that in a random process 99 percent of sents the most severe condition since additional
the peak values will be below a value of 3 times loading suppresses the high-frequency vibration
~the rms value. Accordingly the 1.4 g rms value components generated by the truck platform -
i, translates to a peak value of 4. 2 g. This peak panels. T his is demonstrated by comparisons
~value Is considered acceptable from a fiduciary with Fig. 9(b) and 9(c), and with Fig. 10(b) and ,
" -" i 179
 (a) Zer-lbo-a

1.94

(a) Z300-lb load

10 j
 :1 2
3 1,3

(c 40-bla

I Fig. 10 - Average rms resultants of each position for all speeds.

Bars are placed at the approximate measurement locations and
theirheights represent the rms accelerations (in g) whose values
are indicated at the top of the bars. Vertical lines indicate aver-
age rms values for each row and column-Continued.

10(c). It should be noted that the structural program conducted by the Naval Research
stiffening effect of the steel-weight loads bolted Laboratory under its requirements, and one of
to the floor panels reduces the resulting g-rms its predecessors, BUSHIPS Specification 40T9.
responses considerably on the weight loads
themselves. For the 2300-lb load as shown for Since 194' Shock and Vibration, Test and
position LL (on left load), Fig. 10(b), tne accel- Evaluation Programs at the Naval iesearch
eration level is 0.8 g rms. For the 4500-lb Laboratory have been conducted a develop-
load a value of 0.5 g rms is indicated at the mental rather than "go-no-go" basis. For most
corner of the load assembly near location 2-3, of the damages occurring during the tests,
Fig. 10(c). modifications were made which precluded the

It can be inferred, therefore, that the data same failure. Consequently, as testing pro-
analysis supports the conclusion that the bumps gressed difficulties were gradually worked out
used on the simulator
usedon hedrums,
rum, in
smultor the develop-
inthedeveop- ofmor, o thesin
a design. Ifeqipnt woias ionsowere
indicated modifications were
ment of this proposed specification, generate major, or the equipment was shown to be gen-
adequately a random arandm corresponding to
vibration
adeqateyvbraioncorrspodin to erally unsatisfactory for test requirements, it
was returned to the manufacturer for the
the upper level of severity demonstrated to oc- wa neda
cur in the field. changes.

Ii' Fig. 14 the number of vibration damages

DETERMINATION OF DURATION to 120 equipments tested under this develop-
OF TESTING PROCEDURE mental program according to BUSHII-S Specifi-
cation 40T9, is plotted for the three principal
The duration of testing for the proposed directions of vibration [11]. Fifty of the units
specification is based on experience with the were shock-mounted and seventy were not
application of other specifications. Military shock-mounted. Both medium and lightweight,
Standard MIL-STD-167 (Ships), "Mechanical electronic and nonelectronic equipments were
Vibrations of Shipboard Equipment," was adopted included in the results, and only those equip-
as a principal guide in this determination be- ments which satisfactorily completed the tests
cause of the long-term equipment developmental were considered in compling data.
181
 0.1-

N-
0.01
I!

I-L
0I0! I

1. FREQUENCY (HZ)
(a) Vertical

Fig. 1 1 - "ypical group of spectral density plots obtained for platform

position row ,Z column 3, for ,zero-lb and speed of ,)0 rnph

182!
 0. . 7I;; j &

- h,-

4" ii1'I
(Li
0.001 t'

i- Il I IIIII
10 100
FREQUENCY (HZ)

(b) Transverse

Fig 1 - Typical group of spf-trail density plots obtained for platform

position row 2 column 3, for 4ero- lb and speed of 20 mph--Continued

183
 IFI
I71 "1
JO 100A
U)EUEC (Z
(c) Lo7giudWa

F.001 yulgou fseta eniyposotindfrpafr

positionrow
2 coumn 3, fr zero-l and sped' 0mh-otne

U184

W4 , .I i
 Fig. 12 - Steel-weight arrangement for the nominal 4300-lb truck load

Vibrations in the directions corresponding was concluded that the total testing time should
to the lowest horizontal resonant frequency and be limited to about one-third of the total time
in the vertical direction are seen to be of ap- required for testing in three mutually perpen-
proximately equal severity. Low resonant fre- dicular directions consecutively, in accordance
quencies of the rocking modes of vibration were with tne procedures of MIL-STD-167 (SHIPS),
the chief cause of damages in the first direction, which is 13 hr and 15 min or hr and 25 min in
'4
whereas low individual chassis resonancesprob- each direction up to 33 Hz maximum. The total
ably the great number for the latter. Only the maximum testing time required by the proposed
first mode of vibration in each direction was specification is 3 hr and 48 min, of which 2 hr
excited by the frequency requirements of the is designated as the endurance portion.
testing specifications. The few damages occur-
ring in the last half-hour test period in each
direction indicated that the duration of the en. SUMMARY
durance test, 2-hr in each direction__was ade-
quate to determine with a high degree of confi- 1. A proposed draft of a Wheeled-Vehiele
dence whether or not an equipment displayed Rough-Road Transportation Test Specification
good dynamic characteristics, is presented as Appendix A herewith.
Since the rough-road simulator produces 2. The specification is related to the use of
vibration simultaneously in three directions it a specific machine designated by the descriptIve

185

I
 Fig. 13 - Steel-weight arrangement for the nominal 4500-lb truck load

title of "Rough-Road Simulator for Wheeled directions, only one test-mounting set-up is
Vehicles." normally required versus two or three mount-
ing set-ups incurred in vibration testing with
3. This machine has been placed in suc- conventional vibration machines.
cessful operation and provides a practical and b. The simulator input of simultaneous
economical laboratory device for the simulation random vibration in three directions is a rea-
of the shock and vibration environment encoun- sonably accurate duplication of the wheeled-
tered by equipment transported in wheeled ve- vehicle transportation environment and results
hidles, in r-.iucing the test duration to one-third of that

4. Representative samples of spectral den- required when conventional vibration testing

sity plots (g2 per Hz) for three different loadings machines are used.
and several speed combinations for a 2-1/2-ton, c. The impedance coupling frequencies for
6 v 6 open body truck were obtained to illustrate varying weight loads imposed on the test vehicle
the statistical analysis of the responses. are duplicated precisely without manipulation
for a given road profile.
5. Advantages accruing by use of this ma-
chine and specification are enumerated in the REFERENCES
following subparagraphs.
1. U.S. Military Standard, Environmental Test
a. Since this machine produces trans- Methods, MIL-STD-810B 15 June 1967,
port-like motion simultaneously in three METHOD 514, METHOD 516.

186
 50 METHOD 201A, METHOD 202B, METHOD
DIRECTION OF LOWER 205C, METHOD 213
40
I ~FREQUENCY RESONANT
HORIZONTAL 3
R 30 3. H.M. Forkois, Report of NRL Progress,
I - 20 aO Januaiy 1963, pp. 14-17
i , 4. H.M. Forkois, "Rough Road Simulator for
0 30 60 90 120 Wheeled Vehicles," SAE Journal 71, No. 6,
FREQ, CPS TIME, MIN June 1963
5. H.M. Forkois, "Rough Road Simulator for
Wheeled Vehicles," Naval Engineers Jour-
50 nal, December 1963
VERTICAL DIRECTION -40wI

E 30 6. E.W. Clements, "Analysis of NRL Rough

Road Simulator Outputs," Report of NRL
20 Progress, July 1967

120 7. H.M. Forkois, "Development of a Rough-

52 Road Simulator and Specification for Test-
FREQ,CPS TIME, MIN Ing of Equipment Transported in Wheeled
Vehicles," NRL Memorandum Report No.
2100 "

DIRECTION OF HIGHER 50 8. E.W. Clements, "Measurement and Analysis

HORIZONTAL RESONANT 40 of Acceleration Environments - Measure-
FREQUENCY ment and Analysis of Acceleration Environ-
30 ments Generated by NRL Rough Road Simu-
- 20 lator," NRL Memorandum Report No. 2097

S .L , O.'
0 9. "Standard Load Vibration Test on the Mun-
5 15 230 30 60 90 120 son Test Area," Automotive Division Pub-
FREQ,CPS TIME, MIN lication, Aberdeen Proving Ground, Mary-
land, December 1959
Fig. 14 - Vibration damages to equip-
ments tested according to Bureau of 10. Shock and Vibration Bulletin No. 6, U.S.
Ships. Specification 40T9. Naval Research Laboratory Report S-3200,
November 1947

11. K.E. Woodward and H.M. Forkois, "Design-

2. U.S. Military Standard, Test Methods for ing for Resistance to Shock and Vibration,"
Electronic and Electrical Component Parts, Electrical Manufacturing, July and August
MIL-STD-810B Notice 2, 27 May 1965 1954

Appendix A

A PROPOSED DRAFT OF A WHEELED-VEHICLE ROUGH-ROAD

TRANSPORTATION TEST SPECIFICATION
1. Scope wheeled vehicles, incident to traversing rough
roads and cross-country terrain.
_ 1.1 The purpose of this specification is to
establish tests which can be used to determine 1.2 Test procedures are given which pro-
whether or not an equipment or assembly is suf- vide a simulation of the general vibration en-
ficiently rugged to withstand vibration environ- vironment occurring in wheeled vehicles in
ments when transported in trucks, or other traversing these worst-case surfaces.
F187
 !I

till"

Fig. Al - Rough-road simulator for wheeled vehicles (general arrangement)

2. Rough-Road Simulator or at least once a month, or after each 50 hr of

Test Machine machine operation.

2.1 General Description. The simulator 2.4 The test-machine vehicle shall have no
consists of a test-machine vehicle which is other loads except those of the items or assena-
placed with its wheels on the drum wheels of blies under test and such auxiliary equipment
the test machine (Fig. Al). The surfaces of the as may be necessary for the performance of the
drum wheels are provided with a suitable con- test.
tour by bolting bumps to their surfaces. The
test machine vehicle is constrained from mov- 2.5 Bumps or Drum-Wheel Profiles. Un-
ing appreciable distances fore and aft and side- less otherwise specified the drum-wheel pro-
ways. As the drums are rotated, shock excita- files shall consist ot two bumps of different
tiou is applied to the vehicle. The equivalent eccentricities per drum. They shall be placed
road speed is equal to the surface velocity of 180 degrees apart and they shall be shaped as
the drum wheels. shown in Fig. A2.

2.2 Test-Machine Vehicle. The test ma-

chine vehicle may be any one of several types. 3. Methods of Performance of Tests
Ifthe item to be tested is to be transported, or
installed, in a particular vehicle type, then a 3.1 Location and Mounting of Test Item in
test-machine vehicle similar to this type may the Test Machine Vehicle. The test item shall
be used. If no preference of vehicle is indicated be secured to the test machine vehicle in a
then the test-machine vehicle shall be a 2-1/2- manner which would simulate normal practice.
ton 6 x 6 open body truck. Other test-machine If several different methods of securing the test
vehicles may be jeeps, trailers, or special item are probable, and they involve different
carriers. orientations of the test item, then it may be re-
quired that the test be performed for each of
2.3 Maintenance of Test-Machine Vehicle. the several orientations. The test item shall be
The test-machine vehicle shall be maintained in secured at a location immediately above the
proper operating condition. The tires shall be rear wheels of the test vehicle and as near the
inflated to recommended values before each right or left hand side of the vehicle as is
test. If the vehicle is equipped with shock ab- practical.
sorbers their effectiveness shall be subject to
weekly inspections. The chassis shall be lubri- 3.2 Unsecured Test Items. If the test item
cated as specified in the maintenance manual, is unsecured then the forward part of the cargo

188
 1/2

62S" FEATHER EDGE

CASTEDGES
SHOULD
6E
P APPROX 118* THICK

NOTE LENGTHOF PROFILE AND

SMRNG OF HOLESSME AS
THOSE ON T"HE
FOLOWING FIGURE j
/
/

(a) 3/4-in. eccentricity

Fig. A2 Details of bump contours

space shall be fenced off so that the test item speed that is a multiple of 5 mph, up to the
will be contained in the rear part of the cargo maximum value. In addition there shall be a
space. If the size of the test item permits, the 15-min test at a speed which is deemed by the
fence should divide the cargo space into two ap- test engineer to be most damaging to the test
proximately equal parts. item or assembly. In no case shall the endur-
3.3 Exploratory Test. The first part of ance test be less than 2-hr duration.
the test is to provide an opportunity for short
time observation of the responses of the test 4. Criterion for Acceptable Per-
item at any equivalent road speed. The test ma- formin o Tet tem
chine shall operate so that the drum-wheel
peripheral speed corresponds to an integral 4.1 Equipment Subjected to Infrequent
number of miles per hour from 5 mph up to T s a Eiment . nfequeot
some maximumvalue appropriate
value shall for the vehicle. Transportation Environment. Unless otherwise
The maximum be 35 mph unless specified, a test item in this category will be
m
therwis value
iimu timeshalltest
be 35ms considered to have acceptably passed the rough-
otherwise specified. The time of test shall be 3 road transportation test if after the test there
minutes at each average integral value of speed. is not evidence of major structural damage and
3.4 Endurance Test. The endurance test the equipment can be put into proper operating
shall consist of a minimum of 15 min at each condition with but minor adjustments.

189
I

a7/3T CATHRLI-49VG4
COrWE&

CASTMaSS "'9JAOK
APPAT 1/8' THCK

Mb I -1I/2-in. eccentricity

Fig. A2 -Details of bunip contours -- Continued

'\ 4
4.2 Equipment Mounted in Vehicles and vibrations generated, it should not be expected
Portable Equipment. No significant damage that tests on different test vehicles of the same
shall be acceptable for equipment that is sub- type will be identical, even though the equip-
jected frequently to transportation environments. ments carried and their methods of attachment
Equipment that is required to operate during are the same. However, if the rms values of
transportation shall be in an operating condi- the vibrations derived under these conditions
tion during the test, and if not so required, it with instrumentation having a prescribed band-
shall be specified that the equipment be capable width, deviates appreciably more than 10 per-
of restoration to normal operation with no un- cent frdm each other, there may be reason to
usual delay after the completion of the test, suspect damage or wear to parts of the suspen-
Acceptable down-time should be stated in the sioa system of the vehicle, namely springs and
equipment specification, shock absorbers. Improper inflation of the
tires or brake drag may be other causative
~**.factors
5.Nature of5the Taso~t for large deviations. In some cases,
Envirnmentdesign changes in the suspension systems of
more recent vehicle models may cause sub-
machine vehicle, and the random nature of the values.

190
 LABORATORY CONTROL OF DYNAMIC VEHICLE TESTING (U)

James W. Grant
U. S. Army Tank-Automotive Command
Warren, Michigan 48090

In order to study vehicle suspension and frame dynamics under

controlled and reproducible laboratory conditions, TACOM's
road simulator or "shaker test" was developed. A road simula-
tor is a laboratory test device which imparts dynamic forces
simulating road inputs, on a complex vehicle. It is the
purpose of this study to develop vertical position control
signals for the road simulator so that good correlation
between laboratory test and field results is obtained. As a
result of this study, the design engineer has a more exact
vehicle model than he has had and the test engineer has a
laboratory simulation which has been verified for vertical
dyn?,nic inputs. The combined effect of these two engineering
tools will serve to produce a better prototype vehicle which
in turn will eliminate many of the initial field test failures
which plague new vehicles.

IU1TRODUCTION Laboratory testing of off-road

vehicles offered a new challenge. Due
The road simulator concept of to the large wheel deflections, the
laboratory vehicle testing came into vehicle must be restrained from falling
existence to facilitate studies of off the road simulator. In order to
frame and suspension dynamics. Prior facilitate restraining the vehicle and
to the road simulator, frame and also to allow the addition of longi-
suspension components were divorced tudinal excitation forces, the wheels
from the vehicle for laboratory were removed and the spindles attached
evaluation. In most cases, the con- to the actuator through a multiple
trol or excitation signal for the test degree of freedom assembly. The res-
was some well defined mathematical traints were attached from vehicle to
function whose correspondence validity ground so that their effect on the
to actual field excitation is question- dynamic motion of the sprung mass was
able. Testing then progressed to a minimal.
point where recorded field signals and
shaped random noise were used to con- The present state of the art
trol component tests, includes vehicles in the 5-ton payload
class with up to six vertical and four
Since there is interaction horizontal linear actuators. Currently
between the component being tested being constructed at the U. S. Army
and the vehicle to which it is mounted, Tank-Automotive Command (TACOM) is a
it became apparent that a road simula- road simulator for 1/4-ton class
tor which would test the total vehicle vehicles which has four vertical
system in the laboratory would yield actuators with position control, four
useful results. The early road horizontal actuators with load control
simulators provided vertical inputs of and four rotary hydraulic pumps to be
low amplitude to each wheel of a used as absorption dynamometers with
passenger car. The inputs were accom- torque control. This simtilator, fully
plished using four electro-hydraulic operational, will test the total
linear actuators with pedestals on vehicle system under controlled labora-
which the tires rested, tory conditions.

191
 The analog position or force sig- is not physically realizable. The
nals which control the electro- task, then, is to develop a stable
hydraulic actuators must produce transfer function, the frequency res-
motions or forces in the vehicle which ponse of which approaches the double
can be correlated with those which iare integrator in the desired frequency
recorded during field tests. This band of .5 to 50Hz. This band is
paper will present in detail three within the response limits of most road
different techniques by which valid simulators and also includes the fre-
position control signals may be quencies of interest for suspension
obtained from recorded field data or dynamics studies. The transfer func-
from surveyed terrain elevations. tion chosen to double integrate the
recorded acceleration signal to pro-
duce the position control signal for
CONTROL SIGNAL GENERATION TECHNIQUES the
KS2 road simulator is G2(S) =
The analog signal which controls
the electro-hydraulic actuators of the
road simulator'may be proportional to As can be seen in Figure 1, the
either position or force. The verti- frequency response curve for G2 (S)
cal road input actuators are position when K = 1 asymptotically approaches
controlled and the fore and aft perfect double integration beyond .636
horizontal road input actuators are Hz. The important characteristic of
load controlled. The most readily G2 (S) is that tha low frequency compon-
obtained field data which can be ents of the input signal are
transformed into vertical wheel suppressed. These low frequency
spindle displacements are vertical components, especially zero frequency
accelerations of the wheel spindles, of D.C. offset, are the prime contri-
Terrain profile data can also be butors to unstable double integration.
transformed into vertical wheel This fact is quite clear in Figure 1.
spindle displacement if accurate As frequency approaches zero, GI(S)
vehicle and tire models are on hand. approaches infinity and G2(S) approach-
es zero. The low frequency accuracy
Double Integration of G2 (S) can be theore-ically improved
by shifting the intersection of it
The acceleration signal which is asymptotes to the left. This can be
to be double integrated is the verti- accomplished by decreasing the con-
cal acceleration of the wheel spindle. stant 4 in the denominator to 3, for
The vertical accelerations of each example. However, as thin constant
wheel are recorded simultaneously so approaches zero, G2 (S) approaches
that the control signals generated Gl(S). Another way to increase low
from thase accelerations will have the frequency accuracy is to increase K.
proper phase relationship The Increasing K, however, raises the
acceleration signals thus recorded are whole response curve and the higher
a function of the suspension geometry, frequency accuracy decreases.
the suspension parameters, the tire Stability is also reduced as K is
characteristics and the terrain pro- increased. Either method requires
file. Changes in any of the above trial and error to determine which K
vehicle characteristics would require or which denominator gives acceptable
that , ne,4 test course traverse be response in the desired frequency
made. For the following analysis, range.
assune that an accurate recorded
acce.leration signal is ava-lable. Figure 2 shows the result of
playing a field-recorded acceleration
The integral of well defined signal into G2 (S) with K = 1. The
mathematical functions can be found in acceleration signal was recorded at
any calculus text. Mathematically, the front wheel spindle of an M656 5-
the integral of a continuous random ton 8x8 cargo truck as it traversed
variable such as vertical wheel the Aberdeen Proving Ground Belgian
acceleration is also well defined. In Block Course at an average speed of 15
fact, the double integral of accelera- miles per hour. The displacement sig-
tion which results in displacement is nal peak to peak magnitude of .2 feet
also well defined. The physical (Figure 2) was observed during the U
implementation of double integration,
however, is not well defined.
recording of the acceleration tape.
The accuracy of the approximate
I
It is a well known fact that a double integration depends, of course,
stable perfect double integrator whic] upon the transfer function used to
has the transfer function GI(S) = 1/S perform this operation. The

192 1
correlation between the field In order to apply this control
recorded vertical acceleration of the technique, the vertical wheel spindle
wheel and the laboratory recorded acceleration must be recorded on mag-
vertical acceleration of the wheel can netic tape during field runs. From
be computed to numerically determine this data, a shaping filter for random
the accuracy of the double integration, noise is desired such that the filter
output is a displacement signal
statistically equivalent to the
-20 recorded acceleration signal.
Let the filter to be defined be a
linear time invariant function so that
/ conventional methods of analysis may
be used. The total system is:
y(t) = h(t) x n(t) (1)
-69 Where--
n(t) is the random noise input
h(t) is the filter
-33/ y(t) is the output displacement
Using the convolution integral and

-100 A_ .relationship the fourier transform, as described in

reference 3, page 182, the following
is obtained from equation
.01 0.1 1.0 10 100 ()
FRE3MUECY Hz Sdd(f) = IH(i2o)1 2 Snn(f) (2)
2
FIGURE 1 BODE PLOTS
$2/(S+4) FOR 1/S
4Whr- AND Where-

Sdd(f) is the power spectral density

fu (PSD) of the desired displace-
ment control signal.
_ I Snn(f) is the PSD of random noise.
Sh1 LiThis is a constant and will
P be defined to be unity.

J -- H(j2frf) is the frequency response

Cfunction for the shaping
* --- -- A-
--- - 1 SEC ,-
* filter.
T _T_i F The relationship between dis-
- - placement and acceleration PSD's is
+ ASSaA(f)
defined to be:
(3)

K11. j-
EP
1-
W-AA ~
_06T
!
~
:T
'

A
Substituting Sdd(f) in equation (3)
gives:

FIGURE 2 ACCELERATION AND RESULTING Saa(f) = (27f) IH(j2 f)V

Snn(f)
DISPLACEMENT-TIME TRACES FOR (4)
G 2 (S) '
Since Snn(f) -1 by previous definition:
Filtered Noise 1
The idea of playing random noise,
which has a flat power spectrum,
through a shaping filter to
laboratory road simulator hascontrol
been a obtainEquation (5) will now be used to
suggested previously (references 1 and responsefrom field data the frequency
for the desired filter.
2).

193
 Figure 3 is a PSD curve of the approximations. This technique is
vertical front wheel acceleration of validated using statistical measurement
an M656 5-ton 8x8 cargo truck. The
acceleration signal was recorded during
field tests at Aberdeen Proving Ground
at an average vehicle speed of 14.2
miles per hour. The test courses were
the Belgian Block Course, Three-Inch
Spaced Bump Course, Two-to-Four-Inch
Radial Washboard Course, Imbedded Rock
Course and Two-Inch Washboard Ccurse.
techniques such as histograms, cross
correlation and probability density
functions.
I
from Figure 3 into equation (5)
results in the desired freauency res-
ponse curve shown in Figure 4. The
desired curve was approximated using
an ESIAC algebraic computer by the
following transfer function where
j2o-f is replaced by the Laplace
0.1 I
Operator S: ,01
H(S) = .01 0.1 1.0 10,
15.123 ( 2+ 86.4S + 202S) FREQUEACY Hz
62) FIGURE 4 FREQUENCY RESPONSE CURVES
(6) FOR DESIRED AND ACTUAL

0 FILTERS

SHAPED RANDOM NOISE

40 Terrain Profile
0.1 1.0 10 100 The two previous methods of con-
trol signal gencration required that
FREQUE3CY Hz vehicle dependent acceleration signal
av the wleel spindle be recorded during
FIGURE 3 PSD OF FIELD RECORDED test course traverse. In other words,
now instrumented test runs must be made
VERTICAL WHEEL ACCELERATION for each different vehicle configura-
tion. Consider now the possibility of
The actual frequency response us;-, surveyed terrain profiles to con-
curve for equation (6) is the dashed trol the road simulator system where
curve in Figure 4. Figure 5 is the the excitation is through the wheel
output of the filter with a random spindle as previously stated.
noise input.
Surveyed terrain elevation data
Since PSD is an approximate are readily available, in reference 4,
.,.easurement and the actual filter for example. The major problem to be
response function is an approximation solved then is the transfer function
of the desired filter response function from terrain to the wheel spindle.
the accuracy of this technicue This transfer function represents not
depends upon the accuracy of the only the tire assembly dynamics but is

194
also a function of the suspension facilitated by the availability of the
dynamics and the sprung mass. It is road simulator. The differential
concluded then that a mathematical equations of motion for the vehicle
model of the total vehicle system is are obtained using any of the conven-
required. This model would be pro- tional techniques such as Lagrangin
grammed on an analog or hybrid com- or Newtonian mechanics. The equatioo-s
puter and run in parallel with the are then programmed on an analog or
road simulator to provide the wheel hybrid computer. The computer model
spindle position control signal. The and the road simulator are excited at
block diagram of this system is shown the wheel spindles with identical sig-
in Figure 6. nals and the responses are co.pared.
The response is a combination of sprung
-ompute Wheel Ramass output signals, which could
Terrain Road include, for example, pitch, botnce and
Profile Mode PositionSimulatorl roll displacements. The parameters of
the computer model are adjusted either
manually or automatically such that the
FIGURE 6 SYSTEM BLOCK DIAGRAM error between comparison signals is
minimized. Reference 6 presents a con-
tinuous parameter tracking technique
The system in Figure 6 assumes an which could be extended to attain the
accurate model of both the tire and automatic parameter adjustment.
the vehicle. The tire is a complex
non-linear system which is discussed Once the accurate model is
thoroughly in re--rence 5. A tire obtained, its parameters may be easily
model can be made so complex that it adjusted to maximize some index of per-
is unwieldy or it can be simplified to formance such as driver comfort. The
a second order mass-spring-damper sys- sensitivity of any performance para-
tem. The latter case with a realistic meter to changes in each of the physi-
non-linear spring and point follower, cal vehicle parameters can be
Figure 7, may give satisfactory res- measured. This type of study tells the
ults for the vertical control of a design engineer a range of acceptable
road simulator, values for each physical parameter.
The physical parameters include spring
YT~t)rates, damping coefficients, center of
gravity location, wheel base, etc.

M The above described technique is

an ambitious undertaking currently
being implemented at TACOM. Extensive
K D computer analysis is required, but the
resulting vehicle model will give the
design engineer a new tool with which
to improve vehicle performance.

SUMMARY

_ t Three methods for obtaining the

road simulator control signal (the
input at the wheel spindle of the test
FIGURE 7 SIMPLE TIRE MODEL vehicle) have been presented.
The double integration of field
Where-- recorded vertical wheel spindle
acceleratioi, and the filtered random
m is the unsprung mass noise both require field data acquired
K is the spring rate by a vehicle similar to the test
D is the damping coefficient vehixie. These two methods are well
y(t) is the terrain profile suited to long term durability studies.
yT(t) is the wheel displacement
The third method requires and
NOTE that the velocity profile, y(t), facilitates the development of an
of the terrain is also required. The accurate mathematical model of the
digitized terrain profile is digitally vehicle. The selected terrain profile
differentiated to obtain y(t). is played into the computerized
vehicle model and the wheel displace-
Obtaining an accurate mathemati- ment signals from the model are then
cal model of the vehicle dynamics is used to control the road simulator.

195
The mathematical model resulting from
this technique may now be adjusted to
produce improved ride quality. From
this study, new hardware may be
ride quality.
The first two techniques have
been used at TACOM to control simula-
tors. The third technique is curren-
tly being implemented; we expect to be
using it by November, 1971.
REFERENCES
1. Van Deusen B.D., et al, "Experi-
mental Verification of Surface Vehicle
Dynamics", NASA CR-1399, National
Aeronautics and Space Administration,
Washington, D.Co, September, 1969.
2. Van Deusen, B.D., "A Statistical
Technique for the Dynamic Analysis of
Vehicles Traversing Rough Yielding and
Non-Yielding Surfaces", NASA CR-659,
National Aeronautics and Space Adminis-
tration, Washington, D.C., March, 1967.
3. Davenport, W.B., Jr., and Root,
W.L., "An Introduction to the Theory
of Random Signals - Noise', McGraw-
Hill Book Co., Inc., New York, N.Y.,
1958.

4. Heal, S., and Cicillini, C., "Micro

Terrain Profiles", RRC-9, U.S. Army
Tank-Automotive Center, Warren,
Michigan, 14 August 1964.
5. Schuring, D., and Belsdorf, M.R.,
"Analysis and Simulation of Dynamical
Vehicle-Terrain Interaction', Cornell
Aeronautical Laboratory, Inc., Cornell
University, Buffalo, N.Y., May, 1969.
6. Jackson, G.A., and Grant, J.W.,
"Linear Suspension System Parameter
Identification", SAE 710227, Automotive
Engineering Congress, Detroit,
Michigan, January, 1971.

196
 IDPACT VULNERABILITY OF

TANK CAR HEADS

Jen C. Shang and John E. Everett

General American Research Division
General Arerican Transportation Corporation
Niles, Illinois

An Impact Vulnerability ctudy of tank car heads was undertaken

by means of semi-aralytcal evaluation of head failures through
careful observation of indentat ons and punctures which were
produced in a series of full-scale tests. -ne parameters which
would influence the vulnerability of tank car heads were 'den-
tified. Simple formulas which could be used o determine the
permanent indentation and tho, Inpact ,rce were de-.eloped in
conjunction with a theoretical anriysi of influential dimen-

j
sionless parameters and an application of Hertz' fP)rce-inden-
tation law to collision problem;. Finally, the tark car head
failure criteria were established.

INTRODUCTION

In the past several years, a number of not sacrifice its accuracy in predicting the fail-
railroad derailments resulted in catastrophic ure behavijr due to a collision c' a peojectile
failure of tank cars, either in the form cf and a tank head. Hence, the investigation report-
indentation or puncture, which caused fhe evac- ed herein was undertaken by means of semi-anal-
uation of cities, personal deaths and injuries, ytical evaluation of head failures through care-
and property losses totalling millions of dollars. ful observation of indentations or punctures
Although defects and improper maintenance on which were produced in a series of full-.ca]e
railroads are the major causes cited by govern- tests conductel in the laboratory.
ment safety experts for derailment which have
soared 105% in seven years, a method of zolution The parameters which would influence the
must be sought to protect tank cars to reduce vulnerability of tank car heads were identified
the frequency of head punctures in accidents, an follov'3:
especially for tank cars transporting hazardous (1) head properties: thickness, geometrj,
materials. naterial
A number of tank-car heads are either ( ) commodity: outage, internal pressure,
-weight
indented or even punctured by couplers, side sill, (3) impact characteristics: force and
end sill or other
nents. Applicationflying objects during derail-
of a shfeld, bumper or other Auration, impact velccIty,
location, and orientation impact
possible protective structure to the lower part 04) tank ca, design and attachment construe-
of tank car heads appears to be one feasible tion details
approach to reducing th, number of incidents (5) charac-eristics of draft gear - coupler
involving release of the product, assembly
(6) insulation: material, thic-kness
The first step, protecting tank car heads
from puncture damage, is to determine the impact Simple formulas which could be employed to
valnersbility of tank car heads. However, so far, determirie the permanent indentation and the
no direct observations or instrumentations of coupler impact force were de-eloped in conjunction
head failure phenomena have been made in any with a theoretical analysis of influential diren-
accidents. Furthermore, it was also realized that sionless parameters, and an arplication of Hert?'
it would be extremely difficult or impossible to f.rce-lndentation law to collision problems.
launch a full-scale analytical investigatiun on Finally, the impact failure m.echanism was identi-
this subject within a short period of time. fied ana 'he tank ('ar head failure criteria were
Consequently, the study has to be simple and yet defined.

197
 Despite the fact that a limited number of is no center anchor. Longit'idinal support is
old cars were tested, which had been subjected achieved by welding the tank solid to the cradles.
to fatigue and corrosive environments, the anal- In this construction both steady and dynamic
ysis deduced from the tests yielded results longitudinal train loads pass through one stub
which correlated with failure data within engi- sill, through the tank, and then through the
neering accuracy. The results of this study other sill. Special design requirements are
indicate that (1) filled non-pressurized tank specified by the AAR Committee on Tank Cars for
cars are less susceptible to puncture than pres- this constru,!tion to provide for the safe trans-
surized cars, (2) empty cars are more vulnerable mittal of loads by the tank.
to puncture than liquid filled tank and (3) tank Insulated Cars - A tank car that has a
heads when struck near the knuckle radius are layer of insulation around the tank shell and
more susceptible to puncture than when struck heads. The insulation is covered with a metal
at the center of the head, and (4) the primary jacket, usually 1/8" in thickness and is flashed
mode of failure was attributed to the pl~g for- with metal around all openings for weather-tight-
mation due to the shear force exerted on the ness.
coupler impact impression.
HEAD FAILURE ANALYSIS AND CORRELATION WITH
Correlation of such failure analysis can FAILURE DATA
be used as a guide to develop rational criteria
for designing tank car head protective devices It is realized that it would be extremely
in the future. difficult or even rather impossible to launch
a full-scale analytical Investigation on this
TANK CAR DESIGN REVIEW subject within a short period of time. Conse-
quently, the study must be simple and yet not
The formulation of additional design cri- sacrifice its accuracy in predicting the failure
teria in order to significantly reduce head behavior due to a collision of a projectile and
puncturing must take into account the various a tank head. The study can be commenced by
designs and types of tank cars in service. identifying the problem areas related to the
These are: tank head failures.
Pressure Car - A tank car whose tank is
designed and fabricated under the provisions of Problem areas related to the development
Paragraph 179.100 of the DOT Regulations. These of design criteria involve (1) general physical
cars are characterized by tanks whose designs characteristics of head failures, (2) mechanics
are primarily governed by internal pressure considerations, and (3) the objectives to be
loads. Tank wall thicknesses are therefore accomplished in this task.
greater than in non-pressure car tanks, typically
being between 9/16" and 15/16". Design, fabri- The dissipation of the initial kinetic
cation and inspection requirements for these energy of a flying projectile (coupler, sills,
tanks are also more stringent. etc.) during impacting a tank-car head produces
Non-Pressure Car - A tank car whose tank ic vaiious effects, their nature depending on the
designed and fabricated under the provisions of physical characteristics of colliding bodies
Paragraph 179.200 of the DOT Regulations. These (such as rigidity, mass, material) and the
tanks al required to handle a small amount of marnitudheesh relative velocity. However,
internal pressure, up to 35, 5, and 75 psi in therate essentially two fundamental failure
various cases. Normally, the tank wall thickness mechanisms into which the Pailure behavior of
required to handle these low pressu-,_s 4 s below da.aged tank head can be classified. They are:
that which is roqaired t, handle t)herechanical (1) indentation and (2) puncture.
loads of rail transportation; hence, such tanks
are bull, thicker than requiri- for the pressure Indentation may be defined as the deforma-
load only. Typically, ncn-prssure tank car tion created by a force which exists in the con-
tanks have wall thi-knesses between 7/i(" and tactinp surface of colliding bodies without pro-
- ducing penetration of the xtriking object into
Underframe Car - A tank car bulli with a the head. This action involves either the rela-
center sill running continuously from end t, tive mass ratio, the relative stiffness (rigidity)
end. The tank is attached to this "underfran.e" and the relative velocity of colliding bodies.
by a ctnter "anchor" for longitudinal supprt,
and is held down by steel bands on cradle. over On the :ther hand puncture implies the
each truck for verti-al sl; ; rt. 'The c:ales corplete pi#,rcing (f tank head by the projectile.
provide no lorgi tudinal supp,,rt ,ther than tlhal Th complicated recnar.|,or, ene untered in this
produced by friction. In thi'a eonstructiCon, process has nn, yet be, completely explained cr
steady train loads are transmitte'i directly exat.inel, even thnuigh a consilorable amount of
thriui-h the underfrtne, and lonpitudinal lynansic da.nage data has been collected over the past
loads causling the tatnk t, ae-lerntt,. ny dele, r- years.
ate are transaitted throu'h *h anchor.
Underfraneless Car or Car Without Cnn'iru- For the case f tank-chr heads, it can be
ous ('enter 8111 -A tank car buil t with en I r-.asona,ly predict, i that an Indertation will be
"stub" sill f oniy. 'ht stub oil %r, ilr.liar pro l. el by a pro.'e-'lie, such as coupler, under
to an unierframe in or,, Oe [ l,b nup rge Irl a 1,,w(r itermeliate I-jact Velocity. This is
the taink at a poi In.bctr i f.ai I ruk. Iih, r( b,'auoe the helad with a th: ckress often l rf

198
than 1 inch is considerably more flexible than a
massive coupler head. In addition, the tank head ml + i 2 2 = ll +
is made of ductile material (steel) such that the
initial impact energy would be primarily convert- and v v2 = e (v 1 - v2 )
ed into elastic energy stored in the unyielded
zone of the head and plastic deformation of the where e = coefficient of restitution.
head shell creating an indentation (crater), with
smaller amounts accounting for elastic waves, the When e = 1, the above equations coincide
rebound of the striking object, friction and heat. with those of elastic impact, whereas for e = 0,
it will give formulas for the plastic impact.
Moreover, for the case of pressurized tank
cars, it is understood that the internal pressure For empty tank cars, the failure of the
tends to stiffen the head shell to resist the tank head appears to be described by the plastic
impacting force. However, internal pressure, impact. However, for pressurized cars, more
unfortunately, simultaneously reduces the flex- careful consideration should be given to deter-
ibility to cushion the impact. mine in which mode the tank head had failed,
either plastic or semi-elastic mode.
It becomes obvious that different methods
of approach should be taken to determine the In addition, the momentum-impulse relation-
influencing parameters involved in the foregoing ship gives
failure mechanisms, indentation of piuicture. T
However, the general laws of conservation of m(v - v') F(t) dt
momentum, energy or other laws of mechanics 0
should hold for both cases.
where = impulse duration
T
In the analytical investigation of the fail- = initial velocity of mass m at t = 0
v
ure phenomena, the laws of conservation of momen- = mass velocity at t = T
v'
turn and energy must be satisfied. Thus,
the total change in momentum
of a
For the case of perfect elastic impact, mass, m, during a finite interval of time, T,
there must be no loss in the energy of the is equal to the impulse of the acting force, F,
system. Thus, the following conditions must be on the mass, m, during the same period.
satisfied:
From the work-energy consideration, one also
mI v1 + m2 v2 = m1 v{ + m2 v' obtains the following relation

2 x
1 2 2 2 2 m (v - v) F(x) dx + E'
2ml (Vl2v) 'm (v2- v.
in which Mi, m2 = mass of colliding bodies where x = initial
cation position of the force appli-
vl, v2 = velocity of colliding bodies x
= final position of the force applica-
prior to impact tion
v and E' = energy loss involved in the simul-
= rebound velocity of colliding tneous action of elastic and plastic deformation
bodies after impact, of head, crack formation, spalling, elastic and
Asdfor the case of plastic impact, two :oi- plastic wave propagation, friction and heating,
strain-rate effects and perhaps even shattering
lided bodies will move with a common velocity o the striker.
after impact. This common veloc! ty can be easily
obtained from the momentum consideration, that is This research program is designed to devel-
mlvl + m2 v2 op a "shield", "bumper", or other possible pro-
m + m2 tective structure which can be applied to the
heads of existing and new tank cars in order to
Subsequently from the enere consideration, the significantly reduce the head damage resulting
enerey dissipated during the impact process or from derailments. The technical data to be
converted to othe form of energy, can be eva- developed in this investigation are:
uated by (1) The approximate magnitude of probable
- vn2) al head failure force by reviewing
tank
2(ml + mi) the characteristics of tank car head
failures;
(2) The criteria which would govern the
Under actual conditions, one would expect design of tank car head protective
some deviation from perfect elastic impact; there structure.
always will be some loss in energy of the systeur
during collision. Thus for the case of sem- The study is planned for three major phases
elastic impact, the ,omertuin and enery cu,,ide.r- (,f effort in order to achieve the program goals
ations will give as Identified in the previous section. The

199
effort will be specifically focused on finding have been studied and solutions for various im-
answers to the following questions: pact conditions were found by many investigators.
(1) How can tank head damage, indentation
or puncture be predicted as a result One method which is worth mentioning is
of a projectile striking a tank head, the application of the Hertz force-indentation
if the initial velocities and the law to predict the force developed during a
masses of both colliding cars, the collision of two deformable bodies.
head geometry, the head material and
the commodity condition are known? Some questions have been raised concerning
(2) How can the maximum impacting force the validity of the theory under dynamic condi-
developed during the impact be esti- tions, since the Hertz theory is based on the
mated from the direct observation of quasi-static contact. Nevertheless, the Hertz
head damage? theory has been applied in the impact studies,
(3) How can the impact failure criteria particularly for the case of low-to-moderate
be established, if some analytical velocity impacts (up to 300 fps). Hertz, Ref.
tools such as those obtained in steps (5), extended his solution of two spheres in
(1) and (2) are available? contact to impacting bodies in 1882 and Tsai,
Ref. (6) has proven the validity of the Hertzian
Even though a number of studies, such as impact theory through an analysis of Rayleigh
those reported by Goldsmith, Ref. (1), Fugelso, surface waves in an impacted glass block.
Ref. (2), and Cristescu, Ref. (3), are available
in the field of mechanics of impact and penetra- Davis, Ref. (7) also indicated that for
tion, no problem similar to the case of tank the case of impact of elastic bodies at moderate
head collision was investigated in these refer- velocities, the problem of elastic contact and
ences, elastic impact are in essence identical. Tsai
and Kolsky, Ref. (8), found that the quasi-
Realizing tie difficulty of formulating static treatment approximates very closely the
the problem entir.ly on an analytical basis, a stress in dynamic impacts. Roman, Ref. (9),
method of semi-the,;reticad procedure will be applied the Hertz theory to the investigation
sought. The proced,,': is comprised of a theo- of the coefficient of restitution and found that
retical analysis of influential dimensionless for moderate thicknesses of plate (0.138" - 1"),
parameters upon the test data which will be the theoretical calculated and experimentally
obtained in a short series of tests that will, observed values agreed well.
under laboratory-controlled conditions, some-
what duplicate various types of head failure Finally, Yang concluded in his studies,
on current tank car designs. The tests will Ref. (10,11), that the combined application of
also permit the measurement of head damage the dynamic field equation of solids and togeth-
(indentation of puncture) and the force that er with Hertz' law, to the impact problem of
produces head damage, plates and shells, yielded a reasonable result.
His results confirmed the validity of using the
A theoretical evaluation of test data with Hertz law in the problem of impact rtnd also
an application of the dimensionless analysis supported the conclusion that the application
in terms of various important parameters will of Hertz' law could be extended to the contact
yield results from which other tenk head fail- of visco-elastic bodies.
urea and required impacting force can be esti-
mated. In the Hertz theory, the force-indentation
relationship is given by
A series of full-scale tests was desitne 4 1 5
and conducted by the Engineering Department F = K (d) .
of the Tank Car Division of General American where F = contact force
Transportation Corporation at the Sharon Plant. d = indentation
The test data obtained are tabulated and pre- oC = constant
sented in the report by Everett, Ref. (4), from
which Table I was extracted. The test data The constant.i depends on the geometrical
were then forwarded to the General American and physical properties of tank head, and the
Research Division of General American Transpor- magnitude of internal pressure. It will be
tation Corporation for theoretical evaluation. determined experimentally.

One of the objectives to be accomplished The collisib.n of two bodies may involve
in this task is to develop a procedure which a variety of processes whose existence and rel-
would approximately describe the relation be- ative importance depend almost exclusively on
tween the impact force and the corresponding the shapes, the physical characteristics of the
indentation produced by collision of couplers objects, environmental condition, and, most
and tank heads. Important, on the relative impact velocity.
The relevant mechanical behavior of the materials
During the past decade, increasing atten- is ordinarily classified as beinp elastic, plas-
tion has been focused on the problems atten- tic, viscous, or a combination of these; a quan-
dant to the collisiot of a projectile on a tar- titative description of these properties is
get. Consequently, n.nerous impact probles formulated as a relation between stresses and

200

_
strains and, in the case of time-dependent in thick plates. The perforation is accomplish-
effects, their respective rates. ed by radial expansion of the plate material as
the projectile passes through. Fracturing, such
Impact is differentiated from rapid load- as scabbing, that can be attributed to inter-
ing in which the forces acting at the contact action between transient disturbances generated
point are created and removed in a very short by the projectile may also be present.
time interval. The rapid loading generates
stress waves that subsequently propagate Numerous theoretical and empirical formulae
throughout the entire system. In addition to have been developed in attempts to predict the
the generation of stress waves, the collision perforating ability of a projectile. However,
produces a relative indentation at the area of the behavior of the perforating mechanism is so
contact. complex, and involves so many influential factors
that none of these formulae are completely sat-
It is generally true for low velocity im- isfactory.
pacts that the energy transmitted by wave prop-
agation is small compared to the initial kinetic Since no attempt will be made to conduct
energy and is thus usually neglected relative a complete theoretical investigation on the
to the energy consumed in the local indentation, subject of tank head failure, experimental evi-
Furthermore, the process is often treated as dence will be introduced to assist in identify-
isothermal for the sake of simplicity, so that ing the mechanism under which the tank head may
temperature and other thermodynamic effects fail.
need not be considered.
Figure 2 shows the impression made by the
A major objective of the investigation is striking coupler on a tank head in Test No. 1
to identify the modes of failure which would at 4.3 mph speed. There are clear indications
be created by a striking coupler on a tank head. that excessive shear deformations were produced
To supply the proper background, a short summary (as shown by arrows) at -the impressed areas J
of possible mechanisms for plate perforation which correspond to the upper edge corners of
process is presented. t%, striking coupler when the head was struck
neAr the center of the head. The degree of
The perforation of plates involves the :.near deformation would be increased as the
simultaneous action of crack formation, spalling, striking speed increases. Eventually, the tank
elastic and plastic deformation and wave prop- head would be sheared at more severe impacts.
agatign, friction, heating and even shattering Figure 3 shows a typical shc..r failure when a
of the striker. The plate may fail in a variety tank head was struck at a speed of 15.7 mph.
of ways among which are: 1) plug formation,
2) petal formation or dishing, 3) ductile hole In addition to the striking velocity, the
enlargement and 4) fragmentation. They are vulnerability of tank heads increases with
illustrated in Figure 1. Physically, plate increasing internal pressure. As shown in
failure appears to occur by a combination of Figure 4, internal pressure offers considerable
these various patterns, with one of the mech- resistance to the overall deformation of the
anisms predominatihg. tank head. Consequently, the force applied
would produce a deformation which would be
Plugs are more likely to be found in the highly localized around the area of force appli-
case of hard plates of moderate thickness or cation. As a result -f deformation concentration
in the case of projectiles with sharp edge con- characteristics in the area of force application,
tours. Investigations of the mechanism of plug the tank head would fail wth less overall tank
formation indicate that the plug failure mode head deformation and at a hover energy level as
is produced by plastic deformation along a sur- compared with the case of non-pressurized tank
face of maximum shear, heads. It should be noted that such localized
deformation characteristics will become more
Petal formation (or dishing) occurs where thin severe as the internal pressure increases.
plates are struck by a projectile at low veloc-
ity. The shape of the displaced plate is often A typical failure example of pressurized
assumed to be similar to that obtained under tank head is demonstrated in Test No. l4. In
quasi-static conditions, and an energy solution Test No. "h, a tank head was completely punctured
for a rigid plastic target material can be en- when the tank head was pressurized to 40 psi at
ployed. The energy components considered should 16 mph speed. As observed in Figure 5, the
include the elastic, and plastic works performed primary failure was attributed to the plug for-
in the displacement of the head, the accelera- mation resulting from the great magnitude of
tion of the particles in the deformed part of shear force which was applied at the top edge
the plate, the kinetic energy of the entire of the coupler knuckle impact impression. The
system and the heat produced by friction. The secondary failure which took place at the lower
analysis usually is based on the hole opening edge was due to the dishing and this portion
process with assumptions that the striker will of the tank head failed in tension.
neither disintegrate nor plastically deform.
In addition, the elastic vibration and the wave Based on the primary zode of failure by
effects are also neglected. The ductile type the maximum shear developed around the contour
of failure is the kind most cornonly observed of the impressed area, the magnitude of the

201
maximum force which the tank head can sustain (2) Permanent Indentatiin Estimation P,
will be estimated.d= 5 x i09 'D)2 '21i (-g v e-OO1 eP°
For the case of knuckle impacts, the defor- d g

mation characteristics would be substantially

different from the case of central impact of where d = estimated permanent indentation (in.)
tank heads. Generally the knuckle area is rein- D = diameter of tank head (in.)
forced by a reinforcing plate. In addition, the
transition weldment is attached between the tank h = thickness of tank head (in.)
head and the stub sill. This entire attachment
detail provides a considerable restraint in the
knuckle region against deformation. Thus, the w2 = weight of struck car (lbs.)
tank head would be torn at the transition weld- = str
ment area where an excessive shearing force king velocity
2 (in./sec.)
would be developed. Such phenomenon can be g = 386 in./sec. 2
easily appreciated by observing Figures 6 and 7. P, = internal pressure (psi)
Figure 8 shows the failure mechanisms when the
tank head is struck at the knuckle area. It P0 = 15 psi
indicates that the primary failure of the tank
head is in shear developed at the tank head (3) Maximum Impact Force Estimation
attachment area and the secondary failure of the
tank head is in tension by dishing the area 6Pi + P 0.6
above the knuckle. F = 35 x 10 (d)1 '5 (h)3 ( P .
D P
0
In sharp contrast with the case of internal
pressure, the liquid content in tank cars could where F = estimated maximutm impact force (kips)
serve as an impact cushion medium to absorb a d = permanent indentation (in.)
portion of the impact energy. However, an exact
evaluation of the contribution made by the con- D, h, Pi and PO = as defined in the pre-
tained liquid to the overall strength of the vious formula for con-
tank head would require an extensive investiga- puting d
tion of the interface behavior of the tank head
and the liquid. Thus, for the time being, a (4) Failure Criteria
simplified method of analysis would be sought.
Instead of seeking a solution by means of shell For central imppcting
theory, the solution for circular plates sub-
jected to a concentrated load will be employed
for the first approximation. Taking the first " FChL u
stress is
five-term approximation, the maximum
given by For knuckle impacting
=1.81 F 1.81F
Rh Rh r
where F = total applied load
R = tank radius where F = total impact force (kips)
h = head thickness
C = constant which depends on the envir-
The failure criterion would be established on the onmental condition inside the tank,
basis of the ultimate shear strength of tank taken as Unity for full tanks and
head. That is to say, if T t Z u, the f il- 0.5 for empty tanks, respectively
ure of tank heads would be expected, wre C IC= ultimate shear strength of tank head
is the ultimate shear strength of tank head material ksi)
material. h = head thickness (in.)

Results of the study are summarized as L = perimeter of impressed area (in.)

follows:
Despite the fact that a limited number of
(I) Impulse-Momrtum Relation old cars were tested, which had been subjected
to fatigu and corrosive environments while they
m1 (v1 - v' j T F dt were in service for a rreat number of years, the
analysis deduced from the test yields results
where ml a rmss of striking car which correlate well with experimental failure
T = impulse duration data.
v I = initial impact velocity Table 2 presents t1'e impulse-momentum rela-
v = velocity of striking car at tionship for the striking car. The total impulse
t= T was evaluated by measuring the area confined by
the force-time curve. The momentum change was

202
calculated from the mass and the changing in In conclusion, the results of this study
the measured velocity of the striking car. A indicate t1-at (1) filled non-pressurized tank
thebsered loabty the orrationgo e
2, cars are less susceptible to puncture than pres-
observed in Table 2, the correlation of the surized cars (2) empty cars are more vulnerable
impulse with the momentum change is very satis- to puncture than a liquid fled tank, (3) tank
factory within engineering accuracy. heads when struck near the knuckle radius are
more susceptible to puncture than when struck
Table 3 shows the comparison between the at the center of the head, and (4) the primary
estimated permanent indentation with that meass- mode of failure was attributed to a plug forma-
ured in the tests, whereas in Table 4, the tion due to the shear force exerted on the coup-
results on the maximum impact forces and the ler impact impression.
criteria under which the tank heads would fall
were tabulated. REFERENCES

As can be seen in Tables 3 and 4, the 1. Goldsmith, W., "Impact, The Theory and
analytical results and results obtained in the Physical Behavior of colliding solids,"
test agree well, with the exception of a few Edwa Arnorodo, 1960.
cases in which some unmeasurable factors such Edward Arnold, London, 1960.
as fatigue and corrosion might have played an
important role. It also can be concluded that 2. Fugelso, L.E., "echanics of Penetration,"
the assumption of plug failure due tc shear Vol.
as the primary mechanism appears to be correct. .962.
With only two tests conducted for the case of 3. Cristescu, N., "Dynamic Plasticity," North-
empty tanks, no decisive conclusion can be Holland Publishing Co., Amsterdam, 1967.
drawn with regard to the impact resistance of
the head when tanks are empty. However, there 4. Everett, J., "Impact Tests on Full Scale
is an indication that tanks can only resist Tank Cars," RPI-AAR Report No. 27053,
half of the impact force when empty as compared January, 1971.
with the case when the tank is filled with
liquid. In general, the maximum impact force 5. Hertz, I., 'Ueber die Beruhrung fester
to which tank knuckle heads can be subjected elastischer Korper," Journal fur die reine
is approximately 500 kips for liquid-filled und angowandte Mathematik, Vol. 29, 1882.
tanks and 250 kips for empty tanks, respectively.
For the case of central impact, the ratio of 6. Tsai, Y.M., "Surface Waves Produced by
approximately 1.6 can be multiplied to the Hertziee Impact," NSF-GP-2010/5, Nov., 1966.
maximum endurable impact force estimated for
the case of knuckle impact. The maximum impact 7. Davis, R.M., "The Determination of Static
force which tank heads can sustain decreases and Dynamic Yield Stresses Using Steel Ball,"
with increasing internal pressure by the amount Proc. Royal Society, London, Vol. 197,
Po 0.6 p. 642 99
of (P O )o.6 The results are presented p 16-432, 1949.
, 0 8. Tsai, Y.M. and Kolsky, H., "A Study of the
in Table 5. Fractures Produced in Glass Blocks by
J. Mech.
As for the final and important note, it Impact,"
pp. 263-278, 1967.Phys. Solids, Fol. 15,
appears that the maximum force developed during
impact would literally independent of the tank 9. Raman, C.V., "On Some Applications of
geometry when the indentation formula is sub- Hertz' Theory of Impact," Physics Review,
stituted into the impact force equation. This Vol. 15, pp. 277-2814, 1920.
is misleading for the following reason.
10. Yang, C.S.J., "Application of the Hertz
Fifteen out of seventeen tanks tested were Contact Law to Problems of Impact in Plates,"
equipped with the head plate of 7/16 inch in NOLTR 69-152, U.S. Naval Ordntazce Laboratory,
thickness. Moreover, the geometry of tanks September, 1969.
tested ranges from 78"0 x 1/2" tk. to 88"0 x
7/16" tk. Subsequently, with such narrow var- 11. Yang, C.S.J., "A Study of Hertzian Impact
iation in the geometric parameter designated on Glass and Aluminum Hemispherical Shells
by D/h, the influence of the tank geometry would with Mitigator," NOLTR 70-Il, U.S. Naval
become irrelevant. Since tank car construction Ordnance Laboratory, September, 1969.
allows more freedom for altering the thickness
than the diameter, a large range in the plate
thickness is recommended for the future exper-
iments.

In addition, these formulas should be con-

sidered engineering npproximations since the
exponents applied to all the influential param-
aters included in the indentation and force
formulas contain only two significant digits.

203
 Plug Pefol Ducile Woe Fraomentation

Formation Formation Enlargement

or Dishing

Figure I, POSSIBLE MECHANISMS FOR PLATE

PERFORATION -

Iva heorDefrm-0f
______________

204
P1 2OPfil

PrimaryFilr Mode (Plug Formation by Shear)

Print of Coupler Hs ,_

ilk____________________________________

Figue -TNK
3 RIMAY
H-AD AI1

ofHighSh
_Airo

ForceFoc I

Non-Pressurized Tank Head. Pressurized Tank Hood

Figure 4. SCHEMATIC VIEWS OF TANK HEAD DEFORMATIONS

205
 11 UL- . *

Figure 5 TANK HEAD IMPACT FAILURE (Complete Puncture)

*; Reinforcing Plote

Area of High Shear Draft (stub) Sill

Transition Weidment
Figure 6, SCHEMATIC VIEW OF KNUCKLE DEFORMATION

206
 p, x 10 psi11

S7 INDENTATION OF KNUCKLE AREA

V 216.1 MPH.
Pz 20Opsi

*Secondary Failure Mode (Dishing by Tension)

joj

L_ Figure 8 KNUCKLE- HEAD FAI LURE (Complete Puncture)

207
 TABLE 1
FLU SIZE HEADIW0ACTTEST 1MSULS 5U2#WR
Impct Internal Striking Struck Read Head Damme
Izpact Speed Outage Pressure Car Wt.j Car Wt. Size ImPact Permnent Inl e
Noe (mph) (Pti) (lbs) (Ibs) (in.) Location tatlon (in.) Failure

1 4.3 2 0 128,90o 96000 78 x 1/2 Center 30x 2-11A

1
24 7.2 2 0 " 10730 8o x 7/16 0 x &1/2
6 10.2 2 0 128,900 88 x 7/16 " 60 x n.1/4
8 8.7 2 20 128,200 88 ;c7/16 ' 48:x 8-1/4
1.0 11.0 2 40 " 108)00 80 x 7/16 4 x 7-1/2
32
12 14.o 2 4o - 107,500 83 x 7/16 36 x 8
114 16.0 2 40 " 130,000 88 x 7/16 'S Puincture
15 14.9 2 20 - 128,800 88. 7/16 6r x 16
6
16 15.o 2 20 127,000 88 x 7/1 " 6o x A-1/2
24
17 15.7 2 3$ 127,4D0 88 x 7/16 60 x -1/2 Shear
18 15.7 2 20 '- 107,600 83 7/16 ' 66 x;1i/2 Shear
19 16.0 2 0 107,3 x 7/16 72x 16
20 8.5 100 0 240,9oo 83 x 7/16 " 51 x 9
22 16.0 2 0 " 96,600 78 x 1/2 Knuckle 0 x 12
23 16.1 " 20 108,400 80 x 7/16 * puncture
214 14 2 10 'S 128,300 88 x 7/16 -L Shearl
25 16.1 100 0 28,000
4 88 x 7/16 "Puncture

- Ntodata.
* Threshold puncture.

TABLE 2

IMPULSE - MOMENTUM RELATION

Test

No. Total Impulse (kip-see) Change in Momentum (kip-see)

1 6.52 7.45

4 11.95 15.90

6 16.60 15.20

8 19.48 18.75

t0 24.38 22.40

12 16.60

14 12.02 27.02

i5 42.90 41.70

19 44.28 41.80

FNo data available.

208
 TABE 3
PERHWEN INr*ATION
Internal Head Damage
Test Outage Pressure Permanent Iinfat'lon, d 1n.) HMaM IrenratLon
NO. () Pi (psi) ' Ct ated az/sured Recorded (in.) Failure

1 2 40 2. 2.75 .35
4 2 0 6.35 6.50 5.0
6 2 0 n .6o 11.25 13.0
8 2 20 1 9.65 8.25 S.r
19 2 40 8.05 7.50 12.8
12 P' 4o 10.53 8.00 0
10 02 40 13.00 0 19.4 puncture
15 2 20 16.oo 16.xo 18..6
16 2 20 16.67 14.2 *n
17 2 30 16.70 A__.25 Shear
18 2 20 14.35 15-50 Shar
19 2
10o
lO0 0 15.75 16.Se
7.05 9.00*
22 2 0 10.70 12.00
'23 I 2 20 14,.25 N nctur
24 dta 2 1 10 16.:)o • hear7
25 1 100 0 11-9
, *puncture

*Nlodata.
Threshold puncture.

TABLE 4
MAXIHLI4
I.'ACT FORCEANJtFAIL"V l'IThhIA
Internal MAXin IMUact fore, F (kits) Failure Criteria -
Pressure Ca-l-ated Shear Ltress Uel
Test Outage Pi Based on Ba ed on Mimulm Failure
No~. MA2.. (si)A.d Estimated d esrd Meastred 7re retical Ufltimate P4~de
1 2 0 44 4? 55 2.9 38.5 lnt
4 2 0 100 103 9 7.6 36.5 _ _

6 2 0 169 58 141 . 38.5. _ nt

8 2 20 210 165 1 l7.l 38.5 !1Iint
10 2 40 288 256 2533 22.t 38.5 Dent
12 2 4o0 3531 7=11 ' 454 .J9u N." t
14 2 40 527 7. a,
15 2 20 4e 114,
W " 54.3 05
16 2 20 477 -WI 36.5 38.5 5,

18 2 2o U.9
( 1110- io9
1.5 -

19 2 0 361 41
hl( 27.6 3b15 ._ _ _
20 100 0 i10 139 li.e 13.1'* 38.5 I3.-r
22 2 0 324 P 4. 5.5
164 e.
23 2 20 496 1 IA11 51. 38.5 -ar. Iture
Onlnty
24 2 10 446 * 365 41.9 1,
25 100 Al21 154 ~ i5 uc'r
N~.*

*352data.
' C 0.5 fr empty tans; otherwvie, c 1.,.
* Threshold juncture.

209
 TABLE
FAXLUME FORCM ANDFAILURE WIEES
Estimated Failure Force YAxima "*esured Farce Failure
Teat (ips) (kips) Yode
1 800 55 Dent
4 800 69 Dent
6800 141 Dent
8 i 201 Dent
10 366 283 Dent
12 3616 454 _ Dent
14 366 429 Puncture
15 481 360 Dent
16, 81 375 Dent
17 1411 1465 nicar,,

18 481 4S62 "hear

19 f-00 W Dent
22 507 321 Dent

23 301 314 Puncture

214 367 36
25 256 ___________

* Threshold I.urct.r(-.

DISCUSSION

Mr. Baker (Southwest Research Institute): reactors If a control rod were to be thrown out.
I want to complimentyou on an excellent presen- They ran a number of penetration tests with long
tat'on. I wondered, in looking at these movies, if rod-like objects thrown against steel plates and
you were familiar with some work done at Stan- a group of formulas was developed. You might
lord Research Institute some years ago on the be able to correlate Stanford's results with
problem of penetration of containment shells for whether or not penetrations occur in your study.

210

4-
 .I

A STUDY OF IMPACT TEST E'FECTS UPON FOAMED

PLASTIC CONTAINERS
Don McDaniel
Ground Equipment and Materials Directorate
Directorate for Research, Development, Engineering
and Missile Systems Laboratory
US Army Missile Command
Redstone Arsenal, Alabama 35809

and
Richard M. Wyskida
Industrial and Systems Engin..ering Department
I
The University of Alabama in Huntsville
Huntsville, Alabama 35807

Various types of foamed thermoplastic materials are utilized as

cushioning systems in military containers. The primary aspects
of cushioning design theory currently being utilized by container
designers to predict the response of packaged items when subjec-
ted to free-fall drop tests are discussed. The test results of all
readily available military container designs utilizing foamed
plastic cushioning systems are analyzed to determine the statis-
tical significance of the various factors affecting container
response. The analysis of the factors involved is then compared
with cushioning theory. The superimposed dynamic cushioning
curve technique Is then presented as an Improved method for
presenting the cushioning properties of foamed plastic materials.

INTRODUCTION these findings, the dynamic cushioning curve

technique currently being utilized for response
A container designed for military use, prediction of military containers was Investi-
such as for shipping missiles and missile gated to determine if reaponse predictions for
components, Is required to perform its basic foamed plastic containers were adequate.
functions throughout its operating life.
Current military policy, based on the need for
worldwide deployment, dictates that containers STATE OF THE ART
are to be capable of withstanding the rigors of
a logistic pattern geared to worldwide distri- One of the most important considerations
bution. Consequently, the container and its in the desigp of military shipping containers is
contents, in combination, must withstand all the selection of the materials that are utilized
environments and modes of transportation, In in the container III . The internal packing mate-
addition to the hazards peculiar to the handling rials, frequently called cushioning systems, are
of material in trariit. designed to protect the contents of the containers
from severe shocks Induced when the c,,ntainer
An extensive literature search was is dropped. One oi the most promising types of
conducted of the qualification test programs materials currently being utilized for cushioning
performed by the military on missile containers systems is foamed plastics, frequently called
that utilize foamed plastic cushioning. Data cellular plastics. These are among the most
from those test programs that had sufficient exciting new materials that have emerged from
documentation were then utilized as the basis the chemical laboratories during the past 10
to determine which factors Influenced the years. This versatile family of materials is
response of the packaged item. Based upon now being used for Insulating, filtering,

211
cushioning, floating, and a multitude of uses The static stress is determined by:
that is continually growing. W

Foamed plastics are particularly

attractive for use as low-cost, easily fabrica- where Ss is the static stress (psi), W is the
ted cushioning systems in shipping containers, specimen weight, and A is tie footprint of
Lightweight plastic foams are being used in specimen in the cushion (in. ). A different
conjunction with, and replacing, many of the curve is required for each drop height, thick-
standard packaging materials, because they ness, and type of material. The curves are a
have unique characteristics and properties, good indication of the protection to be expected
are easily fabricated, and are economic when for a particular cushion scheme. Fig. 1
compared to the rising costs of wood, paper, illustrates a series of dynamic test curves for
rubber, and metal products. various thicknesses of 2. 5-pound density
polystyrene foam [2].
have been
To date, foamed plastics
incorporated into military container designs
on a limited basis. One limiting factor is the
difficulty the container designer has in predic- 16o
ting the response of the materials when subjec-
ted to the extreme environmental conditions 140 FOAML
THICKNESS
encountered under military deployment. These 12 T0I .--
conditions are simulated by extensive environ- 120 -- __ \ 1,,./ 3 in.
mental testing programs. Of particular con- - in.
cern in container designs that utilize foamed 0
plastics are impact tests conducted at the tem- 80 -
perature extremes. The thermoplastic nature
of the foamed plastic materials induces a tem- , 60 -- "- -

perature sensitivity
in the container that can
response cause
during variations
these tests. --

In order to fully utilize fom aed plastics 20 -

for military containers, container designers - L

require sufficient information to accurately 0.1 0.2 04 0.8 1.0 2.0 4.0 8010.0
predict response variations. Current practice STATIC STRESS (psi)
is to provide the container designer with
cushioning data for each type and thickness of Fig. 1 - Dynamic cushioning
cushioning material. These data are provided curve for polystyrene foam
in the form of dynamic cushioning curves. For (30-in. drop heights 70'F tern-
any particular container design program, the perature, 2. 5 lb/t density)
designer is generally given a maximum allowa-
ble fragility level which the packaged item is
permitted to experience when packaged in its
container. Also, the particular organization FOAMED PLASTICS FOR MILITARY
involved will have an established testing policy PACKAGING
defining appropriate impact tests, which
generally is given as a number of free-fall Foamed plastics comprise a versatile
drops to be performed at certain prescribed new family of lightweight materials which are
heights and temperatures. These parameters made In a variety of processes remotely
form the basis for the design of the shock resembling the making of bread. These
mitigation characteristics and the selection of synthetic materials are polymers which have
one of the various cushioning schemes available been expanded in volume by a gas so that they
for the container. If foamed plastic cushioning have a uniformly cellular structure which can
is to be utilized, the designer will then generally loo!. like an extremely fine honeycomb or a mass
structure his cushioning system using dynamic of very tiny ping-pong balls fused together. The
cushioning curves for particular materials, cells of some foamed plastics are large enough
to be seen; cells in others are so fine that a
Dynamic cushioning rurves are generated microscope is needed. These materials come
for a particular type and thickness of cushion by in a variety of consistencies ranging from that
performing drop tests using standard weighted of raw cotton to hardwood. Flexible, semi-
specimens that are dropped onto the cushion, rigid, and rigid foams are available in densities

212
of about 0. 1 to 80 lb/ft 3. the blowing agent, the polyol, the catalyst, and
the emulsifier are all preblended and the
Three types of foamed plastics, namely, isocyanate is added at the last minute. When
polystyrene, polyurethane, and polyethylene, this technique is coupled with a still more recent
have been utilized in packaging applications in development in which the ingredients are "frothed"
the military. The most widely accepted foamed prior to dispensing, producers can achieve
plastic material utilized for packaging is greater uniformity of foam densities, lower
expanded polystyrene. This unique material is tempeiatures and pressures, and better flow
manufactured in the form of small translucent characteristics, and can develop thinner skins;
beads or granules containing an expanding agent. thus, less costly jigs and dies can be utilized.
The particles are free flowing, and when expan-
ded by the application of heat they produce an Polyurethane foam is frequently utilized aE
opaque foam which has little or no odor and is a cushioning material in packaging because of its
nontoxic. Expandable polystyrene materials extremely resilient properties. These materials
offer many properties desirable in a packaging are frequently supplied in large buns or slab
material, I.e., lightweight, high strength-to- stock which can be easily cut or shaped into the
weight ratio, low moisture absorption, and good required shapes. Also a foam.,in-place technique
insulating properties, can be used for packaging small items. Here
the item is positioned in the exterior container,
Expandable polystyrene materials are and the polyurethane foam is introduced in a one-
processed in a number of different ways for use shot liquid form into the cavity between the item
in packaging; however, the most common method and the container exterior wall and allowed to
is steam chest molding. The beads or granules "foam up" and capture the item.
are normally "prefoamed," by application of
heat or steam, to the desired density prior to The third foamed plastic, polyethylene
molding. The hydrocarbon foaming agent within foam, is manufactured in slab form, most 3
the individual granules causes the softened commonly at a density of approximately 2. 0 lb/ft ,
material to expand into discrete multicellular which is expanded about 30 times from the solid
foam spheres or pellets. In the molding opera- state of polyethylene. It is a cellular, lightweight
tion, these prefoamed particles are confined material with each cell closed off from its neigh-
in a retaining mold, then subjected to steam bor. Polyethylene foam is a tough, resilient
and pressure, causing them to expand further matorial having good flexibility characteristics.
and knit together to form a unicellular homo-
geneous plastic foam article. The unlinited Polyethylene foam can be fabricated with
variety of shapes and sizes that can be molded conventional hand tools or power tools such as
and the low cost of processing are benofits of band saws and routers. Due to its flexible
interest to the molder and ultimate corsumer nature, polyethylene foam is most easily fabri-
of this packaging material. Steam-chest- cated by tools or equipment with blades or bits
molded expandable polystyrene has excellent having a slicing type action. Polyethylene foams
impact and shock absorbing properties. can als- ie fabricated by electrically heated
resistance wires. It may also be thermally
Polyurethane foam materials have shaped by the use of contoured molds equipped
attracted considerable interest for industrial, for both heating and cooling.
military, and commercial packaging applica-
tions. These materials have extremely Polyethylene foam provides high energy
versatile properties, and formulations can be absorbing characteristics for insert packaging
varied from extremely rigid to very flexible, applications. Its springback rate is slow enough
within a wide range of densities. Polyurethane so that the energy absorbed during a sudden
foams are of two basic types, polyester and shock is.not released with equal rapidity; rebound
polyether materials. Polyester-based occurs at a relatively slow rate. Due to its rela-
urethanes were the first to be developed, but tively predictable properties, it is one of the most
due to relatively high cost are not being utilized frequently utilized packaging materials in military
extensively for packaging. Polyether-based applications.
foams, most ,requently used in packaging, are
made by reacting toluene diisocyanate with
polyglycol materials in a simple mixing
operation. DATA COLLECTION

The greatest technical advance made in In order to establish a meaningful appraisal

recent years has been the development of what of cushioning design theory to foamed plastic
is known as the "one-shot" system, in which containers, a substantial investigation was requi-

213
red. Many military organizations are involved DATA ANALYSIS AND SYNTHESIS
in the design, development, and testing of
containers; and they were solicited for data on The drop test data for each container
the drop testing of containers that utilized experiment tabulated in Table 1 were analyzed
foamed plastic cushioning. Also, a compre- to determine the effect of the factors involved.
hensive literature search was performed by the The test programs conducted by the various
Defense Documentation Center, Alexandria, organizations in their development programs
Virginia, on any container drop tests conducted were not uniform. However, the tests contained
within the last 15 years. Every program of three main treatment effects that can be
drop testing of a container that utilized foamed analyzed: the effect due to change of temperature,
plastic cushioning materials was reviewed, the effect due to change of stress level, and the
effect due to different types of materials used
It became quite apparent at the outset that in the cushioning.
hundreds of container drop test programs would
not provide usable data. The primary difficulty A factorial design was utilized to analyze
was that there was no quantitative measure of the data, and the interactions were included in
container response, such as the G-levels the analysis. A factorial design requires that
experienced by the packaged items. Many drop each treatment be applied at every level of every
test programs were conducted only at ambient other treatment [31. Fortunately, the drop test
temperatures, and many were one-shot tests programs were conducted so that in each test
with no repetition or duplication within the test. each type of drop was conducted at each tempera-
It was also determined that some container ture.
designs were qualified for issue into the field
with no testing at all. In the drop test program, for any particu-
lar container, the fixed factors in the experimont
The results of the investigation provided would be the weight of the container and its con-
15 candidate containers using foamed plastic tents, and the drop height. The variable factors
cushioning that had undergone adequately were of different magnitudes in some of the con-
documented drop testing. Three of the con- tainer tests but all experiments included the
tainers were designed by the Navy for Sparrow, following:
Brighteye, and SUBROC. Six of the containers
are Army designs for the missiles and spare Temperature (Ti) i from -650 to
parts for the Chaparral, Redeye, and Shillelagh 160°F
systems. The remaining five containers were
designed by container manufacturers for use Kind of drop (D4) j corresponds
by the military in missile systems applications. J to bottom, top,
end, 45 deg, etc.
The basic drop test data for each con-
tainer are tabulated in Table 1. The bases for A mathematical equation can be written
the data are the accelerometer measurements that defines one observation value which is
of G-levels experienced by .e packaged item expressed In G-levels. This observation value
when subjected to the type of drop indicated, can be equated to all the variable factors that
Also, the tabulated G-levels are those taken have an effect on the G-levels and is given as
from the accelerometer whose axis is oriented
parallel to the drop axis. These are usually Xij k = TI + D + TDij + ij k
the maximum G-level readings.

Three factors vary throughout the data, where Xijk is an observation value made with
any one of which could have affected the G-level T at level i, D at level J, and a random error
response. The temperature and the attitude of term at level k; c is the error term introduced
the drop both varied within the tests. Also, the during the repetition of the experiment; and TDij
type of cushioning material varied from one
type of container to the next. In the analysis is ar interaction between T and D that affects the
the requirement is to asswss these three main G-levels. From this mathematical equation, four
factors: temperature, type of material, and null hypotheses can be postulated.
stress levels (a function of attitude of drop), to (1) Temperature factor: Changes in
determine
response ofwhich of these factors
the packaged influenced
item when the
subjected ttemperature
emperature ffcton
can have an effect on thgesn
the G-levels.
to the drop tests. It is also important to note that temperature is
a quantitative factor (the levels of the factor can
be expressed In numerical terms) and can have

214
 W! !o NW 0N owO ~ 0 CO09In
o n oC4
.-. O t

- .3A 00C3- 000

mD~ . OWWe2 f0
o r w0 t- 0 00 rwN
t&0 -, ON- Nq

01

o) N. a W o c

xu N NG

N
4.
N~tO.W
co i o
:a
ca 0. c 0 'sNc 0-4 -N-0Or)v) n N O O D - O 0 0
w"
O e
v
~ t

t-4U4 -a0 l N

-d :;- V --- N-N

ON N

40 c.l o - (n -vt oNVqo. w -w 0 O W OO O N O.

oavo
0D NONO NNo co0 44 tOo:;

0 0 -. Or3.tW e D0 C O
oil O ~ O ~ 0O W t0 O... 0 0 O O 0

4,,

(D . 0

So V

215
a significant linear component. polyethylene, polystyrene, and polyurethane foams.
A technique is required for grouping all the drop
Hypothesis 1. Temperature does not tests in order to analyze the type of material effect.
affect recorded G-levels significantly.
Certain limitations regarding the grouping
Hypothesis 2. Temperature does not of the individual drop test data should be recog-
affect recorded G-levels linearly in a signifi- nized at the outset. The drop tests were conduc-
cant manner. ted by a variety of organizations that undoubtedly
used different types of test equipment and response
(2) Kind of drop factor: The attitude monitoring techniques. The procedures have
of the container at impact can have an effect on become relatively straightforward as a result of
the G-levels. extensive testing of this type by the many testing
laboratories over the last few years. However,
Hypothesis 3. The kind of drop certain subtle differences do exist, and these
performed does not have a significant effect on
recorded G-levels.
differences are confounded in G-level response
values that will be operated upon. It is reasonable,
j
however, to assume that these subtle differences
(3) Interactions: The interaction are an order of magnitude less than the effects
between kind of drop D and temperature T can generated by the major variables in the experi-
have an effect on the G-levels. ment such as the type of material.
Hypothesis 4. The (T) x (D) inter- The data from the 15 individual container
action does not have a significant effect on tests were separated according to the type of
recorded G-levels. cushioning and stress level to provide the data of
Table 3 with each observation expressed in terms
The four null hypotheses were tested by of G-level response of the packaged items.
an F-test at 90-percent significance for each Sufficient data existed to provide two G-level
experiment; the results are tabulated in Table readings for each level of each factor. These
2. In each instance marked with "No," the null were considered as replications for analysis
hypothesis stating that the treatment in question purposes.
has no effect was not rejected as a valid conclu-
sion. Those marked with "Rej" signify that the A factorial design can be used to analyze
null hypothesis was rejected, these data since each treatment, in this case,
temperature, type of material, and stress level,
The null hypothesis regarding type of drop was applied at every level of every other treat-
was rejected in all but four of the container tests. ment. The factorial design was selected for this
This indicated that considerable importance is analysis approach because there are four possible
attached to the type of drop, which is consistent interaction effects, and these interactions between
with the emphasis that is placed on this factor the various treatments could be most significant
as the abscissa of dynamic cushioning curves, and should be separated for analysis.
The analysis of the individual container The variables in the experiment included:
experiments provided information on the effects
of stress levels and temperature on container °
Temperature (T) I from -65 to
response. In order to provide information on 1603F
the effect of different types of materials used as
cushioning, an additional analysis is required. Stress levels (D ) j from 1 to 5

The type of material used in a container Type of material (M1) I Is polyurethane,

cushioning system was fixed within the container polyethylene, or
prior to the drop tests discussed earlier. There- polystyrene foam.
fore, the individual analysis did not include
effects of different types of cushioning material The mathematical equation that depicts one obser-
since no significant variation existed in any one vation of the experiment and includes all the
experiment. However, when the 15 separate variable factors involved would be:
container designs were considered as a group,
considerable variation existed in the type of Xijk = T + D + M1 + TD + TM11 + DMjl
material utilized. Included in the group were
container designs that utilized the three main + TDM ij + ijik
types of foamed plastics used in packaging:

216
 TABLE 2

Summary of Drop Test Experiments

Packag"Tmprau size of" uain Treatment of [.tmelot *

Drop Factorial Cush.

Name Container it N eWeiht
Height (7) AId"l Ipfl-
(Ib
tl*)19 (in.) ,00 Aimb Ith
Res meat (T) (D) (TXD) TLIMtsr (Foam
1 86 14 x 14 135 24 .0 70 5 2 ' aS Rej Rej No PU
Brightte 68 x 12 x 13 142 30 .65 70 I60 3 2 3x No Re) No No PS
SUMROC 22 x 23 x 16 13 30 0 70 - 4 3 2 X4 Rej Rej No PC
Monopa 0 x 15 x 13 30 30 -65 - 160 5 2 xS Re) No No PU
Chaparral 15 dia x 22 t0 30 -65 70 160 3 2 3x3 Re No No Re) P3
Chaparral 18 aia x 8 25 30 -65 70 160 2 2 3x3 Re Re) ReJ Rej PE
Wathead
Chaparral 2242ax35 31 30 -65 70 I60 4 2 $x4 Re) Rej ReJ Re) PC
G&C
Shillelagh 13x 153x 2 60 30 -65 70 160 4 2 3x4 Re) Rej ReJ Re) PC
Crate Rite
Shillelagh 56 x 5 x 13 60 0 .65 70 160 5 2 3S Re) Rej Re) Rej PU
Halliburton
Redey. 56X x 10 30 30 -65 70 160 4 2 3x4 Re) Rej No No PS
Unipak
production
Shillelagh 56.25x 37x 17 60 30 -65 70 155 4 2 3x4 Rej Re) Rj Re) PC
37D
RWdeye $6x l0x 15S 30 20 .65 70 160 4 2 2x4 Rej Rej No No PS
R&D
Shillelagh 47 x 13x 13 60 30 -65 70 360 5 2 25 No No No INo PU
Skrdyne
Shillelagh 50x 13x 13 60 30 -65 70 160 5 2 3 a5 ReJ No No No PS
Molded

Shillelagh 50 x 13 4 60 0 .-65 70 160 5 2 3x5 Re) Re) NO ReJ Pj

;letl Drum
rejecitie his
NOTE:Re- of no elfect.
No* cannot reject the hypothesis of no effect.
FS * Polystyrene.
PU Polyurethane.
PC * Polyethylene.

TABLE 3

Combined Analysis Data Matrix

Krss
Levels Polrh Foam Polystyrene Foan Polyethylene roem
ipsi) eplication I Replication 2 Replication I Replication 2 Replication I Replication 2 Average

ow 0.0.2 62.2 82.3 0.6 43.1 27.9 25. 50.1

Temperature 2 0.3.0.4 66.4 73.7 36.3 71.6 73.5 23.5 Co.I
3 0.4.0.6 3.1 51.4 9.9 75.6 36.3 24.1 54.5
4 0.6.0.8 55.3 49.5 42.9 39.6 39.0 48.2 45.8
0.8 up 68.6 73.0 31.2 413. 35.6 63.1 51.7
aw 2 ~ V W
W.1E 31171 M
Ambient 0.0.2. 50.31 35 75.0 40.2 28.4 20.4 38.4
Temperature 2 0.2.0.4 40.2 35.8 85.0 32.2 1..8
36 42.2 40.2
30.4.0.6 4.0 35.6 71.8 84.6 29.4 28.6 46.1
0.6.0.8 40.00 2 3.303.0 26.1 42.3 33.7
0.0 3p
24.5 481.8 2.4 48.6 22.8 6.7 40.3
a1 I W! 7173 3EV M W t1.
High 1 0.0.2 3.3 38.4 36.2 9.6 24.5 44.2 30.7
Temperature 3 0.2.0.4 26.3 35.7 99.0 39.6 54.0 42.7 52.9
3 0.4*0.6 14.8 30.2 75.8 92.2 29.1 22.9 44.1
4 0.6.0.8 26.2 32.6 51.0 34.6 40.0 16.0 40.2
0.9 up 20.0 40.7 3.2 27.2 1.7 17.8 20.7
WI 33S
W!! lI
W.7t F
T&WK
aalue= ?Jat
s O-leoel reading.

217
where Xijlk is one observation made with interaction effect have more impact on
Tat level I, D at levelJ, and M at level 1; e recorded G-levels than the effect due to
is the error term introduced during repetition variations in stress.
at level k; TDij is interaction effect due to The hypothesis regar, -gthe third-order
the interaction of T and D, etc; and TDMijI is interaction could not be rejected. An interaction
that involves three factors such as this one is
an interaction of the three factors T at level I, frequently incorporated into the error term,
D at level J, and M at level 1 (referred to as It occurs as a separate entry here to maintain
a third-order interaction), as much definition as possible in the error term,

Eight null hypotheses were postulated but its mean square value was the smallest and
for this experiment, and F-tests were made could have been incorporated into the error term
at.the 90-percent signficance level to deter- with very little consequence, as can be seen in
mine whether or not to reject the hypotheses. Table 4.
TABLE 4
The hypotheses to be tested were as Combined Data Analysis
follows: ANOVA

Hypothesis 1. Temperature does Degee Wm of Mean 00%sigillrnce

not affect recorded G-levels significantly. Source ofFreedom quarts Square F-Test

Hyp othe sis 2. Temperatu re does Material (M)

Stress() 2
4 5,69.6
30 6.8
2,834.9
76.4
Signilicant
s ignifict
not affect recorded G-levels linearly in a
significant manner (Ti 2 3,392.9 1699.9 gnfcat
(M)x (1) 8 15,169.7 1,971.2 Significant

Hypothesis 3. The variations in (M)x( 4 3,657.8 914.4 Signifiant

static stress levels, as associated with type of (D)x (T) 8 1,441.0 1,0.5N S gnificant

drop, do not affect recorded G-levels signifi- (M) x(D)x(T) 26 1,628.2 102.2 NotSlgotlrcat

cantly. rTLinear
Error jL (1)
45 ] 11,545.6 2,886.7
356.6 Significant

Hypothesis 4. The type of cushioniug r 4 .,545.8 5

material does not affect recorded G-levels OTL 4,111.2

significantly.

Hypothesis 5. The temperature-

stress interaction does not have a significant
effect on recorded G-levels.
SUPERIMPOSED DYNAMIC CUSHIONING
Hypothesis 6. The temperature- CURVE
material interaction does not have a significant
effect on recorded G-levels. The analysis of the available data shows
that changes in temperature that were introduced
Hypothesis 7. The stress-material in the drop test programs to simulate world-
hiteraction does not have a significant effect wide extreme conditions have a significant
on recorded G-levels. effect on the response of the packaged item. In
the analysis of individval drop tests, all but
Hypothesis 8. The third-order two of the 15 container test programs had a
interaction (stress-material-temperature) does significant temperature effect.
not have a significant effect on recorded G-
levels. Temperature effects are not considered
in the dynamic cushioning curve technique that
Table 4 shows that the null hypotheses is currently utilized for presenting information
regarding the type of material effect and the on cushioning materials. The requirement is
material-temperature and material-stress a technique that provides for temperature effects.
level effects could be rejected. Further, the
mean square values indicate that temperature The superimposed dynamic cushioning
and material effects and the material-stress curve technique provides information on the

218
effect of temperature on the response of extreme temperature conditions. A particular
containers that'utilize foamed plastic cushioning static stress condition can be selected for the
systems. The dynamic cushioning curve techni- particular package design in the same manner
que provides information on the response of a as with the standard dynamic cushioning curve.
packaged item for one certain set of conditions: Thib selected static stress condition can then
thickness, type of cushioning material, and drop be projected onto the temperature curves and
height. This same technique can be extended to the results correlated. For example, if the
the superimposed dynamic cushioning curve design conditions required a 30-inch drop
technique, whereby the dynamic cushioning height and polyethylene foam were being consid-
curves that are generated at the military ered in 2-inch thickness, then Fig. 2 would be
temperature extremes (-650 and 1601F) are appropriate. By using the 70°F curve, a static
superimposed on the ambient temperature curve stress of 0.4 psi might be selected to provide
for any one set of conditions; one such curve 40 Gs of protection to the packaged item.
that the authors constructed (Fig. 2) was derived However, when 0.4 psi is projected on the -651F
from information provided by the manufacturer curve, the 40-G requirement is exceeded. The
of polyethylene foam, the Dow Chemical 0.4 psi would then be discarded, another static
Company [41. Fortunately, Dow provided stress would be selected for consideration, and
container designers with separate dynamic the correlation would be repeated until a suitable
cushioning curves for polyethylene foam at -65' response is obtained.
and 155°F as well as at ambient temperature for
a 30-inch drop height.

In Fig. 2 the temperature effects are CONCLUSIONS AND RECOMMENDATIONS

seen as a relocation of the curves. In this
instance the optimum point on the curve shifts The analysis shows that the effect of
horizontally to the right as the temperature changes in temperature introduced into the drop
varies from 1558 to 700 to -65°F. This test program of containers, to simulate world-
demonstrates that increased static stress is wide extreme conditions, was significant.
required to compress the cushion to its optimum Temperature effects were significant in the
when the material stiffens as it gets colder, majority of all individual container tests, and
polyurethane foam containers were extremely
12 -- sensitive to low temperature.
1- 165F
70OF The dynamic cushioning curve technique
100 -645'F- does not provide information on the effects of
temperature. However, present cushioning
theory can be extended to p,-'- vide the super-
2 imposed dynamic cushioning -urves, which do
-- present temperature effects. Fig. 2 demon-
. o strates this technique of constructing super-
W imposed cushioning curves for polyethylene
4 - - / -foam at temperature extremes.

20 Besides temperature effects, the other

two main effects that were analyzed were the
effect of changes in the type of material and
0.1 0.2 0.3 0.01.0 2.0 4.0 8.0 10.0 the stress levels within the cushioning. Poly-
STATIC STRESS (psi) styrene and polyethylene show the same shaped
curves, with polyethylene giving consistently
Fig. 2 - Superimposed dynamic lower G-levels. It can be stated that generally
cushioning curve for a 2-inch the main effect due to the type of material is
thickne3s of polyethylene foam discernible using dynamic cushioning curves
(2 lb/ftc density; -65, 70", and since the curves are segregated by the type of
155°F temperature; 30-inch material. However, the additional information
drop height) on temperature effects that is given in the
superimposed curves is required to assess the
Application of superimposed dynamic material-temperature interaction effects.
cushioning curves provides the container
designer with a tool for prediction of the The stress level effects were shown to be
response of a container design under the significant in the analysis also. The dynamic

219
cushioning curves and the superimposed dynamic REFERENCES
cushioning curve use the stress level as the
abscissa for the curves which provides complete 1. K. Brown, Package Design Engineering,
information on this effect. John Wiley a , Inc., New York,
April 1959.
The superimposed dynamic cushioning
curve is recommended for adoption as the 2. Dyllte Expandable Polystyrene-Principles
technique for presenting information on of Packaging Design, Koppers Co., Plastics
cushioning materials. This extended technique Division, Pittsbiurgh , Pennsylvania, January
can be utilized by manufacturers to present 1969.
information on cushioning materials, while
government agencies can employ the technique 3. Charles R. Hicks, Fundamental Concepts
in evaluating cushioning materials, in the Design of Experiments Hot RineFart
Winston, New York, 1964.

4. Packaging with Ethafoam, Dow Chemical

Co., Plastics Department, Midland,
Michigan, 1966.

DISCUSSION

Mr. Yang (University of Maryland): I Mr. McDaniel: That was the point of this
thought it was a very interesting paper. I have study. Really, fundamental to the container de-
two comments to make. My first comment Is sign, is how that item behaves in the container,
that I would like to speak up for both Navy and and we can take this information that is avail-
Air Force because I have consulted with both, able on the elementary level and apply it. There
and in the area of these energy absorbers, both is a lot of variability in how a particular con-
the Navy and Air Force have done a lot of work in tainer responds when you configure the material
getting properties and doing a lot of tests. As a and test as opposed to just testing, for example,
matter of fact, there are a lot of reports in this raw sheet material.
area. As a result of consultation with the Navy
and Air Force, I myself have five or six papers Mr. Yang: That is true. However, I do
in the area - especially these three materials. thnk lTh,
from this basic research, you can get
One of the papers was referenced by Professor a lot of ideas which can be used. My second
Shang in a previous paper which was presented comment is that I think this temperature effect
two years ago at the Shock and Vibration Sym- is excellent that you were looking at. Also, I
posium. The material was polyurethane, and think another thing you should take into account
also there are many other papers which do what is also the rate. I do not know what rate you are
you have mentioned in your paper. I would be dealing with, but that is another thing you should
very happy to go over some of these papers if look into.
you are interested.
Mr. McDaniel: The viscoelastic property of
Mr. McDaniel: I am familiar with them. a urethane foam does present a rate problem.
What is being done is primarily on a bulk cush- However, in a container that rate is married into
ion in a drop test environment, testing an indi- the test. You do not really have to address each
vidual cushion of a particular thickness as op- individual item if you test the basic configuration.
posed to a container with an item in it. Is that
true?
Mr. Schell (Navalflesearch Laboratory):
Mr. Yang: Yes, the researches to which I I was very much interested in the results on
was reeringh are on a more basic level rather your peak g, static stress curves for various
than on particular cases. They would study den- temperatures, where the optimum actually is
sities and thicknesses of each material at var- better for low temperatures than for higher
ious temperatures and strain rates. These are temperatures. I noticed that, for low static
basic research studies, but I do not know about stresses, the cushion behaved as one would ex-
tests of a particular container. pect, that is, it would stiffen up and give higher

220
g's. For the larger static stresses you got just of view, perhaps because one of the dimensions
the opposite effect. Have you made any conjec- in the space such as velocity of-impact is not
tures as to why this happened? Is there a pos- being included. Perhaps your times relative to
sibility that you are getting material breakdown response times are auch that it should be in-
at the higher static stresses due to embrittle- cluded. You might be overlooking an important
ment or something of this nature? variable. As an illustration, perhaps what you
are showing as a two-dimensional space should
Mr. McDaiiiel: I do not really know. Maybe be a three-dimensional volume. We do not seem
Mr. Yang would like to answer that. Al I know to be talking in terms of these variables. They
is how these containers behave. It would be are certainly present, and you can show theo-
good if we had this type of information available retically that these realms exist. The other re-
to us as designers, but we do not. acticn I had is that you are always putting accel-
erations up in g's. Gravity has nothing to do
Mr. Langhaar (duPont Company): What with it.
maximum thickness of foam did you test, and
did you discover any method of extrapolating the Mr. McDaniel: Well, that is the real world
results to very large thickness, for example, right now. The way we are approached as a con-
24 inches? tainer designer Is: "Given a missile system
with a particular fragility level, design on a
Mr. McDaniel: I think the thickest material particular basis to provide the necessary pro-
we looked at was in one of the Redeye polysty- tection." So, that is what we live with, and
rene versions which is about 8 or 9 inches. whether we can ever get our missile people to
Again, once you select a static stress, and once talk to us in other terms is something for the
you select your material, these are frozen into future. I think you are right.
the design of the container, and there is really
no opportunity to change these things. You can Mr. Gertel (Kinetic Systems): I gathered
change nicely on sheei material and generate a that the drop height, the curves of acceleration
curve, but once you have built it into a container and static stress were for a particular container
you are stuck with it, so to speak. Then is when configuration instead of the sheet material type
you must have made the right choice on the basis of data which you normally find.
of this type of information. No, we had no oppor-
tunity to change anything. We changed drop atti- Mr. McDaniel: The superimposed dynamic
tudes and that is about it. cushioning curve that we showed last was on
sheet material. I got the information from a
Mr. Westine (Southwest Research Institute): manufacturer and compiled it the way It was pre-
From the theoretical point of view, when you sented. It does not perform exactly that way in a
correlate with acceleration and so forth, this container, but it comes close.
means you are in the quasi-static load realm.
That is, thr, durations~of loading are very long Mr. Gertel: That was the question I had
relative to the response time associated with whether you noticed any particular difference
the structure dropped. Throughout all these between the curves for sheet material, and the
presentations we saw the acceleration being put material as actually used in containers.
forth. But if the durations of loading are short
relative to response times, quite frankly, accel- Mr. McDaniel: Yes, significant at times.
eration does not matter. Velocity, not acceler-
ation, becomes important. It puts you in what we Voice: Was there any pattern?
call the impulsive loading realm. You can also
be in a knuckle which gives you both acceleration Mr. McDaniel: No, there was complete vari-
and velocity as the significant initial conditions, ability in what I found. You can not really use
So, in a certain sense you should be talking about sheet material data and extrapolate directly, but
what these period-ratios or these time-ratios are it is a good guide. It is the best we have right
like. When you say, for example, "I make the null now, and I think this type of information would
hypothesis that temperature is not significant. I make it better.
test and find it to be false, therefore, tempera-
ture is significant." That does not necessarily
follow. It may be there is something else that Mr. Leonardi (Picatinny Arsenal): One of
you are ignoring and not treating as a variable, the reasons you can not do this, is that you have
I do not say that it is not, because in making to include parameters such as the degree of con-
that statement, it does not really test it to the finement which you have in a container and the
contrary, does it? From a philosophical point air escapement function which le Lwtys present.

20"1
The response level, which you have mentioned, of cushioning above your product also makes a
is another parameter. What you are doing is difference. that is the relationship of the top
something we also are trying to do. We are try- cushion to the bottom cushion. Also a question-
ing to measare the overall effect of the struc- was that 30 inches of free-fall as compared to a
ture, and sometimes this is the simplest way out. vacuum, or was that dtrlctly a 30 inch free-fall
In the final analysis it gives you an answer - height the way you measure it?
some quantitative value with which to work. Mr. McDaniel: Well, the standard military
Mr. McDaniel: That is right. When you documents present the free-fall drop test in
take sheet-material data and apply it directly, terms of height as a function of the weight of the
the unknowns you mentioned give results that item or sometimes the weight and size or shape
are not predictable from the data used. Just by of the item. It Is a free-fall drop in air. It Is
cutting a piece of foam material into smaller just an anticipated stevedore operation type of
pieces, you change the response because of the thing.
viscoelastic propeftiles-ofair flowing through
the materials. Once you build a container, you Mr. Gaynes: The only reason I bring that
are stuck with a particular design, and you will up-is-because, normally, in conducting free fall
be testing on that basis, tests, the surface upon which you drop is a very
important factor in relation to the resulting g
Mr. Gaynes (Gaynes Testing Laboratories): force.
Others of us have done work in this area, and
we have found variables introduced by the test Mr. McDaniel: That is right. These tests
method. For example, where accelerometers are defined as a drop on a steel plate or on a
are mounted can make a difference. The amount concrete immobile surface.

222
 I

DEVELOPMENT OF A PRODUCT PROTECTION SYSTEM

Dennis E Young
IB.IGeneral Systems Division
Rochester, Minnesota 55901

and
Stephen R Pierce
Michigan State University
East Lansing, Michigan 48823

A workable method for development of a product protection

system, based on hardware and established procedure, is pre-
serted. The date and techniques used in product design
determine the inherent strengths and fragilities of a product.
These characteristics are determined by fragility assessment.
Coincidentally, a continuing program statistically quantifies
the non-use environment through which the product must pass.
This data, along with knowledge of packaging methods and
material characteristics, are combined to engineez the
package. Once engineered, the packaq9 and product are tested
by dynamic simulation of the non-use environment. The
packaging program described indicates that the method
maximizes the packaging engineer's chances of submitting the
ideal economic and protective package the first time.

INTRODUCTION types, communications, and education.

The thrust of this work has been
Since the birth of American industry, to minimize functional degradation
goods in transit from the manufacturer to goods which are subjected to
to the consumer have been exposed hostile, non-use environments.
to damage. From these early beginnings, This paper presents a case study
the problem of damaged goods has of one approach to a solution that
grown along with the nation's industrial seems promising.
performance until today these damage
costs amount to billions of dollars.* At the IBM Rochester, Minnesota,
Coincidentally, experts have realized facility we have a move on to "reconcep-
the need for solving the problem. tualize" the information and action
Among those attacking the problem flow necessary to solve a possible
are professionals in packaging, exposure to goods damage in transit.
transportation, materials handling, The preliminary result was a simple
and associated fields of materials flowchart of the major elements
science, dynamics, testing of various in development of a Product protection
system, as shown in Fig. 1.

* Estimate based on Office of

Policy Review of the US Department
of Transportation.

223
 II" I

ENV IRONMENT
DATA"

I a
PRODUCT ENGINEER TAKAEST
DESIGN PACKAGE P UCT

A +
!1 I DATA

Fig. I - Product Protection System Development I

ELEMENTS OF THE CONCEPT data is available, specific techniques

for application to product protection
Product Design problems are conspicuous by their
absence. In general, packaging
Elements of the mechanical design engineers are not highly trained
of any product determine the inherent in dynamics, and thus the requirement
£engths and fragilities of that is for a straightforward set of
product. The more rugged the initial environmental data that relates
product, the less exotic and expensive well to empirical fragility data.
the package required. Attention A further constraint is that the
to design concepts at this stage environmental quantification be
can reap economic benefits when relatively easy to perform by the
the product is later exposed to engineer. Having this ability,
transit shock and vibration, the persons responsible for the
protection system can assign priorities
Fragility Data to applicable modes of transportation.

Within the constraints of a given User generation of data has some

mechanical design, it becomes necessary disadvantages also. Among them is the
to assess the fragility of the relatively small quantity of data that
product. Fragility may be simply one organization can effectively
defined as the level of dynamic accumulate. Data used for this type
input required to cause minimal of application is a set of graphs of
non-functionality in a product. either deceleration or change in
Recent literdture has emphasized velocity treated statistically.
a method to determine shock fragility
specifications based on parameters Enqineer Package
of deceleration and total change
in velocity [1-31. The result The information on fragility and
of this empirical analysis is called environment is combined with material
a damage boundary. Vibration fragility data and knowledge of packaging
is specified by the natural frequencies methods and processes to engineer
of the product, with emphasis on the product protection system or
the lowest primary resonance [4,51. package. Graphic data on cushion
performance is widely available.
Environment Data Some additional types of information
necessary for designing the package
In r~sponse to a need, continuing are only now under development,
efforts have accounted for wide and so must be generated by the
availability of data on the transit user either for each application
environment 16-91. The forms of or under a continuing program of
data reduction and presentation material evaluation. Some of the
vary considerably. While general most useful forms of this data
are given in Table i.
224
 TABLE 1
Packaging Material Performance Data [10-151

Data X-axis Y-axis

Type Parameter Parameter

1 Percent creep Time

2 Static stress Response deceleratior

3 Static stress Natural frequency

4 Forcing frequency Transmissibility
Natural frequency

5 Strain Stress

Test Package Product PACKAGING CONCEPT APPLICATIONS

Once a prototype of the proposed The Product

design has been built, the product
and its package are subjected to The product to be packaged is a
a series of tests to assess the keyboard subassembly used in the
ability of the package to protect IBM 129 card data recorder. The
the product during transit. This unit is manufactured at Rochester,
testing is an environmental simulation, Minnesota, and then shipped to
with three basic types of inputs: the IBM Toronto, Canada, facility
vibration, horizontal shock inputs, for installation in the host machine.
and vertical shock inputs. Volume of shipment is significant
relative to products of this type.
Feedback The unit (Fig. 2) weighs approximately
14 lb. It sets in the host machine
Throughout the development procedure with the lower mechanical portion
there is feedback to insure the beneath the table top. Construction
proper application of data. If is a combination of plastic and
the stage of product development metal, with the major components
allows, results of environment mounted on a relatively rigid frame
and early fragility tests are given suspended from the cover structure
to the product designer. By identifying by screws. Cabling for connection
critical areas of the product, to the host machine exits from
it is often possible to effectively the rear of the keyboard. Unit
trade off package design versus cost is relatively high, thus no
product modification to the best damage/packaging cost trade offs
economic advantage. A change to are allowed.
the product early in development
can increase its ability to withstand The units are shipped by general
the non-use environment between commodity carrier truck from Rochester
manufacture and ultimate use, without to Toronto, and are handled at
affecting its use environment perform- least five times.
ance, and at minimal cost.
Product Testing
Environmental information sets
the levels for transit simulation Shock Fragility
testing of the packaged product
[16). Results of this testing One of the critical preliminary
are fed back into the package engineer- determinations--prior to shock
ing cycle for possible improvement fragility testing--is to estaolish
of the package design. a damage criteria for the product.
The damage criteria decision was
Thus, the application of the concept simplified because of a standard
shown in Fig. 1 allows an orderly Quality Engineering test performed
and logical progression of data prior to shipment.
and action resulting in the development
of an economic product protection
system.

225
 Fig. 2 - IBM 129 card data recorder keyboard

The unit is mounted on a keyboard which can handle 1500-lb. specimerts,

test adapter (KTA) connected to has horizontal dimensions of 60
an IBM 1130 computing system. by 60 in. An open top allows a
A stored test program is then run. package height limited only by
During the test, all functions stability. The device incorporates
are tested for proper operation. a lifting system to raise the shock
The total test time is only minutes, table to a preset drop height level.
making the repeated dropping and The lifting mechanism is then lowered
testing required for shock fragility away and the table dropped on program-
determination practical. The environ- ming devices. For shock fragility
mental data, discussed in detail testing, pneumatic programmers
below, indicates that the principle charged with nitrogen gas are used.
danger of damage exists in the
vertical direction. Since units The rise time of the pulse is
are shipped in multiples of 36 controlled by changable elastomer pads
(to make use of a standard outer on the bottom of the table; the
ccntainer) the potential for damaging stiffer the pad, the fajter the
horizontal shock or tipovbrs appears rise time of the pulse. When the
slight. Environmental data bore decelerating table force equals
out the opinion that cushion design, the combined pressure of the gas
if required, would apply only to in the six programmers, the programmers
the vertical orientation. This begin to stroke at a constant decelera-
was confirmed later in the environmental tion rate, then rebound at the
simulation phase. same constant rate until the pressure
is again equal to the force. The
Testing for shock fragility is rebound then continues at a rate
an equipment-dependent type determined by the elastomer. The
operation. Although literature resultant pulse is a trapezoid wave
suggests several alternate methods, shape (acceleration vs time). Wave
the testing for this application shaDes affect the results of testing
was done on the shock machine* significantly, and thus the trapezoidal
shown in Fig. 3. The machine, wave is preferred, since it is
the closest practical dpproach
to the ideal rectangular wave (21.
* Hodel 6060 IIKII, minufactured
by ?ITS Systems Corporation,
Minneapolis, Minnesota.

226
 where
Vt = total change in velocity,
Vi = impact velocity,
I cr = system coefficient of
I restitution (0.04c < 1.0), and
r
Vr = rebound velocity.

"__ The first impact was not damaging.

Two consecutive runs were made
on the testing system without deviation.
By iixicreasing the pressure in the
pneumatic programmers, the g force
level was gradually increased with
each successive test. On test
No. 7, no damage was seen at 36 g
input. On test No. 8, the metal plate
between the frame and the cover
bent in two places, displacing
the keys downward and making the
unit inoperative. The g side of
the damage boundary had been tentatively
established at between 36 and 45 g.
To establish the change in velocity
Fig. 3 - MTS shock machine side, the programmer pressure was
increased so that the g fragility
would be exceeded on each drop.
One of the problems associated The first drop of the second series
with this type of testing is that was 85 g and 78 in./sec. change in
most products are sensitive to velocity, with no damage. The
the manner in which the shock pulse second drop was 93 in./sec. with
is applied. Therefore, the method no damage, but the third drop, at
of fixturing becomes important, 102 in./sec., caused the same type
as it must represent the intended of damage as occurred in drop
method of supporting the product No. 8. Since the first series
within the package. In the keyboard was done at a change in velocity
application, the product was supported level that would place the point
by the outer edge of the cover, of the damage boundary established
with the lower frame free of contact on the knee, or curved portion
with the fixture. The keyboard of the curve, the g side was rerun,
was then held firmly in the fixture this time with sufficient drop
to avoid rebound. The shock fragility height to make the point established
test cycle was then started. on the asymtopic portion of the
damaqe boundary. Thus the g side
The first series of tests determined was more clearly defined as 35 g.
the deceleration or g force side The damage boundary for the vertical
of the damage boundary. The first direction of the 129 keyboard was
test was a 9g pulse, from a 7-in. available, and graphed as seen
drop height, with total change in rig. 4. ror most applications,
in velocity of 120 in./sec. Total the more conservative graph, showing
change in velocity is calculated a square corner, has proven easier
as follows. to interpret and use, while maintaining
the ability to establish the curved
AVt = Vi (l+cr) or knee portion if the need arises.

or Vibration rragility
V=V+Vr Since the determination of product
At i+r resonance was important only considerinq
the primary cushion as the springing
member, these tests were performed
with the keyboard mounted in a
prototype holding tray, similar
to the fixture used for shock fragility
227
 4 DAMAGE
W
o K_
35 G
NO DAMAGE

194 rEC-./SEC
TOTAL CHANGE IN VELOCITY

Fig. 4 - 129 keyboard vertical damage boundary

testing. The unit was mounted to an NRZI mode, to provide compatibility

electrodynamic vibrator with a 500 lb. with electronic data processing
vector force rating. The frequency equipment. Data recorded is peak
swept from 10 to 200 Hlz. Audio , measured g force and duration at
was noted at the 25 1iz and at the 75 a preset threshold. Thresholding
Ilzlevels. Use of slower sweeps and minimizes recording of low level,
stroboscopic techniques identified relatively useless data. For the
the lowest primary resonance at 26 trips used in this program, 20 g full
Hz, at which point the entire structure scale and 10% of full scale threshold
hanging beneath the covers resonated. values were selected. The makers S
Testing was at approximately 0.029 in. of TEMARS also provide a standard
displacement, or 1 g, zero to peak. software computer program to print
out recorded data. This standard
Environment Testing program was modified to provide
output data punched in cards for
Our continuing environmental quanti- further processing. These cards
fication program is based on the are the raw data processed through
Transportation Environment Measurinq an IBM 1130 computing system by
and Recording System (TEMARS*). TEMARS, user-written programs to further
as shown in Fig. 5, is a portable, battery reduce the data to graphic form.
puwcred system that records shock The raw data is also converted
inputa experienced in the transit from 9-force data to change-in-
environment. Recordings are made velocity data, and again treated
on a 7-track magnetic tape, in graphically. Fig. 6 shows the
general data processing flow, Figs.
7 and 8 the graphic end result.
* Manufactured by Endevco Corp-
oration, Pasadena, California.

228
 fI

The graphic data is also available

to the user through a television-like
graphic display unit, where the
user specifies information
about the TEMARS trip of interest,
and the computer selects and displays
the appropriate information on
the screen. Data used for this
application was from trips along
a different, shorter route, but
representing the intended environment.
While data so collected is a small
sample of the total environment
population, it is assumed that
the data falls in the middle of
some undefined normal distribution
of values, that is, that both higher
and lower inputs do occur with
some. frequency. We further assumed
that the continuing program will
make available ever increasing
amounts of data that will add signifi-
cantly to the confidence level
of the information.

Fig. 5- Environment sampling device

PROCESS DT

CARDSPROGRAM DATA

PROGRAM

Fig. 6 -Environment data reduction flowchart

229
 12

0
_j 4
W
w

50 I00
PERCENT EXCEEDING INDICATED LEVEL

Fig. 7 -Example 51-force environment data

120

W
V)

Z 80

00

- HIGH ESTIMATE
> 4o
LOW ESTIMATE

S0 100
PERCENT EXCEEDING INDICATED LEVEL
Fig. 8- Example change-in-velocity environment data

Packaging Engineering Process change in velocity to cause damage

was seen in several cases. We
Examination of the environmental estimated that the concurrent g
data available and the fragility force and change in velocity was
curve convinced us that cushioning sufficient to cause damage. Further,
was required to protect the product we noted that damage costs to one
from the most severe inputs expected. bulk load of units was sufficient
Amplification of g level inputs to pay for a month's supply of
were anticipated, thus damaging the proposed packaging. Design
levels could be expected. Enough drop height was based on available

230

---- -- A
environment data. Maximum changes 0.6 and 0.4 = weight distribution
in velocity were seen as possibly factors,
occurring with 0.0 coefficient
of restitution, therefore design M = total unit weight, and
drop height becomes a function
of only the impact velocity. This Ss = desired static stress.
is a most severe case. Calculations
were as follows:
The next problem was to assess
Hd = ViV? the compatibility of the proposed
cushion system with the vibration
2g fragility of the unit. The
standard recommendation is to use
where Hd = design drop height, a cushion with a natural frequency
that will provide 60% reduction,
V. = impact velocity, and or attenuation of the g level at
3. product resonance. As a guideline,
g = gravitational constant this often occurs when the natural
frequency of the cushion system
g = 386.1 in./sec. is one-half of the natural frequency
of the product. In this case,
the product resonated at 26 Hz,
Using an apparent maximum impact so the targeted cushion frequency
velocity of 100 in./sec., a design was 13 Hz. No data is yet available
drop height of 13 in. was assigned. for the natural frequency versus
For a gross bulk weight of about static stress performance o ether
600 lb., this is slightly less urethane. Data is available, however,
than some recommendations [14,16]. on the stress-strain relationship
of this material [10], so the system's
Little information is available natural frequency was estimated
on static stress versus deceleration as follows:
data (type 2 from Fig. 2) on the
12-in. drop height 115]. The 1
18-in. data shows that several fn = - g (1)
cushions give the desired performance. 27t
Two optional approaches were s
identified for trial. One, a molded f 13.9 to 15.5 1z for static
urethane foam end cap, was later dropped n def3.cto of (z 0.0 to .0
because of difficulty in adjusting deflections of (xs) 0.05 to 0.04
for nonsymmetrical weight distribution
and because of cost. The other
initial approach was to use polyether f system natural frequency, and
urethane foam cushions under the n
unit holding tray. Weight distribution
of the unit is 40% front and 60% g = gravitational constant
rear. The size cushions underneath
the front and rear were adjusted g = 386.1 in./sec.
accordingly. Based on a type 2 Another method of estimating static
curve, we decided to use 0.2 psi deflection provided similar results.
as the static loading. This point In a preliminary test, a free fall
falls on the left side of the lowest drop of 8 in. gave an experienced
portion of the curve. The unit force reading of approximately
and holder weigh 15 lb., so the 20 g. Deflection of the system
cushion sizes for front and rear under dynamic load is calculated
were calculated as follows: as:
Ar = 0.6M/S s xt = 211
Af = 0.4M/S s G-2
where Ar and Af = cushion areas for where
rear and front,
respectively, xt = total dynamic deflection,
If = drop height, and

G = experienced g.

231
 By this analysis, total dynamic A prototype set of packaging for
deflection is 0.89 in. Static deflection the individual keyboard was completed
is then given by and preliminary tests were run
to determine the feasiblity of
= tthe design. After some minor dimensional
G adjustments, the basic design was
approved as shown in Fig. 9. Samples
in. were secured for further testing.
0 Since it was necessary to ship
multiple keyboard units, a method
By applying Eq (1), natural frequency was devised to permit each individual
y faplying eq (14 naura l frequentray and keyboard to act separately
again falls between 14 and 15 Hz.
Aon frequency sweep across this area outer protection from
out prtcion rger,
from a larger,
the electrodynamic vibrator
showed that because of the lack multiple unit package. The outer
sowe tat becausy o the ak r package is a standard double cover
of total rigidity in the keyboardtueoaple. Thtbendcs
holding tray, the system did not tube on a pallet. The tube and caps
resonate cleanly. However, the are made of corrugated fibreboard,
point at which the system seemed with inside dimensions of 40 by
to begin to resonate was between 48 by 36 in. high. The cubage
13 and 14
13ande 14sHz,
th thus confirming that
usritconi g That was then divided into four layers
of nine units each, each unit having
range as the critical one. This ison"ac"
is somewhat higher than the ideal its own hatch.
mentioned above. If transmissability Environmental Simulation
performance data were available, El
it would be possible to evaluate B
this increase. From other data Based on available equipment and
1151, it appears that, at worst, data, the following three-part
the natural frequencies identified environmental simu'tion was specified.
would transmit input forces at *Vibration Testing
a ratio of 1 to 1. While this
is less favorable than the recommendation, --Unit level vibration
we decided to try this method and
test the system in the vibration
mode during environment simulation. nominal 0.5 g input at
a sweep rate of approximately
1 octave/min

Ilk
Fig. 9 - 129 keyboard and its protection element

232
 --Multiple. unit vibration: circular Product Protection System Development.
synchronous vibration for 60 min at It is through this process of change
3.0 Hz and improvement that scientific
protective packaging will come
*Horizontal Shock Inputs of age.

--Two impacts each face of

the multi-unit package at 35
in./sec. impact velocity REFERENCES

eVertical Shock Inputs on Step Vel-

ocity Programmers 1. F. C. Bresk and A. Abate, "Cost
Savings Through Reduced Product
--20 Inputs at 30 in./sec. impact Fragility and Optimized Package
velocity Design," presented at AMA National
Packaging Conference, Chicago,
--2 Inputs at 100 in./sec. impact May 1971.
velocity
2. R. E. Newton, "Fragility Assessment
Since there was no damage to the Theory and Test Procedure", Sponsored
units, the package was deemed sufficient by Monterey Research Laboratories,
to ship the product through the Monterey, Calif., 1968.
handling/transit environment.
3. M. Kornhauser, "Prediction
SUMMARY and Evaluation of Sensitivity to
Transient Accelerations", J. Applied
The process detailed in this paper Mech., Vol. 21, No. 4, Dec. 1954.
is a case study of an attempt to
apply existing information and 4. 5 Step Packaging Development.
methodology to the problems of MTS Systems Corp., Minneapolis,
product protection. Since this Minn. 1971.
method was adopted, none of the
several thousand 129 keyboards 5. R. E. Newton, "Package Vibration
have been damaged in transit. Testing," prepared for MTS Systems
The current cost of the packaging Corp., Minneapolis, Minn. 1971.
material and labor for this unit
is significantly less than 1% of 6. F. E. Ostrem and M. L. Rumerman,
the value of the unit. Several "Shock and Vibration Environment
alternate methods of packaging Criteria," Contract NAS 8-11451
have been suggested and investigated, Final Report 1262, General American
with no evidence that adequate Transportation, Research Division,
protection can be offered at less Sept. 1965.
cost. The packaging program described
in this paper is an indication 7. Ibid., Report 1262-2, April
that the method minimizes cut-and- 1967.
try engineering, and maximizes
the packaging engineer's chances 8. F. E. Ostrom, "Survey of Cargo-
of submitting the ideal economic Handling Shock and Vibration Environment,"
and protective package the first Shock and Vibration Bulletin 37,
time. Part 7, Jan. 1968.

Some gaps remain in the theory 9. R. W. Luebke, "Investigation

and application of product protection of Boxcar Vibrations," Contract
system development. Even more DOT-r.R-9-0038, Report No. FRA-
severe gaps exist in the material RT-70-26, Aug. 1970.
data area. The situation in both
of these areas will improve as 10. R. K. Stern, "Package Cushioning
the demand for more information Design," MIL-IIDBK-304, prepared
increases. When data is made available for Department of Defense, Washington,
to the packaging community, as D. C., Nov. 25, 1964.
generated by material users, manufacturers
and independent sources, the rate 11. If.C. Blake III, "Creep Properties
of improvement should increase, of Selected Cushioning Materials,"
School of Packaging, Michigan State
As knowledge and experience increase, University, East Lansing, Mich.
other improved methods will replace Nov. 15, 1964.

233
12. H. C. Blake III, "24-in. Drop
Height Peak Deceleration-Static
Stress Curves for Selected Cushioning
Materials," School of Packaging,
Michigan State University, East
Lansing, Mich., Nov. 1, 1964.
13. H. C. Blake III, "Peak Deceleration-
Static Stress Curves for Selected
Cushioning Materials," School of
Packaging, Michigan State University,
East Lansing, Mich., July 10, 1964.
14. Packaging With Ethafoam, Form
No. 171-458-1014-1266, Dow Chemical
Company, Plastics Department, Midland,
Mich. 1966.
15. P. E. Franklin and M. T. Hatae,
Shock and Vibration Handbook, C.
M. Harris and C. E. Crede, editors,
pp. 41-1 through 41-45, McGraw-Hill,
New York, 1961.
16. A. T. Mickel, IBM. Armonk,
N. Y., correspondence.

234
 MOTION OF FREELY SUSPENDED LOADS DUE TO HORIZONTAL SHIP MOTION
IN RANDOM HEAD SEAS

H. S. Zwibel
Naval Civil Engineering Laboratory
Port Hueneme, California

The theory is developed for the swinging motion induced in a wire sus-
pended load due to the horizontal motion of a ship. An explicit
formula is obtained for the significant amplitude of horizontal load
motion when the ship is exposed to random head seas. Numerical results
are presented for two typical cargo ships in a sea state three. It is
found that very large motions are suffered by the load. For critical
line lengths, resonance effects magnify the ship motion by several
orders of magnitude. In addition, numerical solutions are also
obtained for the pendulation due to simple harmonic motion of the
support while the load is being raised or lowered. The peak displace-
ments are considerably less in this case, however, they are still
undesirably large.

problem tractable the following assumptions are

INTRODUCTION made:

Loading and off-loading from ship-to-ship (a) the line is a fixed length
(or ship-to-pier) is a requirement for the (b) the line is flexible and uniform
Expeditionary Logistics Facility (ELF). In a (c) the load is considered to be a point
wave environment the ships are subjected to mass
forces that generate translational and rotation- (d) the mass of the load is much greater
al accelerations. The horizontal motions due to than the mass of the line
surge and pitch from bow-on wives in a sea state (e) the deviation from equilibrium is
3 have been shown to be quia small [1] (frac- small
tions of feet for the significant surge
amplitude for ocean-going vessels). The in- The coordinate system used to describe the
fluence of thene small motions on a wire sus- dynamics is presented in Figure 1. The
pended load may not, however, be negligible, horizontal position from equilibrium of the
It is the aim of this investigation to determine support is denoted by X. and, as far as the line
the motion of such wire suspended loads induced vibration is concerned, is a given function of
by horizontal ship motion. time. The horizontal displacement from rest
of a piece of line at position z below the
In the next section the constant line support at time t is T(z,t). The mass m, which
length, random motion problem is solved. This is a distance L below the support, therefore
is followed by an analysis of the pendulation translates through a horizontal displacement
while raising and lowering the load assuming of J(L,t).
simple harmonic motion of the support.

FIXED LINE LENGTH

In this section the steady-state pendula-

tion induced by random seas and constant line
length is presented.

Theory

The horizontal motion of the ship is

transferred to the load by means of the lateral -- . . -
vibrations of the lifting line. These vibra- .t,.1,t"o, O.spo t.
tions are due to the motion of the line attach-
ment point which is assumed to be rigidly
connected to the ship. In order to make the Figure 1. Coordinate system.

235
 The dynamics of this system are described Substitution of the linear drag term for
by the one-dimensional wave equation for 1(z,t) the non-linear term in equation (3) gives
for the line and F a ma for the mass. The
equations, including boundary conditions are -;_ae,
given below: - i-L

The steady state response of the system for a

's/i) harmonic motion of the support is readily
obtained. Let the displacement of the support
(2) be given by
m~~~ e) i 3

- o'A- sX(u) is the complex amplitude of the support at

frequency w and Re C 1 means the real part of
the bracketed expression. Then the 1(z,t) that
The
line.first
The equation
velocity is
of the wave ,equation
proa~ation for the
(C) for satisfies equations (1), (2), and (7) is
[>\
transverse vibrations is T/j!, where T is the Rell ij 1 (9)
tension in the line, and giis the mass per unit
length of the line. It Isassumed that m IL, Si K(I j
' L1
rI' CeI(4c
hence T - mg (g is the gravitational occelera-
tion), therefore 5/'T
(4) where
.(10
The second equation imposes the horizontal k (10
motion of the ship, Xs(t), on the upper end of
the line. The third equation is F = ma for the Of prime interest in this study is the displace-
load of mass m. Since l(L,t) is the displace- ment response amplitude operator of the load.
ment of the mass, it appears as a boundary This is given by the complex amplitude of the
condition for the motion of the line. The first load divided by the amplitude of the forcing
term on the right hand bide of equation (3) is function.
the horizontal force exerted on the mass due to i
tension in the line. The second term on the Denoting this complex response operator by
right is the damping force on the mass due to R (W), equation (9) (for z - L) gives
its motion through the air. p is the density of (11)
air; CD is the drag coefficient; A is the cross- L246 _,
sectional area of the load; 6TVbt (L,t) is the " CZ3 L -14) (t f. ) 5,kL
horizontal velocity of the load. It should be
noted that this term is non-linear. The non-
linearity presents several complications. First,
it means that the equations must be solved For practical situations, equation (11) can be
numerically. Second, the linear methods of simplifed somewhat. For frequencies of interest
random analysis are not applicable. In order to and the fact that m >>L, it follows that
avoid these difficulties, this term is linear- k L <<I. It is then possible to let cos k L
ized. The method of equal energy dissipation is 1 and sin (kL) = kL. Introducting this into
used [2]. It is assumed that the driving equation (11) yields
function is harmonic in time, I. e., Xs(t) - s -/ (12)
cos wt. If the damping is sufficiently small, 0" 2 -- :-
=
the system will predominately respond harmoni- I\& I
cally with frequency w. The linearized damping ,f.~j
force is given by
r' ,) (5) where
-
I-b .-- (13)
where

377/ From R(O) one is able to determine the

Xo is an unknown constant and equals the statistics of the load motion when the ship is
amplitude of the motion. Morn will be said exposed to a random sea. Before doing this,
about this later. At this point of the discus- however, it is worthwhile pointing out several
sLon, it is sufficient to regard y as a constant, things. First, the amplitude has a peak at L -
_ o. This is understood by observing that the
Equation symbols are defined where they first resonant frequency of a simple pendulum of
appear and in the rotation listing at the end length L is equal to Vg/L (which is Wo). The
line-mass system therefore acts like a pendulum
due to the large m/pL ratio and the low

236
frequencies considered. Second, the magnitude amplitudes is [3(
at resonance is equal to PaoL/mg)- 1 and the X,--
width is equal to 1/2 (wO yL/mg). This is
shown in Figure 2. For practical situations
yub 2 L/(2mg) << %, so that the response amplitude E is given by
operator is a sharply peaked function. This "
property will be used in the following .
development.
which, from Equations (14, 15) becomes
~(18)

IRU.)(w1
iscalculated from Equation (12); which
when inserted into Equation (17) gives
o A (19)

Due to the extreme narrowness of R(w) compared

to the other functions in integrand great
simplifications are possible. To see this more
clearly let v ((L-wo)/uo. E then becomes
Figure 2. Example of resonance. 11 (20)

It is now assumed that the ship is exposed to = Wo %{#J !s1. (a

random process with a known power spectral t________
density function, S(w). The horizontal motion . 4
of the ship (assuming linear ship theory) is
then also a stationary Gaussian random process
with power spectral density function, Ss(w) The denominator becomes very small for v 0,
given by and rapidly increases for non-zero V. The width
(14) of this resonance is y/(2ma o ) and is much less
than 1. Compared to this narrow resonance, the
S()integrand is a slowly varying function about
v - 0 and can with little error be taken ouitside
Xs(w;h) is the ship's horizontal complex the integral. Equation (2) then becomes
response amplitude operator at a height h above (21)
the center of gravity of the ship.
In terms of Ss(w), the power spectral t) . "-
density function for the load is given by -
3. (15) Furthermore, since the main contribution arises
Rt.ta)IS (i) for vl<<I, v can be ignored when compared to 1.
In addition, the lower limit can be extended to
.-o. With these simplifications E becomes
This assumes that the motion of the load is (22)
linear. Even though the drag term was linear-
ized, the solution at this stage is still not I Jl
1
linear due to the fact that y depends on the ( / s,4
amplitude of the motion for the various
frequencies.
which, at last, is Integrable. The final result
One way to surmount this difficulty is to is
be this (23)
replace y by some
average value; average
it will value. In
be chosen a jlogical71A
Let

fashion after some further development. E . / X(C,); 4)/ cui

The relevant statistical quantities are

obtained from the area under the power spectral
density curve, e. g., if E denotes this area, Using this expression for E, (and Equation (16))
then the significant swing amplitude, X1/3 , the significant swing amplitude of the load is
which is the average of the largest one-third

237
 -1A (24) The inner bracketed expression contains the
response operator for the horizontal motion of
&"j 4 the line support. Itdepends on the dynamic
response of thship to regular waves with fre-
of weldefichd
s isEq e vr
oquencyvW g/L. If the surge and pitch complex
Evnryting is well defined in Equation (24) with response operators for the ship are given by Xa
the exception of y, which is some average value (%),Os,(%) respectively; then, Xs (w0 ;h) is
of y, and must now 'be determined. y from given by
Equation (6) is equal to

and involves the amplitude of the motion. It is where it will be recfflled, h is the vertical
reasonable to let distance from the line support to the ships
_± p/c
A 3 (25) center of gravity.
37"
317 These response operators can be determined
where Xo is an "average" amplitude, either by experiment of by mathematical analysis.
The analysis approach is taken in the report.
The problem now is to select Xo. Recall The standard strip theory method was utilized
that Ro is, in some senbe, a representative to develop the NCEL computer code RELMO (REla-
amplitude. As with all such vaguely defined tive MOtion). A complete presentation of RELHO
quantities there are a variety of possible and is given in reference I. RELMO calculates the
equally logical choices. For example, several heave, surge and pitch response of the ship.
possibilities are: The added mass and damping coefficients are
calculated using subroutine ADMAB (4J and is
(1) Let Xo equal the significant swing based on Grim's [5) method.
amplitude (i. e., Xo - X1 /3 ).
(2) Let equal the most probable swing The last bracketed expression in Equation
amplitude (i. e., X f 0.626 X). (27) contains the Power Spectral Density Func-
tion for the sea. At the present time the most
(3) Let X equal the average swing widely used spectrum is due to Pierson and
amplitude (i. e., Xo - 0.5 X1 /3 ). Moskowitz (23. This is given by

1 eo (29)
Its clear that the list could be expanded. For,
all reasonable choices it would appear, however, 5 t -,/1/0 0
that L (" .,;)J
(26) where If/3 is the significant wave height for
the sea (e.g., in a sea state 3, If1/4 is between
where e' is some constant between one-half and three and one-half and five feet). n terms of
one. Since "the" choice for a is unknown, the L, the last bracket in Equation (27) becomes
general form will be kept intact. In this way f (0
the sensitivity of Xo on the results can be W4 _ ()
evaluated.

When Equation (26) is inserted into This factor is a function of L and is shown in
Equation (24) (via Equation (25)) the resulting Figure 3 for several values of It 3 . The func-
expression can be solved for X1/3 ; the result tion rises from zero (at L - 0) peak at L
of this manipulation is 5.56 11113 and then falls to zero foi large L.
For sea state 3, L is obtained by replacing the
- 3e 4

i
last bracket in Equation (27) by Equation (30)

The solution is written as the product of these L

factors, each one of which represents a It is also of interest to determine the signif-
separate aspect of the problem. These are dis- icant accelerations. These are readily obtained
cussed below. due to the fact that the acceleration response
amplitude operator is simply w 2 times the dis-
The first factor contains the parameters placement response amplitude operator. Let Xl13
that pertain to the viscous drag force and the be acceleration significant amplitude; then one
physical characteristics of the load. These has s a then(32
parameters are independent of the off-loading . " Li, // (2
ship and the ambient sea state. Notice that * / ' 'I
the dependence on ' is quite mild. A change in r L /
a by a factor of two (the reasonable range for %Pm .
C?,og3
0) changes X143 by a factor of only 1.26. The
maximum spre in XI/3 due to a 100% variation The pitch angle is positive for bow down.
in G' is therefore only 26%.

238
This expression is similar to X 13; a plot of and 1.0. In the interest of conservatism o is
the last bracketed expression fo several values chosen to be 0.5.
of H,! is shown in Figure 4.
o3 The containers are considered to be 8'x8'x
20' and have a load capacity of 20 tons. Again,
in order to be conservative, the cross-sectional
area, A is taken to be 64 ft2 and the mass, m,
is 20 tons/g.
6-
For convenience, the assigned parametric
values are listed below:

CD = 2.03
03 ' .5f
S1/3\ o = 0.5
2. \m = 20 tons/g
2
A = 64 ft

o- ,Denoting X and Xt1 3 for this parametric

o 10 20 30 40 so 60 70 to 90 10 choice wit an asterick, the expression for X1 /3
L (feet) and Mt/3 are -

Figure 3. Sea spectrum contribution of X1/3 " )C$ 97' /- a

,u-x ',#/
"' '
p-o ,r (34)

I.e-
1.0 In Equation (33) L must have units of feet.
The units for X and )t are respectively
ft and g (32.2 l17sec2). /k/3 and "11/, for
j.- other choices of these parameters caW~ge obtain-
.1/3 ft ed from X3 and 1 from t following
8.4 3.5it-\ expressions.
. ft-
.5:H3' "/3
)" (35)

0L (feet) 60 £ 100
The two ships selected for study are a C8
Figure 4. Sea spectrum contribution of X1/3' and a C4 and are fairly representative of the
cargo ships to be encountered. Their character-
RESULTS istics are given in the tahlc below:
Due to the large number of parameters in- Displacement (LT) Length Beam Draft
volved (e. g., the mass and cross-sectional area (fully loaded) (ft) (ft) (ft)
of the container, type of ship, height of sup- I
port, etc.) a certain amount of specialization C8 44,428 772 100
is required. For convenience, Equation (31) is C4 22, 630 564 76 32
repeated below ' 3 \ /J
"V,4 , - " X5 (; i, X (V/gL;h) for these ships are plotted in

- , ,Lx 7 Figures 5-6. Each ship exhibits a pronounced

oh ' p/- ct, ," ,J resonance; the C4 at a frequency such that g-L
corresponds to a line length of 41 feet and the
Several of these quantities can be assigned a C8 to a line length of 53 feet. The 'ltion J
constant value. First consider the damping resonance is still quite small; for the C4 and
contributions. The densitz of air as sea level C8 it is respectively 0.17 and 0.11 feet for a
and at a temperature of 6C 7 !s taken as tepre- one foot wave.
sentative; according to Bretschoeider [61 this
value of P is 2.4x10 3 slugs/ftr. From the same
reference the drag coefficient CD is 2.03. The
parameter Y, as noted, is somewhere between 0.5

239
 X/3and X/3for a sea state 3 (111/3
3.5 ft aad 11113 =5.0 ft) as a function of line
length, L, are presented in Figures 7-10. The
general features exhibited by X4,/ are not
surprising. The sea spectrum falls off rapidly
for both large and smsall 1, and cons~quently
so does X1 ,3 . These are peaks in X13due to
the resonance exhibit~ed by Xs and the peak in
the sea spectrum. The extreme responses are
presented in Table 2.

Table 2. Extreme Significant Amplitude X1 /3

111/3 h1 1/3V L at Max (X Max L at Max

ft (ft) (ft) ft (ft) (ft) ____

0.0 3.6 38 .019 3.1 26 .017

3.5 20.0 6.2 37 .034 3.2 28 .019
40.0 8.3 36 .050 3.4 9 .023

5.0
0.0
20.0 12.2
8.6 40
40
.056
.096 7.8
5.3 47
47
f.036
.051
40.0 115.7 1 40 .14 30.3 1 47 [.072

There is one significant fact, namely, the

magnitude of Xj 3 In order to understand thle
effect, thleamp1 itudes must be compared with
the significant amplitudes for the horizontal
motion of the support, (X5 )1 /4, given respec-
tively in columns 5 and 8 oltable The sup-
port displacement is a fraction of. 'cot,
whereas the load swings through many feet. This I

large amplification (a factor of 100 or more) is

evident, such systems are quite efficient at
extracting whatever energy is available at their
resonant frequency.

The curves for X1 13 .'shown in Figures 11-

14. file extremes in the si, ,ficant amplitudes,
for the acceleration are given in Table 3.
Fairly large horivontal amcceleratioens are evi-
dent. ror example, at a 0.4 g horizontal --

acceleration (which is a reasonaible upper limit)

a h~orizontal force of 8 tonsis a. cting on the
loa1d. *The Implication Is thait forces of this Figure 5. Horizontal displatement response
magnitude would be required to restrain the operator for aiC4.
horizontal osc illat ion.
Taible 3. Extreme Significant Ateleration, ~/
1177 It)
Mdx I. at M"x Max 1. at Maix
(it ) 9 (ft) g tt
0 .2 37 .115 22
J3.5 20) .18 36 1i21 22
.10 22
35
_ __
_ ~
L__ 4 0 .24 _

The f orteo F is obta.ined from F MI * In this

p~robl em m -20) 1ons/g amnd am 0.'4,, therefore
-f 20 ton%/g x 1). '#g 8 tort

240
 II
12
.12 1 N1/3 - 5.0 ft

.11 f .
10-1
.10 - -
.09 I II I \i
.08-
I I h 41 /
/ 8I I

_ .07-
-. o II I /1 xQ I
Ii
6 h 40ft
.06I/\ .0I I -40I,
, ft -20
.- I " h h 20ft

0 \ I -

.02 - I "

.01 -
o~uII I C I I
20 40 60 80 100 0 20 40 60 80 100
0
L (feet) L (feel)

Figure 6. Horizontal displacement response Figure 8. Significant load amplitude for C8.
operator for a C8.

I I I I I Hi/ 3 - 3.3 Ii

SI I
I I

I I hI o
iW

• "I
|I , I ,• f

;!\ /'') '

'. i ' /
''I_, . .. I /_
__

Figure 7. Significant load amplitude for C4. Figure 9. Significant load amplitude for C4.

241

IJO
 H1/ 3 =3.5 ft

,1/3
I t / I

Sh = 40 ft
c. 2
,h = 20 ft

h= 0ft , \\\
\'
0
0 20 40 60 80 100
L (feet)
Figure 10. Significant load amplitude for C8.

.4,
I \ H1/3
H1 1 5.0 ft

IC

I~ IIh20 ft
I I
hLtj 0 f t \ - 2

01
0 20 40 60 80 100
L (feet)

Figure 11. Significant load acceleration for C4.

242
 TIME DEPENDENT LINE LENGTH

It's clear from the foregoing analy is

that steady-state response for a constant line
length yields extremely large pendular mition.
In this section the pendulation while raising
and lowering of the load is considered.

" 1*] i I The differential equation describing he

- 5.0 ft
H1I/3 motion is given be'ow:

i ",Sun 9&) (36)

. A0(t) where
= pendulation angle from vertical
h -0 X = horizontal motion of the support

(t)-
SM the horizontal acceleration of the
2 0L (feet) 60 80 100 support
L(t) = length of the line
L(t) - the time derivative of the line
Figure 12. Significant load acceleration for length the pendulum angle with
C8. respect to the vertical

Rather than take the support motion to be random,

I I it is assumed that Xs(t) = Xo Sin (Lt. The line
H 3.5 ft length is to vary linearly with time as follows:
L(t) = Lo + vt where v is the line speed and Lo
l \ is the initial length of the line. The equation
.2
I
I
\
I,^,<
n' 4V1.
,.
t '(;):
to be - 2veD~t
solved is - ,; l) (
o)
20, t
,-h1

Analytical solutions are not available,

h 0 ,,t however, numerical solutions are readily
obtained. The numerical procedure is based on
0 Hamming's modified predictor-corrector method for
0 20 40 (feet)60 80 100 systems of first-order, initial-value differen-
tial equations. [71 This method is forth-order
and utilizes an error criteria whereby the time
Figure 13. Significant load acceleration for C4. increment is adjusted to meet prescribed
accuracy. Details are given in reference (71.

Example Problem

By way of illustration, the ?rotlcm of a

floating crane raising a load from a stationary
platform is presented. It is assumed that the
load is to be raised 30 iect; the original line
length is 100 feet, hence the line is shortened
from 100 feet to 70 feet.
.1Since the purpose of this problem is to
M11 3 ).5 it demonstrate pendulation under conditions of

varying line length, the actual motion of the

boom has been arbitrarily assigned an amplitude
of .1 feet. Results for a line speed of 90 feet/
, h -o40it minute are shown in Figure 15 for three different
S-- 20 t periods of excitation. These results can be
h-0 . 0 partially understocd by observing that the
______
_ _
_'__ I -__ initial line length (100 feet) has a resonant
00 40 60 s0 100 period of 11.1 sec and 9.2 sec for the final
(feet) length (70 feet). One would, therefore, expect
the response to be larger for the 10 second
Figure 14. Significant load acceleration for C8. motion than for the 3 set or 7 set excitation.

243
 H. S. Zwibel

FINDINGS AND CONCLUSIONS

A theory was developed to determine the

steady state horizontal response of a wire
suspended load due to the horizontal motion of
the ship in a random sea. Based on this theory,
calculations were performed for two types of
ships(a C8 and a C4) in sea state 3. It was
found that the motion induced in a wire-suspend-
S d load due to horizontal motion or a ship is
very large. The significant amplitudes for the
horizontal movement of the load (5 to 10 feet
typically) are 100 or more times larger than the
horizontal significant amplitudes for the ship.

The pendulation is much less violent when

one considers time varying line lengths, however,
the motion is still excessive. This study
indicates that anti-pendulation devices will be
required inorder to transfer cargo in regular
waves. A calculation is currently underway to
S ! I I l I l i I I [ extend the analysis to random waves, however,
3 4 5 6 7 S9 11 t)12(Z 13
11).It
It 14 a 1 6 ? 1 this will not completely eliminate the need for

anti-pendulation devices due to the fact that

Figure 15. Pendulation time histories while regular waves (i. e., swell) will also be
raising load from 100 ft to 70 ft. encountered.

REFERENCES
This is brought out more clearly in Figure
16. The largest response per urlt amplitude L13 D.A. Davis and H.S. Zwibel, "The Relative
excitation is plotted versus the excitation Motion Between Ships in Random Head Seas," NCEL
period. The peak response occurs for an Technical Note N-1183, September 1971.
excitation period of about 10 seconds. The
shape of this "resonance" curve is not symmet- (21 M. St. Denis, Floating Hulls Subject to
rical due to the fact that only the first Wave Action, pp. 12-72. McGraw-Hill, New York,
twenty seconds of the motion is considered. 1969.
The longer period excitations reach their peak
somewhat later and hence do not exhibit their [3) W. Frank and N. Salvensen, "The Frank Close-
largest response. It should be noted that the Fit Ship-Motion Program," NSRDC Rept. 3289,
boom motion is amplified by a factor of six for June 1970.
the ten second period excitation. [4) L. Vassilopulos, "The Anayltical Prediction

of Ship Performance in Random Seas," MIT Rept.

No. 64-2, February 1964.

(51 0. Grim, A Method for a More Precise

Computation of Heaving and Pitching Motions
Both in Smooth Water and in Waves, pp.483-518.
Office of Naval Research, Washington, D. C.,
1962.

[61 C. L. Bretschneider, Overwater Wind and Wave

Forces, pp. 12-2 to 12-24, McGraw-Hill, 1969.

[71 System/360 Scientific Subroutine Package

(360A-CM-03X) Version III, pp. 337-342.

LIST OF SYMBOLS

a Acceleration

A Cross sectional area of load (ft)

,, ,, ,. ,, ,.. , , ,.., c Phase speed of transverse wave in wire

Figure 16. Peak response versus CD Diag coefficient

period of excitation.

244
 H. S. Zwibel

Area under power spectral density ON Pitch response operator for ship
curve e(t) Pendulation angle from the vertical
AMass per unit length of line (slugs/
F Linearized drag force ft)
FLD
g Acceleration of gravity (32.2 ft/sec2) V Dimensionless frequency variable

h Height of support above ship's center p Mass density of air (slugs/ft)

of gravity (ft)
wFrequency of i aves (rad/see)
in
k Wave number for transverse waves
wire wo Resonant frequency of load (rad/sec)

L Length of wire (ft)

R(w) Load horizontal displacement amplitude

operator

S(W) Power spectral density of sea

SL(w) Power spectral density of load

S Power spectral density of support

t Time (sec)

T Tension in line (lbs)

x(t) Horizontal displacement of load

Xs(t) Horizontal displacement of support

X0 Amplitude of load motion

X0 Average amplitude

X Amplitude of support motion

X (w;h) Horizontal displacement response

operator for support
Xs (w) Surge displacement operator

x1/3 Significant amplitude of load dis-

placement (ft)

Significant amplitude of load accelera-

1l/3 tion (ft/sec)

1/3 Significant amplitude of load displace-

ment for selected set of parameters (ft)

XI/ 3 Significant amplitude of load accelera-

tion for selected set of parameters (g)

(Xs Significant amplitude for load support

(it)
z Vertical coordinate

r Dimensionless constant

y Linearized drag parameter

y Average linearized drag parameter

l(z,t) Displacement of line at position z

at time t (it)

245