NASA HANDBOOK FOR SPACECRAFT STRUCTURAL DYNAMICS TESTING

Dennis L. Kern(’),Terry D. Scharton(*)

‘‘)Jet Propulsion Laboratoly, California Institute of Technology, 4800 Oak Grove Drive,
MS.157-410, Pasadena, CA 91 109, USA, Email: Dennis.L.Kern@pl.nasa.gov
(2J
Coruultaiit, 213 Homegate Circle, Apex, NC, 27502, USA, Email: Astrobart2@aol.com

ABSTRACT
1.3 Applicability
Recent advances in the area of structural dynamics and This handbook recommends engineering practices for
vibrations, in both methodology and capability, have the NASA programs and projects. It may be cited in
potential to make spacecraft system testing more contracts and program documents as a technical
effective from technical, cost, schedule, and hardware requirement or as a reference for guidance. Determining
safety points of view. However, application of these the suitability of this handbook and its provisions is the
advanced test methods varies widely among the NASA responsibility of program/project management and the
Centers and their contractors. Identification and performing organization. Individual provisions of this
refilicmeiit of the best of these test methndologies and handbook may be tailored (i.e., modified or deleted) by
implementation approaches has been an objective of contract or program specifications to meet specific
efforts by the Jet Propulsion Laboratory on behalf of the progrdproject needs and constraints.
NASA Office of the Chief Engineer. But to develop the
most appropriate overall test program for a flight 2. APPLICABLE / REFERENCE DOCUMENTS
project from the selection of advanced methodologies,
as well as conventional test methods, spacecraft project Pertinent NASA standards and handbook and other
managers and their technical staffs will need overall applicable documents are listed herein. A list of key
guidance and technical rationale. Thus, the Chief reference technical papers and documents that best
Engineer’s Office has recently tasked JPL to prepare a describe new and recently improved testing
NASA Handbook for Spacecraj Structural Dynamics technologies are provided. “Standard Practices”
Testing. An outline of the proposed handbook, with a documents are cited for older, but still relevant test
synopsis of each section, has been developed and is technologies.
presented herein. Comments on the proposed handbook
are solicited from the spacecraft structural dynamics 3. DEFINITIONS AND ACRONYMS
testing community.
Definitions are provided for key terms that are not
1. SCOPE OF THE HANDBOOK uniformly interpreted, such as “limit load”, protoflight
test”, and “primary structure”. Acronym definitions are
1.1 Scope also provided.
This handbook addresses structural dynamics testing of
flight spacecraft and large instruments, and associated 4. TEST DESCRIPTION
dynamic test models and flight structure subsystems for
the mission dynamics environments and loads. The 4.1 Types of Dynamic Tests
handbook concentrates on new dynamics testing The types of Spacecraft structural dynamic tests covered
methodologies, but summarizes and provides key by this document will be vibration, acoustic, and shock..
references for older dynamic and static test Static testing will be discussed only in regard to the role
methodologies. it plays in complementing the dynamic testing in a
complete structural qualification program.
1.2 Purpose
Currently a diversity of testing cultures and approaches 4.1.1 Vibration Tests
exist in NASA and industry. New testing technologies Two types of vibration testing will be discussed: base
need to be disseminated. The handbook summarizes the drive and stinger drive. Base-drive vibration tests are
state of the art in structural dynamics testing, conducted with the test item sitting on a moving
recommends baseline verification programs, and platform that is driven by a vibration generator
describes and compares the advantages and (commonly called “a shaker”). The base-drive
disadvantages of the various test methodology options. configuration is commonly employed to achieve test
levels comparable to the launch environment. Fig. 1
shows a vertical axis base-drive vibration test of the
Mars Exploration Rover (MER) spacecraft mounted on
a platform, which sits on top of the shaker. Fig. 2 shows
a lateral base-drive vibration test of the MER DTM
Rover mounted on a platform that sits on bearings (a
“slip table) and is driven horizontally. Stinger vibration
tests on the other hand are conducted with the test item
either fixed or free, that is with it attached to a massive,
nominally immobile platform so that it is “fixed” at the
base, or alternately with the test suspended from a soft
suspension system so that it is relatively “free” to move.
One or more small shakers are then connected to the
test item with long rods, commonly called “stingers”.
Stinger vibration tests are commonly used for modal
vibration testing where the object of the test is to
generate data for verifying a mathematical model,
which has assumed either fixed or free boundary Fig. 2. MER DTM Rover in Lateral Vibration Test
conditions for the test item. Of course there are
exceptions to the common roles of these two types of 4.1.2 Acoustic Tests
vibration tests in that base-drive tests can be used to Acoustic tests are typically conducted with the test item
generate modal data and stinger vibration tests can be located in a large reverberant chamber, which is excited
used to generate reiativciy high-test !eve!s. One of tl.e witii one or inoie c!ectro-pne-amtic drivers with horns
distinguishing features of the different types of dynamic mounted in the walls of the chamber. The sound waves
tests is the frequency range over which it is useful. For in an acoustic test conducted in a reverberant chamber
large test items such as spacecraft, the useful range of usually have bounced off the chamber walls many times
both base-drive and stinger vibration tests is in the before striking the test item, and this results in an
range of ten to hundreds of Hertz, above which acoustic field that is relatively uniform in frequency and
frequency it is difficult to put much vibration energy space. An alternative acoustic test configuration
into a large test item without excessive motion at the employs a large number of electro-dynamic speakers
drive location. arranged in a circle closely surrounding the test item,
which may be located in a vibration or acoustic test
chamber or in an open space such as a high bay. In this
configuration the test item is in the “direct” acoustic
field of the speakers, which means that most of the
sound waves travel directly from the speakers to the test
item without first striking another surface. Direct
acoustic tests are characterized by relatively large
frequency and spatial variations because of the
constructive and destructive interference of the sound
waves from different speakers at various positions on
the test item. Acoustic tests provide energy at relatively
high frequencies compared to vibration tests. While the
specified frequency spectrum in spacecraft acoustic
tests typically ranges from tens to thousands of Hertz,
most of the energy in an acoustic test is concentrated
above a hundred Hertz.

4.1.3 Shock Tests

At the spacecraft level, shock tests are typically
conducted by initiating the device that causes the shock
environment in flight. The system that separates the
spacecraft from the launch vehicle usually involves a
pyrotechnic charge and is therefore an important shock
source for the spacecraft. This system is commonly
Fig. 1. MER Spacecraft in Vertical Vibration Test
tested by: 1) suspending the spacecraft, 2) firing the
separation charge, and 3) allowing the launch vehicle
adapter section below the separation plane to drop a few
inches onto a soft cushion. The other pyrotechnic
devices on the spacecraft should also be fired, if
possible in the sequence and in the environment the purpose of structural qualification and therefore
(thermal andlor vacuum) that they are fired in flight. involve low frequencies and high levels. Wave forms
The frequency range of spacecraft shock tests is include: a classical half-sine or a modification thereof, a
typically from hundreds to thousands of Hertz, with bundle of sinusoidal cycles of a single frequency and
most of the energy concentrated above a thousand with slowly increasing and decreasing amplitude, or
Hertz. less frequently, a wavelet with many frequencies, or a
complex time history, which may be representative of
4.2 Types of Excitation in Vibration Tests an actual in-flight event.
Three types of excitation are used in spacecraft
vibration tests: sine, random, and transient. Each of 4.3 Control and Limiting of Vibration and Acoustic
these three classes of time history has many variations, Tests
the most commonly used ones of which will be The details of the control in vibration tests are closely
discussed herein. In addition, sometimes two types of related to the type of input being used. However, there
excitation are combined to simulate a particular are some common features of the control and limiting.
environment, e.g., sine on random is often used to First, most of the control is closed loop, which means
simulate gunfire. that the input is adjusted in real time to coincide with
what is desired. Transient testing is the exception,
4.2.1 Sine because there is generally not enough time to adjust the
The most frequently used form of excitation in vibration input in a transient test. Sometimes the control system
tests used to be a swept sinusoid, which involves may be configured to terminate a transient test if the
~ w c ~ p from
i ~ g 2 !owe: frccpency ! h i t to an ’inner
-rr inpnt is not as desired, hut sidden termination of a high
frequency limit at a rate usually specified in level test is in itself problematic. Acoustic tests may be
octaves/minute. For example, a swept sine vibration test conducted open loop with little danger, because there is
of a spacecraft might involve a sinusoid with amplitude very little nonlinearity and only a weak interaction
of one “ G , the acceleration of gravity, swept from 5 Hz between the acoustic field and the test item. However,
to 80 Hz at a rate of four octaveshninute, which would even acoustic tests are commonly conducted with a
take four minutes to complete. Another type of sine test closed loop control system because it speeds up the
is “sine dwell”. In this case, the frequency is fixed and process of equalization to the test specification as the
the test proceeds for a fixed time duration or number of level is increased. Sinusoidal tests are generally
cycles. controlled to a peak or rms level, and random tests to a
PSD level. In both cases there is some preset tolerance
4.2.2 Random and some threshold for automatic shut down. In
A random vibration test is specified by the power addition to control, it is common practice in spacecraft
spectral density (PSD) of the input acceleration, which vibration tests to have some limit channels, which are
defines the distribution of average vibration energy with used to modify the control if these channels start to
frequency, and by the duration of the test. The square exceed their specified limits. In either sinusoidal or
root of the integral of the acceleration PSD over random vibration tests, these limits may be a function of
frequency is the root-mean-square (rms) acceleration. frequency and the input may be reduced, “notched”, at
The most appropriate measure of the severity of a frequencies where the limit is exceeded, which are
random vibration test is the maximum PSD value or the typically those frequencies where the test item has
PSD value at the frequency of the resonances of the test resonances. Acceleration responses measured at various
item. It is a common mistake to use the rms value of the positions on the test item are the most common signals
input as a measure of its severity. The problem with the used for limiting, but the advent of compact and stiff
rms value is that it depends strongly on the values of the triaxial force gages has made limiting the forces
PSD at very high frequencies and on the upper between the shaker and the test item increasingly
frequency limit, which are often irrelevant. popular.

4.2.3 Transient 5. P U R P O S E S A N D COMPARISON OF

All the inputs in vibration tests are transient in the sense DIFFERENT TESTS
that they are of limited duration, but here transient
refers to inputs, which last only a fraction of a second or There are generally three reasons for conducting
so. Many different types of transient waveform may be dynamics tests of spacecraft: qualification,
used for vibration tests of spacecraft, and transient workmanship, and verification.
excitation may be characterized in many ways,
including: waveform, duration, frequency content, level, 5.1 Qualification for Flight Environments
etc. Except for shock tests, which are seldom conducted The primary purpose of most dynamic tests of
with a spacecraft mounted on a shaker, transient spacecraft is the simulation of the flight dynamic
vibration tests of spacecraft are usually conducted for environments, which are typically so severe that they
 would cause failure of many electronic components, the system level tests. (The Cassini spacecraft RTG
mechanisms, optics, and structures were these items not problem mentioned in the last paragraph is an example
designed to survive them. These high levels of vibration of an interface problem, as the RTGs were extensively
and sound are generated by the launch vehicle and other vibration tested at the subsystem level.) It is, however,
sources such as pyrotechnic devices and a spacecraft important that the test levels in workmanship dynamics
landing on Mars. The most straightforward way of tests be low enough so that they do not cause problems
testing for these environments would be to exactly that would not occur in flight.
simulate the flight environment, but this is not
appropriate in most cases. Rather, the tests typically 5.4 Model Verification
represent a simulation of the dynamic environments The third reason for conducting dynamics tests of
defined from a statistical analysis of many missions and spacecraft is to verify dynamics models. This is the
many different operational conditions. In fact, it is justification for modal testing, and tests to verify jitter
common practice to define the flight environments and in-flight vibration models. In these cases it is also
using parameters of the dynamic tests that can be important that the test levels and durations be such that
reasonably conducted, e.g. random vibration PSD’s, the tests are nondestructive.
one-third octave band acoustic levels, shock spectra,
which is the maximum response of a single-degree- 5.5 Roles of Test and Analysis
of-freedom system, etc. It is in this context that In structural dynamics, the roles of testing and analysis
spacecraft designers often complain that they are are complementary, and one cannot overstate the value
designing to pass a test. Of course it is always good of pre- and post-test analysis. Since testing tends to be
practice to perindicz!!y c o n ? p ~ ethe test simulations expensive; it is important to use analysis to plan the
with actual flight data to insure that the conservatisms tests so that they may be conducted efficiently, and after
that invariably creep into test specifications do not the tests to use analysis to extend the test results to other
become excessive. loading and hardware configurations.

5.2 Flight Failures Due to Dynamic Environments 5.6 Advantages and Disadvantages of Various Tests
Since dynamic tests of spacecraft are both expensive The various types of dynamics tests have different
and risky, it is reasonable to ask: “How many flight purposes, different frequency ranges of applicability,
failures have there been due to the dynamic and also different advantages and disadvantages, so it is
environments?” In the beginning of the space program, very important to tailor the test program to fit the needs,
there were probably quite a few, although it is always reliability, schedule, and cost, of each program.
difficult to ascertain the cause of a flight failure with Different organizations, and even different programs
certainty. It is suspected that the JPL Rangers 4 and 6 within organizations, have different approaches to
Spacecraft failures were caused by launch vibration and defining dynamics test programs. All dynamics tests
that the Galileo high gain antenna’s failure to open was are risky in that even handling a built-up spacecraft
caused by the transportation vibration environment. involves some risk, and some tests, like open loop
Other government laboratories and agencies and their transient tests on shakers have proven to be particularly
contractors have experienced similar cases of vibration- risky. In general, acoustic tests are the most benign,
induced problems. For example, the problematic jitter followed by modal vibration tests, random vibration
of the original solar panels on the Hubble Space tests, sinusoidal sweep vibration tests, and finally
Telescope was caused by vibration generated by shaker transient loads tests. On the other hand, acoustic
thermal transients. In addition, it is appropriate to ask: “ tests are basically limited to detecting workmanship and
How many problems have been discovered in spacecraft high frequency problems. Random vibration tests are
dynamics tests that would, or may, have caused flight generally safer than sine sweep tests, because they are
failures? This is also difficult to answer, but there are easier to limit and notch, since one may dwell at lower
probably many. At JPL, the vibration test of the Cassini levels until the control system has adjusted the notches.
spacecraft identified an electrical grounding problem In a sine sweep test on the other hand, the control
between the spacecraft bus and the radioactive isotope system has to put the notch in “on the fly”, and
thermoelectric generators (RTG), which could have sometimes the resonance frequency is passed before the
been a serious problem in flight. notch is fully implemented. Shaker transient loads tests
are the most dangerous because they are of very short
5.3 Workmanship Dynamics Tests duration and open loop, so that over testing may occur
A secondary reason for conducting dynamic tests of before there is any chance of rectifying the situation.
spacecraft is to identify workmanship defects, which if The shaker failure that occurred during the HESSI
undetected would cause problems or failures in flight. spacecraft vibration test is an example of this. However,
Most workmanship defects are detected at lower levels shaker transient tests are still popular because they are
o f assembly, but there are some interface and inexpensive and save schedule compared with the
interconnection problems that can only be detected in extensive static test programs that they can replace.
 6. TEST PROGRAM AND TEST PLAN workmanship and re-work tests, which may be
conducted at or sometimes below flight levels in order
The term “test program” generally refers to the strategy, to identify flaws without risking any failures that would
or plan, for testing all the hardware associated with a not occur in flight.) The ratios of the design and test
given program, whereas the term “test plan” generally levels to the predicted flight level are often called
refers to the plan for testing a specific hardware item, “margins”. For example, for a vibration test, the design
such as the flight spacecraft. Herein both are discussed. margin might be 1.4 and the test margin 1.2. The
The “test procedure”, which generally refers to the amount of margin depends on many factors such as: the
detailed steps of conducting the test, is discussed in the institution, the purpose of the test, the conscquences of
next section, 7.0 Test Implementation. failing the test, and the degree of confidence in the
flight predictions.
6.1 External and Institutional Requirements
The first step in putting together a dynamics test Test specifications for random and sweep sine vibration
program is to assemble the requirements, some of which tests and for acoustic tests should generally be relatively
may flow down from external organizations. For smooth functions of frequency. One should avoid the
example, there may be requirements imposed by the temptation to follow a complex frequency pattern, e.g.,
NASA center responsible for the launch, or from the one associated with a time history recorded for a
launch vehicle contractor, or from the spacecraft particular flight, or a complex frequency response
provider, etc. Some‘ of these requirements may be function predicted by a finite element method (FEM)
difficult to change, and some may be negotiable, but code. It is also good to provide some flexibility, or
they should always be scrutinized to make sure that they room for negotiation in the case of flow-down
are applicable and the best approach for the subject specifications, so that the specification may be adjusted
system. Often each institution has its own institutional if preliminary tests with a mass simulator or with low-
test requirements, which may depend on the ultimate level inputs show that the project would be better served
customer or risk category of the mission. In the past, by modifying the specification.
these requirements were often contained in various
standards and compliance with the standards was 6.4 Baseline Requirements
mandatory. Now there tends to be much more flexibility A set of baseline dynamic test requirements should be
and a willingness to let each project tailor the testing defined at the beginning of each program. (The
requirements to the needs of the specific mission. In the alternative, i.e. to have the baseline program evolve as a
case of commercial spacecraft, the insurers often set the result of descoping later in the program, will usually
test requirements, however NASA is self-insured. involve a non-optimal program and wasted
resources.)The baseline program should include
6.2 Requirements Flow sufficient testing to satisfy the requirements for: 1)
In addition to external and institutional requirements, Qualification (validation) of the ability of the system to
there is the logical requirement that subsequent tests withstand the flight dynamic loads, 2) Workmanship
should be more benign than the ones that preceded it, so testing, and 3) Verification of models used to predict
that the early tests are a proof or masking test. (This is responses to loads and environments, which cannot be
the same philosophy as dad testing a swing before the adequately or reasonably simulated in a test.
child uses it. Of course, if dad breaks the swing, the (Deployment of booms in a zero gravity environment
child will be unhappy, but it’s probably better than might be an example of the latter.) For example, almost
having the swing break later with the child on it.) For everyone would agree that the baseline dynamic test
example, tests conducted on the flight structure are requirements for a spacecraft should include an acoustic
usually at lower levels than those conducted earlier on a test. Most would also include some type of modal test,
qualification structure. Similarly, the tests at higher and many would include a vibration test with the
levels of assembly are usually at lower levels than those spacecraft mounted on a shaker. The number and type
conducted on the unitdcomponents. of system structural dynamics tests depend on the
culture of the institution, the heritage of the spacecraft,
6.3 Design and Test Specifications and Margins the severity and nature of the flight environment, the
The starting point for the test engineer is often a test amount of static testing, the credibility of the analyses,
specification, which may have been provided by and of course, cost and schedule constraints.
someone else responsible for planning the complete
design and test program for the system and its parts. 6.5 Test Options
The design specifications for most systems include a Testing options might be whether the acoustic test is
specification based on the dynamics tests that are conducted in a reverberarit chamber or with speakers in
planned. The test specification is generally lower than a high bay; or whether the modal test is conducted with
the design specification, and higher than that predicted the spacecraft mounted on an inertial mass, on a shaker,
for the flight environment. (An exception are or, typically for verification of in-flight models,
suspended freely; or whether the vibration test uses the experience of the specific operators in conducting
transient, sine, or random excitation. Other test options similar tests.
involve the decision of when and how to test large
subsystems and subassemblies, and the fabrication and 6.9 Instrumentation
testing of dedicated test structures or units, called Instrumentation is discussed briefly here, instead of
structural development models, qualification test under test conduct, because it is often necessary, or at
models, or something similar. least advantageous, to install some of the
instrumentation before the test. The most common form
6.6. Combining Tests of instrumentation for structural dynamic tests is
Sometimes the various types of dynamics tests may be accelerometers, which may be of a variety of size,
combined with considerable savings of cost, schedule sensitivity, frequency rage, etc. depending on the
and handling risk. The specialists in each of these applications. Other types of instrumentation include
disciplines sometimes argue that combining these tests force gages, strain gages, and occasionally temperature
compromises the accuracy and utility of each test, and sensors. Often many of the interior instrumentation
they are strictly speaking correct. However, sometimes locations are only accessible at specific times during the
compromises are necessary. For example, in the build-up of the test item. (These may be removed after
QUIKSCAT program a quasi-static loads test, the test if the test item is partially disassembled, or
frequency identification test, random vibration test, and sometimes the cables are cut and the accelerometers
acoustic test were all four conducted in the space of actually fly.)
approximately one week with the spacecraft mounted to
a shaker in the vibration test cell mef. D]. This saved at 6.10 Fixtures
least a month of schedule compared to a test campaign Dynamics testing usually requires that the test item be
involving separate static, modal, vibration, and acoustic mounted on some type of fixture, which is often
tests. The short schedule of the QuikSCAT program, specific to the test item. The fixture configuration, and
one year from contract initiation to launch readiness, its interfaces to the test item and the vibration source, as
would not accommodate the schedule for conducting well as other ground support equipment, must be
four separate tests, and combining these tests satisfied defined, fabricated, and fit checked in advance of the
the requirements of a baseline program and test.
encompassed all four types of test.
6.11 Schedule
6.7 Hardware Definition The test plan must include a schedule of events leading
The first item discussed in both the test program and the up to the test and a detailed schedule of the conduct of
test plan is usually the test item(s). The extent and the test.
configuration of the test item for the test must be
described. Will it consist of flight hardware, some 6.12 Test Organization Relationships
engineering models, mass simulators, or a combination The single most important step in organizing a
of these? Will any items be missing? Will it be powered spacecraft dynamics test is to have one person identified
in the launch configuration? Usually the plan will as the test director, responsible for the safety of the
include some drawings, solid models pictures, or photos spacecraft during the test, for the instrumentation, for
of the test hardware showing the major components and the test conduct, for data analysis, and alas, for writing
interfaces. The coordinate system(s) and interfaces the test report. All information flow and important
should also be defined. decisions must flow through the test director, or their
delegate for specialized tasks. All of the other
6.8 Facility interrelationships and responsibilities for the test
The test facilities should be identified in the test conduct will depend on the organization of the
program and specifically described in the test plan. The institution(s) responsible for the spacecraft and for
facility must have the capability to safety implement performing the test. The engineering organization
the test specification and meet the cleanliness, handling, responsible for planning the test may have separate
and other test requirements. This is sometimes a groups responsible for structures, environments,
challenge. However, the safety of the hardware is the hardware, safety, etc. Generally the test will be
one thing that should not be compromised. It is a good conducted in a test laboratory, which may be part of the
idea to inquire as regards to the recent use of the facility organization that provides the spacecraft, or may be an
to conduct the corresponding type of dynamics tests of independent or outside organization. The responsibility
similar hardware, Le. of similar weight and size. (For of the test laboratory includes facility safety, which
example, some project managers don’t want to be the means that the shaker and all other systems operate
first customers to use new equipment, and most properly and do not malfunction. The test laboratory
probably would not want to be the last to use old will also provide people to operate the vibration
equipment.) Also it is appropriate to inquire as regards equipment, set-up and run the instrumentation, and
conduct the data reduction. It is the responsibility of the requirements in the test plan. A good test procedure is
test director to coordinate all of these activities. the key to a successful test. As I S 0 9000 says “Say
what you’re going to do, and do what you say.” One of
6.13 Risks: Test Safety, Flight Failure, and Cost I the major purposes and benefits of preparing a test
Schedule procedure is that it forces one to think through the
The three greatest risks to the flight hardware during details of the test in advance. Of course, there are many
dynamic testing of are: 1) handling damage, 2) shaker other benefits, including providing a road map so that
system malfunction, and 3) overtesting. Overrunning of everyone involved in the test can work together
the program cost and schedule are of course also risks, efficiently and know what’s scheduled to happen next.
which are very important to the program, since projects However, it is also important to realize that dynamic
are often descoped, or sometimes even canceled due to testing will always involve some uncertainties and
cost and schedule overruns. These risks of damaging the surprises, so it is good to maintain a certain amount of
hardware during the test and then perhaps overrunning flexibility to accommodate the unexpected.
the cost and schedule must be balanced against the risk
and embarrassment of a flight failure due to the 7.2 Facilities and Personnel
dynamic environment or to a workmanship problem. Good communication and smooth interfacing with the
test facility and its,personnel is very important. In all
6.14 Pre-test Analysis cases, it is essential that good communication and a
Pre-test analysis is one of the most important aspects of harmonious working relationship be established
test planning, because it provides insight into what to between the flight hardware, dynamics engineering, and
expect and how to dedi with it iii d v m c c of ‘?;e actiia! fzci!it)l peop!e. Genera!!y, there is a natural pace or
test. This causes the actual test to go much faster, and rhythm in the conduct of tests, which should be sensed
also allows the attention during the test to focus on new and honored. The chains of command and individual
problems that could not be anticipated. The most responsibilities should be as defined in the test plan, and
common type of pretest analysis consists of a of course the test director should be in charge of all
simulation of the actual test using numerical models, aspects of the test.
which may consist of finite element models (FEM) for
vibration testing, statistical energy analysis (SEA) or 7.3 Instrumentation and Data Analysis
boundary element models (BEM) for acoustic testing, The instrumentation is the heart of a dynamic test and
and appropriate high frequency modeling for shock, may include accelerometers, microphones, force gages,
although the latter are rare. and strain gages. It is important that the instrumentation
be set-up and calibrated, preferably end-to-end before
6.15 Preparation of Written Test Plan the test. Typically, the real time data analysis in
It is useful to distinguish between the test plan and the dynamic tests consists of spectral plots. However, it is
test procedure, which will be addressed in the next also highly recommended that the time history data
section. Generally, the test plan is prepared well in from each run be recorded, so that if there is a problem
advance of the actual test. The test plan serves two it can be investigated later. For example, excessive
major purposes: 1) It provides a description of the what rattling of the test item in a vibration test can be a
is planned, so that others may review it and comment, problem that requires examination of the time histories
and 2) It provides coordination and scheduling of the to resolve. The test director must decide how much data
many activities that must fit together in order for the analysis is to be conducted between each test run and
test item and test facility to be ready and the test to be how much will be done later. As a minimum, sufficient
successful. The test plan will typically cover the topics data analysis must be done after each run to understand
discussed in this section including: 1) Defining test what is going on and to insure that it is safe to proceed.
hardware, 2) Describing the facility and test equipment, A good rule is to not proceed if a significant portion of
3) Defining the test fixture, 4) Defining the the data is not available, not as expected, or not
instrumentation, 5) Defining the test specification and understood. Similarly, it is important that problems with
limits, 6 ) Defining the test runs and intermediate data the test item and test equipment be resolved, and that
analysis, 7) Naming the test director, who often writes the cause of the problem be understood, before the test
the test plan, and other key personnel and defining their proceeds. Sometimes waiting until a problem is
responsibilities, and 8) Describing the safety and remedied and understood takes resolve on the part of
cleanliness requirements and precautions. the test director, as the project personnel often want to
press ahead.
7. TEST IMPLEMENTATION
7.4 Equipment Operation and Control
7.1 Procedures The proper control of a vibration test is a very important
The test procedure is usually prepared by the test aspect of the test program, because overtesting and test
facility organization and flows down from the failures are not that uncommon. The first priority is to
 insure that the shaker does not malfunction and that the might consist of runs at -18 dB, -12 dB,-6 dB, and full
operators do not make an error in operating the level, with some data analysis and review between each
equipment. (One good practice to help prevent the latter run. Sometimes the -18 dB run is conducted without
is to limit the working hours to a standard day, and to and with limiting. It is best in the lower level runs to
avoid doing the most dangerous, high level tests late at have all of the limits scale down with the inputs, so that
night or the very first thing in the morning. (This is any problems may be identified and corrected, by
particularly important in small laboratories where one adjusting the limits, before the full-level test. Typically
operator or instrumentation person may be doing many the lower level runs are conducted for a shorter interval
jobs and may have been working very long hours.) The of time, the only requirement being the time necessary
test equipment should be exercised at full level, plus a to acquire valid data. Thirty seconds is typical for the
margin, before the test item is installed to insure that it lower-level runs
is operating properly. This pretest should include any
test fixturing, and a mass simulator if the weight of the 8. INTERPRETATION OF RESULTS
test item or reaction forces are appreciable, Le., greater
than 50 YOof the shaker capability. All the control There is a tendency to heave a sigh of relief and move
accelerometers in the pretest should be installed in the on after the completion of a system dynamics test. Of
same positions as for the actual test. The purpose of the course it is always good news if nothing as dramatic as
pretest is two fold: 1) to checkout the equipment and 2) a structure failure occurs, but it is always a good idea to
as a “dry run” to prepare all the personnel for the actual ask what has been learned from the test. Some
test. In this regard, it is best to have the pretest as close questions, which may be asked include: 1) Were the test
iii iillie io the actiia! test as the schedu!c wi!! permi?. inpu?s cnrrect? 2) Was there any under or overtesting?
During the actual test, the input to the test should be 3) How could the test procedure be improved for future
reviewed before and after each run, as well as tests? 4) Were there any structural, electrical, or
monitored during the run, to make sure that it is as functional failures of the test item? 5) Was there any
desired and within the test tolerances. significant wear of deterioration of the test item, which
should be remedied or taken into account in future
7.5 Response Limiting and Notching testing or service? 6 ) Are the test data consistent with
It is also important that the input in a vibration test be model predictions, and if not, why not? 7) Was anything
limited at the structural resonances to avoid overtesting. learned from this test, which would effect other testing
This may be accomplished by placing limits on the in the same or other programs? 8) And finally, how
responses, typically accelerations or forces. The limits should the test results be documented? 9) What should
may be on the peak level of the time histories, on the be the form and distribution of the test report? 10)
frequency spectra, or on the overall, that is the integral Should the test data, and perhaps the analytical model,
over all frequencies, of the responses. If there are rattles be incorporated into a database for future use? 11)
or spikes on the data, which interfere with control or Should the results be documented and distributed in a
limiting, it may be necessary to low-pass filter some of meeting presentation or paper in an archival journal?
the data channels. There is always a compromise
between the complexity of the test set-up and operation, 8.1 Structural Integrity
and the numbet of safeguards and limits one may wisely The most notable thing that can happen in a dynamics
implement. The balance depends on the sophistication test is a structural failure. Sometimes a structural failure
of the test hardware, test equipment, and operators. For is accompanied by a loud noise and visual observations
example, if too many limit channels are used, the such as separation of the parts and even pieces falling
vibration controller may be slow to update the input and off. More often a structural failure is observed only
to sense over testing. when the test item fails to operate properly in a post-test
mechanical functionality test, or when the test item is
7.6 Test Runs disassembled and loose parts andor other damage is
The number of test runs depends on the complexity of found. The before-and-after test traces observed in the
the test item, the number of test configurations andor vibration signature tests are seldom identical. It is
axes, and the problems encountered during the test. In usually difficult to make the decision to stop testing or
each configuration, it is common to begin with a low- to disassemble the test item to look for damage on the
level signature, or health monitoring run, which is basis of signature changes. Sometimes, a small change
normally repeated after the full level testing before is cleverly recognized as the indicator of a key
going on to another test axis or configuration. It is also structural failure, while other times the cause of a
good practice to have some sort of functionality check, frequency shift or in some cases even the complete
i.e., electrical, mechanical, optical, etc., between disappearance of a frequency peak is never determined.
configuration changes. There are normally a number of This decision of whether to stop or proceed with testing
low-level tests, before going on to the full level test. For after a signature change usually requires a caucus of the
example, in a random vibration test, a typical sequence technical specialists and the project personnel. When
. 1
i
’

none of these changes occur during the test, no damage square in a random vibration test in order to estimate
is observed in a visual inspection, and the test item the maximum stress that will occur during the test. Even
performs normally in a mechanical functionality test, it though shaker random vibration inputs are clipped at
may be said that the test item maintained it’s structural three sigma, responses can exhibit peaks with higher
integrity, and in that regard passed the test. Of course values of sigma. For example for a part with a high
the item may still have undergone some wear, e.g. the resonance frequency of 500 Hz, the commonly used
ball joints may have loosen up, or the structure may three sigma limit may b e exceeded in a random
have used up some of it’s fatigue life, e.g. through the vibration test after only a few seconds!
growth of a small but undetectable fatigue crack.
Often the cause of the failure remains somewhat
8.2 Functionality ambiguous. In these cases it is recommended that the
Frequently test failures are found after the dynamics suspected cause of failure be verified by retesting the
tests during the test item functionality tests, which may old design with additional instrumentation. If the failure
include electrical, mechanical, optical, or thermal is determined to be associated with a design problem, it
testing or some combination of these. Sometimes the is usually good practice to change the design so that all
electronics are powered on in the launch configuration of the relevant design margins are significantly
and monitored during the dynamics test to insure increased, so that the chances of another failure are very
normal operation of the equipment, which must operate small. Finally, it will be necessary to test the new
during launch, or sometimes just to aid in identifying design to verify that the problem has been fixed. A
intermittencies or failures early before continual review of the test specification is recommended at this
exposure to the dynamic environment causes additionai p i n ? , to insrzre thit the new par! is not overtested.
damage. Electrical failures are perhaps more common
in tests at lower levels of assembly, -wherethe dynamics 8.5 Verification and Validation
test levels are generally higher. Verification testing is usually conducted to check or
more often collaborate an analytical model or to assure
8.3 Post-test Analysis that the design indeed meets the specified requirements.
Post-test analyses may be conducted for a number of An example might be to verify that the test item, say a
reasons, including: 1) to tune the analytical model with spacecraft, has a fundamental axial resonance above 30
the test data, 2) to understand why a structural test Hz. Or in the case of a modal test, the test data may be
failure occurred during the test, 3) to predict the used to improve the finite element model so that it may
dynamic behavior of the test item after a design change, be used with confidence to predict the response of the
and 4) to extrapolate the dynamic response of the test test item to another dynamic environment, for which a
item to a different test or flight environment. Modal test will not be conducted. Also, one of the reasons for
dynamic tests are conducted expressly for the purpose conducting a random vibration or acoustic test is often
of tuning the analytical model, but data from to verify the workmanship of the test item.
environmenta1 base-drive vibration tests and even static
tests are also often used to improve the model. Validation testing is more fundamental than verification
The merging of test and analysis in order to extrapolate testing. Validation implies more of an end-to-end check
dynamic test data to predict the response of a modified of the whole design and fabrication process including
or new test item in a dynamics test is the most the starting points and assumptions. System
challenging type of post-test analysis. qualification tests for a predicted flight dynamic
environment such as random vibration or acoustics are
8.4 Test Failures, Redesign, and Retest examples of validation tests.
The first step in dealing with a test failure is to
determine the root cause of the failure. If the failure is a APPENDIX A. SURVEY OF INSTITUTIONAL
major one, NASA headquarters may appoint a failure PRACTICES
review board to help in this regard. It is very important,
but often difficult to determine the root cause of a APPENDIX B. CASE HISTORIES
structural failure. Without knowledge of the root cause
it is impossible to determine how to correct the problem Big vs. Small, High vs. Moderate Risk, (MER, Cassini,
or whether it is fixed. Because most good structural M U , TES, QuikSCAT, and GALEX)
designs are redundant, many failures occur because of a
cascade of events. For example, a bolt or a restraining ACKNOWLEGEMENTS
pin may back-out and then excessive motion may result
in stresses exceeding the design limit. Other times it is The work described in this paper was carried out by the
just a case of the design margins of a number of parts in Jet Propulsion Laboratory (JPL) California Institute of
a mechanism being too low. A common mistake of this Technology under a contract with the National
kind is the use of too low a multiplier on the root mean Aeronautics and Space Administration (NASA).