THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

345 E. 47 St., New York, N.Y. 10017 82-GT-311

C The Society shall not be responsible for statements or opinions advanced in papers or in
rJ discussion at meetings of the Society or of its Divisions or Sections, or printed in its
M publications. Discussion is printed only if the paper is published In an ASME Journal.
^^ LReleased for general publication upon presentation. Full credit should be given to ASME,
the Technical Division, and the author(s). Papers are available from ASME for nine months
after the meeting.
Printed in USA.
Copyright © 1982 by ASME

C. H. Cook
Structural Methods Manager Damage Tolerant Design of
Turbine Engine Disks

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

C. E. Spaeth
Program Manager
This paper describes a USAF sponsored effort to develop, apply, test, and evaluate
D. T. Hunter Pratt & Whitney's Damage Tolerant Design System for cold-section gas turbine
engine disks. The design system includes a Damage Tolerance Specification
Life Technology Manager
proposed for new USAF engine programs, material characterization for crack-
United Technologies Corporation, growth behavior, design procedures, and analytical life prediction methodology for
Pratt & Whitney Aircraft Group, consideration of large flaws. To evaluate and refine the design system, a current
Government Products Division, engine fan disk was redesigned to operate safely for a specified time after the oc-
West Palm Beach, FL currence of 0. 030-inch (0.76 mm) surface length fatigue cracks. The redesigned disk
was tested to failure while monitoring crack growth and correlating observed
R. J. Hill measurements with analytical prediction. Test results were used to refine the design
system. Current work involves extending Damage Tolerant Design capability to
Program Manager,
hot-section powder-metallurgy disks. The impact of these efforts is twofold; current
Aero Propulsion Laboratory,
Air Force Wright Aeronautical designs will benefit from improved life prediction capability in applying
Laboratories, Retirement-for-Cause philosophy, and future designs can take advantage of the
Wright-Patterson AFB, Ohio Life-Cycle-Cost benefit of designing for damage tolerance.

NOMENCLATURE criterion. Under this criterion, all components of a given popula-

tion are considered to be unsafe as soon as a crack of some finite
A = Crack Size size (defined by the NDE detection capability) has formed in the
B/M = Bill of Material statistically few members of the population which have minimum
DTD = Damage Tolerant Design fatigue properties. No attempt is made to utilize the life as-
LCC = Life Cycle Cost sociated with the remaining population members which have
LCF = Low Cycle Fatigue statistically higher fatigue properties and are therefore not
N = Life Cycles cracked. In addition, no attempt is made to optimize the design
NDE = Nondestructive Evaluation for slow crack growth except for consideration of internal mate-
RFC = Retirement-for-Cause rial flaws (the first disk application of fracture mechanics techni-
AK = Stress Intensity Difference ques).

INTRODUCTION From a safety standpoint, this approach has been generally

very successful since it contains a built-in safety factor by treat-
The design requirements of recent jet engines entering mili- ing all components as if they were "minimum". However, for real
tary service have emphasized increased performance and higher materials and for real design situations, lifetimes based on
thrust/weight ratios, which in turn result in higher stresses and time-to-crack-initiation of the minimum-life member tend to be
more severe environments for many components. These high extremely conservative for a component population. This may be
stress levels have resulted in the introduction into service of a seen by reference to Fig. 1, which illustrates the LCF crack
relatively small number of low life components, where "life" is initiation behavior at 1000 ° F (820 ° K) of Inconel 718, a typical
based on time to crack initiation (inception and growth to an nickel-based superalloy. For engineering materials like Inconel
inspectable size). Unfortunately, the traditional maintenance pol- 718, there is significant scatter associated with the number of
icy associated with this "initiation approach" leads to a large and loading cycles required to initiate a crack at some given stress
costly logistic supply problem caused by the relatively small level. For design purposes, this problem of "material scatter" is
number of low life components. To provide the needed perspec- usually eliminated by statistically debiting from the typical fail-
tive of Damage Tolerant Design philosophy, the following para- ure curve to a conservative "design allowable" level where the
graphs provide further description of the "initiation approach", probability of "failure" (in this case, crack initiation) becomes
introduce a life extension concept (Retirement-for-Cause) for remote. For critical components such as engine disks, this proba-
current designs, and describe an improved life design system bility is usually set at 0.1%; Fig. 1 shows a design allowable curve
(Damage Tolerant Design) for future designs. established via this philosophy. In service, a component manufac-
tured from this material would be used for the number of load
Initiation Approach (fatigue) cycles permitted by this design allowable curve and then
all such components in the population would be retired upon
Traditionally, components whose dominant failure mode is reaching this "design allowable" life. Theoretically, at this compo-
low cycle fatigue (LCF) have been designed to a "crack initiation" nent usage point (cycle count), only one component in a popula-
tion of 1000 would have actually initiated a crack and the
remaining 999 components would have some undefined useful life
to crack initiation remaining. Reference to Fig. 1 shows that in
the case illustrated, the number of cycles to reach the "design
allowable" curve and the population "typical" curve are signifi-
cantly different. At the design allowable limit a "typical" compo-
nent would have consumed approximately 10% of its potential
Contributed by the Gas Turbine Division of the ASME. life to crack initiation. Obviously, under an initiation criterion,
 there is no way to utilize this potential life without accepting a
higher probability of failure of the minimum-life member. L Ac -----------;
w
However, a recently developed method is currently under
evaluation to extend the "usable life" of current designs de- ,

veloped under an initiation design approach. It is a method based J

upon a knowledge of fracture mechanics crack growth rates and U I
Non-Destructive Evaluation limits (NDE). This new method is 4
termed Retirement-for-Cause.

90
0 _,l/
I

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

N 80
Typical Nc
_ 1
U)
Oes . q//o
♦'9n
7o ♦ ♦:6/e 1N 12N1 3N1

d Inspection Interval
a 60 FIGURE 2
RETIREMENT-FOR-CAUSE

50 ' It is clear that not all fatigue-limited components may be

handled in this way, and that each component must be evaluated
individually to determine the economic feasibility of RFC. The
40
inspection interval N I (Fig. 2) must be chosen such that it does
103 10° 105
not place undue constraints on the operation of the component
and also so that the cost of the necessary "teardown" and
Loading Cycles to Crack Initiation inspection does not negate the advantages of the life extension
FIGURE 1 gained. One thousand cycles of crack propagation may represent
INCONEL 718 S-N CURVE many years of service for one component and a much smaller
period for another. It seems unlikely that Retirement-for-Cause
Retirement - for-Cause (RFC)
can be applied to components limited by high cycle (vibratory)
fatigue considerations, but for many high cost components lim-
Under the RFC system (reference 1), additional life can be
ited by low cycle fatigue, such as engine disks, RFC does appear
utilized by adopting a rejection criterion that uses each compo-
to offer significant economic advantages.
nent in a population until it specifically initiates a crack rather
than rejecting the entire population based on the behavior of the
It is also clear that in applying Retirement-for-Cause,
statistical minimum. The development of fracture mechanics
Non-Destructive Evaluation (NDE) becomes a critical factor. The
concepts over the last several years has provided the degree of
inspectable crack length value (A 0 in Fig. 2) determines the
confidence, in the predictability of crack propagation rates, which
residual life of the component, and its detection is limited by the
is considered necessary to implement such an approach on a safe
resolution and reliability of the inspection system employed. In
basis for cold-section components.
many cases, the applicability of Retirement-for-Cause will be
predicated upon the ability of available NDE approaches to
Fig. 2 shows the basic Retirement-for-Cause concept. For a
detect a usable A. with sufficient sensitivity and reliability.
given component, the number of cycles, N, required to propagate
However, because the RFC procedure includes an in-depth stress
a typical crack from an initial detectable size A, to critical size A,
analysis, a component's defect-critical locations can be accurately
can be calculated and verified. An inspection interval is then
predicted and verified. For this reason, NDE techniques can be
established at some fraction of N, designated N I . The value of N I
selected and refined for a particular area rather than attempting
is established by considering the confidence in the NDE
to develop a technique for characterizing the quality of an entire
threshold crack value A 0 , cost-effective overhaul intervals, scatter
component. This inherently tends to increase the sensitivity of
expected between typical and minimum crack-growth life predic-
the NDE system to a level where RFC can be viable if the
tion, and the degree of conservatism desired. It can be seen that,
component design allows cost-effective inspection intervals.
over the interval of time NI, no component containing a crack
equal to or smaller than A. could fail catastrophically by reaching
Note however, that Initiation-based design with a RFC
A.
maintenance philosophy is not an optimum combination. In order
to produce an optimum design, the crack growth characteristics of
In using RFC as a hardware management system, all compo-
a particular configuration must be evaluated during the design
nents would be inspected at the end of the initial NI cycles, and
phase to insure than an adequate crack growth life exists and to
only those components containing detectable cracks (i.e., cracks
insure that cost effective inspection intervals are achieved. Com-
equal to or greater than A 0 ) would be retired. All others would be
ponent design with consideration of crack growth characteristics
returned for additional service. After an additional NI cycles, all
and critical crack size is termed Damage Tolerant Design (DTD).
components would again be inspected and again all components
The following describes a recently completed three year effort
with cracks larger than A. rejected and the remainder returned
(refergnce 2) to develop, evaluate, and refine such a system for
for service. In this way, the crack propagation residual life N, is
cold.section turbine engine disks.
continually reset to a safe value for operation during subsequent
intervals NI. By following this approach, components are only
rejected for crack occurrence (cause) and each component is
DAMAGE TOLERANT DESIGN (DTD)
allowed to operate for its own specific crack initiation life. It
should be noted that if three inspection intervals are provided
A DTD system for cold section (fan and compressor) engine
and a crack is missed during an inspection, another chance exists
disks has been developed and subsequently demonstrated in the
to find a larger crack, A*, before A, is achieved (Fig. 2).

2
design of a functional replacement disk for a current production relative to its meeting crack growth requirements. This is ac-
turbine engine. The damage-tolerant design system was suc- complished by collecting all itemized inputs and processing them
cessfully verified and refined by testing this disk with induced in proper order through appropriate computer programs and
flaws subjected to accelerated mission cycles. One program goal design evaluations.
was to demonstrate a damage tolerant disk design with the
capability of operating for three overhaul periods after the occur- The integration package is schematically shown in Fig. 3 to
rence of 0.030-inch (0.76 mm) surface length cracks. This interrelate the following:
"cold-section" effort built upon a prior program (reference 3) and
is currently being extended to hot-section powder-metallurgy • Criteria
disks. • Initial flaw characteristics
• Engine duty cycle
• Material crack growth model
DTD System Development • Stress intensity factor prediction and crack growth
algorithm methods
Fundamental and basic to the damage tolerant design sys- • Cumulative damage prediction model and cracked

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

tem was the development of a quantitative design specification disk residual strength requirements.
similar to MIL-A-8344 for allowable (inspectable) flaw size,
shape, orientation, and location applicable to cold-section engine The more significant details defining the DTD system and
disk materials and design. During this effort such a design the damage integration package are provided in the subsequent
specification was developed and used to govern the design ap- discussion of evaluation results.
plication and evaluation. It is a specification which after further
USAF and industry revision is proposed for use in future USAF During the course of the program the design system was
engine design programs. applied and then evaluated experimentally. Basically, a func-
tional replacement damage-tolerant cold-section fan disk for the
A DTD methodology was developed which is capable of F100 engine was designed and subjected to ferris wheel ac-
implementing the specification criteria and satisfying program celerated mission cycles equivalent to those encountered in the
goals by means of a damage integration package. The damage F-15 application. The product of a current effort to extend
integration package represents a step-by-step design procedure Damage Tolerant Design to hot-section disk will be evaluated
which, when followed to completion, evaluates a disk design similarly.

Disk D/T Criteria

Stress
1) Bores
Intensity
2) Boltholes
Factor
3) Rims
Surface Bore
Crack Defects
(NDE) (NDE)

Disk Residual
Strength
Requirements
Stress
Analysis

Disk Damage
Integration Engine Duty Cycle
Package
R
Material Crack P
Growth Model M
1) da/dN vs AK (Temp, R)
2) OK TH (HFF) vs R-Ratio ' Time

Disk Useful Life

N_

t6

Cumulative Damage
U
F/M Prediction
Model 10 102 103 104
Remaining Life

FIGURE 3
DAMAGE TOLERANT DESIGN

M
 To accomplish this cold-section evaluation, four major tasks manufacture damage tolerant disks was tracked and the con-
were accomplished as follows: clusion reached is that initial disk cost differences do not figure
prominently in the determination of engine Life Cycle Cost. The
Task I — Engine disk selection and design B/M disk could not be included in the LCC study because prior
contract efforts (Ref. 3) had demonstrated that it had negligible
Task II — Disk fabrication and life cycle cost tolerance for 0.030-inch surface length cracks. Two levels of NDE
evaluation capability were included in the LCC study to establish the LCC
sensitivity to this variable. LCC cost decreased significantly with
Task III — Disk test and design system refinement improvements in NDE capability to detect smaller flaws.
Task IV — Test-Hardware materials selection, Improving NDE capability to reliably detect 0.010-inch (.25
characterization, and subcomponent mm) surface length flaws made the disks comparable on a com-
tests mon basis; i.e., with inspection intervals equal to one-third of
their respective safety limits. On this basis, when the LCC
DTD TASK I — Engine Disk Selection and Design analysis was performed as part of the design selection process, the

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

damage tolerant redesign achieved a 40 percent Life Cycle Cost
The disk selected for damage-tolerant design was the F100 saving relative to the B/M. In addition, improved NDE capability
engine 2nd stage fan. This disk as a "Bill of Material" (B/M) combined with beneficial residual stress treatment, as achieved
component is made of Ti-6A1-2Sn-4Zr-6Mo.The F100 engine is by cryogenic spin treatment, would now allow consideration of
used in the F-15 and F-16 aircraft, and has well-defined duty Ti-6-2-4-6 damage tolerant designs which may meet the DTD
cycles and environments. intent and be lighter than Ti-6-4 designs. Current work with
cryogenic spin treatment of Ti-6-2-4-6 appears promising in this
Utilizing the DTD design system, three analytical disk de- respect (reference 5).
signs were developed which investigated the individual and syn-
ergistic effects of flaw size, shape, orientation, location and mate- After fabrication of three damage-tolerant 'Ti-6-4-disks,
rial data scatter on the disk life. To achieve damage-tolerance, as tests were conducted to allow evaluation and refinement of the
defined by surviving three inspection intervals after occurrence of design system.
0.030-inch (0.76 mm) surface length cracks, these designs utilized
three major redesign features:
DTD TASK III — Disk Test and Design System Refinement
• Substituted Ti-6-4 for Ti-6-2-4-6 to improve crack
growth characteristics Two of the three Ti-6-4 damage-tolerant design disks (De-
sign No. 1 geometry) underwent a one-cycle overload pre-spin to
• Utilized a pre-induced residual stress (pre-spin or introduce beneficial residual stresses. The third disk was desig-
cold-work bolthole) to reduce crack growth rate nated to be a damage tolerant baseline. It utilized the advantage
of material substitution (to Ti-6-4) and stress reduction but did
• Reduced nominal disk stress (added weight) to not receive a pre-spin to induce beneficial residual stress. A
provide overspeed margin for large cracks Ti-6-2-4-6 B/M baseline was unnecessary because prior work
established it as intolerant of 0.030-inch (0.76 mm) cracks.
For the purpose of comparison, a design constraint matrix
(Table I) was developed and included life (all failure modes The two pre-spin disks were accelerated to 120% overspeed
applicable), material type, component cost (recurring and at 500°F (530°K) and held at this condition until sufficient local
nonrecurring), manufacturability, ease of assembly, stress level, inelastic deformation was achieved as required to develop the
inspectability, weight (component and mating hardware), and proper beneficial residual stresses determined in Task I. Local
engine compatibility. Additionally, the matrix elements were beneficial residual stresses occurred at the notches (boltholes and
weighted to indicate impact on the total engine according to the blade-retention slots in rim) when the disks were unloaded. The
F100's experience with production hardware. Design choice of unloading caused the large volume of elastic material surrounding
Design No. 1 was an interpretation of the created matrix in terms the notch to force the inelastic behaving local notch material into
of program goals and Life Cycle Cost (reference 4). compression.
DTD TASK II — Disk Fabrication and Life Cycle Cost A complete "ferris wheel" strain gage survey of each disk
was performed prior to life testing. The disk No. 1 "ferris wheel"
• Three damage-tolerant disks (Design No. 1) were manufac- crack growth test was monitored using advanced acoustic
tured as functional replacements for the B/M disk. The cost to emission technology to detect crack growth. Acoustic emission
Table I. Design Summary Matrix

Damage Damage Damage

Configuration Tolerant Tolerant Tolerant
Factor Baseline Design No. 1 Design No. 2 Design No. 3
Material Ti-6A1-2SN-4Zr-6Mo Ti-6A1-4V Ti-6A1-4V Ti-8A1-lV-IMo
Bolthole Treatment PWA 99-1 PWA 99-1, Hot Spin Cold Work PWA 99-1, Hot Spin
Stress (Critical) °tavg °tavg Vibration °tavg
Functionality Base Spacer Change Spacer Change Spacer Change
Life (Luke Hours) 3,100 >10,000 >10,000 >10,000
Burst Margin 1.275 1.275 1.275+ 1.275 ±2
Weight, lb 14.3 17.3 19.9 17.3
Resonance (Speed and Stress) Base Improved Improved Improved
Life Cycle Cost ($ Millions) 1 22.3 24.4 23.3
Inspectability Base Improved Improved Improved
Maintainability Base Improved Improved Improved
Manufacturability Base Special Spin Cold Work Special Spin

'See discussion of Task II

2 Reflects
minor uncertainty in material characterization

4
 monitoring was supplemented with eddy current and replication preflaws, and to grow the cracks to a 0.030-inch (0.76 mm) surface
methods to correlate crack length with predicted crack growth. length. The disk was then cyclically tested to destruction. This
For disk tests No. 2 and 3, only replication methods were used in required 574 simulated engine mission cycles at accelerated load-
order to accelerate test time and minimize test costs. ing conditions (Figure 5) to achieve the 0.400-inch (10 mm)
critical crack length. Prediction of inspection intervals is based on
To expedite the ferris wheel crack growth test, disk No. 1 typical cracks and Figure 5 illustrates that crack growth scatter
was preflawed with elox slots in 10 boltholes. As illustrated in between the 10 locations was reasonable and that the fastest
Figure 4, disk No. 1 was subsequently cycled in the "ferris wheel" crack only slightly preceded the predicted typical life. The "ac-
under simple sawtooth loading to initiate cracking at the 10 elox celerated loading" used a bolthole concentrated stress which was
130% of engine loading.

aNi
0.35 0 Disk No. 2 was prespun and then preflawed in 10 rim slots.
C o It was sawtooth cycled in an attempt to initiate and grow the
C
Test 0.020-inch (0.50 mm) elox damage to a 0.030-inch (0.76 mm)
aN 0.30
/ (2000) surface length crack, however, after 3500 load cycles there were

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

One Cycle
no detectable initiated cracks from the elox starter damage. It
N0.25
o // was concluded that the disk had demonstrated substantial dam-
o,
C w age tolerance capability, and since the last 2000 load cycles were
J Max
at the limit of the pullbars (110% of engine loading), the test was
U 0.20
discontinued to ensure sufficient funds and pullbar life to corn-
——— plete the No. 3 disk test.
U Mean of 10 Locations
m Min
0.15
Disk No. 3 was not prespun to introduce the deep com-
I- 0 1000 2000 pressive residual stresses. Instead, this test provided a baseline
Precrack Cycles disk which utilized a damage tolerant material, and a reduced
nominal stress disk profile compared to the B/M disk manufac-
tured from Ti 6-2-4-6. The disk was preflawed in 10 boltholes and
FIGURE 4 sawtooth cycled to initiate and grow the cracks to the required
BOLTHOLE PRECRACKING 0.030-inch (0.76mm) starting surface length. The disk was then

10

aResidual = —59 ksi

I CMaximum = 96 ksi
= 70 ksi
O'Nominal
QRange = 155 ksi
100x/

co m
1.0 I

I
C
C
jd /0 1 1 1 1 Two Locations Failed I Predicted "Typical" Failure
Ca (at 574 Missions) (667 Missions)
N 25%

x (Largest Not Failed) }

rn
C
cc
N J
O Simulated Mission 4
J W
One Cycle
U
f0
U V
m
0 0.10
I-
Predicted and
— Experimental
r—Max i
—Mean of 10 Locations

Min

0.01 1
1 10 100 1000 10,000

Simulated Mission Cycles

FIGURE 5
DTD BOLTHOLE LIFE TEST
5
mission cycled using the same accelerated test conditions as disk 6000
test No. 1 and the crack propagated to a near-critical final size in 5600 hr
275 mission cycles.

Following the crack growth tests, representative locations 5000

from the disks were broken open and examined to determine final
crack lengths and depths and variations in crack aspect ratios
(crack length/crack depth). 3 Life Goal
0 4000
x
Life results indicated that the prespin treatment alone w
yielded a 2X life improvement under accelerated life testing, m
which equates to 5X improvement under normal engine operating W
3000
loads. Figure 6 illustrates that the combination of prespin, mate-
rial substitution, and stress reduction (weight increase) was suffi-
cient to achieve 140% of the life goal (4000 engine hours subse-

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

quent to a 0.030-inch (0.76 mm) surface length crack). W 2000

TASK IV — Test Hardware Material Selection, Characteriza- V4.`^•I pI

1100 hr
tion, and Subcomponent Tests
1000 Test No. 1 i
The material chosen for use in the damage tolerant design I •I Test No. 3 Results
was Ti-6A1-4V. Three candidate materials that were considered • Results (574) --I
are: (1) Ti-6A1-2Sn-4Zr-6Mo, (2) Ti-6A1-4V, and (3) 0 11(275)
Ti-8A1-1V-1Mo. 0
I. I I
100 200 300 400 600
Titanium was chosen for this 2nd-stage fan disk because fan Accelerated Test Cycles
weight studies conducted during early F100 engine design showed
that weight penalties of up to 30% would be incurred when FIGURE 6
nickel- or iron-base alloys were substituted for the titanium disks. DAMAGE TOLERANT DISK BOLT HOLE
Of the candidate titanium alloys, Ti-6-4 was chosen because of its
desirable toughness and the fact that its advantageous crack Through this cold-section DTD effort, significant design
propagation characteristics were already well defined and only a system refinements were accomplished and notable lessons were
minimum number of specimen tests from the test disks heat code learned. These included:
would be required.
1. Normal variations in material properties necessitate several
During the design system development, the fact that cracks specific material strength considerations for damage tolerant
would be growing through regions where substantial residual design of disks. These are:
compressive stresses exist, was recognized. From previous work
with Ti-6-4, compressive stresses are known to contribute to and • Manufacturing production post-spin (service
increase the crack growth rates for a constant level of maximum pre-spin) speed is to be selected based on forging
tensile stress. Therefore, this degree of (-) R-ratio degradation for strength qualification test for each heat of disk
various minimum compression and maximum tension stress levels material.
had to be quantified. (R-ratio is the ratio of the minimum stress
to the maximum stress of the test cycle). • Disk growth should be "designed" recognizing vari-
ation between minimum and maximum material
In addition to positive and negative R-ratio specimen tes- properties. Post spin growth measurement is to be
ting, bolthole specimen tests were required to evaluate the crack used to substantiate having achieved the proper
growth model and verify life prediction calibration. Life results residual stress treatment.
from bolthole specimen tests demonstrated significant life im-
provement due to the overload treatment but surfaced the fact • Prediction of local inelastic behavior at
that during Task I disk design, disk stress residuals had been stress-concentration features should evaluate the
overpredicted due to assuming minimum material properties. need to use maximum (upper-bound) stress-strain
With this conclusion, the disk spin treatment was modified to and/or creep behavior to avoid optimistic predic-
recognize typical material properties and to define the spin speed tion of beneficial residual stress.
used in Task III. The design system was modified to account for
the allowable variation in material strength as described in the 2. Stress prediction procedures may require revision to ac-
following results section.
comodate Damage Tolerant and/or Retirement-for-Cause
design philosophy. Specifically:
RESULTS AND SUMMARY

Table I shows the final comparison between designs gener- • Damage Tolerant structures, by nature of their
ated and Figures 5 and 6 show the results of the test of the DTD
benefiting from large residual stresses and achiev-
ing long life, are very life sensitive to small stress
disk produced by the DTD system. As can be seen, the agreement
prediction errors.
between the predictions on crack growth rate and the test results
in Figure 5 are very good for the simulated mission test cycle. It
is felt that this effort demonstrated that turbine engine cold
• Significant stress prediction conservatism such as
section disks can be made damage tolerant by materials selection, may be the case in the broach slot bottom of
3-dimensionally complex rims, is acceptable for
stress reduction (weight increase) and spin treatments that do not
constitute a radical departure from industry state of the art. The "Initiation" design philosophy but causes unaccep-
Life Cycle Cost savings that result are very sensitive to the level table overdesign (weight, costly material, etc.) in
application of "Damage Tolerant" philosophy. Lo-
of NDE technology available and are being evaluated more thor-
cations such as flexible fan rims may require a
oughly as part of a current effort to extend Damage Tolerant
rigorous 3-D stress analysis to minimize con-
Design system capability to hot-section (powder metallurgy)
disks. servatism.

6
 • Costly, inelastic stress-concentration analysis is "n" is influenced by the nature of previous cycles
seldom required. The state of stress/strain at local (n-1), (n-2), etc. The integration of mission cycles
concentration regions can be adequately de- having this complication must be handled by
termined with simple extension of 2-D or 3-D cumulative damage models which incorporate re-
elastic finite-element analyses. To account for local tardation effects.
surface plasticity located at strain concentration
regions, a modified Neuber (reference 6) approach • Models for crack growth retardation due to ov-
should be used to determine the controlling local erloads are available in the literature for use where
surface inelastic residual stress values based on the significant overloads do occur during operation
2-D or 3-D elastic analyses. The term "modified" (reference 7 and 8).
Neuber approach implies the use of a
kinematic-hardening perfect-plasticity assumption However, recognize that the damage tolerant design
for modeling the reverse yield situation. This approach employed in this effort was to apply an
method of approximating plasticity effects has overpowering overload (pre-spin) under controlled

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

been substantiated by correlation with specimen conditions so that crack growth under all subse-
and component crack propagation test. Note that quent cycles would be retarded. It should be noted
this does not preclude requirements for inelastic that the mission cycling of many turbine engine
(plasticity/creep) analysis for overall disk residual disks is like that of the F100 2nd stage fan in that
growth and determining residual stress distribution the mission events produce differing minimum
from bore to rim of disk. stress values but a common maximum stress value.
Under these conditions, simple linear crack growth
3. A sophisticated analytical system of crack growth prediction response is to be expected. On the other hand,
is required. Specific recommendations relating to this effort turbine disk stresses are driven by thermal gra-
include: dients in addition to centrifugally induced stress.
Consequently, the opportunities for crack tip
• The "damage integration package" must recognize "blunting" by mission-related overloads increase
the influence of R-ratio variation (vs. crack depth) the likelihood of having to employ non-linear frac-
due both to local (notch) and bore-to-rim residual ture mechanics techniques increases correspond-
stresses.
ingly. The current hot-section Damage Tolerant
Design effort will address this consideration.
• A material crack-growth model with extension to
CONCLUSION
large values of negative R-ratio is required. This
model must be calibrated to component specimen
testing to ensure proper "blending" of (+) R-ratio A Damage Tolerant Design System for high energy compo-
nents has been developed and proven. This design system which
and (–) R-ratio models.
was successfully demonstrated for a cold section disk, is in-
herently adaptable to hot section disks, other rotating compo-
• To successfully predict crack growth behavior in
complicated disk geometries under cyclic stress, nents, and non-rotating structures. As a tool to assess existing
sophisticated crack tip stress intensity factor (AK) designs it has been successfully applied in a variety of Air Force
prediction methods are required. OK is the param- engines including the F100, TF30 and J52. However, more effort
eter which embodies the effect of the stress field, is required and is being conducted to use DTD with newer, lower
the crack size and shape, and the local geometry. toughness alloys and high temperature powder processed alloys.
The influence function theory (Reference 7 and 8) Figure 7 shows both the trends in design operating stress levels
is recommended in life prediction algorithms for and material crack growth for newer systems. Together these two
complex stress/geometry combinations. Life predic- trends indicate that DTD must be a consideration in the design
tion algorithms for part-through crack geometries of future gas turbine engines; in fact, all new engines developed
by the US Air Force are expected to utilize Damage Tolerant
such as full- or half-elliptical surface and corner
cracks should be developed by contractors engaging Design concepts.
in damage tolerant design of disks.
120 — — Resistance to Crack Growth
_• The empirically derived elastic fracture mechanics F100 r
relationship between crack growth rate (da/dN)
and crack tip stress intensity factor (AK) is numer- 100 0
ically integrated to determine crack size as a func- ♦ TF30-414
tion of mission cycles. In basic form, the relation is:
U Operating
)

A (final) 80 Stress
da ro z TF30 `
Npropagation — J
(initial) f(OK, R-ratio, temp,freq) a J58 0 0
A0
m o ♦ a
60
0J75 a o
where the numerical integration of this expression
will be performed on a cycle by cycle basis to the 0
z
_u_u
defined engine duty cycle simulation. 1955 Time —^ 1981

The integration can be performed in a variety of

ways provided due regard is given to the particular FIGURE 7
MATERIAL AND DESIGN TRENDS
load spectrum and its potential for producing
non-linear crack growth response. An example of ACKNOWLEDGMENTS
this effect is the retardation of crack growth during
cycles subsequent to a cycle containing a significant The authors gratefully acknowledge the significant contribu-
overload. In this connection it is said that the tion made by Messrs. H. E. Johnson (Life Prediction) and
structure has a memory, since its response to cycle M. S. Mills (Testing).

7
 REFERENCES 6. H. Neuber, "Theory of Stress Concentration For Shear
Strained Prismatical Bodies With Arbitrary Nonlinear
1. R. Hill, W. Reimann, and J. Ogg, "A Retirement-for-Cause Stress-Strain Law," Journal of Applied Mechanics, Transac-
Study of an Engine Turbine Disk," AFWAL-TR-81-2094. tion of the ASME, December 1961, pp. 544-550.

2. C. H. Cook, C. E. Spaeth, and H. E. Johnson, "Damage 7. J. R. Rice, "Some Remarks on Elastic Crack-Tip Stress
Tolerant Design For Cold-Section Turbine Engine Disks," Fields," Int. Journal of Solids Structures, 1972, Vol. 8, pp.
AFWAL-TR-81-2045, June 1981. 751-758, Pergamon Press.

3. T. A. Cruse and T. G. Meyer, "Structural Life Prediction 8. H. F. Bueckner, "A Novel Principle for the Computation of
and Analysis Technology," AFAPL-TR-78-166, Dec. 1978. Stress Intensity Factors," Z. Agnew Math Mach 50, 1970,
pp. 526-246.
4. C. E. Meece and C. E. Spaeth, "Damage Tolerant Design:
An Approach to Reduce the Life Cycle Cost of Gas Turbine 9. O. E. Wheeler, "Spectrum Loading and Crack Growth,"

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1982/79603/V005T13A029/2394840/v005t13a029-82-gt-311.pdf by guest on 20 July 2024

Engine Disks," AIAA Publication No. 79-1189, June 1979. Journal of Basic Engineering, March 1972.

5. F. C. Gillette, "Cryogenic Proof Test, A Positive Inspection 10. Lukas, P. and M. Klesnil, "Transient Effects in Fatigue
Technique," Presentation to AFWAL, Sept. 1981. Crack Propagation," Engineering Fracture Mechanics, Vol.
8, No. 4, 1976.