Proceedings of ASME Turbo Expo 2014: Turbine Technical Conference and Exposition

GT2014
June 16 – 20, 2014, Düsseldorf, Germany

GT2014-26423

RESIDUAL LIFE ASSESSMENT OF A CRITICAL COMPONENT OF A GAS

TURBINE– ACHIEVEMENTS AND CHALLENGES

1
Wieslaw Beres , Zhong Zhang, David Dudzinski, W.R. Chen, X.J. Wu
National Research Council Canada
Aerospace Portfolio
Ottawa, Ontario, Canada, K1A 0R6

ABSTRACT collaboration program executed several years ago. An objective

The residual life assessment of a turbine spacer from a gas of this program was to update lives of selected life-limited
turbine engine is presented. The spacer has been identified as rotating components of this engine.
one of the safety critical components of the engine, therefore The initial life of this gas turbine engine was assessed by
the useful life of this component significantly affects economic the Original Equipment Manufacturer (OEM) at the engine
operation of the fleet. design stage. The lifing methodology included analyses and
Numerical analyses of fatigue crack propagation at one testing of various life-limiting locations in critical components.
critical location of the spacer were performed using both three Commercial engine operators have to strictly adhere to the life-
dimensional (3D) finite element based method and the weight limits established by the OEM, but in a non-commercial
function method. These results combined with the material data environment, system managers or life-cycle mangers have
allowed for basic assessment of the damage tolerance of this airworthiness authority. That is, they can decide for their own
component. Experimental validation of the spacer life was assets on engine overhaul policies and the possibility of flying
performed in a spin rig facility. During this validation, two sets engines beyond the OEM recommended life. These
of spacers were tested and the number of cycles to appearance airworthiness decisions have to be made on the basis of sound
of a detectable crack was recorded. Moreover, a fractographic technical data and properly applied risk assessment. Therefore,
study was conducted on the fracture surfaces of two spin rig lifing of critical components of aerospace gas turbine engines
tested spacers using scanning electronic microscopy techniques. has a significant effect on the safety and economics of fleet
It was found that crack nucleation occurred at multiple sites and operation.
crack propagation occurred by a mixed mode of striation Engine fleet operators are currently faced with the need to
formation and faceted fracture. Therefore it was concluded that operate legacy gas turbine engine fleets due to diminishing
the mixed mode interaction should be considered in predicting resources for new equipment. Because of the uncertainties in
the fatigue life of the spacer. residual lives of high-time engine components, the amount of
Finally, the paper describes the challenges and pitfalls time these engines can be kept safely in service is a concern to
encountered during preparation and execution of the analyses engine life-cycle managers. Because maintenance cost
and tests, including availability of engine and operational data constitutes a significant portion of the total life cycle cost of an
and also uncertainties in interpretation of the results. engine, methods that can extend their lives, without costly
component replacements, are of significant benefit to fleet
INTRODUCTION operators. The need to balance risk and escalating costs of
This paper describes the work performed to assess the replacement explains the growing interest in application of life
residual life, referred to as damage tolerance, of a turbine extension technologies for safely extracting maximum life out
spacer from a widely used legacy turboprop engine. The work of safe-life limited parts [1]–[5].
was performed as a follow-on project to an international
1
Corresponding author. Email wieslaw.beres@nrc-cnrc.gc.ca

1 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 Gas turbine components are subjected to a combination of LIFE UPDATE PROGRAM
damage processes induced by the combined effects of The life limits for the critical turbine components of this
mechanical forces, temperatures and environments in service. engine were established in the early 1980’s using stress analyses
One such process is fatigue damage where the majority of life is and spin rig testing. These analyses, although on the cutting
taken up by the nucleation and propagation of a dominant crack edge of the finite element analysis (FEA) for their time, were
[1]. based on relatively coarse meshes, when compared to current
A lifing procedure for life limited rotating components standards. Two-dimensional (2D) analyses were used, along
generally addresses the following issues: (a). Identification of with broadly defined mission profiles, a limited number of time
the critical locations; (b) Profiling the load history; (c) points from thermal analysis, and estimated stress concentration
Identification of the damage mechanism; (d). Definition of factors. 3D analyses were not common in the early 1980’s
crack initiation; (e) Establishing of crack detection limits; (f) because of long computer calculation time and memory
Characterization of crack growth; (h) Determination of critical limitations. Consequently, with coarse meshes in 2D FEA, the
crack length; (i) Verification of life as supported by engine resulting stresses were not as accurate as those obtained by the
data. latest numerical techniques with fine meshed 3D features.
Some theoretical concepts are still not clear, such as how to As described in [14] and [15], in 2001, an international
define crack nucleation, transition of short cracks to collaboration program was formed to comprehensively re-
macroscopically detectable cracks and whether this occurs evaluate and update the low-cycle fatigue (LCF) lives of critical
below or above a fatigue threshold. Addressing these questions turbine rotor components of this engine. This was done using
poses great challenges to implementation of the lifing spin tests, thermal and stress analyses, and the most current life
procedures. In addition, at the theoretical level, other issues, prediction techniques available.
such as models that can bridge the crack nucleation and crack The primary objectives of the program were to provide
growth stages, must be devised and implemented. more accurate LCF crack initiation data, to make crack growth
Recently, significant amount of work has been performed in data available, and to improve the life management procedures
NRC-Aerospace to address these questions. These include of the selected, life-limited, rotating turbine components. This
physics-based modelling of fatigue and creep crack nucleation was achieved using advanced 2D and 3D finite element
and their interaction, with applications of these models to life analyses, fatigue crack growth tools, and spin rig testing.
prediction of aerospace components. In addition, efforts have As a result of the program, the LCF lives of the life-limited
been made to develop diagnostics, prognostics and health components were established, but only limited amount of work
management systems for gas turbine engines, as described in was done at that time to assess damage tolerance of these
[6]–[10]. critical components, i.e. determination of crack propagation
Damage tolerance concepts for gas turbine engines have lives.
emerged due to limitations of the traditional safe life design The current paper describes the follow-on activities
concept, where only 1 out of 1000 components is expected to directed towards assessment of the damage tolerance of one of
develop a detectable fatigue crack at the end of a safe life the life-limited component of the engine, the turbine spacer.
period. The remaining 999 components are retired in a crack- It should be noted that the use of crack growth limits, that
free condition with a large amount of potential life still exceed the safe crack initiation life for these components has
available. not been endorsed by the OEM. However, such established
By applying a safety-by-inspection life-cycle management lives can be used by the non-commercial engine operators, if
approach, which relies on predictions of crack growth life and desired, to support fleet risk management.
non-destructive evaluation (NDE) of components at overhaul,
integrity of these components can be assured and significant STRESS ANALYSIS OF ROTOR COMPONENTS
cost savings can be realized. This is often referred to as the The finite element stress analyses of the entire turbine rotor
“damage tolerance based life-cycle management” (DTLCM) were performed previously as reported in [14]. The main
concept, [11]–[13]. In this philosophy at the end of each Safe purpose of these analyses was to provide the stresses, based on
Inspection Interval (SII), all components are inspected and temperatures supplied by the OEM, for calculation of the low
components with no crack indications are returned to service for cycle fatigue life and determination of spin rig testing
another SII. This procedure is repeated until a crack is found. parameters, in addition to establishing life limiting locations.
With this approach the components are retired based on their This task was achieved by using 2D axisymmetric FE stress
individual condition. DTLCM procedures assume that: (i) flaws models for the turbine rotor with a much finer mesh than that
exist in as-manufactured parts, (ii) they are located in the used many years ago to obtain current LCF lives for the critical
fracture critical locations of the components, and (iii) their sizes components. The purpose of the 2D analysis was threefold—to
are just below the detection limit of the NDE techniques used to provide temperature and stress information for the fatigue life
inspect the components. calculations at the axisymmetric locations, to identify the
In this paper, to protect the intellectual property of the critical stress increments and to provide temperature and
OEM, only relative (non-dimensional) data are reported.

2 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 displacement boundary conditions for the 3D analysis of non- The spacer under investigation is manufactured from
axisymmetric features. IN901 alloy. At the time of the original collaboration program,
The temperature distribution and displacements at the material data for the analyses and testing were delivered by the
model boundaries were extracted from the 2D analyses and OEM. Further to this, material cut from service exposed
interpolated onto the 3D models. components was evaluated in-house to update the material
Since the large number of combinations of 2D geometry properties.
and mission cycles could not practically be extended to the 3D
models, for the purpose of life assessment, only the worst case
combination of geometry and cycle was considered in the 3D
models. In this way, the large number of geometry and cycle
permutations was reduced to a single equivalent stress
amplitude and temperature for the spin rig tests. [14]
The turbine rotor component lives were then analyzed by
the OEM. The life assessments included both LCF and crack
growth predictions in the turbine disc [15].
On the basis of the mission profile analyses as well as
thermal and stress analyses for the entire rotor, four components
were considered to be the most critical ones and thus were
chosen for spin rig testing. One of these components under
investigation was a turbine spacer.
Figure 1. TWO-DIMENSIONAL FE CALCULATION RESULTS OF THE
Subsequently, the detailed stress analyses for these four TURBINE SPACER UNDER SPIN RIG CONDITIONS [14] .
components for spin rig testing were performed. The objective
of these calculations was to provide spin rig operating
conditions that would generate the same Walker equivalent
CRACK GROWTH PREDICTION IN THE SPACER BY
stress range in the life-limiting locations of the component in
FEM
the spin rig as that calculated for engine operating conditions
To assess damage tolerance of the spacer, a 3D FE
[14], [16]:
modelling technique was required to determine the stress
intensity factor (SIF) ranges and predict crack growth.
 *   max (1  R) (1) Therefore, sub-models of these features of interest were
where: generated.
σ* is the Walker equivalent stress, In this study Zencrack [17] was used to predict crack
σmax is the maximum cyclic stress, growth under spin rig conditions. In the Zencrack method,
γ is a constant, special crack elements were inserted into the uncracked mesh
R is the stress ratio, R=σmin/σmax . and replaced old mesh at the fracture critical location. The new
mesh was then submitted for FE analysis with the results
After the values of the Walker stress were defined, 2D extracted and processed automatically by Zencrack.
axisymmetric finite element models of the spacer spin rig Abaqus [18] was used as a main FE solver for calculation
assemblies were generated to analyze the behaviour of the test of the crack progression in the spacer.
rotors. For damage tolerance calculations, an initial semi-circular
Subsequently, sub-models for the spin test were created. crack of 0.010” (0.25 mm) in radius located at the spacer bore
Cyclic symmetries were maintained at the radial cutting planes. corner was assumed. This was the fracture critical location
The loading applied included: assembly loads, thermal loads, established by the FEA. Figure 2 shows the typical mesh
centrifugal loads and interactions between the components. The arrangement around the crack generated by Zencrack and
boundary conditions at the circumferential and radial cutting progression of the crack front.
planes at the sub-models were transferred from the global 2D As mentioned, the fatigue crack growth rate data for the
models. The rotational speeds were adjusted in each analysis Paris regime were generated in-house and compared to the data
such that the equivalent Walker stress ranges under the spin rig provided by the OEM for conditions corresponding to the local
test conditions were equal to those under the engine operating temperature at the critical location and the appropriate stress
conditions. The isothermal test conditions that differ from the ratio. To account for an inherent scatter and also to take into
thermal conditions of the spacers in operation were taken into account differences resulting from the changes in crack
account when performing finite element analyses for the spin rig propagation due to service exposure of the material, scatter was
conditions. Figure 1 provides an example of the stress analysis taken into account by assuming two values of the coefficient C
results for the spacer performed for the spin rig conditions. in the Paris law. i.e. C1 as delivered by the OEM, and C2=2•C1.
These variations in C represent the mean value and the assumed

3 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 upper bound value that take into account variable conditions Distance from the initial center of the crack vs. number of cycles.
such as microstructure, loading, temperature, stress ratio and (Nondimensional)
1.20
natural variability of the fatigue process. The simulated cyclic C = C2
C2 = 2 * C1 C = C1

Distance from the initial crack center (a/amax)

loading used for crack growth predictions was representative of 1.00

the actual spin rig loading conditions. The Zencrack control Critical crack size

parameters were adjusted to achieve numerical progression of 0.80

the crack. Approximately 123 Zencrack iterations and 51 hours

0.60
of clock time were required to complete the crack propagation
interval (CPI) calculations, Figure 3. As mentioned, CPIs are 0.40
presented in the relative, non-dimensional form.
0.20
CPI
ai CPI
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Number of cycles, N/Nmax

Figure 3. CRACK PROPAGATION CURVES FOR CRACKS IN THE BORE

CORNER OF THE TURBINE SPACER FOR TWO VALUES OF THE C
COEFFICIENT OF THE PARIS LAW. RESULTS ARE PRESENTED IN A
NON-DIMENSIONAL FORM.

CRACK GROWTH PREDICTION BY WFM

To augment the crack propagations calculated by the FEM,
a weight function method (WFM) was developed to speed up
predictions.
For this, a weight function was built for 3D components
containing elliptical cracks, such that the stress intensity factor
(SIF) can be evaluated through integration under a given stress
field in the uncracked component calculated only once by the
FEM.
A computer code was developed in Fortran to implement
the above FEM-WFM solution procedure. The program
requires the input of the stress field obtained from the FEM
calculations. The code controls the stress extraction from the
FEM results file, evaluates the SIF through numerical
integration and performs the crack growth simulation.
For the spacer, the stress results from the 3D FEM analysis
Figure 2. CRACK GROWTH IN THE BORE CORNER MODELLED IN for the component with no crack under the given loading and
ZENCRACK. CRACK ORIGINATED FROM A SEMI-CIRCULAR CRACK. boundary conditions were taken to map the stress distribution
FOUR STAGES OF CALCULATIONS ARE REPRESENTED BY DIFFERENT on the potential crack growth plane; usually defined as the
COLOURS. maximum principal stress plane. The SIF value for the initial
crack condition was evaluated using the 2D weight function
integral. Then, subsequent crack growth increment was
calculated according to the material’s crack growth law by the
Paris equation. The crack profile was updated after a prescribed
number of cycles, and the SIF was evaluated for the new crack
profile. This integration process was repeated until the SIF
reached the material’s fracture toughness.
Results of WFM simulations of the crack propagation in
the spacer are shown in Figure 4. An initial semi-elliptical crack
located at the bore corner was assumed based on post-mortem
fractographic examination. At approximately the normalized life
of 0.34, the semi-elliptical crack turned into a corner crack,
when one side of the crack tip met the chamfer. After this point,
crack growth entered into a fast growth region.

4 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 Research on the development of a more efficient WFM, The spacer tests were performed in a back-to-back
that would allow for application to larger cracks is underway. configuration, where two identical spacers were installed on the
test rotor simultaneously and tested. This configuration design
0.6 was based on the earlier rig testing performed by the OEM.
Tests were conducted in vacuum at high temperature. Two tests
0.5 designated A and B included a total of four spacers. Test
conditions are given in [14]. The number of cycles was
Normalized crack size

0.4 c - half major axis

recorded throughout the test.
a - half minor axis At the beginning all spacers were non-destructively
0.3
inspected (NDI) to ensure “the crack free” condition. During
both subsequent spin rig evaluations, tests were periodically
0.2
interrupted for NDI of the spacers. Three NDI techniques used
0.1 to inspect spacers were:
B • Visual inspection technique,
0 • Liquid penetrant inspection (LPI) technique, and
0 0.1 0.2 0.3 0.4 0.5 • Eddy current (EC) technique.
Normalized number of cycles

TEST RESULTS
A summary of the spin rig tests is shown in Table 1. In the
Figure 4. CRACK GROWTH IN THE SPACER SIMULATED BY THE first test, Test A, Spacer 1 showed a significant amount of
WEIGHT FUNCTION METHOD. CRACK ORIGINATED FROM A SEMI- cracking after the test was stopped when the rotor vibration
ELIPTICAL SURFACE CRACK AT THE BORE CORNER.
level increased significantly. Inspections of Spacer 2 installed
on the same rotor did not show any indications of cracking.
SPIN RIG TEST Figure 6 shows the largest crack generated at the bore corner of
Spacer 1. Both spacers were destructively examined after the
tests. In the second spin rig test, Test B, Spacer 3 was
intentionally tested to burst to establish the experimental safety
margin for the spacers. Similar to Test A, in Test B, Spacer 4
did not show any visible indications of damage after the burst of
the neighbouring spacer. Remnants of Spacers 1 and 3 were
subjected to metallurgical analysis.

Table 1. SUMMARY OF THE SPIN RIG TESTS FOR TURBINE SPACERS.

Location
Test Spacer Crack nucleation
tested
Bore
Spacer 1 At the bore corner
corner
Test A
Figure 5. SPIN RIG FACILITY USED FOR TESTING [14]. Bore
Spacer 2 None found
corner
After stress and fracture mechanics analyses were Bore At the bore corner;
Spacer 3
performed for the spin rig test condition, the turbine spacers corner Burst
Test B
were tested in the spin rig facility shown in Figure 5. The spin Bore
Spacer 4 None found
rig, with chamber dimensions of 1.2 m in diameter and 1.2 m in corner
depth allows for testing of components at high temperatures. Two marked points sequentially obtained during the NDE
Tested components are driven by the air turbine [14]. of Test B allowed for rudimentary assessment of the crack
The particular variant of the spacer of interest for this growth rate. The burst point is marked in Figure 4.
program, although still in service in many fleets, is no longer Three uncertainty factors have to be taken into account
produced. Therefore, four used spacers with estimated when comparing numerical predictions and the rig test results.
operational histories were used for testing. At the beginning of The first is the uncertainty in the fatigue crack growth rate for
each test, a thermal survey was performed to ensure that the material of the spacers tested. The second is the uncertainty in
heater configuration was such that the temperature distributions the recording of the total number of cycles since new for these
on both spacers were uniform, since as previously mentioned, spacers. As mentioned, the spacers were taken from operation
rig test conditions are isothermal. with the estimated time since new. This time was marked on the

5 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 spacers in hours and the conversion to a corresponding number be either grain boundaries or twin boundaries. The size of these
of cycles was done using the generic exchange rate common for facets is roughly in the order of grain size, and they are not only
the entire fleet of engines. Therefore the actual number of present in the crack nucleation area but also well into the crack
cycles that the spacers experienced before they entered the test growth region. Therefore, the formation of these facets must
possessed a significant uncertainty. The third factor contributing have occurred by another mechanism than fatigue, such as
to the uncertainty of the results is that the numerical predictions dislocation pile-up.
of the crack propagation interval were obtained by calculating
the number of cycles starting from an initial crack size, ai, while
the rig tests were performed on the spacers that did not show
any indication of damage at the beginning of the test. Therefore,
the spin rig results for the burst spacer represent a combined
interval comprising four stages of component damage
progression: crack nucleation, small crack growth, large crack
growth and unstable fracture. These three factors contribute to
difficulties in interpreting the spin rig test results.
Spacer 2 did not show any indication of damage after
completion of Test A, however it was not tested further due to
business arrangements. Spacer 4, which did not show any
indication of damage, has been put back in the spin rig and the
test is ongoing.
(a)

(b)

Figure 6. LARGEST CRACK REVEALED BY LPI IN SPACER 1 [14].

METALLURGICAL ANALYSES
Metallurgical investigation of remnants of two cracked
spacers was performed.
After the first spin rig test, the largest crack in Spacer 1
(Figure 6) was opened for fractographic examination. Figure 7a
shows the microstructure of IN901 material cut from the spacer. (c)
It was found that the fracture surface (Figure 7b) had a region Figure 7. (a) MICROSTRUCTURE OF IN901 MATERIAL; (b) THE
of discoloration, which was indicative of oxidation of the SURFACE OF CRACK #1 IN FIGURE 6; (c) CRACKING APPEARS TO BE
fracture surface during testing. The darkest color should OF MIXED FRACTURE MODE COMPOSED OF FACETED AND STRIATED
correspond to the crack nucleation area, which was near the FATIGUE FRACTURE FEATURES, WITH A, B, C AND D BEING CRACK
chamfer at the bore and which also happened to be the highest NUCLEATION SITES.
stress region as predicted by the finite element analysis,
Figure 1. Deeper into the material, at the flange, the crack The surrounding striation patterns, going outward from
surface became progressively less discolored, which indicates these areas, indicate that all these facets were possible crack
that oxidation occurred during the later stage of crack growth. nucleation sites. They would form first as subsurface cracks and
SEM examination revealed a faceted and striation mixed mode then break into the surface as surface cracks. A semi-elliptical
of fatigue fracture on the crack surface (Figure 7c). Particularly, crack that enclosed two large connected facets C and D was
areas A, B, C, and D had large facets, which were suspected to assumed as the initial nucleation site in the WFM simulations of
crack propagation shown in Figure 4.

6 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 Based on these observations, it can be postulated that the REMARKS ON DAMAGE TOLERANCE APPROACH
crack nucleation in the spacer may occur by dislocation pile-ups Damage tolerance of critical components requires rigorous
at either grain boundaries or triple junctions, or twin analyses and test validation. Analyses show that not every
boundaries, as a result of near surface plastic deformation due component in aging engines is a good candidate for application
to low cycle fatigue loading. The dislocation pile-ups may turn of the damage tolerance lifing methodology. A good candidate
into subsurface wedge cracks, which then propagate along grain for damage tolerance is a component made of material that
or twin boundaries, breaking into the surface and finally exhibits slow fatigue crack propagation rates and is
forming a surface crack, such as shown in Figure 7c. Inside the conservative in geometrical design. Generally, components of
material, the twin boundary cracks may coalesce with the the old engines had only a few clearly identified critical life
propagating fatigue crack, leading to a mixed-mode crack limiting areas and their overall safety was achieved by declaring
propagation. It can be postulated that there could be numerous a safe-life based on crack nucleation. Analyses supported by
short cracks along the bore corner, which is the primary crack operational experience show that they were designed with a
nucleation location. large amount of conservatism. The current trend however is to
To summarize, although some cracks might nucleate design components to minimize engine weight and to fully use
randomly along the highly stressed bore and chamfer, the the material capacity. As a result, the components of the new
presence of subsurface crack formation via grain boundaries or engines are fully stressed and the safe-life at each critical
twin boundaries may be more dangerous and this affects location is fully utilized. Analyses show that crack propagation
damage tolerance of the component. The subsurface cracks intervals (CPI) for these components are short and therefore a
subsequently propagated under fatigue loading, as evidenced by damage tolerance approach based on periodic inspections is not
striations, which would intermittently coalesce with inner economically justified.
fractured facets along the path, resulting in a mixed fracture Our experience in execution of this research showed that
mode. there is a range of technical and managerial difficulties in
The above findings impose two questions or challenges on application of the damage tolerance methodology to extending
lifing of the spacer. lives of legacy engines by users. They can be summarized as
Crack nucleation is probabilistic in nature and mostly follows:
occur subsurface, depending on the microstructure under stress, • Difficulty in obtaining engine design data, operational
i.e., whether it has grain boundaries or twins favorably oriented temperature data and mission profiles.
to allow crack formation at the identified location. Because of • Difficulty in obtaining service exposed components for both
the small crack size during subsurface crack nucleation, this material property testing and validation testing.
poses a challenge to nondestructive inspection and also to the • Material properties, such as fatigue crack growth rate data
initial flaw size assumption for damage tolerance analysis. are costly to generate.
Crack nucleation life prediction models for service-exposed • Gas turbine component lifing methodologies are considered
components should consider both dislocation pile-ups at grain by the OEM as proprietary and usually are not shared with the
boundaries or triple junctions and dislocation pile-ups at twin users.
boundaries, as well as the possible effect of prior service • Costs of such a research in terms of money and time are
exposure. high. A cost-benefit analysis should be performed first.
• As mentioned, the use of crack growth limits, that exceed
SUMMARY AND CONCLUSIONS the safe crack initiation life for spacers of this turbine engine is
The residual life for a spacer from a turbo prop engine was not endorsed by the OEM.
assessed through numerical predictions of crack growth, spin In the research presented here, these difficulties were
rig testing and metallurgical examinations. Numerical overcome through close working with the OEM during the
predictions using both the finite element method and the weight course of the international program, through verification of the
function method for the fatigue crack from the bore corner of material properties data in-house, and also through use of in-
the spacer were performed. Two spin rig tests performed for house developed lifing algorithms and methods.
four spacers in back-to-back configurations showed significant
variability in the component life. In both tests, one component
was severely damaged while the second one showed no ACKNOWLEDGMENTS
indication of damage. The test revealed that there is not much The work presented in this paper was partially funded by
damage tolerance in the spacer and that the crack, once DND-DRDC-AVRS, DND-DAEPM(TH) and DND-DTAES of
nucleated, will propagate rapidly. Metallurgical investigations Canada. Significant contributions of Ms. Sandi Robertson to
revealed that the subsurface cracks may have been major factors spin rig testing are acknowledged. Finite element calculations of
in generation and propagation of engineering cracks in the the crack growth were performed by a co-op student Mr. Alex
spacer. Murzionak who painstakingly mastered intricacies of Zencrack.

7 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 NOMENCLATURE [8] Wu, X.J., Beres, W. and Yandt, S., 2008. “Challenges
CPI Crack Propagation Interval in life prediction of gas turbine critical components,” Canadian
DT Damage Tolerance Aeronautical and Space Journal, 54, pp. 31-39.
FEM Finite Element Method [9] Bird, J., Wu, X., Patnaik, P., Dadouche, A.,
IN901 Incoloy™ 901 Létourneau, S. and Mrad, N., 2008. “A multidisciplinary
LCF Low Cycle Fatigue approach to diagnostics prognosis and health management of
NDE Non-destructive Evaluation military gas turbine engines,” Ensured Military Platform
OEM Original Equipment Manufacturer Availability, NATO-RTO-AVT Symposium AVT-157, Paper No.
SEM Scanning Electron Microscope 27, Montreal, QC, Canada, October 13-17.
SIF Stress Intensity Factor [10] Wu, X.J., 2010. “Life prediction of gas turbine
WFM Weight Function Method materials,” Chapter 9, Gas Turbines, ed. Injeti Gurrappa, Sciyo.
2D Two-dimensional [11] Beres, W., Dudzinski, D., Robertson, S., Prentis, C.,
3D Three-dimensional 2008. “Damage tolerance assessment of ageing gas turbine
engines through analyses and testing,” NATO RTO Symposium
REFERENCES AVT-157 “Ensured Military Platform Availability,” Montreal,
[1] “Recommended practices for monitoring gas turbine Canada, October 13-17, 2008, Paper 19, pp. 19-1–19-22.
engine life consumption,” 2000. NATO RTO Working Group [12] Beres, W., Dudzinski, D., Murzionak, A., 2009.
AVT-017 Report, RTO Technical Report RTO-TR-28. “Fatigue crack growth rate evaluation in a turbine disc after spin
[2] Immarigeon, J-P., Koul, A.K., Beres, W., Au, P., rig testing,” 12th International Conference on Fracture, Ottawa,
Fahr, A., Wallace, W., Patnaik, P. and Thamburaj, R., 2000. ON, July 12-17, 2009, Paper 00587, pp. 1-10.
“The aging of engines: an operator’s perspective,” NATO RTO [13] Beres, W., Kearsey, R.M., Wu, X.J., Yang, Q.,
Lecture Series 218 on “Aging Engines, Avionics, Subsystems Dudzinski, D., Tsang, J., 2012. “Research on advanced high
and Helicopter,” RTO-EN-14, pp. 2.1–2.20. temperature materials and protective coatings for new and
[3] Beres, W. and Koul, A.K., 2002. "Stress intensity legacy gas turbine engines for military platforms,” Design,
factor calculations for cracks emanating from bolt holes in a jet Modelling, Lifing and Validation of Advanced Materials in
engine compressor disc," 23rd International Congress of Extreme Military Environments, NATO STO Applied Vehicle
Aeronautical Sciences, Toronto, Canada, September 8-13, 2002, Technology Panel (AVT) Symposium held in Biarritz, France
Paper 0433, pp.1-8. 15-18 October 2012, MP-AVT-187, Paper 6, pp 6.1–6.28.
[4] Immarigeon, J-P., Beres, W., Au, P., Fahr, A., [14] Beres, W., Fread, D., Harris, L., Haupt, P., Kappas J.,
Wallace, W., Koul, A.K., P. Patnaik and Thamburaj, R., 2003. Olson R., Reineke, P., Robertson, S., Stocks G., 2008. “Critical
“Life cycle management strategies for aging engines,” NATO components life update for gas turbine engines—case study of
RTO Workshop on Life Management Techniques for Aging an international collaboration,” ASME Turbo Expo 2008:
Engines, Manchester, UK, October 2001. “Aging Mechanisms Power for Land, Sea and Air, June 9-13, 2008, Berlin,
and Control, Specialists’ Meeting on Life Management Germany, GT2008-50655.
Techniques for Ageing Air Vehicles,” RTO-MP-079(II), pp. [15] Hou, J., Dubke, J., Barlow, K., Slater, S., Harris, L.,
17.1–17.16. Calcuttawala, S. and Beres, W., 2008. “3D Crack growth
[5] Beres, W. and Koul, A.K., 2003. “Damage tolerance analysis and its correlation with experiments for critical turbine
based life assessment for aging Nene 10 turbine discs,” NATO components under an international collaborative program,”
RTO Workshop on Life Management Techniques for Aging ASME Turbo Expo 2008: Power for Land, Sea and Air, June 9-
Engines, Manchester, UK, October 2001; “Aging Mechanisms 13, 2008, Berlin, Germany, GT2008-50548.
and Control, Symposium B: Monitoring and Management of [16] Dowling, N.E., 1993. “Mechanical Behaviour of
Gas Turbine Fleets for Extended Life and Reduced Costs,” Materials,” Prentice-Hall.
RTO-MP-079(I), pp. 14.1–14.16. [17] “User Manual for ZENCRACK 7.5,” 2007. Zentech
[6] Wu, X., 2007. “Evolution of life in metallic materials International Ltd.
– a Hollistic approach to material life prediction,” NRC, [18] ABAQUS Ver. 6.8, 2008. Dassault Systèmes.
Institute for Aerospace Research, Structures and Materials
Performance Laboratory Report No. AN-SMPL-2007-0070, SELECTED BIBLIOGRAPHY
Ottawa. [19] Annis, C., 2004. “Probabilistic life prediction isn’t as
[7] Bird, J., Wu, X., Patnaik, P., Dadouche, A., easy as it looks,” Journal of ASTM International, 1(2), pp. 3-4.
Létourneau, S. and Mrad, N., 2007. “A framework of prognosis [20] Asquith, G., Pickard, A.C. “Fatigue testing of gas
and health management—a multidisciplinary approach,” ASME turbine components,” in Full scale fatigue testing of
Turbo Expo 2007: Power for Land, Sea and Air GT2007- components and structures,” ed. by K.J. Marsh, Butterworths,
27953, May 14-17, 2007, Montreal, QC, Canada. pp. 210-234, 1988.

8 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo

 [21] Blüml, P., and Broede, J., 1999. “Critical Parts’ Life [37] Kappas, J., 2002. “Review of risk and reliability
Extension Based on Fracture Mechanics,” Qualification of Life methods for aircraft gas turbine engines,” DSTO Australia,
Extension Schemes for Engine Components, RTO-AVT MP-17, Report DSTO-TR-1306.
pp. 19-1–19-7. [38] Leverant, G.R, et al., 2004. “A new tool for design
[22] Boyd-Lee, A.D. and Harrison, G.F., 1999. “The and certification of aircraft turbine rotors,” ASME, Journal of
development of life extension methods for fracture critical-aero- Engineering for Gas Turbine and Powers, 126, pp. 155-159.
engine components,” Qualification of Life Extension Schemes [39] McClung, R.C., et al., 2004. “A software framework
for Engine Components, RTO-MP-17, pp.18-1–18-10. for probabilistic fatigue life assessment of gas turbine engine
[23] Boyd-Lee, A.D., Harrison, G.F., and Henderson, rotors,” Journal of ASTM International, 1, pp. 199-215.
M.B. 2001. “evaluation of standard life assessment procedures [40] McClung, R.C., et al., 2011. “Integration of NASA-
and life extension methodologies for fracture-critical developed lifing technology for PM alloys into DARWIN®,”
components,” Int. Journal of Fatigue, 23, pp. S11-S19. NASA/CR-2011-216977.
[24] Carter, J.A. et al., 2013. “Probabilistic risk [41] Wicks, B.J., et al,. 2001. “The Inadequacy of Safe-
assessment of a turbine disk,” Fatigue of Materials II Advances Life Prediction: Aero-Engine Fan and Compressor Disk
and Emergences in understanding, TMS, pp.71-85. Cracking,” Ageing Mechanisms and Control, Part B –
[25] Christodoulou, L, and Larsen, J.M. “Materials Monitoring and Management of Gas Turbine Fleets for
Damage Prognosis: A Revolution in Assets Management,” Extended Life and Reduced Costs, NATO RTO MP 079(I), pp.
Materials Damage Prognosis, Ed by J.M. Larsen, L. 22-3–22-9.
Christodoulou, et al., TMS, 2005, pp. 3-10. [42] Waldhart, C., Enright, M.P., and Fitch, S.H.K., 2004.
[26] Cruse, T.A., 2001. “Non-deterministic, non- “Methodology for application of 3-D stresses to surface
Traditional methods (NDNTM),” NASA/CR—2001-210976. damage-based risk assessment,” AIAA Paper 2004-1599.
[27] “Engine Structural Integrity Program (ENSIP),” [43] Zhuang, W.Z., 2000. “Prediction of crack growth
2002. MIL-HDBK-1783B, U.S. Department of Defense. from bolt holes in a disc,” Int. Journal of Fatigue, 22, pp. 241-
[28] Enright, M.P., et al., 2003. “Probabilistic surface 250.
damage tolerance assessment of aircraft turbine rotors,” ASME
Turbo Expo 2003, Paper GT2003-38731.
[29] Enright, M.P., McClung, R.C. and Hudak, S.J., 2006.
“Nonparametric risk assessment of gas turbine engines,” AIAA,
Paper 2006-1725.
[30] Enright, M.P. and McClung, R.C., 2007.
“Probabilistic mission identification of aircraft engine usage
using nonparametric density estimation techniques,” ASME
Turbo Expo 2007, paper GT2007-27176.
[31] Enright, M.P. and McClung, R.C., 2010. “A
probabilistic framework for gas turbine engine materials with
multiple types of anomalies,” ASME Turbo Expo 2010, Paper
GT2010-23618.
[32] Evans W.J., Screech J.E., Williams S.J., 2008.
“Thermo-mechanical fatigue and fracture of INCO718,” Int.
Journal of Fatigue, 30, pp. 257-267.
[33] Federal Aviation Administration, 2001. “Damage
tolerance for high energy turbine engine rotors,” Advisory
Circular, AC.33.14-1.
[34] Gorelik, M. et al., 2005. “Development of
probabilistic lifing system for advanced turbine rotor alloys,”
ASME Turbo Expo 2005, Paper GT2005-68770.
[35] Harrison, G.F. and Evans, W.J., 2000. “A Review of
Fatigue Life Assessment Procedures for Aeroengine Fracture
Critical Components,” Fatigue 2000, pp. 15-30.
[36] Jameel, A., 2005. “Surface damage tolerance analysis
of a gas turbine engine rotor,” ASME Turbo Expo 2005, Paper
GT2005-68760.

9 Copyright © 2014 by Her Majesty the Queen in Right of Canada

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/80944/ on 02/24/2017 Terms of Use: http://www.asme.org/abo