94ÖÖÖ0Z2QI3

DISTRIBUTION STATEMENT A
Approved for Public Release
Distribution Unlimited
A Service of:
National Aeronautics and
Space Administration

DEC QuALIJT IlfEPBCflD 4

20000821 018 SCIENTIFIC &
TECHNICAL INFORMATION
 RTO-TR-28
AC/323(AVT)TP/22
NORTH ATLANTIC TREATY ORGANIZATION
00
CM
■
a,
6

RESEARCH AND TECHNOLOGY ORGANIZATION

BP 25, 7 RUE ANCELLE, F-92201 NEUILLY-SUR-SEINE CEDEX, FRANCE

© RTO/NATO 2000
Single copies of this publication or of a part of it may be made for individual use only. The approval
of the RTA information Policy Executive is required for more than one copy to be made or an extract
included in another publication. Requests to do so should be sent to the address above.

RTO TECHNICAL REPORT 28

Recommended Practices for Monitoring

Gas Turbine Engine Life Consumption
(Pratiques recommandees pour le controle du vieillissement
des turbomoteurs)

Report of the Applied Vehicle Technology Panel Task Group AW-017.

Published April 2000

 The Research and Technology
Organization (RTO) of NATO
RTO is the single focus in NATO for Defence Research and Technology activities. Its mission is to conduct and promote
cooperative research and information exchange. The objective is to support the development and effective use of national
defence research and technology and to meet the military needs of the Alliance, to maintain a technological lead, and to
provide advice to NATO and national decision makers. The RTO performs its mission with the support of an extensive
network of national experts. It also ensures effective coordination with other NATO bodies involved in R&T activities.
RTO reports both to the Military Committee of NATO and to the Conference of National Armament Directors. It comprises a
Research and Technology Board (RTB) as the highest level of national representation and the Research and Technology
Agency (RTA), a dedicated staff with its headquarters in Neuilly, near Paris, France. In order to facilitate contacts with the
military users and other NATO activities, a small part of the RTA staff is located in NATO Headquarters in Brussels. The
Brussels staff also coordinates RTO's cooperation with nations in Middle and Eastern Europe, to which RTO attaches
particular importance especially as working together in the field of research is one of the more promising areas of initial
cooperation.
The total spectrum of R&T activities is covered by 7 Panels, dealing with:
• SAS Studies, Analysis and Simulation
• SCI Systems Concepts and Integration
• SET Sensors and Electronics Technology
• 1ST Information Systems Technology
• AVT Applied Vehicle Technology
• HFM Human Factors and Medicine
• MSG Modelling and Simulation
These Panels are made up of national representatives as well as generally recognised 'world class' scientists. The Panels also
provide a communication link to military users and other NATO bodies. RTO's scientific and technological work is carried
out by Technical Teams, created for specific activities and with a specific duration. Such Technical Teams can organise
workshops, symposia, field trials, lecture series and training courses. An important function of these Technical Teams is to
ensure the continuity of the expert networks.
RTO builds upon earlier cooperation in defence research and technology as set-up under the Advisory Group for Aerospace
Research and Development (AGARD) and the Defence Research Group (DRG). AGARD and the DRG share common roots
in that they were both established at the initiative of Dr Theodore von Kärmän, a leading aerospace scientist, who early on
recognised the importance of scientific support for the Allied Armed Forces. RTO is capitalising on these common roots in
order to provide the Alliance and the NATO nations with a strong scientific and technological basis that will guarantee a
solid base for the future.

The content of this publication has been reproduced

directly from material supplied by RTO or the authors.

Published April 2000

Copyright © RTO/NATO 2000

All Rights Reserved

ISBN 92-837-1032-0

Printed by Canada Communication Group Inc.

(A St. Joseph Corporation Company)
45 Sacre-Coeur Blvd., Hull (Quebec), Canada K1A 0S7
 Recommended Practices for Monitoring Gas Turbine
Engine Life Consumption
(RTO TR-28)

Executive Summary
The main reason for usage monitoring is to ensure that gas-turbine engine components do not fail in
service. Early maintenance policies were based on a "hard-time" philosophy, measured in hours of use.
More recently, technical advances have permitted a trend towards a condition-based philosophy. Life
monitoring systems have evolved to automate measurement of the life used in modern engines. In
many cases, the new assessment and monitoring methods have been applied to ageing fleets with great
success. This document describes the design and operational factors that should be considered prior to
the application of these techniques to an aged engine fleet.
Turbine disks are the most safety critical parts, due to the amount of energy released should they fail.
Moreover, future improvements in engine performance directly depend on increases in component
stress levels. These demands for better performance must be met within the regulatory requirements
for safety standards. This situation presents one of the most challenging areas of gas turbine design.
Satisfying these conflicting demands carries a cost that reaches far beyond the development and
purchase costs of a particular engine design. Although turbine disks are particularly expensive to make,
over 90% of them are thrown away when less than 50% of their life has been consumed. Physical use,
component life and safety have to be managed on a statistical basis because of variations in material
properties.
Because of the overwhelming prevalence of mature fleets, consideration was restricted to aircraft that
entered service prior to 1990. This keeps it in the realm of feasible and relevant solutions for operators
with ageing fleets. The document remains relevant for new projects begun after 1990. The aim has
been:
• To cover technical and managerial issues in a manner that allows readers to concentrate on those
parts that interest them. This has necessitated some repetition between chapters;
• To show the parallel regulatory developments in the USA and Europe;
• To make clear the difference between 'safe life' and 'damage tolerance' design philosophies;
• To show the impact of differing approaches to maintenance practice;
• To show the results obtained from existing monitoring systems;
• To give practical help and guidance to operators, with mature fleets.
Within the forum, the leading design and life management practitioners from the world's major engine
design companies worked with representatives from most of the NATO aircraft operators.
 Pratiques recommandees pour le contröle du
vieillissement des turbomoteurs
(RTO TR-28)

Synthese
L'objectif principal du controle du vieillissement est d'eviter la defaillance en vol des composants des
turbopropulseurs. Les premieres politiques de maintenance suivaient une philosophic de « cycles »,
mesures en heures d'exploitation. Plus recemment, les avancees technologiques realisees ont permis
l'adoption d'une politique basee sur les conditions d'utilisation. L'evolution des systemes de controle
du vieillissement a ete teile qu'ils sont desormais en mesure de calculer automatiquement le nombre
d'heures de fonctionnement restant pour les moteurs modernes. Dans de nombreux cas, ces nouvelles
methodes devaluation et de controle du vieillissement ont ete appliquees avec beaucoup de succes ä
des flottes aeriennes vieillissantes. Ce document decrit les criteres de conception et les facteurs
operationnels dont il faut tenir compte avant d'appliquer ces techniques ä de telles flottes d'aeronefs.
Les disques de turbine sont les elements les plus critiques du point de vue de la securite, en raison de
l'energie qui est liberee en cas de rupture. Ainsi, 1'amelioration des performances des moteurs ä
l'avenir est directement liee au degre de resistance aux contraintes de leurs organes. II faudra de plus
repondre ä ces sollicitations d'amelioration de performances en respectant les exigences reglementaires
en matiere de securite. Cette situation represente Tun des plus grands defis dans le domaine de la
conception des turbopropulseurs.
La satisfaction de ces exigences contradictoires comporte un coüt qui depasse largement les coüts de
developpement et d'achat d'un moteur particulier. Bien que les disques de turbine soient
particulierement coüteux ä fabriquer, plus de 90% de ces elements sont mis au rebut ä moins de 50%
de leur vie utile. L'utilisation physique, le cycle de vie des organes moteur et la securite doivent Stre
geres statistiquement en raison des variations des caracteristiques physiques des materiaux.
Etant donnee la predominance des flottes d'aeronefs en service depuis de nombreuses annees, il a ete
decide de ne considerer que les aeronefs etant entres en service avant 1990. Le domaine se trouve ainsi
reduit ä des solutions pratiques et appropries pour des exploitants responsables de flottes vieillissantes.
Le document demeure toutefois applicable aux nouveaux projets datant de moins de dix ans. L'objectif
a ete de:
• couvrir les questions techniques et administratives d'une maniere qui permet au lecteur de porter son
attention uniquement sur les elements qui l'interesse. II s'en est suivi un certain nombre de
repetitions entre les differents chapitres;
• Montrer les developpements reglementaires paralleles aux Etats-Unis et en Europe;
• Faire la distinction entre les philosophies de conception dite « ä vie sure », et « ä tolerance ä
l'endommagement»;
• Illustrer 1'impact des differentes approches en matiere de pratique de la maintenance;
• Presenter les resultats obtenus par differents systemes de controle du vieillissement;
• Fournir des conseils pratiques et des directives pour les exploitants de flottes vieillissantes.
Au sein de ce groupe, les principaux concepteurs et gestionnaires de cycle de vie des principaux
motoristes dans le monde ont pu travailler avec des representants de la majorite des exploitants
d'aeronefs de l'OTAN.
 Contents
Page

Executive Summary üi
Synthese iv
Members of the AVT-017 Task Group vi
Contributors *"•
Publications of the RTO Applied Vehicle Technology Panel ix

Chapter 1 - Introduction to Engine Usage Monitoring 1-1

An overview of the reasons for usage monitoring is given. The history of usage monitoring systems is
described. 'Safe life' and 'damage tolerance' are briefly described. The practical value of usage
monitors for mature fleets is illustrated. An example of the cost savings for a mature fleet is provided.
2
Chapter 2 - Civil and Military Practices "1
The history of regulatory controls for engine development and engine qualification testing is
described. International requirements are overviewed. The contemporary technology and its enabling
effects are examined, and examples of real engines and their systems are given.
Chapter 3 - Maintenance Policies and Procedures 3-1
A maintenance-based view of usage monitoring is described. Differences between military and civil
practices are examined, and conclusions drawn. The physical basis for inspection and reuse are
examined, and a detailed description of ENSIP practices is given. Strategies for retirement and
maintenance are described.
Chapter 4 - Modes of Gas Turbine Component Life Consumption 4-1
Physical failure mechanisms are described in an operational context. Photographic examples are
provided.
Chapter 5 - Mechanics of Materials Failure 5-1
Detailed analytical models for creep and fatigue are introduced. Extensive references are provided.
Chapter 6 - Translation of Service Usage into Component Life Consumption 6-1
The techniques used to determine component usage, with the safety levels required by the regulatory
authorities are described. The safe life and damage tolerance methods are discussed in detail. ENSIP
and database methods are compared. A full methodology for usage modelling is described.
References provide supporting materials data.
Chapter 7 - Lifing Procedures and Monitoring System Verification and Validation 7-1
The detailed procedures for ensuring that complete life-monitoring systems function correctly are
described. The need to ensure that reduced order algorithms, used in operational systems, accurately
reflect the full design model is emphasised.
Chapter 8 - Usage Survey and Mission Analysis 8-1
A wide-ranging view of operational management considerations is provided. A number of different
life-tracking methods are described in detail, and the advantages and disadvantages discussed.
Chapter 9 - Usage Data from Operational Monitoring Systems 9-1
How to safely determine the number of engines to monitor in a fleet is described. Usage monitoring
results from different engine types are shown, and conclusions are drawn. The benefits of individual
engine monitoring, rather than fleet sampling are quantified.
Chapter 10 - Conclusions and Recommendations 10-1
Overall conclusions and highlights from all chapters are drawn together.
Appendix 1 - US Military Engine Tracking and Operational Usage Methods Al-1
A formal description of Total Accumulated Cycles (TACs) is provided.
Appendix 2 - Maintenance Policies and Procedures A2-1
An overview of current military-engine lifing-practice with in-service aircraft is given. Eleven NATO
nations are included.
Appendix 3 - Mechanics of Materials Failure A3-1
This appendix contains a further development of the physical materials models described in chapter 5.
Glossary G
 Members of the AVT-017 Task Group
Chairman Technical Author and Editor
Mr. G. Lazalier Mr. M. Sapsard
Sverdrup Technology, Inc. Implement Ltd
872 avenue E 9 Green Dell Way
Arnold AFB, TN 37389-5051, USA Hemel Hempstead
Herts HP3 8PX, UK

CANADA Mr. G. Panagakis

Dr. W. Beres Hellenic Aerospace Industry
National Research Council of Canada P.O. Box 23
1500 Montreal Road, Bldg. M-7 Tanagra
Ottawa K1A 0R6 32009 Schimatari
Mr. D. Rudnitski ITAT V
National Research Council of Canada Ing. E. Campo
1500 Montreal Road, Bldg. M-7 FIATAVIO s.p.a.
Ottawa K1A 0R6 corso Ferrucci, 112
10138 Torino
FRANCE
Mr. C. Colette Mr. C. Vinci
Turbomeca FIATAVIO s.p.a.
64411 Bordes via Nizza 312
10127 Torino
Mr. O. Etchevers
SPAe, 26, boulevard Victor NETHERLANDS
00460 Armees Mr. L. ten Have
Mr. P.R. Mosser National Aerospace Laboratory NLR
SNECMA, Direction Technique P.O. Box 153
Aerodrome de Villaroche 8300 AD Emmeloord
77550 Moissy Cramayel
PORTUGAL
GERMANY Lt. Col. M. G. Chambel
Dr. J. Broede CLAFA/DMA
MTU München GmbH avenida de Forca Aerea
Dept. EMB, Postfach 50 06 40 2720 Alfragide
D-80976 München
TURKEY
Prof. K. Boichhausen
Major C. Erel
MTU München GmbH
lnic Hava Danal Bakim Merkez
Hauptabteilungsleiter Komutanligi, 26030 Eskisehir
Stromungsmachinenberechnung-EW
Postfach 50 06 40 Major E. Gunes
80976 München lnic Hava Ikmal Bakim Merkez
Komutanligi, 26030 Eskisehir
GREECE
Capt. D. Adamopoulos UNITED KINGDOM
Hellenic Air Force Research Center Sqn. Ldr. C. Eady
Terpsithea Post Office Logs (OR) le. F10 Royal Air Force
16501 Glyfada RAF Brampton, Huntingdon
Cambs PE18 8QL
Prof. P. Kotsiopoulos
Hellenic Air Force Academy Mr. R.V. Cottington
c/o Nazliou 35-39+ Byzadiou 70 Propulsion Department
17122 - N. Smyrni, Athens Building 304
DTEO Pyestock
Hants, GU14 0LS
Mr. P. Everitt Mr. O. Davenport
Rolls-Royce plc ASC/LP
P.O. Box 3 2145 Monahan Way
Filton, Bristol BS34 7QE Wright Patterson AFB
OH 45433-7017
Prof. G. Harrison
DERA Mr. R. Holmes
Griffith Building (AT) United Technologies - Pratt and Whitney
MSS, Farnborough, Hants GU14 OLX P.O. Box 109600
West Palm Beach, FL 33410-9600
Group Capt. J.K. Newton
FS(Air) 4, MOD PE (UK) Mr. P. Maletta
Maple Oa #45 Abbey Wood, Bristol BS34 8JH GE Airdraft Engines
Applied Mechanical technology
Mr. J. Nurse 1 Neuman Way, Mail Stop Q105
FS(Air) MOD PE (UK) Evandale, OH 45215 6301
Maple Oa #45 Abbey Wood, Bristol BS34 8JH
Mr. C. Meece
Mr. F. Tufnell Turbine Engine Component Center
FS(Air) MOD PE (UK) Pratt and Whitney Aircraft
Maple Oa #45 Abbey Wood, Bristol BS34 8JH P.O.Box 109600, M/S/711-31
West Palm Beach, FL 33410-9600
UNITED STATES
Mr. A. Cifone Mr. P. Zimmerman
Propulsion and Power Engineering Naval Air Warfare Center, MS3
Bldg 106, NAWCAD Aircraft Div. 22119 James Road
22195 Elmer Road, Unit 4.4c Patuxent River. MD 20670 - 5304
Patuxent River, MD 20670-1534
 Contributors
Chairman: Glen Lazalier, Sverdrup Technology, AAFB, USA

Technical author and editor: Michael Sapsard, Implement Ltd, UK

Lead authors:
Chapter 1 Michael Sapsard, Implement Ltd, UK
Chapter 2 Richard Holmes, Pratt & Whitney, USA
Chapter 3 Otha Davenport, ASC/LP, WPAFB, USA
Chapter 4 Christopher Eady, RAF, who succeeded Owen Barnes, RAF, UK
Chapter 5 Wieslaw Beres, NRC, CA
Chapter 6 George Harrison, DERA, UK
Chapter 7 Pierre-Ettienne Mosser, SNECMA, FR
Peter Everitt, Rolls-Royce, UK
Chapter 8: Jürgen Broede, MTU, GE
Chapter 9: Michael Sapsard, Implement, Ltd, UK
Appendix 1 Paul Maletta, GE Aircraft Engines, USA
Appendix 2 Otha Davenport, ASC/LP, WPAFB, USA
Appendix 3 Wieslaw Beres, NRC, CA

Contributors:

Experts:
Lex ten Have: National Aerospace Laboratory, NL
Paul Zimmerman: Naval Air Systems Command, USA
Carl Meece: Pratt & Whitney, USA
Johnny Adamson: Pratt & Whitney, USA
Dimitris Adamopoulos: Hellenic Airforce Research Centre, GR
Fred Tufnell: FS(Air), MOD(PE), UK
Jim Nurse: FS(Air), MOD(PE), UK
John Newton: FS(Air), MOD(PE), UK
Christophe Colette: Turbomeca, FR
IPA Etchevers: SPAe, FR

Readers:
William Wallace: NRC, Canada to whom a special debt of gratitude is due.
Bernie J Rezy: Allison USA
Kalyan Harpalani Pratt & Whitney, Canada

AVT/Former PEP Panel members:

C Erel: TU
A CiFone: USA
C Vinci: IT
A Kottarakos: GR
R Cottington: UK
D Rudnitski: CA
 Publications of the RTO
Applied Vehicle Technology Panel

MEETING PROCEEDINGS (MP)

Design for Low Cost Operation and Support
MP-37, Spring 2000
Structural Aspects of Flexible Aircraft Control
MP-36, Spring 2000
Aerodynamic Design and Optimization of Flight Vehicles in a Concurrent Multi-Disciplinary Environment
MP-35" Spring 2000
Gas Turbine Operation and Technology for Land, Sea and Air Propulsion and Power Systems (Unclassified)
MP-34, Spring 2000
New Metallic Materials for the Structure of Aging Aircraft
MP-25, April 2000
Small Rocket Motors and Gas Generators for Land, Sea and Air Launched Weapons Systems
MP-23, April 2000
Application of Damage Tolerance Principles for Improved Airworthiness of Rotorcraft
MP-24, Januar}' 2000
Gas Turbine Engine Combustion, Emissions and Alternative Fuels
MP-14, June 1999
Fatigue in the Presence of Corrosion
MP-18, March 1999
Qualification of Life Extension Schemes for Engine Components
MP-17, March 1999
Fluid Dynamics Problems of Vehicles Operation Near or in the Air-Sea Interface
MP-15. February 1999
Design Principles and Methods for Aircraft Gas Turbine Engines
MP-8. February 1999
Airframe Inspection Reliability under Field/Depot Conditions
MP-10, November 1998
Intelligent Processing of High Performance Materials
MP-9, November 1998
Exploitation of Structural Loads/Health Data for Reduced Cycle Costs
MP-7, November 1998
Missile Aerodynamics
MP-5, November 1998

EDUCATIONAL NOTES
Measurement Techniques for High Enthalpy and Plasma Flows
EN-8, April 2000
Development and Operation of UAVs for Military and Civil Applications
EN-9, April 2000
Planar Optical Measurements Methods for Gas Turbine Engine Life
EN-6, September 1999
High Order Methods for Computational Physics (published jointly with Springer-Verlag, Germany)
EN-5, March 1999
Fluid Dynamics Research on Supersonic Aircraft
EN-4, November 1998
Integrated Multidisciplinary Design of High Pressure Multistage Compressor Systems
EN-1, September 1998
TECHNICAL REPORTS
Recommended Practices for Monitoring Gas Turbine Engine Life Consumption
TR-28, April 2000
Verification and Validation Data for Computational Unsteady Aerodynamics
TR-26, Spring 2000
A Feasibility Study of Collaborative Multi-facility Windtunnel Testing for CFD Validation
TR-27, December 1999
 1-1

Chapter 1
Introduction to Engine Usage Monitoring
by
(M. Sap sard)

Page

1. Introduction 1-3

2. Lifing Philosophies '-3

2.1. 'Safe Life'Design 1-3
2.2. 'Damage Tolerance' Design 1-3

3. Scope I"3
3.1. Classification of Engines 1-4
3.1.1. First Generation l"4
3.1.2. Second Generation 1-4
3.1.3. Third Generation 1-4
3.1.4. Fourth Generation 1-4
3.2. Scope of Effort l"4
3.2.1. Engines in Use 1-4
3.2.2. Technology '"4

4. Maximising the Benefits of Usage Monitoring 1-5

5. Cost savings '""

6. Summary 1~°

7. References 1 -6
1-2
 1-3

that they are tolerant of cracks. However, components

1. INTRODUCTION must still be retired before they become dangerous. This
Engine usage monitoring systems exist for the following approach carries a weight penalty and the additional costs
reasons: of a highly structured and organised inspection system.
This is known as the 'damage tolerance' or '2/3
• To allow aircraft to be operated safely; dysfunction' approach. This lifing method demonstrates
• To ensure that as much life as possible is extracted that the component is capable of continued safe operation
from life limited parts, hence reducing costs; during crack growth. This is if the cracks grow
• To allow performance levels to be pushed to higher sufficiently slowly during service to allow their growth to
levels with lighter, smaller components; be reliably detected, and perhaps even monitored, through
• To provide information for next generation engine regularly scheduled inspections.
design.
Both methods aim to provide similar safety factors, and
frequently use very similar detail design techniques. The
This is because, if the right counter-measures and safety
cost effectiveness of each approach depends on the
precautions are not in place, the highly stressed
industrial and operating environment in which the
components in modern engines can fail in service -
equipment is designed, built, operated and maintained.
catastrophically, and without warning. These preventive
processes begin during design and continue throughout Associated terms, which may cause confusion, are:
the life of an engine. The failure mechanism of most
concern, and discussed in detail, is low cycle fatigue. • Fracture mechanics - the science of predicting crack
growth;
We aim to show what is being done and describe in detail • ENSIP - the engine structural integrity program,
the fundamentals of engine monitoring systems. After which is a US attempt to close the design loop
reading this document, managers of older aircraft and through extensive component testing to the point of
specifiers of new aircraft should be in a good position to failure.
ask the right questions and make the right decisions about
the systems they have, should have, or will have. ENSIP embraces the damage tolerance approach. Only by
testing until components fail is it possible to be certain of
2. LIFING PHILOSOPHIES their characteristics and failure modes. If the tests were
There are two main approaches to the design of life only conducted on laboratory models, the results would
critical parts in gas turbines. In all components, be more dependent on the assumptions made. Because
metallurgical defects may become the initiation point for components are regularly inspected for cracks, and the
a crack. These may in turn lead to component failure. The crack growth may be monitored, the loop is closed
first approach aims to retire components before the crack throughout the life of the fleet.
is detectable. The second approach tolerates a growing Damage tolerant design does not necessarily mean that an
crack and aims to remove the component when the crack ENSIP type of program must be followed. In the UK, a
is detectable. In theory, crack growth may be monitored. database approach to materials characterisation has been
In general practice, safety-critical components are used successfully for components that have been
removed as soon as a crack is detected. designed to be damage tolerant.
2.1. 'SAFE LIFE'DESIGN
3. SCOPE
The first lifing approach is to assume that when a crack
appears, the component has failed, and to ensure that all Whichever lifing policy is used it is necessary to know
components are retired before the first crack appears. This how engines are used in service, in order to calculate how
clearly carries a penalty in that many components will be much life has been consumed. This can be done either by
retired prematurely. This approach is known as 'Safe monitoring the usage of a representative sample of the
Life' or 'Life to First Crack'. The lifing method fleet and hence calculating the fleetwide usage or by
employed for these parts follows a "cycles-to-crack monitoring engines on an individual basis and treating
initiation" criterion where a minimum life capability is them accordingly. The latter approach requires a higher
defined statistically for simulated service conditions initial investment, but allows a higher proportion of life to
through extensive coupon testing and component test be consumed safely. This is largely because there is no
verification in a spin pit test rig. The statistical minimum need to allow for the statistical variation between the
is usually based on the probability that no more than, say, sample and the fleet. This aspect is described in detail in
1 in 1000 components will have developed a detectable chapter 9.
crack (typically chosen as a crack of 0.8 mm in length or Monitoring systems depend on the design data and rules
0.38mm in depth). It should be mentioned that the value of the systems they monitor. This knowledge was used by
'1 in 1000' is chosen here arbitrarily. Different the working group to categorise engines into four
manufacturers use different values that depend on the risk generations. All generations are in use and will continue
that they are prepared to take. For example, 1 in 980, or 1 to be in use for many years. It should be noted that many
in 750 components. real life systems inevitably straddle the generations that
are loosely defined below, but this did not reduce the
2.2. 'DAMAGE TOLERANCE' DESIGN usefulness of the categories as working group guidelines.
The second lifing approach is to design components so
1-4

This categorisation had a major influence on the scope length or depth;

and arrangement of the subsequent work. It is included • Every critical feature is individually tracked;
here because it provides historical context for the engine • Lifing is by life units that are not hours;
lifing methods that are in use today, and a justification for • The monitoring system is integrated into the engine
the self imposed constraints that the group agreed. design and development process.
The final scope of the work included lifing and usage
monitoring of gas turbine engines that were designed up 3.2. SCOPE OF EFFORT
to the middle or end of the 1980s. How this was decided To see whether there were any underlying systematic
is indicated below. characteristics, the engines known to those present were
considered in two ways. Firstly by 'start of design' date,
3.1. CLASSIFICATION OF ENGINES and secondly by the technology used to build the engine.
3.1.1. FIRST GENERATION The results of both are shown below.
Simple materials data is used; 3.2.1. ENGINES IN USE
Simple design rules are used - disks and blades only; An examination of engines introduced into service since
Components are lifed to first crack using primitive the 1950s produced figure 1. The dates shown are the
crack detection techniques (hot oil and chalk); estimated 'start of design' dates which was taken to be a
Simple mission profiles are used for design purposes; better indicator of design philosophy than 'entry into
Broad usage assumptions are made in terms of so service'. Because of the commercial competition and
many cycles per hour; proprietary information surrounding the 1990s engines,
Life is measured in operating or flying hours; these were excluded from further consideration. All of the
Fleetwide life is extended by inspection of lead engines are believed to be still in sendee. This suggests
engines. that some engines are likely to achieve much longer
3.1.2. SECOND GENERATION service lives than were ever imagined when they were
conceived. Indeed some have had the life of their critical
Design data and rules remain unchanged; components re-estimated using modern design tools.
Spin pit testing is introduced;
Tape based recorder systems are used; 3.2.2. TECHNOLOGY
Most of the available aircraft and engine data is Consideration of the control technology used and the
recorded; differing maintenance and economic drivers led to the
Only a small part of the fleet is monitored; generation of figure 2. The four generations were
Ground based analysis is done
at the manufacturer's plant; 1950s 1960s 1970s 1980s 1990s
• Only small sample data is
available to help assess hourly J57 TF41 F100 Fl 10 EJ200
usage rates. J75 J85 F101 F100 F414
TF33 TF30 F108 F200 M88
3.1.3. THIRD GENERATION Dart TF34 CFM56 F220
Monitoring equipment is fitted Conway TF39 T700 F118
to all aircraft/engines in the T56 Spey Makila AE2100
fleet; J69 Adour TFE125 RTM322
Lifing by life units is converted Tyne Pegasus RB199 F117
to hours to conform to the J79 Gem F404 F119
airworthiness documents; T58 Turmo M53 F120
Data is analysed on-board; Astazou RD33 MTR390
Flight by flight results are ALF502 PT6
downloaded; Allison 250
Fleet usage management is by T64
component rather than engine; ATAR
Tracking of individual parts is Larzac
normal; Gnome
Data is used to feed into new Figure 1 - Engine generations by design year
engine specifications; typically identified and decisions made as to which
Some fracture mechanics methods may be used. generations should fall within the scope of the report.
3.1.4. FOURTH GENERATION Engines that started design between approximately 1950
Design is fully dependent on the materials database; and 1989 were considered Tn-scope'. This encompassed
Fracture mechanics based design techniques are all of the old, mid-life and new engines operated outside
extensively used; of their home territories, but covers most practices up to
Comprehensive time domain stress models are the current time.
developed;
Life is to 2/3 dysfunction or limiting acceptable crack
 1-5

4. MAXIMISING THE BENEFITS OF USAGE individual component monitoring on a fleetwide basis can
MONITORING be achieved.
Figure 6 shows information about engine number 4029,
To know the life remaining for each component is clearly
the first step towards
1950s 1960s 1970s 1980s 1990s
being able to run each
component in service for Control Mech Hydro/ Mech Electro/ Mech Elec/ Mech/ Fadec
Fadec
as long as possible. A
Monitor Manual Manual EHM EHM EHM
manual log might be EHM
sufficient for the Maint. Hard Life Hard Life Hard Life/ FMRFC FMRFC
operator to be in full Policy RFC Hard Life
control. However, in a Drivers Safety Safety Safety/ Safety/ Safety/
busy operational Economic Economic Economic
Generation 1 1 2 3 4
environment the
management problems Report Old Engines New & Mid-Life Future
are complex. Aircraft In Scope Yes Yes No
come and go and
engines are frequently RFC - Retirement for cause.
managed on a modular FM. - Fracture Mechanics.
basis. FADEC - Full Authority Digital Engine Control.
DT - Damage Tolerance
This means that a
component might be Figure 2 - Concurrent technology and maintenance philosophies
flying in three different its position, its running time and its flight time. Life
aircraft at three different bases within a period that is as remaining for each of the main stress features in the low-
short as six months. The only way to manage this is to pressure compressor (NDV), intermediate compressor
run a comprehensive parts life-tracking system. (MDV), the high-pressure compressor (HDV), and the
To put the frequency of engine changes in a mature fleet
into perspective, figure 3 shows Adour Mk 104 engine Adour Mk 104 ECU Rejections
rejections from the RAF Hawk fleet over a sixteen-month Fuel Lea c
period. 43% were due to cracking or engine component StowAca I
Vibration
life limits. This indicates that engine removals to replace Probe D;f
| Idle Oe
life expired components is a major nuisance and cost I Over 7 amp
driver in mature engine fleets. Every fleet manager would EFDC Failure
Lowlirusl
presumably welcome the tools for managing component R/HC elects
MaxFPMDef
lives as effectively as possible. Oil D3f
=J Sort Def
Just by using hourly lifing methods, it is possible to Surge
Miscellaneous
create charts like figure 4. That shows the remaining life
for various turbine blade sets. As the remaining life
reduces, the distribution for each group spreads. This
illustrates a phenomenon that is typical of many 10 20 30 40
mechanical systems. Number of arisings In period

By adding more information, namely the usage rate for Figure 3 - Typical engine rejections
each engine, figure 5 can be constructed. The example is
for Adour Mk 104 HP Turbine disks. The significance of
this chart is that a tidal wave of life expired components
Remaining Life in Hours
is only 4 to 5 quarters away. With a lead-time of two to Adour Mk 104 - G09 HP Turbine Sets
three years for new components, this type of information
is essential for the supply organisations. 30

This approach is of value to both first line and logistic

planning, At first line aircraft can be assigned duties so
«> 20
that availability is maximised. At HQ, purchase orders
can be placed, so that sudden increases in the number of
life expired components do not ground the fleet. I 10
To maximise these gains it is important that the lives of
the many components on an engine are synchronised to
maximise the on-wing period. Tools for doing this
Hours Remaining - Finite Life
already exist, and examples are the UK Pegasus
(PRAMS) system, in figure 5, and the GE Tornado
Figure 4 - Typical component life distribution
OLMOS system, which is shown below, in figure 6. Once
tools like these are in place, the full benefits from
1-6

corresponding turbines are also shown. A graphical

Predicted Lifex Arislngs
display of the remaining life, compared to the fleet Mour kk 104 • HP Turbine Disc
average is also shown.

5. COST SAVINGS
Although every costing exercise is unique due to
differences in accounting, operational and commercial
practices, some cost studies have been done on
retrospective fitting of life monitoring equipment. Figure
7 shows the estimated benefits that will be enjoyed by the
RAF as a result of fitting simple engine life recorders to
Expected Rejection by Quarter Year
the RBI99 engines in the ADV Tornado aircraft. The
algorithms are speed related only, with engine speeds Figure 5 - Typical component life expiry
sampled eight times per second.
Figure 7 clearly puts the time-scales for ■iiimmmMUHiHj
adding monitoring systems to ageing fleets 01/04
Konto-Restleben
into perspective. In this case, the financial TAKZ : 4632 HS: 1Z
benefits were about £20M after thirteen Twk-Restleben/Std. 524:08:00
4029
NDV-Restleben/Std.
KDV-Restleben/Std.
924:08:00
5743:12:00
SN Mt
years, and the breakeven point was about Twk -Variante
Position
CA
I
HDV-Jestleben/Std.
HDT-Restleben/Std.
1915:26:00
2495:50:00
eight years from the beginning of the TT Betrieb
TI Flug IWk
145:28:00
111:00:00
HDT-Jestleben/Std.
NDT-testleben/Std.
2746:10:00
8400:35:00
project. The slowdown in 2006 is due to RHL-Kestleben/Std. 1550:00:00

replacement of some aircraft by Eurofighter. Konten Restleben relativ

NDV 041
This takes no account of the safety benefits NDV 043
that should accrue from a better knowledge NDV
NDV
045
047
of fleet operation. Even if this amount of HDV 049
HDV 051
money appears trivial, to dismiss other KW 053
programs as not worthwhile, because of this HDV
HDV
055
057
cost study, would be shortsighted. Every HDV 059
HDV 061
situation should be treated on an individual HDV 063
HDV 065
basis. HDV 067
HDV 069
The thirty-year old aeroplane is a fact. When F 1
purchases are made, of new or used aircraft, F2

monitoring systems should be included in Figure 6 - Typical display of engine usage information
the negotiations. If sampling is considered,
then figure 7 indicates that a full fleet fit
saves more than twice as much as a half-
fleet fit over just thirteen years. Referring to Cost/Saving Analysis -Including Maintenance
figure 1, it may be seen that some engines
may remain in service for 40 to 50 years!
Therefore, the benefits that should accrue
from a modern monitoring and management
system, especially if fitted before entry into
service, should be considerable. Strong
supporting evidence for this is in chapter 3.

6. SUMMARY
This introduction has attempted to provide
an overview of the safety and financial --♦— Cumulative Total Fleet - - Cumulative Half Fleet
requirements that make the subject of
component life monitoring so important. Figure 7 - Cost benefit analysis for simple life counters on Tornado
The topics touched upon have been diverse,
09522854 01. Published by Future Systems (Air)42,
with the intention of providing an appreciation of the
MOD(UK).
breadth of the subject.
Figure 6 - Courtesy of MTU.
In the following chapters, many different aspects of life Figure 7 - Courtesy of Rolls-Royce and FS(Air)42,
usage monitoring are discussed in detail, ranging through MOD(UK)
regulatory, managerial, technical and operational
perspectives. Here only the surface has been touched.

7. REFERENCES
The Story - People and Eucams. Jan 1994, ISBN
 2-1

Chapter 2
Civil and Military Practices
by
(R. Holmes)

Page
1. Introduction 2-3
1.1. Control Systems Technology 2-3
1.1.1. Hydro-Mechanical Development 2-3
1.1.2. Analogue Electronic Controls 2-3
1.1.3. Digital Electronic Controls 2-3
1.1.4. Coincident Development of ENSIP and Electronic Controls 2-4
2. Structural Integrity Requirements 2-5
2.1. Airframes - US 2-5
2.2. Engines - US 2-5
3. Regulatory Developments 2-6
3.1. Damage Tolerance 2-6
3.2. Development of General Procedures 2-7
3.2.1. MIL-E-5007E (Military Specification) 2-7
3.2.2. MIL-STD-1783 (ENSIP) 2-7
3.2.3. MIL-E-8593A (Military Standard) 2-8
3.2.4. FAR 29 - Civil Standard 2-8
3.2.5. FAR 33 - Civil Standard 2-8
3.2.6. UK DEF STAN 00-971 2-9
3.2.7. JAR-E - Commercial Standard 2-9
3.2.8. Canadian Standards 2-9
3.2.9. Other European Standards 2-9
4. The Application of Standards 2-9
5. Fighter Engine Applications 2-9
5.1. Hydro-Mechanical Controls 2-9
5.1.1. France 2-9
5.1.2. UK 2-10
5.1.3. US 2-10
5.2. Hybrid Control Examples 2-11
5.2.1. France 2-11
5.3. Digital Control Examples 2-11
5.3.1. France 2-11
5.3.2. UK 2-11
5.3.3. US 2-11
6. Transport Engine Applications 2-11
6.1.1. Hydro-Mechanical Controls 2-11
6.1.2. Digital Control Applications 2-11
7. Helicopter Engine Applications 2-12
7.1. Hydro-Mechanical Control Examples 2-12
7.1.1. France 2-12
7.1.2. UK 12
7.1.3. US 12
7.2. Hybrid Control Examples 12
7.2.1. France 12
7.3. Digital Control Applications 12
7.3.1. France 12
7.3.2. International 12
7.3.3. UK 2-13
8. Significant Issues of Cost and Safety 2-13
8.1. Parts Classification Distinctions 13
8.2. Economic Life Decisions on a Component or Major Structures Basis 13
9. Summary 13
10. Conclusions 14
11. Recommendations 14
19 Rpfprpnrps 14
2-2
 2-3

the flight envelope (Mach number, temperature, pressure,

1. INTRODUCTION altitude, etc.), and engine requirements to provide service
Design practices, life prediction methods, life usage bleed air. This data can be used to provide automatic
monitoring systems, and control systems for gas-turbine thrust rating control, engine limit protection, transient
engines have all improved since they were in their control and engine starting. These electronic systems are
infancy in the 1940's. Then, a 25-hour test was the main simpler to operate by the flight crew, save fuel, are easier
pass or fail criteria for a new engine entering service. As and less expensive to maintain, and improve reliability
with many engineering practices, the developments standards when compared to the hydro-mechanical
evolved in response to durability problems and a desire to controls of the past. The implementation history of
do a better job of producing and maintaining gas turbine electronic controls at, for instance, Pratt and Whitney
engines. dates back to the 1950's when an early attempt was made,
using vacuum tube technology, to develop a system for
The need for engine reliability and durability was the J57 engine. The harsh vibration and acoustic
recognised in the US MIL-E-5007C specification, as long environment proved too much for these early controls.
ago as 1966. Although lacking the rigorous analysis, test
and demonstration aspects of today's specifications, MIL- 1.1.2. ANALOGUE ELECTRONIC CONTROLS
E-5007C called for a reliability of 500 hours between
The Rolls-Royce Gnome helicopter engine (developed
power loss events, and an inherent engine life of 5000
under licence from the General Electric T58) entered
hours within the specified environment. MIL-E-5007C
service in 1961 with an analogue electronic control
also required a 150-hour life demonstration when tested
system that was designed to reduce the helicopter pilot
in accordance with the MIL-E-5009 specification. In the
workload. The system primarily governed the power
UK, D. Eng RD 2100 and 2300 provided design
turbine output shaft speed by controlling engine fuel
requirements and guidance for all aspects of gas turbine
development, from the early 1950's until very recently. flow. It also provided temperature limiting, droop slope
matching, and throttle rate control functions. If the
As time passed these increasingly took account of the
electronic system failed, the control functions reverted to
practices adopted in the US.
the pilot, who controlled engine fuel flow using
Today the 'Safe-life' and the 'Damage Tolerance' information from the cockpit gauges.
approaches are the two most widely used design methods
The Concorde aircraft with Rolls-Royce and SNECMA
for producing components that meet life requirements.
The processes which are codified in specifications such as Olympus 593 engines has a two-lane full-authority
analogue engine control as the original design standard.
MIL-STD-1783 (ENSIP), MIL-E-5007E, FAR 33, MIL-
The first flight was in 1969 with entry into service in
E-8593A and JAR-E cover transport, helicopter, and
fighter engines. Their scope addresses design, but also 1976.
material characterisation, engine testing, and gathering of 1.1.3. DIGITAL ELECTRONIC CONTROLS
information on usage, maintenance and inspection. In this
chapter the evolution of engine controls, the regulations The advent of monolithic integrated circuits that were
governing structural integrity and methodologies for developed in the mid-1960's addressed control system
determining the life of components that ensure gas reliability problems and made electronic controls more
turbine engine durability are discussed. practical. In 1968 Pratt and Whitney was competing for
the FX engine program with General Electric and elected
1.1. CONTROL SYSTEMS TECHNOLOGY to use electronic technology in the proposed engine.
In chapter l, the technology factors that were used to Because the reliability of electronic technology had not
decide the scope of this document were discussed. One of yet been proven, the engine was to operate with hydro-
the major enabling factors for real-time present day mechanical controls and would include a supplementary
systems is the ready availability of the necessary data electronic unit.
from the instrumentation that already exists for engine The hydro-mechanical unit controlled all engine functions
and aircraft control purposes. Hence, it is relevant to while the electronic unit, known as the supervisory
consider control systems and their development to the control, operated several electric motors that trimmed
current state of the art. exhaust gas temperature and engine speed to maximise
performance. Supervisory control systems have an
1.1.1. HYDRO-MECHANICAL DEVELOPMENT electronic unit, which is superimposed on a hydro-
The 1950's and 1960's saw an era dominated by the use mechanical system, to govern power management. If the
of hydro-mechanical controls which demonstrated electronic unit fails, the system reverts to hydro-
reliability rates as high as 300,000 hours between in- mechanical control.
flight-shutdowns. Engines did not need the precise or
In the 1970's, the United States government funded
high-speed response of electronic controls, which was
electronic control research, culminating in the Digital
rather fortunate!
Electronic Engine Control (DEEC) which was
The basic function of modern electronic engine control incorporated on the F100-PW-220 in 1985 (Kuhlberg et
systems is to adjust variable geometry, control fuel flow, al, 1989). This full-authority electronic control for
trim, and regulate other actions such as the application of military engines was two and half times more complex
anti-icing systems. The logic for controlling these than the preceding supervisory control, and cost the same
functions varies with throttle movement, position within as the control system that it replaced.
2-4

1 a single set of sensors and

computers. If a disabling
EVOLUTION OF ELECTRONIC CONTROLS failure occurred in any of
these elements, the system
IN JET ENGINES reverted to a hydro-
mechanical operation. Even
the DEEC can revert to a
simple backup mechanical
system in the advent of a
failure.
Control Digital Electronic Engine
Functions Control systems were flight
tested on Rolls-Royce Gem
helicopter engines from 1979
and entered service in 1985.
These systems normally
provided engine control and
2000 governing of the power
turbine (and aircraft rotor)
speed. If a system failure
Figure 1 - Evolution of Electronic Controls occurred, the pilot took
manual control of engine fuel
flow.
F100

SAB
Sat The military efforts spawned
MIL-E-5007 R*vl*w commercial efforts, which in
No
Formal
USAF
I 1 The
Currant
ENSIP
the US featured a capability
Structural MIL-STD-17M demonstration and a reliability
Requiremenle
' Leuons
/ Lttrnad
/ / > (USAF) demonstration. The capability
/ / / / demonstration was carried out
Production Engine Development Programa
JM TF-30 F1M
in 1975 on a JT8D engine and
JM TF-33 in 1978 on a JT9D engine.
Both the JT8D and JT9D
| tt | 70 | 71 I 72 | 73 | 74 | 75 | 7« ) 77 | 7» | 7» | 80 | »1 | u | rpri systems were supervisory
Calendar Year
electronic plus a hydro-
mechanical control. The
reliability demonstration was
Evolution of the ENSIP Philotopky
carried out on several, airline
Figure 2 - Changing US Durability Requirements operated, Boeing 727 aircraft
' equipped with JT8D engines.
While the electronic system did not control any
engine functions, it did carry out error detection and
HISTORICAL PERSPECTIVE diagnostic tests in a closed loop system by addressing
outputs and inputs. Over a six-year period covering
MANY ENGINE DEVELOPMENTS IN PAST DID NOT REVEAL 400,000 flying hours, these electronic controls
SERVICE RELATED PROBLEMS demonstrated reliability levels of 20,000 hours mean
TYPICAL 150 HOUR TEST
time between failures (Kuhlberg et al, 1989).
MIL
A full authority, dual channel, Engine Electronic
Control (EEC) was certified on a PW2037 engine in
MOT TYPE 1983. Electronic engine control is a major advance in
ENOUHANCE gas-turbine engine control-systems that provide
CYCLE
significant benefits in operability, supportability,
reliability and safety over hydro-mechanical controls.
IDLE
6 HOURS - REPEAT 26 TIMES Figure 1 shows how the growth of control functions
CYCLE DEFINITION
IHHn 2,000 MR has increased over time.
TEaVT USAGE
IDLE-MIL-IDLE MO 10,000 1.1.4. COINCIDENT DEVELOPMENT OF
20 1,000
O-MU.-0
ENSIP AND ELECTRONIC CONTROLS
Figure 3 Typical 150-Hr test The first implementation of full digital electronic
engine controls roughly coincided with the
Full authority control systems perform all control
introduction of United States Air Force MIL-STD-1783
functions electronically and do not interact with a hydro-
or (ENSIP). ENSIP was issued in 1984 and digital
mechanical system. Each of these electronic systems had
 2-5

controls were introduced on the

F100-PW-220 in 1985. Both efforts
aimed at improving engine
reliability.
DISTRIBUTION OF CRACK
INITIATION TIMES
2. STRUCTURAL CRACK
SIZE
INTEGRITY
REQUIREMENTS

2.1. AlRFRAMES-US
ENGINE USAGE (CYCLES, EFH, ETC.)
The United States Air Force
approach to structural-fatigue CRITERIA: LOW CYCLE FATIGUE (LCF) LIMIT BASED ON LOWER BOUNO
certification was outlined during the (-3« OR 1/1000) DISTRIBUTION OF CRACK INITIATION TIME

1959 Symposium on Fatigue of ACTION: 100% PART REPLACEMENT AT LCF LIMIT

Aircraft Structures. Because of the
increasing number of fatigue
failures/incidents that occurred Figure 4 - Safe life approach
during the 1950's, a large-scale
effort was conducted in 1958 to
determine the steps that were
necessary to improve aircraft
service life. This culminated in the w///w/ss//yw/s/sssss/ss(sssss/rss/s/s//sss//s/s.
CRITICAL
Aircraft Structural Integrity
Program (ASIP) which was
published in 1969 and contained
changes in the following areas:
Structural development INITIAL

process;
Structural testing practices and IST 3ND 3R0
INSPECTION INSPECTION INSPECTION
policies;
EXPOSURE (CYCLES. EFH.)
Monitoring the actual aircraft
usage; CRITERIA: TIME FOR LARGEST FLAW TO GROW TO CRITICAL SIZE
ACTION: INSPECT OR PROOF TEST
Air Force Fatigue
Certification Program.
Figure 5 - Damage tolerance approach
2.2. ENGINES - US
Engine requirements for structural
integrity followed a similar path to EWUmONOfASP/EHSIP
those of the airframe program, a short DAMAGE TOLERANCE ASSESSMENTS
|CYJ1 CY» CT7T |CY71 |om|cni |cm | cm CTO| CTM|Cm|CTH |CtI7
while later. Figure 2 shows how the CT7I |CY7J |CY74

■Ml
engine durability requirements came 1
t 1' Km
about for US Air Force gas-turbine c 1 II!
engines. Prior to 1969, early structural 1 r.l >i
r~ ,....,J F«UH
assessments were based on engine 1 1 11 1
testing which in 1946 was 25 hours and 1 Icuu "*"~1 MS IfT |

increased to 150 hours in 1952. This | ■ "■ ";ii »-in 1

A «CRAFT
1 ' C-HU
cycle was very simple, as shown in 1 Z2 MW
figure 3, and unfortunately, it did not L 3i»

reveal service-related problems. Before Z} ix

1969 engine specifications were 1 1 3'-»

deficient in such areas as life n in Mll~~*

1 |U1»

requirements, duty cycles, analysis, and L _.. lew

L.
testing that was not mission based. 1 1 U-T1

1 1""," .'.'l«-»**
™, "". ■) »-«
Lessons were learned by reviewing the
TF30 and other engine designs and in
1973, the U.S. Navy issued the MIL-E- ENGINES C 1 F-1M i
c TTM
5007D specification, which established
rigorous standards for ensuring engine Figure 6 - Evolution of ENS! P
durability. This document included the
Safe Life concept, and recognised that components should life. See figure 4.
be retired at the minimum calculated and validated fatigue
2-6

Commercial regulations for gas-

turbine reliability practise the
safe-life approach. Review of EVOUmONOFASV/ENSIP
these regulations is pertinent to DAMAGE TOLERANCE DESIGNS
the discussion of military
CY7Z 1CY73 1CT74 | CT75 | CT» | CY77 | CY7I | CYT» | CYW | CY»1 I CYK ] CYM | CYM | CYM | CYM_ CYM
transport, helicopter, and fighter ei? r
applications as many of these 3ca«

commercial engines have found 3i-i» Till

their way into military ni«
applications. In some cases, as AIRCRAFT *w C"
discussed elsewhere, engines that
were designed to civil
specifications are further
scrutinised under a separate and
quite different set of military ENGINES 3 mi on
standards.
| I FID»» ICO«)

3. REGULATORY JMEMTFE I

DEVELOPMENTS Tm»
noptPi u
The history of regulatory Fill IK C
developments is:
Prior to 1965, the civil ' ■■■' ' «««m/nmB»in>
regulations, both the US
Civil Air Regulations (CAR) Figure 7 - Application of damage tolerant design
and the British Civil Air
Regulations (BCAR), dealt
with durability issues
through testing. ffSTIMfOR EN6ME WRABIUTY
In 1965, in the United ACCELERATED MISSION TESTING (AMT)
States, the Federal Air
Regulations (FAR) were AMT ELIMINATES NONDAMAGING TESTING
codified. These defined
FULL MISSION
operating limitations for (TYPICAL)
disks and spacers in terms of
service life.
By 1970, FAR 33.14 placed
restrictions on allowable life
of a component through test
demonstration.
In 1976, a Scientific
Advisory Board was
convened to review engine
durability issues. Findings
from the panel included:
1. '...we need to apply a ET
system of discipline to our TIME (HB)
development process'; BUT RETAINS DAMAGING CYCLES AND HOT TIME
2. '...Air Force should
Figure 8 - ENSIP Accelerated Mission Testing
define an aggressive program
for engine mechanical and during a similar period and Change 9 permitted
structural integrity and durability'; analytical life prediction.
3. '...this program should be required by
regulation'; 3.1. DAMAGE TOLERANCE
4. '...durability and damage tolerance assessment Damage Tolerance is the ability of a component to resist
should be performed on fleet engines analogous to failure due to the presence of flaws, cracks, or other
those being performed on several weapon system damage. This method is illustrated in figure 5. Following
airframes;' the recommendations of the advisory board data was
5. By 1980, the FARs permitted approved collected from various engines and structural assessments
procedures to be used. These covered both analytical were conducted to determine what the impact of applying
methods and testing. In Europe, the BCARs were the damage tolerance procedures would be. These
combined into the Joint Air Regulations-E (JAR-E) structural assessments included the following:
 2-7

• Damage tolerance assessments of components of 122 percent of the maximum allowable steady-
previously not reviewed with fracture mechanics state speed;
analysis; • Engine static load tests;
• Updated low cycle fatigue analysis with improved • Substantiation of durability with a Durability Proof
mission definitions; Test consisting of 300 hours of accelerated mission-
• Evaluation of inspection procedures and their oriented endurance testing that is preceded and
implementation; followed by stair-step/Bodie test schedules.
• Establishment of maintenance plans that incorporate • Substantiation of durability by Accelerated
non-destructive evaluation; Simulated Mission Endurance Testing (ASMET)
• Development of improved durability manufacturing equal to at least 1000 hours or one half of the cold
methods. parts lives;
• Low cycle fatigue testing of three sets of critical
Other related efforts concentrated on the generation of engine components via spin pit testing.
crack growth data on titanium and nickel base alloys, and
the testing of cracked components in spin rigs and in Components are monitored in the field, design mission
engines. In 1984 the Engine Structural Integrity Program profiles are validated and the operating conditions arc
Standard (MIL-STD-1783), which is the current standard recorded. By monitoring this operating experience, and
for designing and maintaining gas turbine engines providing feedback into the life analysis process, new
utilising damage tolerance procedures, was released by lives arc determined for components in a service
the United States Air Force. Some airframes and engines environment. To further improve life assessments,
assessed with the damage tolerance process and those that inspections of fracture-critical in-service parts are carried
have been subsequently designed with these procedures out on an opportunistic basis.
are illustrated in figures 6 and 7 (Ogg, 1998).
3.2.2. MIL-STD-1783 (ENSIP)
3.2. DEVELOPMENT OF GENERAL PROCEDURES MIL-STD-1783, the Engine Structural Integrity Program
A brief overview of each of the military and commercial (ENSIP), was issued in 1984 by the United States Air
practices used to produce durable gas turbine designs for Force and followed the development of MIL-E-5007 to
military fighter, helicopter, and transport applications which it has many similarities. ENSIP is an organised and
follows. Many have common features. disciplined approach to the structural design, analysis,
development, production, and life management. Areas
3.2.1. MIL-E-5007E (MILITARY SPECIFICATION) stressed in ENSIP include:
The current 5007E specification was issued in 1983 and Development plans;
superseded 5007D, which was issued in 1973. This was Operational requirements;
the original military specification that placed strict cyclic Design analysis, material characterisation, and
life requirements, maintenance, testing, and design development tests;
processes on gas turbine rotating hardware. It is most
Component tests;
widely used by the US Navy. The 5007E specification
requires that the safe-life approach be practised on critical Core and engine tests;
parts. These parts are minimally defined as the engine Ground engine tests;
pressure casing and all rotor stages. Flight engine tests;
Engine life management.
Military specification MIL-E-5007E includes the
following design requirements. ENSIP emphasises the damage tolerance approach, with
• All engine critical parts are to be designed to twice crack-growth resistance-criteria added to maintain
the life requirement. reliability in the presence of manufacturing and
• Conventional structural criteria are defined to processing defects in the materials. Components arc
ensure durability from such failure modes as burst classified into the following categories:
strength, vibration, low cycle fatigue, high cycle • Safety Critical for components that will result in the
fatigue, and creep. probable loss of the aircraft, or a hazard to
• Materials are to be characterised with a three-sigma personnel due to direct part failure or by causing
standard deviation, and fracture toughness of the other progressive part failures;
material is to be considered. • Mission Critical for components whose failure will
• Consideration is to be given to engine-airframe generate a significant operational impact by
inter-actions. degrading mission cap-ability to the extent of
creating an indirect safety impact on the aircraft.
Military specification MIL-E-5007E Structural requires Items whose failure will cause a non-recoverable in-
verification testing and measurements as follows: flight shutdown on multi-engined aircraft are
considered mission critical.
• Turbine and compressor rotor overspeed;
• Durability Critical components are functionally or
• Over-temperature of the first stage turbine rotor by
structurally significant, and in the event of failure
81 degrees Fahrenheit;
possess the potential for generating a significant
• Disc burst testing of all rotating discs to a minimum economic impact but will not necessarily impair
2-8

safety or mission capability; 3.2.4. FAR 29 - CIVIL STANDARD

• Durability Non-critical component failures result in FAR 29 or 'Airworthiness Standards -Transport Category
a minor economic impact or minor degradation of Rotorcraft Sub-part E' refers to powerplants. Section
mission capability, and may initiate unscheduled 29.903 discusses the engines and the need to minimise
maintenance actions; hazards to the rotorcraft in the event of an engine failure.
• Other. It states that systems associated with engine control
devices, systems, and instrumentation must be designed
Other key features of MIL-STD-1783 include the to give reasonable assurance that those engine operating
following, many of which already existed in MIL-E- limitations that adversely affect engine-rotor structural
5007E: integrity will not be exceeded in service.
• Structural design must be based on realistic mission 3.2.5. FAR 33 - CIVIL STANDARD
usage and service life requirements;
• Conventional structural criteria are issued to ensure United States Federal Air Regulations (FAR) ensure
durability from such failure modes as vibration, low engine durability and are similar to military specifications
cycle fatigue, high cycle
fatigue, and creep; EHSIP DURABILITY REQUIREMENT
• Damage tolerance (crack ACCELERATED MISSION TEST (AM) DURATION
growth resistance) criteria
are added to maintain INITIAL FULL INITIAL OPERATIONAL
reliability in the presence FL16HT FLIGHT SERVICE CAPABILITY
of materials, RELEASE RELEASE RELEASE RELEASE
(IF«) (FFR) (ISR) (OCR)
manufacturing, and (5007 PFRT) (5007 HQTI (IOC)
processing defects. Parts
classified as safety or
mission critical must
receive damage tolerance
AMT I 2X FLIGHT TEST I
analysis. FLIGHT CLEARANCE IMSICX
SHYICE LIFE
• Compliance may be (BSD
achieved through analysis AMT II
or verification; DESIGN VERIFICATION
1-2 »051
• New materials and AMT III
processes are characterised; DESIGN VERIFICATION
• The internal operating INCLUDING CHANGE
environment must be AMT IV
defined from analysis and OPERATIONAL USAGE
engine Figure 9 - ENSIP durability requirement
thermal/performance test
data; in a number of ways. Requirements include:
• Durability must be substantiated by engine testing
or an Accelerated Mission Test (AMT) as illustrated • A safe-life practice which must set allowable
in figure 8. As the mission and engine testing operational requirements. These must be approved
progresses the design is released for increasing by the Federal Aviation Administration (FAA);
levels of use, as illustrated in figure 9. • Components must be classified as to whether they
can be contained, or not, in the event of a fracture;
3.2.3. MIL-E-8593A (MILITARY STANDARD)
• Components judged non-contained require
The general specification for aircraft with turbo-shaft and appropriate life limits and tracking and include
turbo-prop engines was released in 1975. Its preamble disks.
states 'This specification establishes the performance, • Components which may cause significant secondary
operating characteristics, design features, detailed damage are also to be life-limited and tracked and
interface configuration definitions, and installation include: shafts, seals, and fan blades;
envelopes for turbo-shaft and turbo-prop engines. It also • Component lives may be set through analytical
establishes the demonstrations, tests, reports, inspection prediction and/or component testing.
procedures, and other data required for satisfactory • Initial certification may be to 33% of the predicted
completion and acceptance, by the using service, of the safe-life by component testing. Additional testing is
Preliminary Flight Rating Tests (PFRT) and the permitted to extend the safe-life (life extensions
Qualification Tests (QT) for the engines.' MIL-E-8593A may be up to 1 cycle extension for every 2 cycles
mirrors the structural integrity requirements of Mil-E tested);
5007E. The differences between the two documents exist
• The Certified Life is an allowed percentage of the
in paragraphs pertaining to requirements other than life or analytically predicted safe-life;
structural performance.
• Mission definitions may be defined by the
manufacturer;
• The required suitability and durability of materials
 2-9

is defined; on military transport engines that were originally

• The acceptability of the materials test data base designed in accordance with safe-life practices.
must be demonstrated/proven; Transport engines generally start their design history
• Conformance must be controlled by approved under civil regulations which may include (FAR or JAR-
specification; E). As damage-tolerance standards have become
• The engine design and construction must minimise necessary for United States Air Force applications MIL-
the development of an unsafe condition of the STD 1783 has been applied to these engines. Adherence
engine between overhaul periods. to this standard may lead to the application of additional
testing, analysis, and inspection requirements. In some
3.2.6. UK DEF STAN 00-971 cases, limited component redesign may also be necessary
This document has superseded Def Stan 05-2100, which to meet the new design safety requirements.
superseded D Eng RD 2100 and 2300. The requirements Fighter engines are not often designed to civil
tend to be harmonised with other standards, but include requirements, but can include both safe-life and damage
special requirements based on UK military experience. tolerance design approaches, as encompassed in the MIL-
E-5007E specification and MIL-STD-1783 standard
3.2.7. JAR-E - COMMERCIAL STANDARD respectively. It should be noted that the United States Air
Joint Air Regulations are a harmonisation of existing Force and Navy requirements are different. The Air Force
national regulations into an international requirement. practices the damage tolerance philosophy and the Navy
Most nations with aviation design and operating practices the safe-life approach.
requirements are involved in their development. Without
Since 1995, the US Navy has introduced refinements to
too much loss of accuracy, they may be considered as a
the M1L-E-5007E practices that include damage tolerance
combination of the US FAR, MIL-STD-5007E and the
analysis. In this case, fracture critical components are
British CAR regulations. For civil usage they will replace
designed to LCF and damage tolerance, while managed to
the BCAR requirements.
crack initiation.
3.2.8. CANADIAN STANDARDS Helicopter gas turbine engines are procured for United
In Canada, two documents that relate to airworthiness States forces predominantly by the Army and Navy. Safe-
were produced in 1996. The first is entitled 'Review of life practices are followed for these applications with
the Regulatory Requirements for Aero Engine Design'. It representative specifications including MIL-E-8593A for
contains a review of the airworthiness certification turbo-shaft engines. On the civil side both FAR and JAR-
processes and requirements for aeroengine design E specifications are applied to helicopter engines.
changes of Transport Canada (TC), the United States
Federal Aviation Authority (FAA), the Department of 5. FIGHTER ENGINE APPLICATIONS
Defense, the Civil Aviation Authority (CAA) in the UK, The earliest aero gas-turbine engines were designed for
and the European Joint Airworthiness Authority (JAA). fighter applications. Prior to the introduction of military-
The second document, 'Qualification Methodology for engine durability standards, engine life was determined
Advanced Gas Turbine Repair/Rework', provides gas by testing. Initially, lives were so short that flight duration
turbine engine operators with qualification methodologies was limited by the failure of turbine blades, and examples
for airworthiness certification, design and repair and from this period can be seen in the Deutches Museen
rework changes. The airworthiness certifications are (Museum) in Munich. Within a few years gas turbines
based on requirements of Transport Canada for design had developed to the point, where lifing was based on
approval applications. The document includes Failure engine-starts and flying hours, but frequent inspections
Mode, Effects and Criticality Analysis (FMECA) which were still required. Major improvements in engine life
will enable design authorities to assess component began to appear in the 1950s and 1960s with the advent
criticality, typical failure modes and failure probability. of new materials, and improved engine designs. The
following control-system comments pertain to engine
3.2.9. OTHER EUROPEAN STANDARDS usage and monitoring systems.
Most multinational-European military-aircraft projects
create their own airworthiness standards, for both the 5.1. HYDRO-MECHANICAL CONTROLS
aircraft and the engines. Examples are Tornado and RB 5.1.1. FRANCE
199, Eurofighter and EJ 200, Alphajet and Larzac.
5.1.1.1. ATAR
4. THE APPLICATION OF STANDARDS AT AR is a jet-engine family developed since the 50's and
during the 60's. The different models equipped the
The transport, fighter and helicopter applications are each
Mirage 3, 4, V and Fl family, the Etendard and the Super
influenced by durability requirements that are placed on
them by the government or civil entity that is procuring Etendard, the Cheetah and the Panthera. The early models
were controlled by a hydro-mechanical system; the latest
them. It is not unusual for further requirements to be
mark (9K50), which powers the Mirage Fl and which
placed on a particular gas turbine design, if it is purchased
by an agency that was not involved in the original entered service in 1969 is controlled by a hybrid analogue
procurement or certification process. A good example of computer and hydro-mechanical system.
this is the imposition of damage tolerance requirements Initially the mechanical loads were limited by over-speed
2-10

resistance and long-term creep elongation. The life limit etc.) to an appropriate place in the airframe, where a
was fixed in Engine Flying Hours (EFH). Cyclic limits monitoring unit can be fitted.
were determined by engine ageing testing and
observations made on operational engines. For the newest 5.1.3. US
marks a complete verification of LCF stress was 5.1.3.1. AIR FORCE
undertaken and used current numerical methods
The TF30 engine entered production in the mid-60's, on
correlated with photo-elastic models. the Fl 11 and A-7P aircraft (See Portugal in App 2) when
5.1.1.2. LARZAC durability standards were handled by engine test and any
durability requirements were included within the engine
The 04-C5 Larzac engine entered service in 1979 to specification (Taylor, 1969). The TF-41 engine also
power the Alpha Jet. In 1984 the 04-C20 version with entered service at the same time on the A-7 aircraft. As
higher thrust and temperature was developed. Both
shown in figure 2 lessons learned from the TF30 program
versions use a hydro-mechanical regulation driven by an
were used to help develop MIL-E-5007. Life extension,
analogue computer, and were designed to safe-life
and a retirement for cause evaluation on the TF30 have
standards. In mid-85, some engines of the French Air required that a damage tolerance analysis, a focused
Force were equipped with a tape recorder to monitor
inspection, and refurbishment or modification of
flight parameters. These parameters were replayed in a components be conducted to extend the life of critical
ground station to calculate the life consumption
components. Under the life extension plan, the
associated with each recorded flight. This allowed the
components are allowed to operate for one inspection
correlation of all flight profiles with their damage counts
interval following the refurbishment prior to being
and the determination of a cyclic exchange rate between
replaced or retired.
engine flying hours and the different flight profile
categories. The cyclic exchange rates for other operators The F100-PW-100 was in production from 1976 to 1986
were determined using their theoretical mission profiles. and used on the F-15 fighter. This engine was developed
Life extensions are authorised after the evaluation of disc using safe-life procedures and was the subject of a
residual life performed on normally serviced parts using structural assessment, conducted from 1978 through
spin pit tests. 1982, prior to the development of the ENSIP (MIL-STD-
1783) specification. This study sought to determine the
5.1.2. UK impact of applying damage tolerant procedures to an
The Pegasus engine entered UK operational service in engine that had been designed using safe-life procedures,
1969 in the Harrier GR1 and in the US in 1972 in the and what would be involved in implementing such a
AV8A (Mk 402). The basic engine design dates from the design approach on an existing engine.
late 1950's. It is based on safe-life procedures. This has
not changed through subsequent up-rated engines. The 5.1.3.2. NAVY
lives of critical parts are defined from specimen and The J52 engine, entered operational service in 1965 in the
component testing using a 1 in 750 probability of A6 (J52-P-6) aircraft and made its first flight on an A4E
detection of a 0.38-mm crack length with 95% (J52-P-6A) aircraft in 1961 (Taylor, 1969). However, the
confidence. The lives in operational service are controlled J52 started out its life as an air-to-surface cruise-missile
in terms of flight hours using cycles-to-hours exchange engine where durability was not a concern. The engine
rates. These exchange rates for the United Kingdom was subsequently updated to the safe-life process and
Services are based on measurements from a small sample over 2,600 engines were produced. The J52 uses a hydro-
of operational aircraft using the EUMS equipment and for mechanical engine control.
other operators on assumed mission profiles. The Engine The TF30 entered service on the F-14 fighter in 1972. It
Usage Monitoring System (EUMS) records the basic
was developed using safe-life procedures that included
engine parameters during normal flying and these
the MIL-E-5007 standard. The TF30 uses a hydro-
recordings are replayed in a ground station to analyse the
mechanical engine control. After several years of
LCF cycles consumed. operational service, TF30 rotors were found to have LCF
The Adour engine entered service in 1974 in the Jaguar cracks, well before the design life requirement was
aircraft and critical parts lives were controlled in the same reached. In an effort to keep the fleet flying, whilst a
way as with the Pegasus. In 1976, the engine entered redesigned disk could be qualified, component lives were
service in the Hawk aircraft. Since 1980, it has been used re-analysed using damage tolerance procedures. (1/1000
by the Red Arrows display team. The display flying initiation + life to rupture from a 1/32-in crack -
carried out by the Red Arrows was known to generate a approx.0.8-mm - on a 1/1000 basis) Components were
wide variety of engine usage rates (a factor of 40 between inspected at overhaul with normal inspection procedures
extremes). In order to mange this a simple low cycle (FPI), returned to service and flown to one half of the
fatigue usage counter was fitted to the Red Arrow Hawk crack propagation life. By this time, the redesigned disk
aircraft and has been supplied to several other fleets. An was qualified and began to replace the high flight-time
updated version of this unit is currently being applied to components in the field.
the whole RAF Hawk fleet. The retrospective application
This is a good example of the use of damage tolerance
of a usage monitor to a 'pre-electronic' engine is analyses and field inspections to mitigate risk until a
relatively difficult. This is because aircraft wiring
redesign could be deployed.
changes are needed to get the signals (speed, temperature,
 2-11

5.2. HYBRID CONTROL EXAMPLES flying and these recordings were replayed in a ground
station to analyse the LCF cycles consumed.
5.2.1. FRANCE
With the introduction of the Harrier GR Mk5 the Engine
5.2.1.1. M53-5 Monitoring System was introduced. This was fitted on a
The M53-5 entered service in 1978 to power the Mirage fleet wide basis and computed life usage on selected
2000. The engine control system is composed of an components in near real-time. At the end of each flight
analogue full-authority regulator with redundancy. A the results were downloaded and transferred to the
hydro-mechanical system that covers the full flight Harrier Information Monitoring System (HIMS). This
envelope is provided for the safety mode. The engine has provided fleet life management tools.
a modular design. Life is determined using safe-life
procedures. The main damaging parameters are fatigue 5.3.3. US
and creep. Individual component damage-tracking, based The introduction of electronic engine controls in gas
on the flight time, is performed on each module. turbine engines coincided with the application of damage-
tolerance design practices in fighter applications for the
5.3. DIGITAL CONTROL EXAMPLES United States Air Force. The F100-PW-220 turbofan was
one of the first engines designed to the MIL-STD-1783
5.3.1. FRANCE
damage tolerance standard, and was introduced into
5.3.17. M53P2 operational service in 1986.
The M53-P2 entered service in 1985, with a full-authority The F100-PW-229 turbofan is a higher thrust derivative
digital control Engine Control System (ECS). The engine that followed the F100-PW-220 program, and was
engines operated by the French Air Force since 1987 designed to meet damage tolerance standards. It contains
have been individually equipped with an Engine Life a second-generation digital electronic engine control, and
Monitoring System which calculates the fatigue and the is used in F-15 and F-16 aircraft as well.
creep damage and stores the residual life for about 20
critical part locations. After engine initial build or rebuild The F110-GE-129 includes a single channel Digital
the on-board system is loaded with the engine Engine Control (DEC) with a hydro-mechanical backup.
configuration i.e. serial numbers and residual life of the The F110-GE-129 was designed in accordance with
monitored components. If the component is new, the damage tolerance standards. Used in the F16C and D
residual life is the initial life. The life calculation is fighters, this engine was introduced in the late 80's.
performed in real-time by a dedicated system using the
flight parameters transmitted by the engine control 6. TRANSPORT ENGINE APPLICATIONS
system. Flight-by-flight life consumption can be
displayed to the flight mechanics. But most of the time it 6.1.1. HYDRO-MECHANICAL CONTROLS
appears that a periodic, typically once a week, 6.1.1.1. us
downloading of the life consumption to the on-ground
As with fighter engines, prior to the introduction of
component life database is adequate. It allows a precise
military standards and specifications on engine durability,
enough monitoring time-step while reducing maintenance
engine testing to determine life was the rule.
workload. Since that time the ELMS has evolved several
times because of the acquired experience of the engine One of the earliest US gas turbine engines was the J57 on
behaviour in operation. the B52 bomber (Models C, D, F, G) and the KC-135A
Tanker. This engine was introduced in early the 50's and
The fatigue life and creep lives, and the inspection
is similar to the JT3 engine used on commercial Boeing
schedule of the monitored components are based on limits
707 models (Taylor, 1969). This engine uses mechanical
expressed in a specific flight-damage unit consistent with
controls.
the on-board damage calculation. In the 1990's the on-
board ELMS was introduced as a standard engine The TF33 has replaced the J57 on the B52 bomber
component and is now used by several customers. (Model H) and is also used on the Lockheed C-141 B,
Boeing E-3A, KC-135E and C-18A
5.3.2. UK
The Pegasus engine entered operational service in 1969 in 6.1.2. DIGITAL CONTROL APPLICATIONS
the Harrier GR1 and in 1972 in the AV8A (Mk 402). The 6.1.2.1. us
current Pegasus Mk 408 has a digital electronic control
system. The engine design is based on safe-life Three US civil gas turbine engines that were originally
procedures. The lives in operational service are controlled designed to safe-life standards (i.e. FAR) have since been
assessed to military damage tolerant standards (i.e. MIL-
in terms of flight hours using cycles-to-hours exchange
STD-1783). The assessment and subsequent actions to
rates. These exchange rates for the UK Services are based
on measurements from a small sample of operational meet the damage tolerance standards do involve
aircraft using the EUMS equipment and for other additional work and cost but can pay benefits later in
operators on assumed mission profiles. terms of life cycle cost. The engines that have been
scrutinised with damage tolerance standards are:
Early aircraft were fitted on a sample basis with the
Engine Usage Monitoring System (EUMS) which • The JT15D-5B turbofan for the Jayhawk Tl A;
recorded the basic engine parameters during normal • The F117-PW-100 turbofan for the C-17 transport;
2-12

Each of these engines features electronic controls, which control process similar to that used for the Pegasus.
have been standard from the start on the PW2000 series
of engines from which the F117-PW-100 turbofan is 7.1.3. US
derived. The electronic controls were later added to the Engines are designed to safe-life standards, (FAR 29, 33,
JT15D-5B engine, which began its derivative and JAR-E) and MIL -E-8593A (Aircraft Engine Turbo-
development from the original JT15D-1 engine in the late shaft/Turboprop). The Army and Navy are the primary
80's. These engines range in capability from 3,000 to purchasing agencies and set the requirements.
60,000 pounds of thrust which shows that damage
All of the turbo-shaft engines currently in the United
tolerance procedures can be applied to any gas turbine
States Navy inventory completed design and qualification
application regardless of size. Each program had to:
prior to 1985. They were designed using safe-life
Identify critical components; procedures. If, in the future, component life limits are
Conduct thermal and stress analyses; reviewed and updated for these engines, damage
Perform durability and damage tolerance analyses; tolerance may be assessed. Two points of caution should
Characterise materials for crack growth behaviour; be considered. First, the engine components were not
Qualify inspection capabilities. originally designed with damage tolerance in mind and
the likelihood of usable crack growth life existing is
In the case of the JT15-D Accelerated Mission Testing considered small. Second, it may be more cost effective
(AMT) was conducted to a cycle that was very different to simply replace components with a new (zero time) part
from the commercial cycle for which it was originally than to conduct periodic inspections.
designed. Additionally, the following tasks were required:
7.2. HYBRID CONTROL EXAMPLES
• Tracking to update the operational mission
definition is planned; 7.2.1. FRANCE
• A life management plan had to be developed; The MAKILA turbo-shaft engine is installed on the
• A plan to modify hardware, if necessary, had to be SUPER PUMA helicopter. This engine was designed
put in place; using safe-life procedures. It uses a hydro-mechanical
• Additional inspections, over and above those engine control unit for the generator and an electronic
required for commercial applications, could be engine control unit for the power shaft. The LCF counting
called for. philosophy is the same as for the ARRIEL 1.

7.3. DIGITAL CONTROL APPLICATIONS

7. HELICOPTER ENGINE APPLICATIONS
7.3.1. FRANCE
7.1. HYDRO-MECHANICAL CONTROL 7.3.1.1. ARRIEL2
EXAMPLES
The TURBOMECA ARRIEL2 turbo-shaft engine powers
Like transport aircraft engines, helicopter engines in this the S76C+, AS365N3, and AS350B3 helicopters, and is
category were generally designed to safe-life standards. equipped with a single channel FADEC. Certification is
in accordance with the JAR-E regulation and its design is
7.1.1. FRANCE based on a safe-life analysis. The initially authorised in-
The TURBOMECA ARRIEL 1 turbo-shaft engine was service life limits of the rotating parts are to be
developed in the 70's, entered in service in the early 80's progressively extended following a life management plan.
and is installed on the AS365N2, AS350B2, S76c, and The in-service low-cycle fatigue counting is carried out
Agusta A109K helicopters. It uses a hydro-mechanical according to one of the following methods:
engine control. The in-service low-cycle fatigue counting
is carried out according to either the 'lump' or the • The 'lump' method;
'recommended' methods. The latter is based on the • The recommended method based on partial cycle
counting and level of complete and partial cycles. There counting.
is some benefit using the recommended method because
it counts fewer cycles than the 'lump' method. The recommended method is programmed into the
Electronic Engine Control Unit. It uses rainflow
An electronic device is proposed as an aid to cycle decomposition and is the primary counting method.
counting. This aid uses an algorithm based on the
recommended method, with rainflow decomposition of The ARRIUS2 engine design and control philosophy is
the engine speed cycles. The rainflow technique identifies the same as that for the ARRIEL2. In some applications,
and pairs the peaks and troughs of the engine rotor speeds the use of emergency rating (One Engine Inoperative) is
into cycles. The damage is then computed for each cycle also electronically counted.
and added to the cumulative total.
7.3.2. INTERNATIONAL
7.1.2. UK 7.3.2.1. FRANCE AND UK
The Rolls-Royce Gem engine entered service in 1976 in The Rolls-Royce Turbomeca RTM322 turbo-shaft engine
the Westland Lynx helicopter and used a hydro- was developed at the beginning of the 80's. Engine
mechanical control system. The critical parts were version 01/1 was certified in accordance with the JAR-E
designed using safe-life procedures with the overall life-
 2-13

regulations. It is also qualified in accordance with the UK components for the ENSIP process but the procedure for
DEF STAN 00-971 for the EH 101 and MERLIN the FMECA is actually described in MIL-STD-1629.
helicopter applications. The RTM322 is to be mounted on FMECA looks at whether or not a component failure will
the NH90 (JAR-E) and APACHE (DEFSTAN 00-971) jeopardise safety or completion of the mission, and
helicopters. It includes a FADEC with a twin channel accumulates the risk of failure at the component level so
electronic engine control. Based on the safe-life concept, that it can be evaluated on a system basis. The effect of
the low cycle fatigue life counting for the rotating parts is the more restrictive classification procedure in ENSIP
calculated by the helicopter control unit using a rainflow varies from application to application, with single versus
algorithm. multiple engine applications showing the most striking
differences. An example of the level of components
7.3.3. UK classified as fracture critical is 59% for the single engine
Variants of the Rolls Royce Gem engine (from 1984) in application against 21% for a similar multiple engine
the Lynx, WG 30 and Augusta A129 aircraft were fitted application. Interestingly, the rotating components and
with a digital electronic control system. The critical parts cases named in FAR 33 and 5007E will be found as a
were designed using safe-life procedures. subset of the MIL-STD-1783 classification and will be
considered fracture critical.
The Augusta A129 helicopter incorporates a central on-
board computer system that carries out several avionics 8.2. ECONOMIC LIFE DECISIONS ON A
functions. Since all of the necessary signals are available COMPONENT OR MAJOR STRUCTURES
in suitable electronic form, this central computer is used BASIS
to carry out the engine usage monitoring function.
Under the ENSIP philosophy a classification is forced on
In the new JAR OPS regulation, single or twin engine, every component. This places it into one of the categories
class 2 helicopters are no longer allowed to take off in described earlier. A fixed period of unrepaired service,
hostile areas. In some countries, exceptions may either safety limit or inspection interval, for safety and
temporarily be accepted provided that precautions, mission critical components is required. Special
including automatic cycle counting, are taken to keep the procurement and tracking procedures are applied
probability of in-flight shut down during the exposure according to the durability-critical classification level of
time lower than a specified value. each component. This process forces appropriate
maintenance actions down to component level. Flight
8. SIGNIFICANT ISSUES OF COST AND safety and mission capability are the paramount concerns,
SAFETY while cost is emphasised to a lesser degree in this
approach to maintenance actions.
The transport, fighter and helicopter applications are each
influenced by durability requirements. The impact of For those following safe-life procedures, components are
different durability standards is discussed elsewhere in not classified in as much detail. When maintenance is
terms of the following. required, all parts are not necessarily disassembled and
inspected. In this case, inspections are normally
• Parts classification procedures; conducted only when the aircraft, engine, or component is
• The on-condition maintenance approach versus a available for disassembly. Components are only replaced
more intensive inspection and preventative once a minimum fatigue life is reached and an overhaul
maintenance. required. Components can also be classified with
different life limits depending on whether they are in the
8.1. PARTS CLASSIFICATION DISTINCTIONS hot or cold section of the engine. This influences the
Varying levels of classification exist between period at which a component or engine module is brought
specifications, which include: in for overhaul and inspection. Frequently, the hot section
has half the life-between-overhaul capability requirement
• (ENSIP) which produces five categories of parts of the cold section. While flight safety is the paramount
(safety, mission, durability, durability non-critical, concern, cost is also an important consideration. Because
other) depending on the impact to the aircraft and its inspection facilities are not required in the infrastructure
performance, and cost. as part of the lifing policy (unlike ENSIP), their non-
• MIL-STD-5007E which considers rotating availability may influence maintenance decisions that
components and casings. These parts are further directly affect safety. For example, a lack of fleet
differentiated as hot parts, cold parts and critical management tools may lead to optimistic decision making
parts. because adequate information is not available.
• FAR 33 requires that all parts, such as disks, that
would be uncontained in the event of a fracture have 9. SUMMARY
life limits published in a table of limits that is
Design practices that emphasise durability and the advent
approved by the FAA. Parts, such as shafts and seals
of electronic controls have improved the reliability of
that would be contained but would cause significant
helicopter, transport, and fighter gas turbine engines.
secondary damage also have life limits.
These classifications have worked successfully for their Developed as a response to engine structural problems,
intended applications. A failure mode effects and damage tolerance and safe-life practices arc part of wider
criticality analysis (FMECA) is used to classify specifications to ensure engine durability. The advent of
2-14

these improved design methods coincides with the Disadvantages of the safe-life method include:
introduction of electronic engine controls, which have
Fewer fleet management options;
contributed to improved engine reliability and safety, by
providing better data to usage monitoring systems. There No focused inspection process available if cracking
has been a continuing effort, since durability problems problems develop;
were recognised, to improve engine design, test, and The premature retirement of 999 out of 1000
maintenance activities. components.
The need for readily available replacement parts.
Safe-life design procedures practised by commercial
engine manufacturers parallel those of the military but the Modifications to the safe-life practice offer opportunities
associated maintenance procedures can differ greatly. to extend the life of components through fracture
Generally, when a gas turbine design is originally mechanics, after a minimum safe-life is reached.
developed for the commercial sector and later adopted for Unfortunately, the margin is generally smaller than for a
military use, it requires: component that is designed to the damage tolerance
Additional analysis; standard. Damage tolerance design practices have these
Modified maintenance practices; advantages:
More focused inspection requirements; • Increased use of parts beyond LCF limits may be
Mission assessment and testing; permitted, subject to regular inspection;
Redesign to improve durability. • More thorough testing and verification of designs;
• It provides a more readily available, focused
These requirements are especially true when changing inspection capability to address cracking problems.
from a safe-life to a damage tolerance approach. Damage
tolerance practices do offer advantages over safe-life The disadvantages of imposing damage tolerance
methods in terms of life cycle cost but may be difficult to practices include:
implement by some military services because of the • It is more costly to implement than the safe-life
logistics involved in both facilities and inspection process;
equipment. • It adds weight to the design;
• It requires a larger infrastructure;
10. CONCLUSIONS • Parts' handling is increased
Improved durability design practices and the introduction • It may dictate the only viable management system for
of electronic engine controls to gas turbine engines introduction of mature fleets.
coincide historically. 'Safe-life' and 'Damage Tolerance'
design practices are the most commonly followed Commercial engine designs have been certified to a
procedures to ensure reliable engine design. Either military ENSIP standard but this practice carries extra
method can be chosen by the agency procuring the cost, such as:
design. The decision is usually based on a number of
Additional analysis;
factors such as:
Modified maintenance practices;
Logistics; More focused inspection requirements;
Infrastructure investments; Mission assessment and testing;
Common operational procedures; Redesign.
Common maintenance procedures;
Common training; 11. RECOMMENDATIONS
Contracting requirements;
Recent experience. Neither safe-life nor damage tolerance, is right for every
agency on every occasion. When choosing a design
Based on the number of engines reviewed, the safe-life practice, consider the following:
method is more widely practised than the damage • Fleet utilisation;
tolerance method. However, the competitive nature of the
• Access to inspection facilities;
engine market place is driving all new engine designs
• Life cycle cost comparison (i.e. safe-life versus
towards a damage tolerance approach for highly stressed
damage tolerance approaches);
components.
• Choose the best approach based on a valid
Comparison of the safe-life and ENSIP based processes comparison.
shows that each offers advantages and disadvantages to
the services that practice them. The advantages of a safe-
life practice are:
12. REFERENCES
Kuhlberg, JF, Battle, RG, Glomb, WL, FADEC: A
• The emphasis on the least maintenance possible; Continuing Technology Success Story, The Australian
• Maximisation of 'on wing' time without inspection; Aeronautical Conference, Melbourne 9-11, October 1989.
• A decrease in the facilities and equipment needed to
conduct maintenance inspection actions. Ogg, JS, Engine Durability & Damage Tolerance
Assessments, Proceedings of the 1987 Aircraft/Engine
 2-15

Structural Integrity Program (ASIP/ENSIP) Conference.

MIL-E-OO5007E, Military Specification for Engines,
Aircraft, Turbojet and Turbofan, General Specification
for, Prepared by United States Navel Propulsion Center, 1
September 1983.
MIL-STD-1783, Engine Structural Integrity Program
(ENSIP), Prepared for the United States Air Force by
Aerospace Industries Associates of America, Inc, 30
November 1984.
Taylor, JWR, Combat Aircraft of the World from 1909 to
Present, GP Putnam's Sons, 200 Madison Avenue New
York, NY 10016, 1969, pp 458, 496-498.
 3-1

Chapter 3
Maintenance Policies and Procedures
by
(0. Davenport)

Page

1. Introduction 3-3
1.1. Engine Classifications 3-3
1.1.1. Fighter Engines 3-3
1.1.2. Transport Engines 3-3
1.1.3. Helicopter Engines 3-4

2. Commercial and Military Differences 3-4

3. Maintenance Policy & Support Costs 3-4

3.1. Field Studies 3-4
3.1.1. RAF Tristar - 500 3-5
3.1.2. UK Engine Monitoring Retrofit Working Group 3-5
3.1.3. F-16 C/D with F100-PW-220 3-6
3.1.4. PW2000 Approach 3-6
3.2. Damage Tolerance Cost/Benefits 3-7
3.3. Reliability Centred Maintenance 3-7
3.4. Life Management Plan 3-8
3.5. Maintenance Planning Working Group 3-9

4. Physical Basis for Inspection and Reuse 3-9

4.1. Safe-Life 3-9
4.2. Damage Tolerance (Fracture Mechanics) 3-10
4.2.1. Non-Destructive Inspection 3-11
4.2.2. Inspection Requirements for Damage Tolerance 3-11

5. Usage Monitoring in Life Determination 3-12

5.1. Fleet Usage Monitoring 3-12
5.2. Parts Life Tracking 3-12

6. Retirement Strategies 3-12

6.1. Retirement at Predetermined Intervals 3-13
6.2. Retirement for Cause 3-13
6.3. Inspection 3-13

7. Damage Tolerance for Risk Management 3-13

8. Overview of Current Maintenance Strategies 3-14

9. Recommendations 3-14

10. References 3-14

3-2
 3-3

1.1. ENGINE CLASSIFICATIONS

1. INTRODUCTION
For our purposes, engines may be classified into three
There is a multitude of maintenance policies and main groups: Fighter, Transport and Helicopter. These
procedures that may apply to the management of fracture have different operating regimes and weight sensitivities.
critical rotating parts in aero gas-turbine engines. Frequently, the greater the potential gains from a life-
Maintenance policy is influenced by: usage monitoring system may be the greater the difficulty
in finding the weight budget, and the space. For
• The era in which the engine was developed; psychological reasons, the weight sensitivities sometimes
• The sophistication of the aircraft in which the increase dramatically when additional weight in the form
engine is installed; of a life monitoring system is proposed. Because the
• The information technology base of the operator; operational requirements for similar aircraft types also
• Design analysis improvements by the engine vary markedly between operators there are likely to be
manufacturer; small but significant differences between the maintenance
• The influence that the operator has, with the aircraft policies of different operators with the same aircraft and
and engine manufacturers; engines. An example of this diversity may be found in the
• The inherent life of the engine components. European Tornado aircraft. The UK records data on a
very small sample of aircraft, Germany monitors the
There does not appear to be a single solution to this most whole fleet, and Italy is changing from no monitoring to
complex situation and it is incumbent on the using the German OLMOS system.
manufacturer, airworthiness authority and operator to
select appropriately from the broad array of possible 1.1.1. FIGHTER ENGINES
solutions. The challenge is difficult in an era of declining Today's fighter engines have undergone remarkable
defence budgets and the dramatic expansion of improvements in the operability of the engine across the
information and sensor technology. Still, this decision flight regime. Most aircraft now have unrestricted throttle
will significantly affect the cost, safety, and readiness movement with stall-free operation at all points of the
attributes of the weapon system. flight envelope as a standard feature. They also have
slam-acceleration times of a few seconds, rather than the
There are three primary criteria for the success of a 30 to 60 seconds of the earliest operational gas turbines.
maintenance policy for fracture critical rotating engine This results in the potential accumulation of many types
components: of cycles resulting from throttle changes. For many
• It must ensure the airworthiness of the aero engine current engines usage is measured parametrically on an
by ensuring that the parts are removed from service equivalent or reference cycle basis. For USA engines a
in time to prevent catastrophic failures; common parameter is "Total Accumulated Cycles"
• It must be affordable; (TACs). These may relate to throttle movements or spool
• The engines must provide the combat readiness speed, and are defined in Appendix l. Some modern
needed to support the national security strategy. engines also limit the effect of rotor transients by
modulating thrust using the variable vanes, exhaust
The first is clearly achieved when there are no nozzle position and fuel flow to temporarily maintain a
catastrophic mishaps that are caused by this class of constant rotor speed. While still in use in many NATO
components. The second criterion is much more obscure. countries, older engines have operating limitations that
How does one know when the lowest maintenance costs may limit throttle movements. Therefore, these engines
have been achieved? Because of its wide influence and do not experience the same myriad variety of cycles and
the repercussions of any serious shortcomings combat cycle types as the more modern engines.
readiness, in many cases, overrides the drain on the
national treasuries. This means that affordability is often 1.1.2. TRANSPORT ENGINES
expressed in the availability of resources to purchase, Turbine engine powered transport aircraft may have
stock, and issue replacement parts and components. either turboprop or turbofan/turbojet engines. In general,
When the policies become unaffordable, readiness and the age of the system determines the engine maintenance
sometimes airworthiness may be degraded. policy. Transport aircraft tend to fly with far fewer
Older systems are generally maintained in accordance throttle movements than fighter aircraft. However, this is
with one of the following policies: not always true. The following cases are possible
exceptions:
• Fixed hourly or cyclic interval;
• On-condition through management of turbine • When automatic landing systems are in use a large
temperature margin either calculated or measured; number of throttle excursions may be used;
• A more complex system of data recording and • Tanking (in-flight re-fuelling) manoeuvres;
analysis. • Low level freight drops.

Newer systems may use any of the above and additionally Turboprop engines are typically designed to be
may include damage tolerance techniques for lifing or life maintained either on-condition or at a fixed hourly
extension. interval. Typically, on-board condition monitoring and
life measurement equipment is not installed in transport
aircraft, although the UK VC10 and TriStar fleets are
3-4

notable exceptions. practices can vary, and here it is up to the manufacturers

and the countries to which the engines are sold to arrive at
1.1.3. HELICOPTER ENGINES the best process. Exchange of information in the form of
Helicopters have been fitted with turbo-shaft engines for maintenance orders from the parent country in which the
many years. Indeed, the installation of gas turbines into engine was designed and manufactured and the country
helicopters transformed their military effectiveness in the that purchased it are the rule.
1960s. Again, the age of the system and the design Usage of the aircraft is another variable. Countries that
requirements determine the engine maintenance policy. are smaller geographically tend to have more severe
Turbo-shaft engines are typically designed with cyclic- missions (more low level activity with heavy weapons
life limited components and are maintained to a hard-time loads and little or no cruise time to engagement). Thus,
hourly limit. Numerous rotating components are life reduced durability and an increased frequency of
limited due to the rotor-speed related stresses that they maintenance action may result from introducing the
endure. Power turbines in turbo-shaft engines are engine into a harsher environment than the one for which
effectively life limited by thermal cycling because they it was designed.
operate at a constant rotor speed for power settings above
idle. However the power turbine disc and shaft still 3. MAINTENANCE POLICY & SUPPORT
experience one major LCF stress cycle per flight, and
COSTS
torque levels vary frequently. Special considerations for
helicopter engines are: The general relationship between the maintenance
strategy and the operating and support costs is simple.
• Torque-matching of engine power outputs to protect
helicopter gearboxes; • The acquisition cost of the system accounts for
• Multiple engine control system adjustment to ensure approximately half the life cycle cost;
that one engine does not use excessive life, compared • The maintenance and repair costs account for the
to the others; remainder.
• The 'automatic' engine throttle movements resulting
from Power Turbine Governor system operation, to Many factors affect costs both directly and indirectly.
keep the helicopter rotor speed constant despite Design choices directly affect the acquisition costs, and
collective and cyclic pitch inputs from the pilot. are a significant factor in the maintenance and repair
costs, especially when viewed on a cost per flight hour
2. COMMERCIAL AND MILITARY
basis. The inherent reliability of a design is demonstrated
DIFFERENCES by the initial capability of the machine. Maintenance
The operational mission that commercial engines are practices and policies affect this reliability once
subjected to is much less complicated than the military maintenance actions begin. If the performance is not
cycle. This difference is best illustrated by the operational restored during maintenance, lower reliability and higher
cycle of the military F117-PW-100 engine, contrasted operational costs are the inevitable result. If the
with that of its commercial cousin, the PW2037 engine. performance is improved following a maintenance action,
The Fl 17 experiences one or more major cycles with then this is almost certain to be due to the correction of a
each practice touch and go, cycles due to airdrops, cycles deficiency in the original design.
due to use of thrust reversers for backing and turning, etc.
The cyclic accumulation rate is about 2-3 times the 3.1. FIELD STUDIES
PW2037 on a cycle/hour basis. Clearly, adapting a civil There appears to be an optimal relationship between the
engine to a military application can require design material replaced at each shop visit and operating and
modifications, increased inspection requirements, and support costs. Commercial experience with CF6-80A
alternative maintenance approaches. An exception to the engines on the A310 aircraft has conclusively shown that
above is that engines on regional jet transports operating and maintenance costs are optimised by
accumulate cycles faster than similar engines in long minimising the cost per flying hour, rather than shop visit
range endurance such as reconnaissance or maritime and material replacement costs. Figure 1 shows a typical
patrol applications. relationship of material cost per shop visit versus cost per
engine flight hour. This was achieved in practice by
Commercial operators tend to see to the care of their own synchronising maintenance so that a complete engine
engines once they are purchased from a manufacturer. with accessories is placed in service as a unit. This
Therefore, the overhaul and repair facility capabilities, practice, rather than making module changes on the wing,
and the quality of the approach to maintenance vary has produced dramatic reductions in operating and
greatly. Usage of the engines and aircraft also vary as to maintenance costs.
the geographical region served and utilisation (short
versus long flights). Government regulations do enforce a This finding is of immense importance to all operators,
degree of uniformity in maintaining engines and their safe and efforts should be made to determine whether a similar
operation but each carrier is a separate entity, and relationship exists for all fleets. Operators with older
responsible for its own actions. engines that are changed as complete Engine Change
Units (ECU) will already be following this practice, and
In the use of military engines and their maintenance by should thoroughly investigate the real cost of changing to
different countries, there are some similarities to the a modular engine philosophy, should this be proposed.
commercial situation. Again differences in maintenance
 3-5

These benefits are derived from a harmonised Preservation of the integrity of the structure from
maintenance schedule that incorporates both life limited undetected damage due to operational incidents;
components and performance limited components that are Protection of the Controllerate of Aircraft (CA)
typically in the engine core i.e. compressor and turbine Release fatigue usage assumptions.
flow path components.
Figure 2 shows the
potential for engine Work Scope - Value Relationship
removals when the
maintenance is not
$400
synchronised. It is easy
to observe from figure
2 that a maintenance $300
program that aligns the
module maintenance tperEFH
requirements - life 200
limits or performance -
offers the greatest
potential for reducing $100

operating and support

costs. 1
$300K
1
$400K $500K
CF6-80A
3.1.1. RAF Net Material Cost Per Shop Visit A310 Data
TRISTAR -
30%Increase in Mat'l Cost/Shop Visit = 55%Lower Cost/EFH
500
Figure 1 - Optimising engine flying hours cost
The RAF acquired six
Lockheed Tristar -500
aircraft in 1984. These
had a usage monitoring Fan --X- X X
facility fitted. This Comp -X- -X--_,
system is much more
capable than a
Comb
dedicated engine Turb x- X X
component usage Sys
monitoring system.
Benefits reported from
QEC X X
T/R -X
the system, after
X
extensive analysis of
several years service
were classified as
f
New or Overhaul
>Each Module independent
»Doesn't synchronize module life
Time

economic, and safety.

>Leads to more removals than RCM
Economic benefits
Figure 2 - Synchronising module and engine lives
accrued from the
following. When multiplied to represent potential savings in a large
• Avoidance of secondary damage, due to detection fleet these savings represent many million pounds per
of incipient engine failure; year.
• Avoidance of unnecessary removals;
3.1.2. UK ENGINE MONITORING RETROFIT
• Avoidance of maintenance actions and service
WORKING GROUP
disruptions;
• Fuel savings due to improved variable inlet guide An extensive study report by the UK Engine Monitoring
vane adjustment; Retrofit Working Group on the benefits of fitting engine
• Planning of engine changes to occur at base. life counters included the following conclusions:
A conservative estimate of savings, that excluded any • Lifing practice was investigated and it was
manpower considerations, showed a reduction of £72 per concluded that for future implementations Thermal
aircraft flying hour or £24 per engine operating hour at Transient methods are the only methods acceptable
1984 prices. if full Predicted Safe Cyclic Life (PSCL) is to be
Safety benefits accrued from the following. achieved.
• The cost benefit studies undertaken indicate that on
• Preservation and protection of the margins in the Hawk aircraft fleetwide monitoring will produce
operational envelope of the Tristar; significant cost savings from parts and labour.
• Preservation and protection of the margins in the Management benefits are more difficult to quantify
structural lifing of the Tristar; but it is likely that examination of Minimum Issue
3-6

Service Life (MISL) policy could give rise to the initial product feasibility study or definition up to its
further cost benefits. retirement; the same process is performed to analyse any
design change, life or performance improvement
The Working Group considers that the fleetwide fit of requirement, mission profile variation, support and
engine monitoring units is supported on safety grounds maintenance concept variation. Figure 7 shows, under the
alone. The statistical analysis undertaken as part of the support activity, the recurrences of the Life Support Costs
Harrier/Pegasus data study indicates that small sample versus Cost of Ownership analysis in the various
monitoring is inherently inaccurate. Application of this programme phases.
statistical method on Hawk Adour and Tornado RBI99 The above analysis originated the pie chart of figure 8.
data indicated that considerable factors would need to be This represents a product for military application, and
applied to average exchange rates derived from small highlights that inside the system Life Cycle Cost, the
samples. Independently, recent agreements on Adour and relationship between Acquisition Costs of a System and
RBI99 data reflect the uncertainty of the small sample its Operations and Support Costs is generally one to one.
methodology, which has resulted in a 30% worsening of
the exchange rates. The other LCC elements, Development and Retirement &
Disposal Costs, are subject to various deviations
3.1.3. F-16C/DWITHF100-PW-220 originated by technical and design solutions, material
The result of a specific on-condition-maintenance utilised etc, that can largely influence their weight on the
program that focused on restoring functionality at the Total Life Cycle Cost.
right time is shown, in figure 3, for the F100-PW-220. Note: In the figure 8 piechart, acquisition cost includes
When this engine entered into service the Mean Time research & development, production investment, and
Between Repairs (MTBR) approached 800 engine flight initial support investment. This is not applicable to
hours, and the operation and supports cost were
quite low. As the engine aged and the component
failure rate increased, maintenance was Projected Operational Cost vs
performed at the module level without regard for Time Between Engine Removals
the overall engine performance and coherence.
Consequently, the operator found that the $900
\ Current MTBR
reliability decreased with a commensurate
increase in maintenance costs.
A programme, which incorporates the best
u.
UJ
$750

$800
~\s \. Optimum cost effectiveness
practices from the commercial fleet, to manage
the life of the engines more effectively is Q.
$450
X. /
currently underway in the USAF. Engine life o
$300
^ JLL ■—■""""""
management is a high payoff, relatively low cost o $150 -
process that significantly improves both
readiness and supportability. 0 200 400 £00 800 1000 1200 1400

3.1.4. PW2000 APPROACH EFH Between Removals (MTBR)

The PW 2000 modular maintenance approach of Figure 3 - Minimising Operational Cost for the F100-PW-220
the mid 80s, figure 5, was evolved due to
commercial customer demands for improvement commercial products where acquisition costs are market
in time-on-wing, reliability, performance and operating driven and recovered via specific commercial agreements
costs. The need for an Engine Management Plan (EMP) (Flight Hour/ Shop Visit fees etc.)
which optimised engine-management practices based on
service experience and development efforts was The experiences, gained in the past via several Cost of
established. The EMP embodies a 'whole engine' systems Ownership Analyses, have highlighted that the number of
level (RCM) approach to maintenance, which is designed variables to be considered in the military applications is
to optimise maintenance costs per EFH. The more recent greater than in commercial ones. In fact, the typical
Fl 17 takes full advantage of this commercial experience military approach to "System Effectiveness" may drive
as well as previous military lessons learned on the TF39 Product Cost Effectiveness to be "Optimised" instead of
and other engines. An EMP has been established for the "Minimised".
Fl 17, which weighs all pertinent engine data in order to Figure 9 shows the relationships and dependencies
make the best maintenance decision, figure 6. From this between Life Cycle Costs and Systems Effectiveness. An
plan, a customised work package is established for each example of the application of this philosophy is shown in
Fl 17 returned for overhaul. The Fl 17 EMP also takes Figure 10.
into account the current spares level, shop load and
economic factors in establishing the detailed build The final aim of the study was to compare the influence
requirements for each engine. on Operations and Support Costs of three different
Engine Thrust settings (performed via Engine Control
This approach is not limited to new products but is Unit) that were requested by the operator to expand the
performed for all products in operational use, both licence mission profile flight envelope.
and own development. The philosophy is applied from
 3-7

The result is a 'Product Handling

Programme' that provides utilisation $#>_--- PW 2000
'windows' so that the operator can control Engine Maintenance Evolution
costs or performance according to current
needs. Evolution of tire PW2000 WtoleEiigiiie (Systems) Mobiteiunc« Appioacli
Moduli
UjinUnjnc
19_M
3.2. DAMAGE TOLERANCE PW20Q0 ENTRY WTO SERVICE

COST/BENEFITS 1989

A comparison of the basic differences

between MIL-STD-5007E, a 1990 ENGINE MANAGEMENT PLAN(EWP) A PROCESS TO GATHER AHD DISTRIBUTE FLEET
EXPERIENCE ANOPW RE COMMENDATIONS.

predominately safe-life approach, and

1991
MIL-STD-1783, a damage tolerance NEED: IMPLEMENTATION OF EMPENGINE MAINTENANCE RECOMMENDTIONS

approach is made in chapter 2. Fracture 1991 COMMERtCAL OCgELOPEDNEW OR SUPPLEMENTED EXISTING ENGINE MAKTENANCE

mechanics material characterisation, parts PIAS

classification, and sub-system damage 1992

PWCUSTOMER DEVELOPED PVCO00 WORKS COfE
tolerance verification testing create extra
1996
effort in developing the damage tolerant Whol« Engina
R17-PW100 CUSTOMIZED EMP
(System) MMntftninct
design. These efforts include:
Figure 1
• More refined finite element analysis
Figure 4 - Modular maintenance example
to define stress states for crack
growth analysis;
• Crack growth life computations;
r
• Crack growth characterisation of =—~= -
materials;
• Verification testing of a few F117 Approach to Effectivi
F117EHM
prototype parts;
N.
• Increased inspection costs (focused
FPI and eddy current) on each ^ Peftinent

engine;
MOB Test Cell F117Data
V-
• Inspection facility acquisition.

As discussed elsewhere, very large safety

- UAPNCT

G0S1

• Cnginc History
}/ F117
Individual
Whole Engine
Workscope
\

/
1 .

and operational benefits go along with Cycle, Time

Generic
/
these increased costs. The primary result Woikscope
being the virtual elimination of LCF F117CEMSIV

failures in the fleet. Not all components • Trending Report? Economically

Refurbished
are defect free when manufactured. The
anomalies can cause premature failures / Engine
F117EMP
and a safe-life approach does not
recognise these defects. Further, if RFC is Fi gure 5 -Eng ne man igement plan
used for the life management approach a
large life cycle cost savings may be
achieved over the life of the engine. ~~ v
Operational safety is the primary
justification of the ENSIP costs. EFFECT WE MAINTENANCE MANAGEMENT.
Assesses all pertinent engine data

3.3. RELIABILITY CENTRED EHM TEST CELL

INSPECTION
LINE MAINTENANCE REPORTS
PBnrcn IRCS
MAINTENANCE ■ CMGINC PCRFBASC - BORCSOOPC - CRCWSO.UAWKJ
-TRCKD PLOTS
■ FLCCTRANKM ■ DCTAIL PCRF INFO 1 - CDDYCURRCMT /
The reliability centred maintenance • FUSCQNG
- MAIWTCNANCC ACTIONS
■GROUND RUN*
(RCM) concept was first employed in the ENGINE WRKSCOPE 1 K
OiFCRiCNCC - VISUAL IMSPCCTOM»
early 1980's. The philosophy is based on TF«M nFTFBMIMATinN 1
• THOUBl£*HOOTIMG RCtULT*
an examination of the operational CUSTOMER • »PARC**miAT>OM
DEVELOPED - SHOP LOAD
consequences of failure for each WORKSCOPE ■ COOMOMKS
component. If the consequences of failure 1
are unacceptable, a maintenance task is EMP \
established for the component. The result "BEST" MAINTENANCE DECISION

is a program that achieves safety and - GCMCRALCC OK 1MIC* - INOIUDUAL'WHOLC CNGINC V OR (PACIUCC
■ DCTAILCDBUILDBCQUIBCMCI )T E
- OPERATION [3<P CRICMCC
operational goals while minimising costs, - MODULE INTI auCTION
maintenance man-hours and material
usage. RCM uses engineering decision F'igure 6 - Fl 17Inc iviciual work packaging
logic to evaluate each failure mode and
3-8

mathematically determines the risk over the life of a can benefit from a life management plan that lays the
system or engine. basis for the process appropriate to those engines. Typical
life-management plans provide a structured, logical
In general, the RCM analysis process begins by progression of activities that integrate the design,
determining if a component is structurally or functionally development, manufacture, use and repair or upgrade of
significant. A separate RCM decision tree exists for each an engine throughout its life. Such a plan will typically
classification of part. This analysis is normally begin shortly after the beginning of development and
accomplished as part of the Failure Modes and Effects include:
Criticality Analysis. The second step is to determine the
failure consequences. The consequences are classified as: • Design Maintenance Concept
• Field Maintenance Capability
• Safety;
• Depot Maintenance Capability
• Economic/operational;
• RCM Analysis Results
• Non-safety hidden
failure; ^ PHASES DEFINITION DEVELOPMENT PRODUCTION PRODUCTION SUPPORT RETIREMENT«.
• Safety-hidden. ACTIVITIES^.
PROPOSAL
(A)
(B) INVESTMENT
(C)
(D) (E) DISPOSAL
(F|

After the failure consequence DEFINITION (0|

PROPOSAL
is determined, actions to avoid
subsequent failures are DEVELOPMENT (1)
x
proposed, and evaluated for PRODUCTION
applicability. The most INVESTMENT (2|

appropriate is chosen. Cost- PRODUCTION |3|

.-.-.v.-y^.
benefit analysis may be
conducted to determine the SUPPORT (4) ■■s ■"'•'■*'
; ■ \.' \\\

optimum mix of new parts RETIREMENT &

DISPOSAL (5)
(old design), new parts (new (PHASED-OUT)
design) or repaired/reworked
parts. The maintenance Figure 7 - Recurring cost of ownership
options are:
• Do nothing;
• Redesign; Typical Life Cycle Cost
• Perform preventative
maintenance tasks. «~. Development
17%
The preventative maintenance \ ^0$M:MW^»W^»^^^
tasks include & SyppcR /
servicing/lubrication, on-condition '"* k
inspections, hard-time rework or
discard, and a combination of the
previously mentioned options. ft
i 11
\^
After applying RCM, the US ^"~~-~~-^. -iil ^|».-,...;...v ■ _„ i\ y..;;;vAjjv -,;. ■■■\-. ' ■- '^„.K*^-*"

Navy classifies turbine engine life •arnS in&.äJ;nvssim§r4

limited parts as structurally 41^

significant items, which are Figure 8 - Typical life cycle costs

managed through the preventive
maintenance task of hard time
discard actions, based on a safe-
life limit.
A more complete explanation of
the RCM analysis process is
documented in ATA 143 (MSGIII
SYSTEM
Recommended Maintenance
Practices) and MIL-STD-2173. EFFECTIVENESS
This document should be
thoroughly reviewed before
implementing the RCM process in 8UPP0RTABILJ1YANALYBJ
any engine program.
JpCE> CD
3.4. LIFE MANAGEMENT
PLAN Figure 9 - Cost effectiveness dependencies
Each engine type, model, or series
 3-9

• Initial Life Estimates components that may threaten operational safety, should
• Age Exploration Plans they deteriorate. The lifing procedure employed to
• Spares Requirement Process establish life limits varies with component type, and the
form of damage. In modern gas turbines, discs and
spacers are normally designed to withstand LCF, burst by
3.5. MAINTENANCE PLANNING WORKING overspeed, and creep. This applies to both the hot and so-
GROUP. called cold sections of the engine as engine pressure
The life management plan will need to be updated ratios and compressor exit and combustor exit
periodically, as new data becomes available from either temperatures rise with more advanced designs. Turbine
advanced analysis tools or field and test experience. Any blades and vanes are designed to withstand creep as well
changes in the maintenance concepts will also require the as thermomechanical fatigue and high-cycle fatigue.
plan to be updated.
Engine Cost of Ownership relative to Thrust options
PHYSICAL BASIS FOR
INSPECTION AND REUSE
300%
Switch
There are two competing strategies for life ra
■o Posn A2 t
c
management, the safe-life approach and the » 250% /
damage tolerance approach. Within the damage w /
c /
tolerance approach, there are two methods: £ 200% /
standard ENSIP, and 'retirement for cause'. 3 /
O /
The life defined in hours using the safe-life u>
3
150% /
c01 /
approach is often called 'hard time' or 'hard /
> 100% /
life'. It is based on a calculated or demonstrated Switch
life for a nominal component, given an estimate / o 'Posn Al
in /
of the life history and material properties. *•*IA 50%- /
o Switch
Conservative factors are applied to the material o iPosn A0
properties or the demonstrated test life to allow n 0%-
1 ,--1 l 1
a safety margin for the minimum-material- _ -10% "■" 0% 10% 20%
Q
property component and extremes of operation. -50%-

Engine thrust setting options

The damage tolerance approach is dependent
on the component being designed to have a Figure 10 - The incremental cost of thrust
relatively long crack growth (residual life)
period, between a detectable flaw and rupture.
This period determines the basis for the interval at which
inspections must be performed, to ensure that the
component is still serviceable, and to retire components 50
(0
that have flaws. The standard ENSIP method sets o
inspection intervals, but the ultimate retirement life is still 3
a "hard time" life as detailed above.
-3 sigma / \
TO
LL
The "retirement for cause" method includes damage
tolerance analysis to set inspection intervals. However,
£ / \
retirement life is based on repetitive inspections until a
flaw is found, or the part is retired for economic reasons.
0 s V
At some point in the life of a component, economics will 100
dictate replacement of a component based on its "hard
% Nominal life
time". The mature engines addressed in this report, that
were designed and developed prior to 1980, are managed Figure 11 - Residual fleet life when first component
to the safe-life approach except for the F100 engine that is retired
uses damage tolerance.
The life limits for these components are initially based on
Two types of flaw may exist. These are surface cracks estimates of in-service damage accumulation rates. The
and buried flaws. Surface cracks are more easily detected limits are later revised when field experience has been
by NDI methods. They may have a different inspection accumulated. For discs, the most common procedure for
methodology and periodicity from buried cracks. establishing safe life limits follows a 'time or cycles to
Manufacturing processes and defect reduction techniques crack-initiation' criterion. For turbine blades and vanes,
such as hot-isostatic pressing are the primary controls to life limits may be prescribed by engine designers. In
reduce the frequency and severity of internal porosity particular, turbine blades and vanes in aero engines are
(precursor to cracking) in blades and disks. seldom lifed because of the difficulties associated with
predicting the service behaviour of metallurgically
4.1. SAFE-LIFE
complex material systems under conditions that can vary
Engine designers' work hard to establish life limits for widely with user practice. Furthermore, in the case of
3-10

coated components, these difficulties are made worse by the cracks growing sufficiently slowly during service to
coating substrate interactions that are not usually well allow their growth to be reliably detected, and if
understood. A 'life on-condition' approach is sometimes necessary, monitored through regularly scheduled
employed, for instance, where creep growth or untwist of inspections. In practice, crack growth has only been
airfoils is measured and distortion limits are used as 'monitored' when it has been necessary to overcome a
retirement criteria. shortfall in the required service life. It also appears that
the 'monitoring' is usually restricted to a life extension
As indicated, rotating parts such as compressor and for components that do not have a detectable crack.
turbine discs are usually life-limited due to LCF damage
accumulation. The most common lifing method employed In their most elementary forms, these alternative damage
for these parts follows a 'cycles to crack-initiation' tolerance based lifing procedures, known as Life-On-
criterion. A minimum life capability is statistically Condition, Retirement-for-Cause or simply Fracture
defined for simulated service conditions through Mechanics lifing, assume the following.
extensive testing. This normally includes standard coupon
• That the fracture critical locations of a component
testing and component test verification in spin-pit test
contain crack nucleation sites of a size that lie just
rigs. The statistical minimum is usually based on the
below the detection limit of the NDI technique used
probability that 1 in 1000 components will have
to inspect the component;
developed a detectable crack (typically chosen as a crack
0.8 mm long). • At some point in the life of the component it is
assumed that a crack may have started, and regular
The approach has been criticised as being overly inspections are begun;
conservative and costly on the ground that disks are • The crack then initiates, and grows during service in
usually discarded with significant amount of useful a manner that can be predicted by linear elastic
residual life. There are two major concerns when a cycle- fracture mechanics, or other acceptable methods;
to-crack-initiation rejection criterion is used to life • All cracks above a critical size will be detected, and
rotating parts. The first one, by implication, is that 99.9% monitored;
of the components will be retired before any detectable • When the cracks reach a predetermined size the
crack has formed. These parts may have a much longer component will be retired. In some cases, this will
service life before they develop a 0.8-mm crack. be the detection size.
Secondly, the components may be capable of tolerating
crack sizes much greater than 0.8-mm mentioned above. It is clear that the successful application of damage
This limit reflects the sensitivity and reliability of current tolerance based procedures depends on the supporting
NDI methods rather than the mechanical tolerance of a technologies. These technologies include non-destructive
particular part to cracks. inspection, mechanical testing of test coupons and
As can be seen from figure 11 a large percentage of components, structural analysis, mission profile analysis
components have useful life remaining (potentially many and condition monitoring of components. Well-directed
thousands of cycles) depending on the material and usage and extensive materials testing at room and elevated
scatter) when the first component reaches its retirement temperatures must be performed to obtain crack growth-
life. Economical considerations would suggest that rate data, for the application of deterministic and
procedures which allow the use of components to attain probabilistic fracture-mechanics based life-prediction
their individual life rather than the life of the shortest concepts.
lived component would greatly reduce the need for The basic hypothesis that allows a retirement-for-cause
replacement components. Likewise total dependence on approach is that each component, as manufactured,
rejection by inspection for components with long contains inherent metallurgical features provide crack
manufacturing lead times and demand based purely on initiation sites, which must be allowed for in the design
part rejection will lead to readiness issues at fleet lives process. This assumption does not imply that the parts are
approaching the nominal. The secret to cost efficient defective, rather that natural crystallographic features
maintenance policies is to achieve a balance between occur when a component is cast or forged. Acceptance
parts life usage and readiness. and understanding of this fact has been hindered by the
It is quite clear that the maximum life cannot be safely inferences that components from a vendor or engine
extracted from the whole fleet of components unless each manufacturer are "defective" or "flawed". In fact, all
component is considered, and has its life-usage tracked materials have microporosity, lattice vacancies, grain
on an individual basis. boundary inclusions, etc. It is the size and frequency of
these characteristics that separate a "good" component
4.2. DAMAGE TOLERANCE (FRACTURE from a "defective" component.
MECHANICS)
To ensure that the life potential of rotating hardware is The Engine Structural Integrity Program (ENSIP) for the
more fully used, alternative lifing procedures have been USAF takes account of these factors, because it:
suggested. These are based on damage tolerance and the • Recognises the engineering nature of the materials;
application of fracture mechanics principles. The • Characterises the inherent defect distributions;
philosophy behind these alternative lifing procedures • Provides for their inclusion into the design process;
assumes that the component may be capable of continued
• Establishes the level of inspection at manufacture to
safe operation during crack growth. This is dependent on
 assure the control of the processes that generate If'retirement for cause' is used then additional and more
these defects; sophisticated ultrasonic and eddy current inspections will
• Establishes the residual life for each component. be necessary to ensure safety. Finally, it is important to
combine the concepts of "hard time" with "retirement for
cause" to minimise readiness concerns, and maximise
4.2.1. NON-DESTRUCTIVE INSPECTION
safety. This means that all components, regardless of their
The primary constituent of a life management policy crack-life history should have an over-riding 'hard life' at
based on inspection and reuse of serviceable components which they will be retired. In most cases, it is preferable
is to use an appropriate tool from the vast array of non- to retire a fraction of the population before its actual life
destructive inspection processes available today. While is consumed to preclude any in-service critical component
no single method is universally applicable, eddy current failures and to provide production lead-time for
and fluorescent penetrant methods have found wide replacement components. A rough guide would be to set a
acceptance. X-ray, magnetic particle and neutron hard time at the point where a Weibull analysis shows
radiography, etc have practical but limited application. As about 10% of the fleet would be rejected.
can be seen in figure 12, once the initial flaw size has
been determined for the as-manufactured condition an 4.2.2. INSPECTION REQUIREMENTS FOR
interval of failure free operation can be determined. This DAMAGE TOLERANCE
may be for either an actual initial material defect MIL-STD-1783 or ENSIP requires non-destructive
distribution (probabilistic) or assumed flaw distribution evaluations (NDE) of critical components during
(deterministic). For many materials and processes, defect
distributions have been developed from specimens cut up • Manufacturing;
and analysed with microscopes. Assumed initial flaws are • Overhaul;
typically set at the stage 1 facet size based on historical • Field inspection.
observations.
Based on the crack growth characteristics of the material, Designing with inspection in mind, and relating and
environment, and loading, a crack growth forecast is then limiting the (design) period of un-inspectcd service of all
calculated. An interval is established to allow the non- safety critical parts to the demonstrated inspection
destructive inspection (NDI) operation to just miss a flaw capability are parts of the ENSIP process. All safety and
and provide a safe growth interval (one-half the crack mission critical parts receive special inspection, compared
growth life). Subsequent inspections determine whether to only rotating parts and cases under MIL-STD-5007E
such a flaw existed. If it did not, the part is returned to and rotating disks, shafts and spacers under FAR33.
service for another interval. If the inspection detects a Ultrasonic, fluorescent penetrant inspections (FPI), and
flaw, the part is retired. eddy current inspection methods are relied on and
Any inspection process inherently has a finite probability qualified for the three different maintenance levels. NDE
of finding a crack and therefore a finite chance of missing reliability demonstration programs to identify confidence
a crack. It is essential to be able to quantify the limits, based on testing inspectors against known
probability of detection
(POD) to establish a
reasonably effective C
inspection process for r Inspection
specific component critical a Interval
areas, such as bolt holes, c 1/2 Crack Growth Life
broached features, webs and k
other areas. The emphasis of NDI Minimum Retire Part
this approach is on the L Detectable Flailv
largest crack that could be e
missed rather than the n
smallest crack that can be g
found. Inspection methods t
should be chosen with this in h TIME
mind. Eddy current is a
Figure 12 - Defect growth monitoring
primary inspection technique
for areas that rely on small
standards, are required under ENSIP. Verification of the
flaw detection, and for those areas which have high
inspection methods capability at each inspection site
residual compressive stresses. Other methods of
produces a probability of detection (POD) curve.
inspection that are suitable for service use include
ultrasonic and fluorescent penetrant inspection. The US experience is that FPI is capable of finding a 0.035
Reliability Centred Maintenance (RCM) process inch deep by 0.070 inch long flaw with good reliability
addresses the relationship between the probability of for whole field inspections and a 0.020 inch deep by
failure, its progression rate and the severity of its 0.040 inch long crack under focused inspection
consequences, the probability of crack detection and the procedures. This is normally quantified in such terms as a
inspection interval required. 90% probability of detection, with a 95% confidence
3-12

level. and cases of thermal distress that have arisen in fleet

operations was not predicted.
In a response to field related durability problems it is
possible for the safe life practitioners to use enhanced Early and repeated (-3-5 year) monitoring of the fleet
inspections, such as eddy current, however this is not the usage, with simultaneous recordings of significant engine
standard practice for inspecting components. The parameters (rotor speeds, operating temperatures, etc.)
verification and use of enhanced eddy current inspection and aircraft parameters (Mach number, altitude, attitude,
procedures for flaw sizes (0.005 inch deep by 0.010 inch etc.) must be done to identify operational changes. This is
length) is a major difference between using the damage so that timely adjustment to life analysis and the
tolerance process and using the safe-life management subsequent maintenance and logistics provisions may be
procedures. made.

5. USAGE MONITORING IN LIFE 5.2. PARTS LIFE TRACKING

DETERMINATION There are many methods used throughout the NATO
nations for tracking the life of fracture critical rotating
There are several uses for life-usage monitoring components.
equipment. The most basic is simply to prevent the failure
of components in service, and to maximise use for a The simplest is to maintain a paper record of the hours of
given cost. Secondary uses are to refine and adjust the operation of the engine in which the component is
design mission(s), mission profiles, mission mixes, installed. The remaining life of the components is
operating conditions, etc. from those originally assumed recorded as the engine is assembled and then debited at
by the designer. Better knowledge will always bring one periodic intervals as the engine is operated. When the
of the following: stated life is reached, the engine is disassembled and the
component replaced and retired.
• Improved safety, through a reduction in usage;
• Reduced costs, through an increase in life; A more complex, but still simple method accumulates
• Security, in the knowledge that the life limits are cycles by a conversion of operating time and mission to
correctly set. cycles by a fixed algorithm. The conversion factors are
sometimes known as ß (beta) factors. The process is then
the same as for operating time.
5.1. FLEET USAGE MONITORING
It is a common experience for an operator to discover that A more complex method is for an on-board processor to
a new weapons system with a new engine has much more measure cycles and time in certain temperature bands.
capability than its predecessor and for new tactics, roles, This is then downloaded to a ground based computer
missions, etc. to be developed for the aircraft. This is system that maintains the configuration records and debits
especially true for fighter/bomber aircraft and helicopters, the life remaining on the tracked parts. Operational
and occasionally so for transport aircraft, leading to much difficulties may occur with extended deployments away
shorter or longer lives than predicted by the designer. from the ground system and lead to a loss of cycles.
There are many examples that support this. A further enhancement is for the data to be recorded in
The F-16 is a lightweight multipurpose fighter/fighter bulk, and then to be downloaded for analysis on a ground
attack aircraft and is in service in many NATO countries. based system. This allows the full manufacturers design
The aircraft was developed with a version of the Pratt & model to be used for analysis if required. This provides
Whitney F100 engine. In the original version, with the the most technically accurate approach.
F100-PW-200, certain flight characteristics such as range, The final enhancement is for the records to be updated in
payload, operability, etc. were established. When the near real time by the on-board processing system as well
alternate fighter engines were developed, some as maintaining the configuration records. This is
characteristics were changed. With an F110-GE-100 periodically downloaded to ground systems for planning
engine, the number of practice bombing runs was use and avoids the deployment data loss. Usually a
increased due to reduced fuel consumption. However, the reduced version of the manufacturers design model will
Fl 10 was designed using the mission profiles, mixes, etc. be used to extract the major and minor cycles and sum
from the Pratt & Whitney engine and some early them for each monitored stress feature.
problems arose. These were traced to the operational
differences made possible by the better fuel consumption While the objective of these systems is to protect flight
and subsequent changes to the training mission profiles. safety through the retirement of life limited parts prior to
fleet failure, it is unclear that any one demonstrates a
A fleet survey was accomplished, somewhat belatedly, to dramatic advantage in reducing the cost of operations.
identify these characteristics and create a new definition The operating risks, however, change significantly and
of the usage for accelerated mission testing (AMT). This chapter 9 provides a tool that will help operators to
usage definition was then applied to both the increased quantify these changes.
performance engines - F100-PW-229 and the F110-GE-
129. It has proven to be a good indicator for the F16 6. RETIREMENT STRATEGIES
aircraft, but not for the F100-PW-299 engine installed in
the heavier, ground attack F-15E. In this case, the correct Retirement strategies generally fall into two categories;
amount of time at high power settings was not included predetermined intervals and retirement for cause.
 3-13

6.1. RETIREMENT AT PREDETERMINED 6.3. INSPECTION

INTERVALS Developing inspection processes for RFC components is
This strategy is a conventional fatigue design approach a major undertaking. The inspection techniques must be
for part retirement. It is based on low cycle fatigue (LCF) able to find cracks emanating from internal inclusions and
life. The LCF limit is established based on a lower be detectable from any orientation in a finished part. In
bound(l in 750 or 1 in 1000) distribution of crack order to achieve this, new inspection standards, from
initiation times. The use of the lower bound minimises which to measure the adequacy of new inspection
the occurrence of cracking and the need for component techniques, must be designed and fabricated with buried
repair. Implementation of this conventional strategy flaws. The USAF has an inspection system and a set of
requires 100% hardware replacement at the lower bound standards, which were developed for this purpose.
LCF limit. Damage tolerance control philosophies could An alternative is to use RFC for surface flaws only and to
also be combined with this strategy as a means to ensure retire hardware at the buried-flaw inspection life-limit.
hardware structural integrity during the LCF life. This approach will still yield a significant extension in
life over the historic minimum LCF life criteria. To
6.2. RETIREMENT FOR CAUSE maximise the gains it will be necessary to ensure that the
With the previous approach, only 1 component in a buried flaw life of the weakest component in the fleet is
population of 1000 components would typically have controlled by design.
initiated a crack, and the remaining 999 components
would be retired with substantial useful life. By using the It is clear that in applying RFC, the non-destructive
retirement for cause (RFC) methodologies, this useful life inspection capabilities are critical in defining the
can be made available. This because each component is economic and physical inspection feasibility. As shown
used until a crack is found in it - or a predetermined earlier, the crack length determines the residual life.
percentage of components have been rejected and one is Hence, the inspection interval and crack detection is
effectively climbing the right side of the reliability limited by the capability and reliability of the inspection
bathtub curve. Substantial development of fracture system. Inspection capability and economics are equally
mechanics concepts over the last several years makes it important considerations when implementing RFC on a
possible to predict crack propagation rates accurately particular component.
enough to use these techniques.
7. DAMAGE TOLERANCE FOR RISK
RFC methodologies apply the fracture control based MANAGEMENT
philosophy described in section [7.3]: All components
would be inspected first at the end of a safety limit period Damage tolerance analysis may be used, for components
divided by an appropriate safety margin (SL/SM) which operating under safe life methods, for two reasons:
would maintain the equivalent risk used in the • Life extension or the usage of components beyond
conventional design life. It is important to note that in calculated low cycle fatigue limits;
order to maintain an equivalent risk, there will be an
• To manage a life shortfall
increase in the number of areas to be inspected over the
design life safety limit inspections. This is a natural
The use of damage tolerance to extend life limits was
outcome of the part reaching its predicted LCF life, and
discussed in the previous section. With a life shortfall,
increased risk would occur if additional areas were not
crack growth lives are calculated from the capability of
inspected. Only those components containing defects
the inspection method (eddy current, ultrasonic, or FPI).
equal to or greater than a, (initial crack length) would be
A risk management plan is developed to keep risk to an
retired. All other components would be returned to
service with the assumption that, if they have a flaw, it is acceptable level based on:
less than a, and thus good for another inspection interval. • How many cycles the fleet has operated;
This ensures that the crack propagation life is continually • How many failures are likely to occur and have
reset to a safe value, and that components are rejected occurred;
only for cause (cracks or defects). Each component is • When the component in question can be inspected;
allowed to operate up to its own specific crack initiation • How long it can be safely operated between
life (or a limit based on the fleet history of space inspections and before a repair or component
cracking). If a crack is missed at the inspection interval, replacement is required.
another chance will exist to find the crack when it is
larger, without the component failing prior to that Because of ENSIP requirements, the application of these
inspection. Clearly not all fatigue-limited hardware can damage tolerance methods is part of the process and
be handled in this fashion: therefore easier than for a safe life design. Under ENSIP,
• Components must be designed to have a slow crack provisions are made for addressing such problems
propagation rate; through damage tolerance analysis, focused inspection,
• The inspection interval must be long enough not to ease of engine disassembly, maintenance implementation
put undue constraints on weapon system readiness; plans, and risk assessment.
• The cost of the overhaul, and inspection of the
hardware, must be economical and not negate the
advantage of the life extension.
3-14

8. OVERVIEW OF CURRENT MAINTENANCE

STRATEGIES
Appendix 1 provides information that is current at the
time of writing, and which does not reflect advances that
will surely occur for the newest engines including the
F119-PW-100, Fl 19 derivatives, F120-GE-100 and the
EJ200.

9. RECOMMENDATIONS
• A systematic approach to component life
management and maintenance is essential if the cost
of ownership is to be reduced.
• Lessons can be learned from other operators.
Conclusions should not be reached without
investigation.
• Tailor the management of component life to the
maintenance concept established by the end user in
consultation with the engine manufacturer and the
services logistics system.
• Engines that have a digital control system or digital
condition monitoring system can benefit from an
on-board life calculation algorithm presuming the
range of operation is within the correlation range of
the algorithm. Off-board processing may allow for a
more complex algorithm.
• Usage monitoring equipment is particularly useful
in determining differences in life consumption
between operational bases, missions and aircraft
series. Monitoring is more accurate than mission
logs, sortie pattern factors, etc. that generalise the
mission profiles. Experience has shown that
differences in usage - which may have the same
cyclic content - can precipitate failure of bearings,
blades, etc. that would have been avoided with
usage monitoring and corrective action when
differences were detected.
• TACs collect more coarsely granulated data than
usage monitoring systems, which catch and account
for all turning points. However, when TACs are
combined with the ENSIP approach the in-service
safety levels of the two approaches - damage
tolerance and safe life - appear to be virtually
identical.
10. REFERENCES
Reliability-Centered Maintenance. F Stanley Nowlan and
Howard F Heap of United Air Lines. The reference
number is MDA903-75-C-0349, dated 1978
Maintenance Management - A New Paradigm was
published in leading maintenance journals in the United
Kingdom, the United States and Australia in 1996. It is
based on a paper of the same name presented at the
annual conference of The American Society of
Maintenance and Reliability Professionals in October
1995.
Reliability-centred Maintenance by John Mowbray,
Butterworth-Heinemann Ltd, 1991, ISBN 0 7506 0230 9
 4-1

Chapter 4
Modes of Gas Turbine Component Life Consumption
by
(C. Eady)

Page

1. Introduction 4-3

2. Gas Turbine Engine Failure Mechanisms 4-3

2.1. Mechanical Fatigue 4-3
2.1.1. Low Cycle Fatigue 4-4
2.2. High Cycle Fatigue (HCF) 4-5
2.2.1. Engine Configuration 4-5
2.2.2. Intake Effects 4-6
2.2.3. Out of Balance 4-6
2.3. Thermo-Mechanical Fatigue 4-6
2.4. Creep 4-7
2.4.1. Stress Rupture 4-7
4 7
2.5. Corrosion "
2.5.1. Corrosion Fatigue 4-8
2.6. Erosion 4-8
2.7. Fretting, Galling and Wear 4-9

3. External Causes of Life Consumption 4-10

3.1. Manufacturing and Material Defects 4-10
3.1.1. Components not to Drawing 4-10
3.1.2. Manufacturing Process Control 4-10
3.1.3. Surface Conditioning 4-10
3.1.4. Stress-Raisers 4-11
3.2. Build or Maintenance Errors 4-11
3.3. Foreign Object Damage 4-11
3.3.1. Limit Exceedance 4-11

4. Summary 4-1*.

5. References 4-12
4-2
 4-3

under the control of engine manufacturers, operators and

1. INTRODUCTION
maintainers, they could be described as avoidable. In
In this chapter the main failure modes of mechanical reality, however, this is difficult to achieve.
components are shown in pictures of real components, and
The accuracy of the process for determining safe
described in simple physical terms.
component life and the monitoring of its consumption may
Critical components, for which safe operating lives are have a major impact upon the safety of the engine or on its
defined, include the major rotating parts, the turbo- life cycle cost. Engine design-features and operational
machinery disks and shafts, and structural casings environments that most undermine safety or arc most
subjected to high loads. In addition to the major costly, financially or in maintenance effort, must be
components, other components in the rotating assemblies, identified. Only then can managers assess the potential
such as spacers, cover-plates and seals, may also be benefit of proposed modifications to the design or changes
designated as critical. To avoid unacceptable risk of to operating practices. Even minor changes to operating
catastrophic failure it is necessary to monitor the life usage practice, which may have little or no operational impact,
of critical components and retire them from service before can significantly lower life usage rates and bring the
their allocated life has been exceeded. benefits of higher availability and reduced life cycle cost.
A thorough understanding of the failure mechanisms Monitoring systems can only monitor phenomena that
affecting gas turbine components is essential if the failure they can detect and measure. Therefore, great care is
modes, the safe-life, and the life usage of each component required during the design and introduction of any system,
are to be accurately determined and safely monitored. to ensure that it meets the needs of people at all levels in
Those failure mechanisms are: the operator's organisation.
Low cycle fatigue; 2. GAS TURBINE ENGINE FAILURE
High cycle fatigue;
MECHANISMS
Thermo-mechanical fatigue;
Creep; 2.1. MECHANICAL FATIGUE
Overstress; In most practical aerospace engineering situations, the
Corrosion; applied stresses are not steady. Instead, the loading
Erosion; fluctuates, often in a random manner. Under conditions of
Fretting and Wear. cyclic stress, it is frequently found that failure occurs at
lower stress levels than would be expected, were a steady
The ability of a component to resist the effects of any of stress applied. This phenomenon is fatigue and causes the
these failure mechanisms is a function of the material vast majority of in-service failures.
properties, the component design and the operating
environment. These features of the component are The process of fatigue is usually divided into the
effectively fixed by the design and application of the following 3 phases:
engine and cannot be affected by
the way in which the engine is Crack Propagation
operated or maintained. As such (Striations)
the chosen engine cycle, its
configuration and the selected Overload Zone
operating environment may be
considered to be intrinsic factors Crack initiation
which impact on the rate of (Nucleation)
component life usage.
Conversely, there are external
factors which also have an
influence on the rate of
component life consumption but
which can be reduced during
engine manufacture, operation
and maintenance. Best
Figure 1 - Characteristic fatigue fracture
maintenance practice, no matter
how good, can at best only • Crack initiation (primary stage);
restore component and system performance to the levels • Crack propagation (secondary stage);
that were intrinsic to the original design. • Unstable crack propagation (tertiary stage).
The external factors affecting life usage rate are:
Figure 1 shows the features of a characteristic fatigue
• Manufacturing and material defects; fracture. The number of stress cycles to failure is
• Build and maintenance errors; dependent upon the material type, component processing,
• Foreign Object Damage (FOD); load history, design, and the degree of loading.
• Limit exceedances.
Since these four engine damage sources are effectively The fatigue behaviour of a material is normally plotted on
4-4

an S-N graph. Here S is the average

stress range applied (in simple
harmonic motion), and N the number of
cycles to failure, at that combination of Failure Distributions

mean stress and stress range. Fatigue

failure is a statistical phenomenon and
accurate measurement of the fatigue
properties of a material requires a large
number of tests to ascertain the failure
probability distribution at each applied
mean stress and fluctuating stress
amplitude. A characteristic S-N plot for
zero mean stress, showing the mean
failure curve and the failure
distributions, is given in figure 2.
Component fatigue life is not only {Cycles to Failure) N
dependent upon the choice of material Figure 2 - S-N diagram.
and the magnitude of the mean and
cyclic stresses. Materials production
and processing can have a significant impact
on component performance in service.
Microstructural variables such as changes in
grain size, alloying and the presence of non-
metallic inclusions will affect the fatigue life.
Fatigue behaviour is also very sensitive to:
• Random stress fluctuations;
• Stress concentrations;
• Surface finish;
• Residual stresses;
• Corrosive environments.

The effect of the above factors is that severe

fatigue life penalties can be imposed through
inadequate component design or
inappropriate material selection for a given
operating environment.
The design loads for gas turbine components
are mainly in the elastic region of the
stress/strain curve of the material. In some
cases, cyclic softening can cause highly the number of times that a cyclic load is applied that is of
localised plastic deformation. This happens when high particular importance. Rather the amount of damage
stress concentrations cause yielding, which then reduces accumulated during each application of the load. For
the local stress to levels where yielding no longer occurs. critical gas-turbine components, it is appropriate to
The traditional approach to fatigue, for which the S-N differentiate between LCF and HCF by identifying the
curve was designed, chiefly concerned itself with failure driver of the cyclic loading. LCF is typically driven by
after a large number of cycles. At higher stresses and pilot demands and the application of relatively large loads
lower cycles, the fatigue life reduces progressively and the whereas HCF is typically driven by sources of vibration
scale of the plastic deformation creates difficulty in and the application of relatively small loads.
interpreting test results. In this region of Low Cycle
Fatigue (LCF), sometimes termed high-strain fatigue, the
2.1.1. Low CYCLE FATIGUE
materials tests are based on constant strain cycling rather If the amplitude of the cyclic stresses applied to a
than constant stress cycling. component is very high (LCF), the accumulated strain
energy per reversal will be significantly higher than for an
The transition point from LCF to High Cycle Fatigue HCF loading cycle. For this reason, under LCF, a
(HCF) is generally assumed to occur where the total strain component will spend a very small proportion of its life in
comprises equal proportions of elastic and plastic strain. the primary, crack initiation stage and the majority of its
However, for convenience, LCF is often considered to life in the secondary, crack propagation phase of fatigue
lead to failure in less than 105 cycles and HCF to lead to failure. As a result, since crack initiation is often governed
failure after more than 107 cycles. The intermediate range by random imperfections in the microstructure, the scatter
may be considered to fall into either region, depending on in LCF data is lower than that for HCF data, as can be
the design application. It must be remembered that it is not seen in figure 2. Under such high loading, it is also
 4-5

possible that plastic deformation around a concentration each time they pass. The passing frequency depends upon
feature will redistribute ('shake down') the stress to a less the number of obstructions and the rotational speed of the
damaging amplitude and leave the component under stage. For example, each blade on a turbine assembly
residual compressive stress whilst stationary. rotating at 19400 RPM behind a set of 34 nozzle guide
Most of the critical components within gas-
turbine engines, the turbo-machinery disks
and shafts, and pressure vessels such as the
combustion chamber outer casing, are
subjected to very high loading cycles and are,
therefore, life limited by LCF. The
mechanical loading on these components is
caused by:
• Centrifugal forces and thermal loads on
the disks;
• Torsion and bending forces on the shafts;
• High pressures within the casings;
• Thermal gradients within components

The cyclic nature of these forces is due to

variations in engine power setting. One
complete, or major, cycle is experienced
when the engine is accelerated from standstill
to maximum engine rotational speed and then
returned to standstill. Minor cycles of varying
size are experienced for all other throttle
movements. LCF for a gas turbine can
therefore be characterised as loading cycles Figure 4 - Fatigue crack grovs-h
caused by variations in rotational speed,
temperature distribution in parts, or engine internal
pressure, which are most often due to throttle
movement. The appearance of typical LCF failures
is shown in figures 3, 4 and 5.

2.2. HIGH CYCLE FATIGUE (HCF)

Fatigue failure occurs when either the material
fracture toughness is exceeded by the combination
of applied stress and crack size, or a critical crack
size is attained in a highly stressed region of a
component. The process to achieve either of these
conditions involves crack initiation and sub-critical
crack propagation. Whilst components under an
LCF regime spend the majority of their lives in the
crack propagation phase, for HCF, crack initiation
often tends to be the time-consuming process. Low
amplitude, high frequency loading cycles (HCF)
could quickly propagate a LCF initiated crack to
failure.
Figure 5 - Faticue failure in an LP blade attachment lug
HCF is primarily a function of engine design. A
component that fails in FICF does so because it has been vanes will experience 39.6 million loading cycles per
subjected to a large number (>107) of stress cycles (where hour. It is, therefore, important that the stress amplitude is
compressor blades are subjected to high cycle very small so that the blades are operating below the
aerodynamic loading this is known as flutter). A number fatigue limit set by the curve in figure 2.
of factors can cause high frequency loading of Campbell Spoke Diagrams are used to identify regions
components. These are generally known as drivers and where problems might occur. A blade design has its
their causes are listed below. natural frequency calculated or measured. Corrections are
applied to account for a natural frequency drop due to
2.2.1. ENGINE CONFIGURATION temperature rise and the stiffening effect due to radial
The layout of the engine itself is the cause of many HCF centrifugal forces. The overall effect is usually for the
drivers. The presence of both upstream and downstream natural frequency of the blade to rise as engine speed
obstructions in the gas path can create perturbations in the increases.
flow that will cause the turbo-machinery blades to deflect
4-6

An illustrative Campbell Spoke Diagram is at figure 6. It 2.3. THERMO-MECHANICAL FATIGUE

shows the frequency of vibration of various components Thermomechanical fatigue damage is caused by a
against engine speed. Included in the diagram are the first combination of the external loads and cyclic compressive
seven Engine Orders (EO) and the first flap (IF) and and tensile loads induced by thermal gradients across
second flap (2F) vibration modes of a typical blade. Due components.
to the higher energy involved, engine speeds above 75%
(5355 RPM) and the first 3 engine orders give
the greatest cause for concern. In this example,
the first flap and the second engine order
interact between the 100% (7140 RPM) and the
110% overspeed (7854 RPM) conditions. This ] yS

is therefore not a viable blade and re-design is |

necessary to alter the natural frequency of the / /^
--"♦j
jf .--y^-^'f*" • 1EO
blade.
- - "/ // ^
2EO
3EO
The potential resonance caused by interaction 500
1
«—463
♦-663
between the second flap mode and the 6EO and 1 •-660
— 7B3
5EO spokes is unlikely to be of high energy. Freqiency{Hz) ^^. - IFMtte

However, it may be significant if there is a lX>^ .... -

strong upstream excitation. In which case, the
resonance could be driven to high energy levels
by splitters or support struts in sets of 5 or 6 or
sub-multiples thereof. 0 10CO 20TO 3000 4000 SK0

Engine Speed (RPM)

2.2.2. INTAKE EFFECTS Figure 6 - Campbell 'spoke' diagram
One of the strongest HCF drivers can be caused
by intake distortion of the airflow which
creates a high energy, normally first engine
order (low-pressure shaft speed) driver.
Intake distortion can be inherent in the
design or due to the operation of the aircraft.
Bifurcated intakes often create uneven
pressure and velocity distributions at the
compressor face that induce flutter in
addition to second engine order excitation.
In operation, aerodynamic effects such as
intake masking during high alpha (angle of
attack) manoeuvres or during high yaw or
side-slip manoeuvres can also grossly distort
the inlet flow. Supersonic intakes, when
operating at off-design conditions, also cause
distortion at the front face of the engine.
Alternatively, distortion can be caused by rtgure - Interg! :ep failure i
physical blockage. Such might occur in icing This effect is particularly significant for turbine blades and
conditions, or if a large object is ingested. Fortunately vanes, and especially cooled blades and vanes. Taking an
these events are normally of short duration and hence uncooled HP turbine blade as an example, before first
rarely lead to engine failure. However designers must engine start it has no residual stresses and is at uniform
allow for some element of running with distorted intake temperature. Upon engine start the blade experiences high
airflows and operators must try to avoid these situations temperatures on its outer surface whilst taking longer to
wherever possible. reach operating temperature at its core. This leads to
excessive compressive forces particularly at the leading
2.2.3. OUT OF BALANCE edge and the material plastically deforms to relieve the
It is impossible to perfectly balance the rotating stress.
components in a gas turbine engine. A mechanical system
that has heavy components rotating at high speed will As the blade core reaches steady state temperature, the
experience vibration. An out-of-balance rotor will create a outer surface goes into tension due to the earlier
vibration driver proportional to the shaft speed and will compressive deformation. The centrifugal loading on the
excite components across a wide range of frequencies in blade increases this tensile loading. As the throttle is
normal operating conditions. Components must therefore retarded, the centrifugal forces reduce and the outer
be designed to avoid the natural frequencies that are likely surface cools faster than the core causing the outer surface
to be generated by an out-of-balance rotor. to try to contract. This contraction is opposed by the core,
which holds the outer surface in tension.
 4-7

When the engine is shutdown and reaches the ambient forms part of the total extension reached in a given
temperature throughout, there remains a tensile stress in time and so affects the choice of clearances.
the outer surface of many components. This process • The Secondary Stage shows creep occurring at a
enhances the level of tensile stress cycling experienced by relatively constant rate; this is the important part of
the blade, particularly in the thinner sections such as the the curve for most applications.
leading and trailing edges, where temperature changes • The Tertiary Stage shows an acceleration of the creep
occur most rapidly. rate to failure; this stage should be avoided in
operation but the transition from the Secondary Stage
It can therefore be seen that the blade goes through a stress to the Tertiary can be difficult to predict.
sequence with every change of temperature. The scale of
the load imposed is proportional to the temperature 2.4.1. STRESS RUPTURE
gradient induced in the blade and this is a function of the For design purposes, so-called stress rupture curves are
rate of throttle movement. Therefore, for long engine life used to simplify creep modelling. These are obtained by
throttle movements should be made as slowly as possible. measuring the time to failure of a specimen under constant
Thrust producing engines, such as turbo-jets or turbo-fans, load and temperature conditions. The curves are simply a
plot of the log of time to failure against temperature for
operate at varying speeds and temperatures. The thermal
and mechanical load cycles tend to happen various loads. The result is usually a straight line and this
simultaneously. However, since the thermal loading representation simplifies the calculations.
affects the degree of mechanical loading,
engine life usage counters really need Tertiary creep
Thermal Transient Algorithms (TTA) in Primary creep
order to most accurately monitor fatigue life
consumption.
Some torque producing engines such as
turbo-props and turbo-shafts, may have one
shaft running at constant speed whilst the
temperature varies. In such cases, the
mechanical load cycles arc easily monitored.
On the other hand, changes in torque
demand are achieved by altering engine-
running temperature. Hence, thermal fatigue
effects predominate and must be monitored, Log time
to ensure safe and efficient engine operation. Figure 8 - The three stages of creep
For critical components such as turbo-
machinery disks, the transient thermal effects may only be 2.5. CORROSION
significant during major throttle modulations. However, Irreversible damage can be caused to engine components
the disks do suffer a temperature gradient from bore to in aggressive environments. Corrosive elements can be
rim. This gradient is usually a cooler bore to a hotter rim, introduced to the engine in the fuel (eg sulphur) and in the
especially for turbines and the HP sections of inlet airflow (eg airborne pollutants or sodium chloride
compressors, but it may go the other way for LP from marine environments).
compressors with anti-icing air or balancing air within the
shaft. Whichever way the thermal gradient operates, it will Corrosion can be controlled through careful selection of
also effect material properties. The thermal gradient will materials, the application of coatings, and the maintenance
produce a stress field that varies across the radius due to of the surface finish throughout the life of the engine.
the temperature differences and the coefficient of Regular water washing after each sortie, if necessary,
expansion. should be considered for engines operated in corrosive
atmospheres.
2.4. CREEP
Corrosion may reduce the life of an aero engine in 3 ways.
Creep is a time-dependent inelastic deformation of metals
and alloys, which occurs under stress, at temperatures • By reducing the load bearing area. Corrosion causes
above about 0.5 of the melting point temperature. The loss of material, and can weaken a component so
higher the temperature and the greater the load, the faster much that it fails under normal loading. Corrosion
will be the rate of deformation. In a gas turbine the on this scale is usually easy to detect during routine
components which suffer most from creep are the hot inspection. Much more difficult to detect is a small
parts of turbine disks and the turbine blades. However, corrosion pit that acts as a stress raiser hence causing
blades tend to fail due to impact with adjacent rows or a premature LCF or HCF failure. Any breakdown in
casings rather than purely due to creep. Figure 7 shows a sealing effectiveness within the turbine section may
typical failure. also permit hot gas ingress to the disk void areas and
result in material degradation.
Figure 8 shows the three stages of creep.
• By reducing aerodynamic efficiency. If aerodynamic
• The Primary Stage depicts rapid extension at a surfaces, particularly blades and vanes and to a
decreasing rate; this is of interest to the designer as it lesser extent walls and diffusers, suffer corrosion
4-8

they are roughened. The effect of this is to alter the aircraft that routinely operate in dusty/sandy conditions.
nature of the airflow over them and hence reduce Helicopter engines are the worst affected, since by their
their efficiency. Any reduction in efficiency of the very nature, they are prone to operate in dust clouds
turbo-machinery will result in a loss of thrust at a generated by the rotor downwash.
given cycle temperature or require that temperature The typical effects of erosion within a gas-turbine engine
to be raised to maintain thrust. In either event, the
are:
life of the engine will be reduced due to premature
rejection for low thrust or excessive operating • To remove material from the rotor tips increasing tip
temperature. clearance and reducing performance;
• By causing a blockage. With aluminium based • To remove material from the leading edge of turbine
materials, the products of corrosion can occupy six blades as shown in figure 12;
times the volume of the non-corroded material. • To reduce the chord width of the blades thus
Therefore, in extreme cases, the corrosion will result reducing aerodynamic performance.
in severe restrictions to cooling passages and
possible bursting of thin-walled sections. The
disruption of cooling flows may also cause hot spots
to develop and lead to early creep failure.
Although corrosive action can remove enough material to
directly affect the performance of a component, the
mechanical integrity of high strength components is likely
to be jeopardised long before any visible evidence of
corrosive attack. This is due to the creation of sites from
which fatigue cracks can propagate. Examples of the
effects of corrosion are shown in figures 9, 10 and 11.

2.5.1. CORROSION FATIGUE

The response of different alloys to combined stress and
corrosion varies greatly but those that fail usually do so
without any gross corrosion. When there is a combination
of a corrosive environment and a fluctuating stress, a
hazardous condition described as corrosion fatigue can
occur. The effect of corrosion fatigue is usually to reduce
the fatigue life much more severely than would be
expected by the separate consideration of the two
mechanisms.
The first stage of corrosion fatigue is usually pitting of the
surface layer of the material. This weakening allows micro
cracks to form and the second stage establishes corrosion
in these cracks. In this way, both the crack initiation phase
and the propagation phase of fatigue failure are
accelerated by the corrosive process as good material is
removed and the cracks are wedged with the products of
corrosion. Electrolysis at a crack tip may cause atomic Figure ■ Salt corrosion of PT blade of aero-
hydrogen, which can penetrate the metal lattice and cause derivative gas turbine in marine
hydrogen embrittlement. This may further accelerate crack environment
growth.

2.6. EROSION
Erosion is the cumulative damage to
components in the airstream caused by small
hard particles carried in the gas path. What
differentiates erosion from any other damage
mechanism is the scale and nature of each
individual damage event. Individually each
event is inconsequential in terms of
performance loss or reduction in mechanical
strength. However, when very many events
take place at one particular area of a
component then the performance or strength
of that component can be severely
compromised. Erosion is often most apparent
in compressors, particularly those used in Figure 10 - Corrosion in nickel superaiioy
 4-9

• Due to the vulnerability of helicopter

engines to erosion, many are fitted with
inlet particle separators, which, at the cost
of some engine performance, remove
many of the potentially eroding particles
from the air before they reach the
compressor. Both helicopter engines and
turbo-fan engines in aircraft that fly low
over the sea suffer from erosion due to the
continual impact of salt crystals on the
leading edges of fan and compressor
blades.

Erosion rate is a function of the following.

• Operating environment;
• Component materials;
• Design;
• Protective coatings.

Predominately, erosion by inlet-airborne

material is a compressor problem but it also Figure 11 - Corrosion and spalling of thermal barrier coating in
affects other components in the airstream such marine gas turbine caused by poor combustion
as those in the by-pass duct. As particulates pass
through the compressor, they are graded and the
finer elements may be drawn into the cooling system. Not
only does this process erode the internal air-washed parts,
but it may lead to the glassification of sand particles in the
turbine. When this happens molten silica solidifies on
suitably 'cooler' surfaces, impedes cooling flows and may
cause component failure due to overheating.
Particulates are frequently generated within the engine
from:
• Erosion of abradable seal coatings by rotor blades;
• Hard carbon produced in the combustion section.

Although the critical components are not generally

affected by the particulates carried in the mainstream flow,
carry-through of hard particles into the cooling passages
will remove material, particularly surface treatments,
exposing the parent metal to corrosion.

2.7. FRETTING, GALLING AND WEAR

There are two sources of damage to contacting surfaces;
fretting and wear. Fretting, shown magnified in figure 13,
is due to oscillatory motion of very small amplitude, much
smaller than may occur with normal wear. This can occur
in joints that are bolted, keyed, press fitted, shrunk or
riveted. Fretting also affects splines, couplings, clutches,
spindles and seals. The effects of fretting are accelerated
because the debris produced is trapped at the wear site
and, whilst it may not lead to component failure, fretting
Figure 12 - Erosion of marine GT PT blade due to
produces small surface cracks which can propagate to a
carbon in the gas stream from poor
conventional fatigue failure. The start of such a crack is
combustion
shown in figure 14.
Wear will occur whenever two or more contacting Relative hardness of contacting materials. The softer
components experience relative motion. Unlike fretting, material will wear more rapidly;
wear is the result of large scale, unavoidable movement Contact force. The higher the contact-force, the
between two components. Typical examples of wear greater the wear rate;
include a piston in the bore of an actuator, bearings, air Lubrication. Lubrication is normally provided to
and oil seals, and blade tips on abradable seals. The rate of control wear in components where relative motion is
wear is determined by a number of factors:
4-10

intentional but areas where fretting occurs

are rarely lubricated;
• Temperature. As temperature increases the
wear rate tends to increase because
materials become softer, lubrication
becomes less effective and clearances tend
to reduce, leading to an increased contact
force.

The majority of fretting can be overcome by the

application of anti-fretting compounds,
elimination of the relative movement (often
difficult), surface hardening, separation of the
components or reduction of the applied stress.
However, wear is inevitable between contacting
surfaces in motion and must be controlled
initially by design and then by maintenance.
Fretting or wear should not be confused with
galling (or spalling). Galling is a fatigue *'isure 13 - Fretting in a titanium compressor disk doveta.
mechanism. It is caused by cyclic loading, of a
small area of the surface, of a component. It
typically occurs in components such as
bearings, where the passage of the rolling
elements over the races induces a cyclic load.
This can cause small cracks to grow into the
surface, and then propagate parallel to the
surface before re-emerging at the surface and
releasing a small particle.
Fretting and galling are apparent on many
blade/disk attachments. Galling (deep groves)
and fretting (light groves) have a very different
effect on fatigue life. Fretting tends to reduce
HCF life and galling tends to reduce LCF life IliililP^
3. EXTERNAL CAUSES OF LIFE
CONSUMPTION Figure 14 ■■ Fretting crack with fatigue extension in training
aircraft starter turbine
3.1. MANUFACTURING AND MATERIAL
DEFECTS
3.1.2. MANUFACTURING PROCESS CONTROL
All metallic materials have finite fatigue lives and these
are calculated based on the assumption that components The component specification, which includes geometry,
will be manufactured to drawing and with material to the heat treatment, and hot and cold working, is vital in
correct specification. However, in situations where the determining the service life of a component. A small
component does not comply with the drawing, the material alteration in the shape, composition, heat treatment cycle,
is of incorrect composition, or heat treatment/surface or manufacturing process can have a major impact on life.
conditioning has been incorrectly carried out then the Minor alterations in the chemical or mechanical
achievable life can be greatly reduced. These issues are composition of an alloy will alter its strength, fracture
dealt with in more detail below: toughness and creep resistance.
Failure to heat treat at the required temperature or for the
3.1.1. COMPONENTS NOT To DRAWING correct length of time may well result in a component that
If a component is not to drawing, then a number of has the wrong grain structure or that has been inadequately
problems might arise. If the section of a component is too stress relieved. Finally, a failure of the manufacturing
small then the stress for a given load will be increased. If a process could lead to components having an incorrect
section is too thick it will be too stiff, its mass will be metallurgical specification. For example, conditions
increased and, at a given operating condition, the loading required to create a single crystal turbine blade need to be
that it exerts on other components might be increased. If a very tightly controlled if the crystal structure is to be
filet is of too small a radius then stress in that region will properly aligned to the span of the blade.
be increased or if the surface finish is too rough then
cracks will be more likely to form early in the life of the 3.1.3. SURFACE CONDITIONING
component. Surface conditioning includes all processes such as cold
working and shot peening which alter the material
 4-11

properties of the surface of the material. Most commonly, • Use of reverse thrust;
in gas turbine components, surface conditioning is used to • Helicopter downwash;
impart a residual compressive stress into the surface of the • Inlet vortices.
component in areas where cracks are most likely to form.
The effect of this is to inhibit crack initiation and thus There are other ways in which FOD can occur:
extend life. If the chosen technique is not carried out
sufficiently then the full benefit is not achieved. If it is • Objects left in intakes during maintenance;
over done then the surface can become excessively work • Objects, particularly fasteners, from either the parent
hardened and brittle, or sub-surface tensile stresses may be aircraft or an aircraft in close formation, typically
generated, making it more prone to cracking. during air-to-air refuelling;
• Bird ingestion;
3.1.4. STRESS-RAISERS • Ice ingestion.
Stress-raisers are features at the surface of, or embedded
within, a component that cause stress contours to deviate Helicopters, due to their downwash and operation from
from their normal (design) orientation, parallel to the unprepared sites, and combat aircraft, due to high specific
direction of load, and to intensify around the tip of the thrust, low by-pass ratios and intake positioning, are worst
feature. This is shown at figure 15. At the tip of the affected by FOD.
feature, in this case shown as a notch in the surface, there Once damage is caused it will fall into one of these
is a high stress concentration. The fatigue life will categories:
therefore be much reduced and a crack will start to
propagate from this point. • Acceptable with no maintenance necessary;
• Acceptable but only after the damaged component
Stress-raisers can be caused in a large number of ways. has been blended or cropped to remove a stress
• They can be present at manufacture as inclusions in raiser;
the metal melt; • Unacceptable, either the damaged blade will have to
• Tools can leave scratches during machining; be changed in-situ, where this is possible, or the
Lack of care during transport, storage or handling. engine will have to be removed for repair;
• Catastrophic, leading to in-flight loss of an engine.
3.2. BUILD OR MAINTENANCE ERRORS This occasionally happens when aerodynamically
One of the most effective ways to shorten the life of an induced blade flutter causes crack initiation and a
engine right from the beginning is to build it incorrectly. HCF blade failure occurs.
Many mistakes can be made during the build or
maintenance of an aero-engine. Some of these are listed Whatever the category of the FOD it may reduce the life
below. of the engine. Even if acceptable, it will contribute to
performance loss or cumulative damage and hence
Incorrect tolerances; premature engine rejection.
Incorrect clearances;
Incorrect torque loading; 3.3.1. LIMIT EXCEEDANCE
Insufficient cleanliness leading to blocked oil ways Typical limit exceedances are listed below.
or dirt in bearings;
Failure to apply lubricants; • Absolute Temperature. Any exceedance of a
temperature limit will result in life being consumed
Incorrect assembly;
at a greater than planned rate. An exceedance of an
Incorrect adjustment.
absolute temperature limit may result in damage to
the turbine and/or combustion section.
All of these can, and have, led to reduced engine life
through early component failure or performance loss. All • Time at Temperature. Most engines have
build and maintenance errors are avoidable but still occur temperature settings (ratings) that should only be
used for limited periods. These include take-off,
regularly.
climb, combat, and emergency limits. The stress
3.3. FOREIGN OBJECT DAMAGE rupture life of the turbine blades will have been
calculated on the assumption that these limits will
Foreign Object Damage (FOD) is an inevitable part of not be exceeded. The longer the time spent at these
engine operations. Gas turbines have a high air mass-flow- elevated temperatures and hence rotational speeds
rate and loose articles that are close to the intake of the (in the case of turbo-jets, turbo-fans and turbo-prop
engine are likely to be ingested. A gas turbine is able to gas generators) the greater will be the rate of creep
lift objects from the ground in some circumstances. life consumption. It should be noted that current
However, ingestion is much more likely to occur when a generation FADECs do not prevent time-at-
particle is already in motion. This motion can be due to a
temperature limits being exceeded.
number of factors the most common of which are listed
• Rotational speeds. Whilst engines are designed to
below.
survive overspeeds of up to 122%, lifing calculations
• Jet efflux of another aircraft; are based on speed limits being observed. Therefore,
• Thrown up by aircraft wheels; overspeeds result in increased dynamic loads that
• High winds; consume fatigue and creep life at a very high rate.
4-12

• Pressures. Over pressure results in increased loading

on casings whose fatigue lives are therefore reduced.

The latest generation of digital engine-controls makes

exceedance of engine operating limits unlikely. However,
on older engines, particularly those with hydro-mechanical
fuel controllers, limit exceedance is much more frequent.
It is important that exceedances are recorded for
subsequent maintenance action as defined by the
manufacturer.

4. SUMMARY
The decision to apply a safe-life to an aero gas-turbine
engine component is an expensive one. It is generally
reserved for those components assessed as having failure
modes that would hazard the integrity of the engine and
the safety of the aircraft. These critical components
typically include the major rotating assemblies and
structural casings. The ability of engine components to
resist failure is highly dependent on material properties
that must be carefully selected to offer the optimum
combination relevant to the engine type and application.
Detail design considerations must also be taken to avoid
stress concentrations and forcing functions that would
promote failure.
Critical component life will be consumed in terms of low
cycle fatigue, high cycle fatigue, thermomechanical
fatigue and creep damage and it is essential to understand
how the engine usage relates to the life consumption rate.
Significant financial penalties are associated with
excessively early retirement of critical components.
Conversely, disastrous airworthiness consequences may
be associated with late retirement. The situation is further
complicated by other damage mechanisms which are
common in gas-turbine usage and abusage, such as over-
stress, corrosion, erosion, fretting, wear and impact
damage, and which can reduce material properties and
promote early failure of critical components.
During operation, there is little which can be done to
increase component lives since the available life and the
expected usage rate are inherent in the engine design.
However, attention to component handling and assembly
during maintenance, preservation and restoration of
surface coatings, and avoidance of foreign object damage
will all assist in achieving the published lives and avoiding
expensive early failures.

5. REFERENCES
Photographs supplied by Dr G Harrison of DERA.
 5-1

Chapter 5
Mechanics of Materials Failure
by
(W. Beres)

Page

1. Introduction 5-3

2. Modelling Material Failures 5-3

3. Stress States in Gas Turbine Engine Components 5-4

3.1. Thermal Analysis 5-4
3.2. Stress Analysis 5-4
3.2.1. Finite Element Methods 5-5
3.2.2. Fatigue and Fracture Analysis for Life Prediction 5-5
5 5
3.2.3. Crack Growth "
3.2.4. Stress Intensity Factor 5-6
3.2.5. Stress Concentration Factor 5-7

4. Properties of Gas Turbine Engine Component Materials 5-7

4.1. Static Properties 5-7
4.1.1. Monotonie Stress-Strain Curve 5-7
4.2. Flow Properties 5-7
4.2.1. True-Stress-True-Strain Curve 5-7
4.2.2. Effect of Strain Rate on Flow Properties 5-7
4.2.3. Effect of Temperature on Flow Properties 5-7
4.3. Creep Properties 5-8
4.4. Cyclic Properties 5-9
4.4.1. Cyclic Stress-Strain Curve 5-9
4.4.2. High Cycle Fatigue (HCF) 5-10
4.4.3. Low Cycle Fatigue (LCF) 5-10
4.4.4. Thermo-Mechanical Fatigue (TMF) 5-11
4.5. Fracture Mechanics Properties 5-11
4.5.1. Fatigue Crack Growth Rate (FCGR) 5-11
4.5.2. Fracture Toughness 5-12
4.5.3. Creep Crack Growth Rate 5-13
4.6. Influence of Mean Stress on Fatigue and Fracture 5-13
4.6.1. Influence of Mean Stress on HCF 5-13
4.6.2. Influence of Mean Stress on LCF 5-13
4.7. Influence of Multi-Axial Stresses on Fatigue and Fracture 5-13
4.8. Environmental Resistance 5-13
4.8.1. Oxidation Resistance 5-14
4.8.2. Corrosion Fatigue and Stress-Corrosion Cracking 5-14

5. Analytical Techniques for Damage Characterisation 5-14

5.1. Cumulative Fatigue Damage 5-14
5.1.1. Palmgren-Miner Rule 5-14
5.2. Creep Damage 5-15
5.3. Creep-Fatigue Interaction 5-15

6. Conclusion 5-16

7. References 5-16
5-2
 5-3

• Internal damage is related to micro-structural

1. INTRODUCTION
changes that occur slowly at a rate which is strongly
This chapter develops the theme of chapter 4, and influenced by service temperature and stresses,
introduces analytical descriptions of the basic physics that which in turn depend greatly on user practice. These
describes the failure modes. Much of the content is micro-structural changes can be associated partly
further described in Appendix 3, and equations referred to with plastic damage accumulation and partly with
as for example (A-20) may be found there. metallurgical ageing. Depending on the component,
alloy composition and service condition, the plastic
2. MODELLING MATERIAL FAILURES damage may be due to creep, low cycle fatigue
(LCF) or creep-fatigue interactions. As internal
Let's define the terms Failure and Failure Mode: damage builds up, the resistance of components to
• Failure of a gas turbine engine component is deformation under static (creep) or cyclic (LCF)
defined as any change in the size, shape, or loading is reduced. Extensive loss of resistance to
mechanical properties of the component that makes plastic deformation can lead to cavitation and
it unable to satisfactorily perform its design internal cracking under creep conditions or ductility
functions. The term may also mean that all of the exhaustion and crack initiation under LCF or
life is used up. thermo-mechanical fatigue (TMF) conditions. Both
• Failure mode is defined as a physical or chemical lead to failure.
process or processes, which take place or combine
their effects to produce component failure. The primary failure modes that are assessed for life usage
purposes are LCF and creep. A summary of all basic

Engine Component Failure Mode

Section
Compressor Blades ER-COR, HCF
Vanes ER-COR, HCF
Discs LCF, C, HCF
Spacer LCF, C, HCF
Turbine Blades TF, C, HC, LCF, HCF
Vanes TF, HC, C, HCF —
Discs LCF, C, HCF
Torque Ring LCF
Combustor Case LCF, C, HC —
Shaft LCF, WR
Compressor discharge case ER, LCF,COR
(diffuser)
Rotating seal LCF, C,HCF
ER Erosion LCF Low Cycle Fatigue
C Creep HCF High Cycle Fatigue
COR Corrosion HC Hot Corrosion
TMF Thermo-mechanical Fatigue WR Wear

Compressor disks and spacers in 1950-1970's engines are often steel and may therefore suffer corrosion,
which reduces the LCF life.
Table 1 - Failure Modes and Life Limiting Properties for Turbine Engine Components

During long-term service in a corrosive environment failure modes for gas turbine engine components is
under high stresses and at high temperatures, turbine shown in Table 1.
engine components such as discs, blades or vanes suffer LCF damage is dominant in bores, bolt hole and fillet
from cumulative damage. This gradually degrades their areas of compressor and turbine discs. It is caused by
mechanical properties, and may lead to a component stress cycling associated with engine start-up and
failure, or loss of engine structural integrity. Engine shutdown and with engine speed excursions during
components can be subjected to both surface and internal service. While the driving force for LCF damage
damage. accumulation in compressor discs is primarily
• Surface damage due to erosion, corrosion, mechanical, in hot section areas of the engine it may also
oxidation, wear, fretting or impact can promote be associated with thermal cycling. Thermal fatigue
crack nucleation at stress concentration areas. These damage in vanes and high-performance blades is often a
cracks can then propagate under the influence of life-limiting factor in modern gas turbine engines.
mechanical or thermal cyclic loads during service Creep damage is usually predominant in the mid-airfoil
and eventually cause component failure. section of a turbine blade, and in some cases in the rims
5-4

of turbine discs, where the stresses and temperatures may proviso that the cracks grow sufficiently slowly during
be sufficiently high to cause time-dependent inelastic service to allow their growth to be reliably detected and
deformation. perhaps even monitored through regularly scheduled
inspections.
Engine designers expend a great deal of effort to establish
life limits for components whose deterioration may 3. STRESS STATES IN GAS TURBINE
threaten operational safety. The lifing procedures
ENGINE COMPONENTS
employed to establish life limits vary with components
and the predominant damage mode. In modern turbines, 3.1. THERMAL ANALYSIS
discs and spacers are normally designed to withstand LCF
and burst due to overspeed as well as creep for the hot The following three things must be known before the
section side of the engine. Turbine blades and vanes are service life of aerospace components can be assessed:
designed to withstand creep as well as thermo-mechanical • The environment and loads to which a component is
fatigue and high-cycle fatigue (HCF). subjected. This includes temperatures, flow rates,
The life limits for these components are initially corrosive or erosive conditions, etc.
established based on service estimates of damage • The macroscopic response of the component to
accumulation rates. They are then revised as field applied loads and environmental stresses and
experience is accumulated. Typical lifing procedures can strains.
be grouped as follows. • The microscopic response of the component
material to the local stress, temperature, and
• For rotating components the most common lifing environment conditions.
procedure employed to establish safe life limits
follows a time or a cycle to crack initiation criterion. The magnitudes, variations and exposure times for the
This is factored down to represent a minimum stresses and temperatures experienced are the major
property component. factors controlling the lives of gas turbine engine
• For turbine blades and vanes, life limits is usually components in service. The role of thermal and stress
be prescribed by engine designers. In particular, analysis is to calculate these quantities so that component
turbine blades and vanes in aero-engines are seldom service lives can be predicted. Particularly, application of
lifed because of the difficulties associated with damage tolerance based methodologies based on fracture
predicting the service behaviour of metallurgically mechanics (crack growth) principles requires detailed
complex systems under conditions that can vary knowledge of stress fields in uncracked and cracked
widely with user practice. components. In addition, since superalloy materials
• In the case of coated components, these difficulties applied in gas turbine engine components are sensitive to
are compounded due to coating substrate inter- temperature, accurate determination of temperature
actions that are usually not well understood. A "life conditions is very important.
on-condition" approach is sometimes employed
where, for instance, creep growth or untwist of Temperature and stress gradients across rotating turbine
airfoils is measured and distortion limits are used as discs and spacers are dependent on conductive heat
retirement criteria. transfer within the component material and on convective
and radiative heat flow at the component surfaces. Engine
As indicated, rotating parts such as compressor and thermal analysis is usually performed using commercial
turbine discs are usually life-limited due to LCF damage or in-house finite-element or finite-difference based
accumulation. The most common lifing, and traditional, software. The associated internal air flow models for an
method employed for these parts follows a "cycles-to- engine take into account spool speeds, torque, shaft-loads,
crack initiation" criterion. In this, a minimum life primary flow parameters (such as gas temperatures, mass
capability is defined statistically for simulated sendee flow, and pressures), and secondary flow analysis in
conditions through extensive coupon testing and cooling and pressure balancing air and oil flows.
component test verification in a spin-pit test-rig. The
statistical minimum is usually based on the probability 3.2. STRESS ANALYSIS
that no more than 1 in 1000 components will have Finite element based stress analysis procedures produce
developed a detectable crack (typically chosen as a crack transient and steady state stresses in every location of a
of 0.8 mm in length). It should be mentioned that the component taking into account external loads, such as the
value '1 in 1000' is chosen arbitrarily. Different following.
manufacturers use different values, e.g. 1 in 980, or 1 in
Centrifugal loads of discs and blades;
750 components.
Gas pressure generated loads;
Because many modern components have demonstrated Torque in shafts;
high levels of resistance to crack growth, and modern Thermal expansion stresses;
materials have different characteristics, alternative lifing Assembly loads, e.g. originating from bolt
procedures based on damage tolerance and the application clamping.
of fracture mechanics principles are now in use. The
philosophy behind these alternative lifing procedures Generally, full 3D finite element analysis of components
assumes that the component may be capable of continued is required, but for an axi-symmetric structure, there is no
safe operation during crack growth. This is subject to the
 5-5

dependency on one direction this allowing a so-called 2D These elements are connected at specific points called
analysis. 3D analysis is usually focused on the critical nodes.
location or critical feature rather than the whole critical Variations of the unknown field variable inside the
part. For example, analysis of bolted joints usually continuum are described by approximating functions,
requires 3D finite element analysis. Body forces are which are also known as shape functions. These functions
calculated based on component mass and rotational are defined in terms of the values of the field variable at
speed. Temperature gradients that create thermal stresses the nodal points. Field equations for every individual
in components are taken from the thermal analysis element are written and the set of unknowns representing
described in the previous section. Materials data are the field values at the nodes is created. After assembly of
usually taken from comprehensive materials databases, the nodal values coming from different nodes, a set of
which include data on changes of properties with linear and usually banded equations is created. Unknown
temperature. In the main, linear static and dynamic nodal values of a field variable of interest are obtained by
calculations are performed, but for critical parts in critical solving this system of linear equations. Usually, for stress
locations temperature-dependent, non-linear-elastic, analysis problems a set of displacements is obtained as a
perfectly plastic behaviour of material is taken into primary solution to the system of linear equations. Then,
account. In non-linear analysis, it is assumed that the in the next step, the stresses are calculated from the
structure responds in a linear elastic manner after stress displacements.
redistribution due to plastic flow has been calculated
within the component. Principal or von Mises stresses Usually the entire process of dividing the domain into
obtained from the calculation results are used to assess finite elements, creating the system of equations, solving
the integrity of a component and to identify fracture and interpreting the results is significantly simplified in
critical locations. widely used commercial finite element packages. The
suitability of these should be assessed on a case-by-case
Potentially critical LCF locations can be identified by a basis for each task. Often aerospace companies use in-
relatively simple 2D analysis at steady state, which will house finite element programs tailored for specific
identify high bulk-stress locations. These bulk stresses requirements such as rotor dynamics applications.
can then be multiplied by stress concentration factors to
estimate peak stresses. The life-limiting feature on a Typical finite elements used for thermal and mechanical
particular critical part is most likely the one with the stress analysis include:
smallest complex geometry in the area of relatively high
Axi-symmetric 3 and 4 node elements;
average stress. Identification of these critical features is
one of the significant outputs of stress analysis. These Plane strain 3 and 4 node elements;
features are then analysed for LCF life capability in terms 8 and 20 node bricks elements;
of simple zero-max-zero cycles. 6 and 15 node wedge elements;
10 node tetrahedral elements;
Combined with mission analysis the outputs of stress 2 node spring elements;
calculations may be used to describe stress as a function 2 node gap elements.
of time and identify the maximum stress points in the
mission (critical mission points) and to identify critical Frequently, only segments of the rotating components
and life limiting locations. such as discs and spacers are modelled to decrease the
Stress results can be presented as 'snapshot in time' size of finite element models. Application of sliding
stresses on a part, feature or subsection of a feature, boundary conditions on the cut out surfaces assures
These represent the stresses at that specific point in the proper representation of displacement and stress fields in
cycle. Results can be also presented as a function of time, the disc segment.
as the stress variation in a specific period for a specific
critical feature or a portion ofthat feature. 3.2.2. FATIGUE AND FRACTURE ANALYSIS FOR
LIFE PREDICTION
3.2.1. FINITE ELEMENT METHODS Results of the finite-element stress analysis are used to
Since in real life problems exact temperatures and stress identify fracture critical locations in the components
solutions are rarely obtained analytically, numerical analysed.
methods are used to obtain approximate solutions.
Various numerical methods are used in engineering The expected life is calculated, based on stress and
practice for solving boundary value problems in solid temperature history in critical areas that are identified
mechanics, including the finite element method, the from the design mission. Cycles are commonly assessed
boundary element method and the finite difference with respect to their range and mean value, although
method. Of these methods, the finite element method is dwell time may be an important factor for some materials
the most widely used. that may undergo corrosion, oxidation or creep.

The fundamental concept of the finite element method is 3.2.3. CRACK GROWTH
that any continuous field variable, such as stress, pressure Crack growth prediction for components in service can be
or temperature can be approximated by a discrete model. based upon various techniques with different degrees of
The discrete model is composed of a set of piecewise sophistication and varying life prediction accuracy. In
continuous sections, which are defined over a finite rising order of complexity these are:
number of sub-domains of finite size known as elements.
5-6

• Rough estimates, where textbook stress intensity • Energy required to induce plastic deformation of the
factor solutions are combined with stress material at the crack tip;
predictions from finite element analysis; • Kinetic energy;
• Analytical methods combined with the finite • Energy required to create two new fracture surfaces.
element method where tabulated stress
concentration factors are used together with finite The kinetic energy term and fracture-surface energy term
element calculations on coarse meshes; are usually considered to be small and are neglected.
• Sophisticated crack modelling in 3D, using Therefore, only the remaining first three energy
automatic mesh-generation software, which components are considered in the analysis. A further
remeshes the component as the calculated crack simplification occurs if the size of the plastic zone at the
propagates. This is an iterative process, in which a tip of the crack is small compared to the total volume of
3D crack is modelled in the structure, the stress the body. In this case, the strain energy released is a good
intensity distribution along the crack front is approximation to the change in the elastic strain energy of
calculated using FEM, and a new crack front due to a cracked body.
crack growth is made. The component is then The stress intensity approach is a more general approach
remeshed and the stress calculation process starts to characterise crack propagation. An essential feature of
again. fracture mechanics is to characterise the local stress and
• Fracture mechanics analyses are performed mainly deformation fields near a crack tip. An elementary
in using linear-elastic fracture-mechanics methods fracture analysis of flawed components may be
with a stress intensity factor as the critical parameter performed by stress analysis based on the theory of linear
governing fatigue crack growth rate. Plastic non- elasticity. The stress field surrounding a crack tip can be
linear analysis for fracture mechanics or life classified according to three major modes of loading
extension uses the contour integral called ./-integral, which incorporate different crack surface displacements:
Anderson (1991), Broek (1989), to correlate crack
growth rates. Crack propagation laws, as described • Mode I - Opening mode, where the crack surfaces
in Section 5.5.1, typically include range, move directly apart in direction normal to the crack
temperature dependency, threshold effects and faces.
different R-ratio defined as a ratio of minimum to • Mode II - Sliding or in-plane shear mode, where the
maximum stresses in a loading cycle. crack surfaces slide over one another in a direction
3.2.4. STRESS INTENSITY FACTOR perpendicular to the leading edge of the crack.
• Mode III - Tearing or anti-plane shear mode, where
The fracture behaviour of a given structure or material
the crack surfaces move relative to one another and
will depend on: parallel to the leading edge of the crack. Mode III
• Stress level; may occur when the component is subjected to
• Material properties; shear or torsional loading.
• The mechanism or mechanisms by which the
fracture progresses. In the majority of engineering configurations involving
components with cracks, Mode I loading is experienced.
The most successful approach in prediction and Mixed mode I-II loading is sometimes encountered when
prevention of fracture has been to model the behaviour of a crack inclined to the load path exists.
the crack tip as simply as possible but including all The magnitude of the singular stress field is described, in
significant parameters, such as: linear-elastic fracture mechanics (LEFM), by the stress
• Crack length; intensity factor K.. The unit for stress intensity factor is
• State of stress; the product of the stress and the square root of length.
• Fracture toughness. This normally expressed as Force/Length3'2, usually
MPam"2. Because fatigue crack initiation is in general a
The effects of temperature, environment, loading rate, surface phenomenon the stress-intensity factors for a
and fluctuating loads must be taken into account by surface crack, or a corner crack in a plate or at a hole, are
quantifying their influence on the basic parameters. To solutions that are needed to analyse small-crack growth.
characterise the propagation of cracks in structures an For some more complex geometry and loading, SIF
energy-based approach and a stress-based approach are solutions for a large number of crack configurations have
both used. been generated and presented in handbooks, e.g. Rooke
and Cartwright (1976), Tada et al. (1985), Murakami
In an energy-based approach, unstable crack propagation (1987). For very complex geometry and loading patterns,
is postulated to occur when the energy that could be such as those experienced in gas turbine engine
supplied to the crack tip, during an incremental crack components, SIF solutions do not exist. In such cases, SIF
extension, is greater than or equal to the energy required should be estimated by numerical analysis.
for the crack to advance. The five energy components
involved in an incremental crack extension are: The use of stress intensity factor to correlate fatigue crack
growth is meaningful only when small-scale yielding
• Strain energy released; conditions exist because plasticity and non-linear effects
• Energy supplied to the body by external work; can change the stress distribution around a crack front
 5-7

significantly. Among others, strain energy release rate the test temperature is increased. However, the general
and ./-integral are quantities that describe a stress field at behaviour may be modified by microstructural changes.
the crack tip when material plasticity is taken into account
Anderson (1991), Broek (1982), (1989), Hertzberg 4.2. FLOW PROPERTIES
(1983). 4.2.1. TRUE-STRESS-TRUE-STRAIN CURVE
3.2.5. STRESS CONCENTRATION FACTOR The engineering stress-strain curve does not give a true
indication of the deformation characteristics of a metal
The stress intensity factor described above is related to
crack growth. It should not be confused with stress because it is based entirely on the original dimensions of
the specimen, and these dimensions change continuously
concentration factors. Stress concentration factors are
during the test. In addition, ductile metals pulled in
applied to relatively crude stress calculations to take into
tension become unstable and neck down during the test.
account local features which disturb the local macro
Therefore, measures of stress, and strain based on the
stress field.
instantaneous dimensions are used. The true stress is the
4. PROPERTIES OF GAS TURBINE ENGINE
load at any instant divided by the cross sectional area
over which it acts, while the engineering stress, or
COMPONENT MATERIALS
conventional stress, is the load divided by the original
This section describes the properties of materials used to area.
manufacture gas turbine engine components. In Because during tension the cross sectional area of the
particular, attention has been paid to the materials and the specimen is decreasing sharply after necking starts, the
properties that most influence component in-service life. load required to continue deformation decreases.
Material properties can generally be divided into two Therefore, the engineering stress based on original area
main categories: mechanical (e.g. tensile, fatigue, creep) decreases after the point of maximum load has been
and physical (e.g. magnetic constant, thermal reached. In reality the metal may continue to strain
conductivity, thermal expansion coefficient). At low harden all the way up to fracture, so that the stress
temperatures, mechanical properties are moderately required to produce further deformation should also
affected by temperature changes, but time is of little increase. If the true stress, based on the actual cross-
importance. At high temperatures material behaviour sectional area of the specimen is used, it is obvious that
becomes a function of both time and temperature. The the true stress-strain curve increases continuously up to
properties of materials have been divided according to fracture. If the strain measurement is also based on
different criteria in the following sections. instantaneous strain, the curve is known as a true-stress -
true-strain curve, or flow curve, since it represents the
4.1. STATIC PROPERTIES plastic-flow characteristics of the material. In the region
of plastic deformation, the flow curve for metals may be
4.1.1. MONOTONIC STRESS-STRAIN CURVE
expressed as a simple power law.
The engineering tensile test is widely used to provide
basic design information on the strength of materials. In 4.2.2. EFFECT OF STRAIN RATE ON FLOW
this test, a specimen is subjected to a continually PROPERTIES
increasing uni-axial tensile force while the elongation of
An increase in strain rate, defined as E=de/dt increases the
the specimen is recorded. Strain in this test is usually
flow stress. This dependence is more pronounced with
defined as an average strain. This is the ratio of the
increases in temperature, as illustrated in figure 1, in
change in length of the specimen to its original length, L.
which the plot is drawn in a double logarithmic scale.
Similarly, an average stress is the ratio of the tensile
Strain rate sensitivity is a function of the base material
force to the initial cross sectional area of the specimen, A.
and microstructure.
Since both the stress and the strain are obtained by
dividing the load and elongation by constant factors, the 4.2.3. EFFECT OF TEMPERATURE ON FLOW
load-elongation curve has the same shape as the PROPERTIES
engineering stress-strain curve. The main factors that The stress-strain curve and the flow and fracture
influence the detailed shape and particular values on the properties derived from the tension test are strongly
stress-strain curve for a metal are metal composition, heat dependent on the temperature at which the test was
treatment, prior history of plastic deformation, strain rate, conducted. In general, strength decreases and ductility
temperature and the state of stress applied during loading. increases as the test temperature is increased. However,
It should be notes that strength increases with increasing this general behaviour may be modified by structural
strain rate. From a tensile test, the main material changes, precipitation, strain ageing or recrystalisation,
properties of interest: the tensile strength, yield strength which may occur in certain temperature ranges. At high
(yield point), percent elongation and area reduction, are temperatures and/or long exposure times structural
derived, as shown in Appendix 3. changes occur, resulting in time-dependent deformation
The stress-strain-curve and the flow and fracture or creep. The best way to compare the mechanical
properties derived from the tension test are strongly properties of different metals at various temperatures is to
dependant on the temperature at which the test was use the homologous temperature. This is the ratio of the
conducted and on the applied strain rate used in the test. test temperature to the melting point temperature
In general, strength decreases and ductility increases as expressed in Kelvin.
5-8

4.3. CREEP PROPERTIES

Creep is an important manifestation of inelastic time-
dependent deformation behaviour. It is a time-dependent
deformation of materials under stress at elevated
temperatures, relative to their melting point. Creep
behaviour is one of the most critical factors determining
the integrity of elevated-temperature components,
because creeping materials can slowly and continuously
deform under constant stress over a period. Because of
such deformation, unacceptable dimensional changes and
distortions, ending in final rupture of the component, can
occur. However, creep can have a positive effect on high
temperature components because local creep phenomena
can redistribute and thus reduce local stresses.
In its simplest form, creep is the progressive Log of strain rate s'
accumulation of plastic strain, in a component under Figure 1 - Effect of temperature on strain rate
stress at elevated temperature over a period of time.
Creep failure occurs when the accumulated creep-strain
results in a deformation of a component that exceeds the , T, < T, < T, < T4
design limit. Creep rupture, used sometimes o, < a, < a, < a,
interchangeably with the term stress rupture, is an
extension of the creep process to the limiting condition
where the stressed component breaks. The interaction of o4 (IJ
creep with cyclic stressing and the fatigue process is of u / °3 (U
a
great importance in aircraft gas turbine technology.
Creep properties are generally determined by means of a 0 / / a, (TJ
test in which a constant uni-axial load or stress is applied
to the specimen at constant temperature, and the resulting
strain is recorded as a function of time. A creep curve is a ^^^ ^—^^ G (T
'
1%
plot of creep strain sc against time /, but sometimes only "/£■
I 1
time to rupture, tr, is recorded during a creep test. A II
■ >

complete description of the creep behaviour of a metal Time, t

necessitates measurements of several families of creep Figure 2 - Creep curves dependence on stress and
curves for various stress levels and temperatures. temperature
The most general shape of a creep curve is shown in
figure 2. This curve can conventionally be split up into /N

three or four sections as shown in figure 3. It should be tl)

remembered that for a particular material under given c: C"OT™

V
SecorwoivCfeep | *££
o
conditions, not all creep stages are necessarily exhibited. tn
For example, tertiary creep is predominant in super-alloys D
used in gas turbine engines. In figure 3, OA is the region
of intial elastic deformation, which occurs as the full load
is applied. It may be termed an incubation period, because
it happens prior to the other creep stages. AB is called the
^_^~-~~-~~^ r
primary creep stage. It is a period of decreasing creep rate B
where work hardening processes dominate. BC represents A
the secondary or steady state region of creep deformation.
It is frequently the longest portion of a creep curve with e0
equilibrium between strain-hardening and thermal
softening. The final stage, CD, is termed the tertiary O Time , t '
Figure 3 - Typical creep curve showing the three
region. This period of accelerating creep rate occurs due
stages of creep
to internal cavitation, and cracking and leads to complete
fracture of the specimen.
where ec is the total creep strain, a is the operating stress,
A useful empirical model, which accurately represents the t is time and T is the operating temperature, and the other
form of the creep curves obtained for most aero-engine values are constants. Although this set of equations is
materials and which applies over wide ranges of stress based on a phenomenological approach to creep
and temperature, is the equation developed in DERA, UK modelling, the expression describes fully the primary,
Harrison (1994), secondary and tertiary stages of the creep process. Of all
the parameters pertaining to the creep curve, the most
£, = Y^c^fi QIRT (1)
important for engineering applications are Sc and tr,
 5-9

specifically their dependence on temperature and applied progressive failure phenomenon that proceeds by the
stress. This depenedence varies with the applicable creep initiation and propagation of cracks. When the cracks are
mechanisms. large enough the component becomes structurally
unstable and catastrophic instantaneous failure occurs.
Although for polycrystalline materials, the creep curve is
conventionally divided into three regimes: primary, The damage done during the fatigue process is
secondary and tertiary creep, in single crystal materials an cumulative and generally unrecoverable. Vibrational
'incubation period', is often observed before the onset of stress on turbine blades, alternating bending loads on
the primary creep. This is really a period of initial creep blades and fluctuating thermal stresses during starting and
acceleration but with extremely low inelastic strain. stopping or due to power changes are examples of cyclic
Depending on the material and the test conditions such as loading that occur in gas turbine engine components.
load, environment and temperature some forms of Fatigue failure investigations over the years have led to
behaviour may be suppressed while another or others the observation that the fatigue process includes two
become predominant. For example, in single crystal domains. One is of cyclic stressing and one of straining.
super-alloys tertiary creep predominates the entire creep These are significantly different in character, and in each
life while primary creep is negligible and secondary creep of which failure is produced by different physical
is not observed at all (Cary and Strudel, 1977 and 1978). mechanisms. One domain of cyclic loading is that for
When sendee exposed IN738, which is an investment which significant plastic strain occurs during each cycle.
cast polycrystalline material, is tested at 899°C and 90 This domain is associated with high loads and low
MPa the entire creep curve is in the transient regime number of cycles to produce failure and is commonly
(Primary plus secondary creep). This is due to the effects referred to as low-cycle fatigue (LCF) or cyclic strain-
of oxidation and precipitation of grain boundary carbides. controlled fatigue. The other domain of cyclic loading is
Tertiary creep does not appear to occur. (Castillo, and that for which the strain cycles remains predominantly in
Koul, 1988). To predict behaviour of a component in the elastic range. This domain is associated with lower
service and prevent the occurrence of the most dangerous loads and high number of cycles to produce fatigue
mode, the engine designer and user must understand the failure, and is commonly referred to as high cycle fatigue
physical mechanisms of creep and know their operating (HCF).
ranges of temperature and stress.
Transient creep is defined as the creep regime where the 4.4.1. CYCLIC STRESS-STRAIN CURVE
creep rate decreases with time until a minimum value, Under strain controlled cyclic loading a material may
smi„, is reached. For a polycrystalline material the creep remain cyclically stable or it may exhibit cyclic strain
rate, including the minimum creep rate, e„„„ , in this hardening or softening. The associated stress-strain
regime is often found to be sensitively dependent on the hysteresis loops tend to stabilise after a relatively small
grain size and grain boundary micro-structural features, number of cycles although as temperature increases, this
such as grain boundary serrations and grain boundary 'shakedown condition' is correspondingly delayed and
precipitate distribution. These observations indicate that may never be reached. A cyclic stress-strain curve can
grain boundary sliding (GBS) is the dominant then be constructed from peak shakedown stresses and
deformation mechanism. GBS occurs as the result of corresponding strains, each LCF test providing a single
dislocation glide along the grain boundary plane, which is pair of points. A power law similar to the curve for a
more easily activated at the interfaces of micro-structural monotonic stress-strain relationship may approximate the
discontinuities, such as grain boundaries, upon resulting cyclic stress-strain curve, see figure 4.
application of a creep load (Langdon, 1970), (Wu and Cyclic strain controlled fatigue as opposed to our
Koul, 1995). This mode of deformation can promote
previous discussion of cyclic stress controlled fatigue,
grain boundary cavitation and oxidation. The
occurs when the strain amplitude is held constant during
accumulation of this damage contributes to tertiary creep.
cyclic loading. LCF is found both in thermal cyclic
In a single crystal material this mode of
fatigue, where a component expands and contracts in
deformation/damage accumulation is absent, and transient
response to fluctuations in the operating temperature, and
creep is caused by dislocation multiplication and
in reversed mechanical bending between fixed
hardening due to the formation of a dislocation network.
displacements.
Tertiary creep is defined as the creep regime where the
The localised plastic strain in notches subjected to either
creep rate increases with time, leading to the final rupture.
cyclic stress or strain conditions results in strain
The acceleration in creep rate may be caused by either or
controlled conditions near the root of the notch. This is
both dislocation multiplication (McLean, 1983), Dyson
due to the constraint effect of the larger surrounding mass
and Gibbons, 1987) or damage accumulation (Ashby and
of essentially elastically deformed material.
Dyson, 1985). Usually they happen together.
A metal may undergo cyclic strain hardening, cyclic
4.4. CYCLIC PROPERTIES strain softening or remain cyclically stable. This means
Any material subjected to repetitive or fluctuating stresses that the material mass become either more or less
will fail at a stress much lower than that required for resistant to the applied stresses or strains.
failure, on a single application of load in a tensile test. The mechanism of cyclic hardening and softening is
Failures occurring under such conditions are termed related to the nature and stability of the dislocation
fatigue failures. Fatigue may be characterised as a substructure of the material. For an initially soft material,
5-10

the dislocation density is low. During plastic strain 4.4.3. Low CYCLE FATIGUE (LCF)
cycling the dislocation density increases significantly In gas turbine engine blades or discs, large mechanical
leading to significant strain hardening. For initially hard stress and thermal gradients occur during operation.
material, strain cycling causes a rearrangement of These may give rise to significant damage accumulation
dislocations into a new configuration, which offers less in only a few thousands of these large cycles during the
resistance to deformation. This is strain softening. In both design lifetime, so that LCF design and analysis methods
. a are of great importance. For many components, the
Cyclic response of a material in the failure critical location was
found to be strain dependent. It is assumed that smooth
^/^Z^(\ L Monotonie specimens tested under strain-control can simulate fatigue
S^^T \ J- **" / damage of the real components. This is because
S i^l / / equivalent fatigue damage is assumed to occur in the
* s\ 1\ /1 /1 material at the notch root, and in the smooth specimen,
• 11/ / when both are subjected to identical stress-strain
histories. In this fatigue domain the cyclic loads are
relatively high, significant plastic strain is introduced
during each cycle, and short lives or a low number of
III £ cycles to failure are exhibited.
The LCF test procedure results in the plot of the plastic
/
1 1 I \
/ 1 1 s
s strain range, Aep, vs. the number of cycles, N, which is
/ \ \ \ s
1 1 1 Ji'' / similar to the S-N plot. Figure 5 shows that a straight line
1 r T 1 sSS is obtained when the test results are plotted on a double
logarithmic scale:
A*.
Figure 4 - Monotonie- and cyclic stress-strain curves '■£- = e',(2N)e (3)
for a cyclically hardening metal
where Ae/2 is the plastic strain amplitude, ^ is the fatigue
cases the newly generated dislocations assume a stable ductility coefficient defined by the strain intercept at
configuration for a given material and for a given 2N=1, 2N is the number of strain reversals to failure (one
magnitude of cyclic strain imposed.

4.4.2. HIGH CYCLE FATIGUE (HCF)

The basic method for presenting HCF data is by means of
S-N curves, figure 4, in which the number of cycles to
failure is plotted against the applied stress range. N is
usually taken to denote the number of cycles applied at
constant stress ratio, R, for complete specimen fracture.
For some engineering materials, the S-N curve becomes
almost flat at low stress, indicating a threshold value of
stress below which failure will not occur for practical 10 10 10
purposes. The threshold value of stress is defined as the Cycles (o Failure
fatigue limit. The S-N curve does not distinguish between Figure 5 - Low cycle fatigue curve for steel
crack initiation and crack propagation.
The number of cycles corresponding to the fatigue limit, cycle is two reversals), c is the fatigue ductility exponent,
i.e., at low stress, relates primarily to crack initiation, which varies between -0.5 and -0.7 for many metals. This
whereas for at high stresses the fatigue life corresponds equation is known as the Coffin-Manson relation. A
primarily to crack propagation. Many variables in the test smaller value of c results in a larger value of a fatigue
procedure, such as type of loading, mean stress, life. The last equation describes the relation between the
temperature, environment, specimen size, specimen plastic strain component and fatigue life in the low-cycle-
surface condition, and stress concentrations, affect the fatigue regime.
fatigue life. The elastic component of strain is often described in
The S-N curve in the high-cycle region is sometimes terms of a relationship between the true stress amplitude
described by the Basquin equation: and the number of load reversals in a manner equivalent
to the Basquin formula:
Napa = C
(2) Agg-£ ; (4)
(7a - or (2Nr)
where N is the number of cycles to failure, <ja is the stress
amplitude and p and C are empirical constants.
where AeJ2 is the elastic stress amplitude, E is the
modulus of elasticity, aa is the stress amplitude, crr is the
 5-11

fatigue strength coefficient, defined by the stress intercept involve thermo-mechanical fatigue processes. Hot section
at one load reversal (2Nj~l), Nfis the number of cycles to components of gas turbine engines are subjected to cyclic
failure, 2Nf is the number of load reversals to failure, b is temperatures simultaneously with cyclic stress. Analysis
the fatigue strength exponent, which varies in the range - of these loadings and consideration of the associated
0.5 and-0.12 for metals. fatigue damage becomes very complex, and many
simplifications have to be introduced. Historically,
The fatigue resistance of a material subjected to a strain thermal fatigue was considered as iso-thermal low-cycle-
range could be estimated by superposition of elastic and fatigue at the maximum temperature of the thermal cycle
plastic components. Summing Eq (3) and (4) results in to which a component was subjected. However recent
the expression: advances in test systems and numerical calculation
methods have made it possible to conduct thermo-
'£, ■p <Jf
■+■ _
(2Nr)b + Er(2Nf)c mechanical fatigue tests under well controlled conditions
and also to analyse numerically complex thermo-
(5) mechanical cycles. It has been found that TMF loading
The total strain life curve would approach the plastic can be more damaging than pure LCF loading for the
strain life curve at large strain amplitudes and approach same total strain range applied.
the elastic strain life curve at low total strain amplitudes,
4.5. FRACTURE MECHANICS PROPERTIES
figure 6.
4.5.1. FATIGUE CRACK GROWTH RATE (FCGR)
4.4.4. THERMO-MECHANICAL FATIGUE (TMF)
Fatigue cracks usually originate at geometric stress
Aircraft engine components, particularly hot section concentrations (holes and fillets) or at manufacturing
components, experience severe operating conditions flaws. The fatigue life of a component is divided into
involving complex combinations of cyclic mechanical several phases: crack nucleation, micro-crack growth,
stress, cyclic temperature and varying environmental macro-crack growth and failure. Crack nucleation is
conditions. Although the stress or strain that causes associated with cyclic slip and is controlled by the local
fatigue is often produced mechanically, it is perhaps even stress or strain concentrations. Micro-crack growth, also
referred to as small crack growth, is the growth of cracks
VV1' from inclusions, voids or slip bands, in the range of 1 to
10 um in length. Fracture mechanics parameters
successfully correlate and predict fatigue crack growth
and fracture in the region of macro-crack growth and

f
failure. They can be applied once the crack has developed
o.
through several grains.
<E . Total
c ^ Reliable crack-propagation models permit the
S
tn
"hi _
-—^ implementation of damage tolerant designs. These
o Elastic'
D)
1 ^ recognise the inevitability of cracks in engineering
V^~ PiasJ«^
structures and aim at determining the critical crack load
and length, which will preclude failure in a conservatively
N ;
estimated service life.
Crack length increases with the number of cycles and the
Log of Reversals to Failure stress level. This may be expressed by a general plot of
da/dN versus a Stress Intensity Factor (SIF) range, AK.
Figure 6 Fatigue strain life curve obtained by The Fatigue Crack Growth Rate (FCGR) da/dN, is the
superposition of elastic and plastic crack extension Aa, during a small number of fatigue
strain-life equations cycles AN. FCGR is a function of the stress intensity
factor range, AK, the load ratio, R, temperature, and the
more usual to find the cyclic strain produced by a cyclic
environment.
thermal field. If the natural thermal expansions and
contractions of a part are either wholly or partially Plots of da/dN vs AK for a given R are usually presented
constrained as often happens when a thermal gradient is on double logarithmic graphs as shown in figure 7. This
applied to the component, then cyclic strains and stress FCGR curve has a sigmoidal shape that divides the curve
result. These cyclic strains produce fatigue failure, just as into three regions.
if external mechanical loading produced them.
Component cracking can be thus induced, not only Region I is bounded by a threshold value AK,,„ below
mechanically but may be also caused or accelerated by which there is no observable fatigue crack growth. At
thermally induced stress. The damage incurred by stresses below AK,h cracks behave as non-propagating
components under such conditions is known as thenno- cracks.
mechanical fatigue (TMF). Thermal fatigue problems Region II is characterised by a linear log-log relationship
involve not only the complexities of mechanical loading, between da/dN and AK. Crack growth rate in this region
but all the temperature induced problems as well. is influenced by material microstructures, mean stress,
Most low cycle fatigue problems at high temperatures and environment. For a fixed load ratio, R, and a wide
5-12

range of AK values in region II, the data can be Region III is a region of accelerating crack growth. Here
represented as a straight line on the double logarithmic Km(ix approaches Kc, the fracture toughness of the material.
plot. The Paris equation describes crack growth in region In Region III, the fatigue crack growth rate rises to an
II as follows: infinite slope. FCGR is strongly influenced by
microstructure, mean stress and specimen thickness in
this region.
100
Increasing the mean stress in the fatigue cycle as
6
5 • In air described by a stress ratio causes an increase in the crack
4 A In argon growth rates in all regions of the FCGR curve. Generally
3 ■ Prediction line (m=2) the effect of increasing R is less in Region II than in
— Oxidation effect Regions I and III. The Forman relation usually gives the
influence of a stress ratio on the Paris formula:
10-1 da A(AK)>
\ IN (l-R)Kc-*K (7)
E
E where Kc is the fracture toughness for the material and
XI thickness, and R is the stress ratio.
"Ü As mentioned before, the use of the stress intensity factor
10-2 to correlate fatigue crack growth has meaning only when
small-scale-yielding conditions exist at the crack tip. This
6
5 is because plasticity and non-linear effects can change the
4 stress distribution around the crack front significantly. A
3 simple approach has been developed to modify the elastic
2 stress-intensity factor to "correct" for plastic yielding at
the crack tip. The approach was to add a plastic zone
radius to the crack length and thus calculate stress
10-3 ■■■■■'■•■'

6 7 e s 102 2 intensity factor at the effective crack length (a+ry). Other

K(MPamV2) non-linear phenomena in crack growth behaviour include
crack closure and the effect of constraints at the crack tip,
Figure 8 Crack growth rate v. stress intensity factor but analysis of these effects definitely goes beyond the
for Udimet. After Wu and Koul (1996) scope of this report.
da The engineering problem of a crack existing in a structure
= C(AKy (6) can be partially solved by linear fracture mechanics. In
dN particular, the following questions can be answered:
where C and n are the empirical constants determined What is the residual strength of a structure, as a
experimentally for a given material. The effects of load function of crack size under a given load?
frequency, temperature, and operating environment are What is the maximum permissible crack size that a
empirically contained in these constants. Note that in this structure can tolerate?
formulation C has the unit of m/cycle(MPa t/n/"" and n is How long does it take a crack to grow from its
a dimensionless exponent which can be thought of as the initial to the maximum permissible size?
slope of the fatigue crack growth rate curve. For most
What is the service life of a structure when a certain
aerospace materials, the value of« ranges from 2 to 4.
flaw size is assumed to pre-exist?
Special care should be taken if data from literature During the period available for crack detection, how
sources are used for component life prediction to ensure often should a structure be inspected for cracks?
that the data corresponds to the actual environment. Most 4.5.2. FRACTURE TOUGHNESS
da/dN vs AK curves are generated using R * 0 and
The theory of Linear Elastic Fracture Mechanics (LEFM)
room temperature. Interpolation of these curves to the
provides a means for estimation of the fracture loads of
actual conditions of intended application should be
structures containing sharp flaws of known size and
performed with great caution. Extrapolation beyond the
location. An important observation from studying fracture
testing region is not safe.
behaviour is that the magnitude of the nominal applied
Eq. (6) provides a link between fracture mechanics and stress, which causes fracture, is related to the size of the
fatigue, because the fatigue life of the component can be crack. For central through-the-thickness cracks, normal to
obtained by the integration of this equation between the the applied tensile stress, as the tensile loading on the pre-
limits of initial crack size and the final crack size. This packed plates is slowly increased, the crack length
depends on knowing the stress intensity factor, K, for the slowly increases and then abruptly extends to failure.
cracked component as a function of component loading.
Experience has shown that the abrupt change from slow
Such integration usually has to be done by an iterative
crack propagation to rapid crack propagation establishes
process.
an important material property named fracture toughness.
 5-n

The fracture toughness is used as a design criterion in (l-m)

fracture prevention. It is defined as the critical stress
_2R_
a
(A*\) ~ G
(A=\) (8)
intensity factor for material, Kk. While regarded as a \-R
material property, the fracture toughness may be strongly
The Walker constant, m, having a value between 0 and 1
affected by operating temperature, heat treatment, and
can be assumed to be a certain value based on a
constraints. Therefore, care must be taken in design to use
knowledge of the material. The Walker method is used to
a value of fracture toughness that closely represents the
cover mean stress effects without generating full design
expected operating temperature and environments for the
curves for other mean stress conditions. The assumption
actual component. may be conservative, i.e. it results in a higher derived
ASTM has developed a standard test procedure for stress than actual testing would show. It can be also
determining plane strain fracture toughness. The compact verified by regression analysis of fatigue test data
tension specimen is the most commonly used test generated at various mean stresses, i.e. R ratios.
specimen geometry. Application of this correction requires substantially less
testing than would be necessary to generate a full design
4.5.3. CREEP CRACK GROWTH RATE curve for the means stress conditions of interest.
Sustained load crack growth is time-dependant sub-
critical crack growth occurring under a stress well below 4.7. INFLUENCE OF MULTI-AXIAL STRESSES ON
tensile failure. Creep crack growth usually occurs only at FATIGUE AND FRACTURE
temperatures greater than 50% of the melting point and is An effect of the multi-axial state of stress and strain fields
not a failure process generally associated with fracture on crack initiation and propagation is of paramount
critical components. importance since most laboratory research involves tests
To find the creep crack growth properties for the material, on simple geometry specimens subjected only to either
different cracked specimens are subjected to a constant axial, torsional or bending loading. Such forces are rarely
load at elevated temperature and the crack extension is applied in combination in a laboratory environment.
measured as a function of time. Examples of crack However in the real world complex cyclic forces varying
growth against time curves are shown in figure 8. These in time may be imposed over a wide range of conditions.
curves indicate an increasing crack propagation rate with An example is that of thermal and mechanical stress
increase in time after the crack growth has become well transients in a gas turbine disc which create a bi-axial or
established. This is mainly because under constant load tri-axial stress field. The stress in most engineering
conditions, the magnitude of the stresses generated at the applications is bi-axial, while the state of strain is usually
crack tip in the specimens increases with crack extension. tri-axial. Miller and Brown (1984) and Ellyin (1997)
presented a brief review of multi-axial fatigue problems.
4.6. INFLUENCE OF MEAN STRESS ON FATIGUE A typical approach to account for the effect of multi-
AND FRACTURE axiality on the LCF life, is to define an equivalent total
4.6.1. INFLUENCE OF MEAN STRESS ON HCF strain range. This is calculated from the components of
strain. An estimate of the life of a component subjected to
Most of the fatigue data published have been determined multi-axial strain can be made from uniaxial low cycle
for conditions of completely reversed stress cycles, oir=0. fatigue data, expressed as total strain range vs number of
However, in engineering practice components are often cycles to failure.
loaded by superimposing alternating stress on a non-zero
constant mean-stress. 4.8. ENVIRONMENTAL RESISTANCE
There are several possible methods of determining and Environmental effects on the fatigue of metals may be
presenting an S-N curve where the mean stress is not severe. Quantitative fatigue-life predictions are often not
equal to zero. The two most commonly used methods are possible because of the many environmental factors that
presented in figures 9a and 9b. In figure 9a, the maximum influence fatigue behaviour, and lack of suitable data. For
stress is plotted against N or logN for constant values of example:
stress ratio R. In the case of completely reversed stress R=
• In corrosion fatigue, frequency effects can be quite
-1.0. In figure 9b the alternating stress is plotted vs
substantial;
number of cycles to failure at a constant value of mean
stress. Other methods of plotting these data include the • In non-corrosive environments frequency is usually
maximum stress vs cycles to failure at constant mean a second order concern;
stress and maximum stress vs cycles to failure at constant • At elevated temperatures mean stress effects are
maximum stress. extremely complex because of interactions between
creep and fatigue, and the environment;
4.6.2. INFLUENCE OF MEAN STRESS ON LCF • At elevated temperatures, the linear elastic stress
The Walker method used for correction for non-zero intensity factor, K, has more limitations because of
mean stress in the FCGR analysis is also applied as a way appreciable plasticity;
of deriving material LCF data for stress conditions other • Substantial reductions in fracture toughness can
than zero minimum stress (A *1). The following formula occur at low temperatures, which reduce the critical
is used: crack sizes at fracture.

Design with environmental considerations places great

5-14

emphasis on real-life product testing, inspection, service simple, is very difficult to implement in practice. The
history analysis, and experience. proper assessment of the amount of damage incurred by a
component operating at any given stress level Sj for a
4.8.1. OXIDATION RESISTANCE specified number of cycles ^ is not easy. Many different
Oxidation plays an important role in high temperature cumulative fatigue damage theories have been proposed
fatigue and creep. Protective oxide formation is a major
factor in the fatigue resistance of a given material. These
protective oxide films can be broken down by reversed I1 /
slip, which causes much shorter high temperature crack 1 //
initiation life. Crack propagation rates may also be
0
+-
j /
i /
//
//
Ö
accelerated by high temperature oxidation, and freshly IX ////
exposed surfaces produced by local plasticity can oxidise
rapidly.
Grain boundaries may be selectively attacked by oxygen. 0
Aj°=C(AKr
Tests at high temperature in a vacuum or inert atmosphere
have shown substantial increases in fatigue and creep of

log (day
Crack
resistance compared to high temperature tests in air or
combustion gases. Thus, local oxidation is a primary
factor in the degradation of fatigue and creep resistance at
high temperatures. Frequency effects are also
substantially reduced at high temperatures in a vacuum or 0)
atmosphere.
Ö)
-i—
4.8.2. CORROSION FATIGUE AND STRESS- 0 Region 1 Regi on II Region III
CORROSION CRACKING LL
Many environments affect fatigue behaviour, and most
components of gas turbine engines interact with air, i 1
water, salt water and hot gases. Corrosion fatigue refers to ArC AK„
the joint interaction of a corrosive environment and Stress Intensity Factor Range
repeated stressing. The combination of these two acting
together is more detrimental than either acting separately. log(AK)
That is, repeated stressing accelerates the corrosive action Figure 7 Fatigue crack growth rate model
and the corrosive action accelerates the mechanical
fatigue mechanism. Corrosive environments may be for assessing fatigue damage and fatigue failure caused
detrimental even under static loads, particularly in higher by component spectrum loading. Of these theories, the
strength alloys. Stress corrosion cracking is one form of Palmgren-Miner rule is presented here and a double linear
environmentally assisted fracture that may occur after a damage rule is described in Appendix 3.
sustained static load.
5.1.1. PALMGREN-MINER RULE
5. ANALYTICAL TECHNIQUES FOR DAMAGE One of the first cumulative fatigue damage theories was
CHARACTERISATION proposed by Palmgren (1924) and later developed by
Miner (1945). This linear theory is generally referred to
5.1. CUMULATIVE FATIGUE DAMAGE as the Palmgren-Miner rule or linear damage rule. It
In gas turbine engine operation, the alternating stress postulates that the damage fraction at any stress level Si,
amplitude varies during component service life, forming is linearly proportional to the ratio of number of cycles of
so-called spectrum loading. Therefore the direct use of operation at this stress level, ni, to the total number of
standard S-N curves is very difficult in engine fatigue cycles that would produce failure at that stress level, Ni:
assessment because these curves were developed and
presented for constant stress amplitude operation.. It is D,= (9)
thus important to have a theory or hypothesis, verified by N,
testing, that will permit good fatigue-life estimates to be
made for components subjected to spectrum loading Failure is predicted to occur if:
using standard S-N curves.
The basic postulate used for spectrum-loading fatigue (10)
evaluation is that any loading cycle at any given mean pttj
stress level produces fatigue damage. It is further Miner's rule simply sums the fractional life consumed by
postulated that the damage incurred is permanent and that each type of damaging cycle identified, e.g. stop-max,
operation of a component at several different stresses in idle-max. The number of cycles to crack initiation for
sequence will result in an accumulation of total damage. each major and minor damaging cycle is determined by
When the total accumulated damage reaches a critical referring to the appropriate stress conditions on the
value, a fatigue failure occurs. This concept, however minimum design S-N curves. The number of cycles or
 5-15

hours required to accumulate enough cycles for the sum changes in conditions during engine starting, operation
to reach 1 is defined as the life of the component for that and shutdown result in large transient temperature
gradients. If these transients are repeated, the differential
type of cycle.
thermal expansion during each transient results in
The main advantage of the Palmgren-Miner rule is it's thermally induced cyclic stresses. The extent of the
simplicity. This rule, sometimes known as Miner's rule is resulting fatigue damage depends on the nature and
widely accepted and used in the industry for LCF frequency of the transients, the thermal gradients in the
summation. It is simple, and fairly accurate when component and the material properties.
damaging minor cycles are of the same order of
magnitude. It is best suited for LCF in engine components
because there is not a great diversity of cycles. There are
two significant shortcomings of these linear theories. The
order of application of various stress levels is not taken
into account. Damage is assumed to accumulate at a
constant rate for a given stress level, regardless of
component loading history.

5.2. CREEP DAMAGE

The design lifetimes of engineering components are often
based on time to a specific strain or time to rupture. For
design purposes it is usually more convenient to re-plot
the data as shown in figures 10 and 11. Secondary creep-
rate, time to specific strain (say 1%), or rupture life can 10 = to6 10'
then be read off at any desired stress and temperature. Cycles to Failure
Data presented in figure 10 can be used for extrapolating a)
to other stresses at a given temperature while data in
figure 11 can be used for extrapolating to other
temperatures at a particular stress. Straight lines, with the
slopes shown, will be obtained in the figures when the
data can be correlated by the stress and temperature terms
given in Equation (A-20) of Appendix 3.
A simple straight-line relationship is usually not obtained
on plots like those in figures 10 and 11, when attempts
are made to include a wide spread of data. This is because
different values of n and Q are needed to describe
different creep mechanisms. Consequently, to achieve
more reliable extrapolations, time-temperature creep
parameters have been devised for superimposing all of
the results on the so-called 'master curve' for a material. 10 = 10s 107
The basis of the creep parameters is that time and Cycles to Failure
b)
temperature have similar effects. The same creep
behaviour is obtained at the same stress in a short time at Figure 9 - Two methods of representing fatigue data
high temperature, as is attained in a long time at low for non-zero mean stress. After Dieter
temperature. This is clearly a simplification but it does (1986)
result in satisfactory extrapolations of creep data.
Components that are subject to thermally induced stress
The need for extrapolation techniques that permit generally operate within the creep range so that damage
estimation of the long-term creep and rupture strengths of due to both fatigue and creep has to be taken into account.
materials based on short-duration tests is thus a very real Gas turbine blades and disks are particularly subject to
and important one in design, quality control, and life severe thermal gradients during start-up and shut down.
prediction. Parametric techniques incorporate time, stress During engine shut down, tensile stresses may be
and temperature test data into a single expression. One of produced near cooling component surfaces, and may
the parameters of concern is the time to rupture or time to create thermal-shock cracks. In view of the importance of
specific strain. combined creep and fatigue damage with respect to
Three parametric techniques are described in Appendix 3: component reliability, many attempts have been made to
the Larson-Miller parameters, the Monkman-Grant develop damage rules that will help in design and in
relationship, and the projection concept. component life prediction under creep-fatigue conditions.
To develop these damage rules, several types of
5.3. CREEP-FATIGUE INTERACTION
laboratory test are used:
In life assessment of high-temperature components, the
influence of interaction between failure modes has been • Strain controlled tests with hold periods at constant
considered. Hot section components in gas-turbine strain;
engines operate at high temperatures. Additionally,
5-16

• Creep tests under cyclic stress or strain;

• Alternating creep and fatigue tests; . \
• Strain-controlled tests under isothermal conditions;
• Strain-controlled tests under thermal and
mechanical cycling.
, y Slope=(-Q/R)
The first test is the most common, while the last three
tests are generally known as thermo-mechanical fatigue
tests.
The principal method of studying creep-fatigue
interaction has been to conduct strain-controlled fatigue C/2>0-\
tests with variable frequencies with and without a dwell
period (hold time) during some portion of the test. The *-
lower frequencies and the longer dwell periods can allow 1/T K
creep to take place. In pure fatigue tests, at higher Figure 10 - Creep properties v. inverse of temp-
frequencies and short dwell periods, the fatigue mode erature at different stresses. After
dominates and failures start near the surface and Webster and Ainsworth (1994)
propagate trans-granularly. As the dwell period is
increased, or the frequency decreased, the creep
component begins to play a role with increasing creep-
fatigue interaction. In this region, fractures are of a mixed o
mode, involving both fatigue cracking and creep
Slope=(-n)
cavitation. With prolonged dwell periods and occasional
interspersed cycles, creep processes dominate, so that the
damage mechanism can be treated as almost pure creep.
In instances where oxidation effects contribute
significantly to the creep fatigue interaction, the situation
is more complex.
Several damage rules are used for estimating cumulative
damage under creep-fatigue conditions. They include:
• The linear damage-summation method , logt,
• The strain range partitioning method , -tog ES
• The ductility exhaustion method.
Figure 11 - Stress dependence of creep properties at
These are described in more detail in Appendix 3. different temperatures. After Webster
and Ainsworth (1994)
6. CONCLUSION
1995.
A thorough understanding of the fundamentals of Ashby, M.F. and Dyson, B.F., Creep Damage
materials science is essential for practitioners in the field Mechanisms and Micromechanisms, Advances in
of usage monitoring. This chapter, together with appendix Fracture Research, Vol. 9, Pergamon Press, 1985.
3, seeks to provide a theoretical basis for further study. Bammann, D. and Krieg, R.D., Summary and Critique,
The following references provide a thorough supporting Unified Constitutive Equations for Creep and Plasticity,
bibliography, and form the basis of this chapter. Ed. by A.K. Miller, Elsevier Applied Science, pp. 303-
336, 1987.
7. REFERENCeS Banks-Sills, L., Application of the finite element method
AGARD Engine Disc Co-operative Test Programme, Ed. to linear elastic fracture mechanics, Applied Mechanics
by A.J.A. Mom and M.D. Raizenne, AGARD SMP, Reviews, Vol. 44, No. 4, pp. 447-461, 1991.
AGARD-R-766, 1988. Bannantine, J.A., Corner, J.J. and Handrock, J.L.,
AGARD Engine Disc Co-operative Test Programme, Fundamentals of Metal Fatigue Analysis, Prentice Hall,
AGARD SMP, AGARD-R-766 (Addendum), 1993. 1990.
Agatanovic, P., and Heidenreich, R., An Integrated Barsom, J.M., Structural Problems in Search of Fracture
Approach to Creep-Fatigue Life Prediction, Low Cycle Mechanics Solutions, Fracture Mechanics; Twenty-Third
Fatigue, ASTM STP 942, pp. 500-518, 1988. Symposium, ASTM, STP 1189, pp. 5-34, 1993.
Anderson, T.L., Fracture Mechanics, CRC Press, 1991. Beres, W., NSIFS Postprocessor for Displacement-Based
Annis, CG., ]r.,An Engineering Approach to Cumulative Stress Intensity Factor Extractions from NISA II
Damage Fracture Mechanics in Gas Turbine Disks, Calculations Result, Report of the Institute for Aerospace
ASME Paper 79-GT-134, 1979. Research, National Research Council Canada, Ottawa,
Arnold, S.M. and Freed, A.D., Constitutive Equations for ST-727, 1994.
Solid Mechanics, Computational Structures Technology, Bernstein, H.L., An Evaluation of Four Creep-Fatigue
Ed. by A.K. Noor and S.L. Venneri, ASME, pp. 1-38, Models for a Nickel-Base Superalloy, Low Cycle Fatigue
 5-17

and Life Prediction, ASTM STP 770, pp. 105-134, 1982. Damage Tolerant Design of Turbine Engine Disks,
Bernstein, H.L., at at., Prediction of Thermal-Mechanical ASME Paper 82-GT-311, 1982.
Fatigue Life for Gas Turbine Blades in Electric Power Cook, T.S., Laflen, J.H., Van Stone, R.H. and Wright,
Generation, Thermomechanical Fatigue Behavior of P.K., TMF Experience with Gas Turbine Engine
Materials, ASTM STP 1186, pp. 212-238, 1993. Materials. Thermal Mechanical Fatigue of Aircraft
Besuner, P.M., Probabilistic Fracture Mechanics, Engine Materials, AGARD SMP, AGARD-CP-569, pp.
Probabilistic Fracture Mechanics and Reliability, 16-1-16-12, 1996.
Martinus Nijhoff Publishers, pp. 387-459, 1987. Cook, T.S. and Laflen, J.H., Considerations for Damage
Bodner, S.R. Review of Unified Elastic-Viscoplastic Analysis of Gas Turbine Hot Section Components, ASME
Theory>, Unified Constitutive Equations for Creep and Paper 84-PVP-77, 1984.
Plasticity, Ed. by, Miller, A.K., Elsevier Applied Science, Cowles, B.A., Life Prediction in High Temperature
pp. 273-301, 1987. Environment: Overview of a Current Gas Turbine Engine
Bogdanoff, J.L. and Kozin, F., Probabilistic Models of Approach, Materials Science and Engineering, Vol.
Cumulative Damage, John Wiley and Sons, 1985. A103, pp. 63-69, 1988.
Boire-Lavigne, S., Gendron, S., Marchand, N.J. and Cowles, B.A., High Cycle Fatigue in Aircraft Gas
Immarigeon, J.-P., Modelling of Thermomechanical Turbines-An Industry Perspective, International Journal
Fatigue Including Phase Transformations, Thermal of Fracture, Vol. 80, No. 2-3, pp. 147-163, 1996.
Mechanical Fatigue of Aircraft Engine Materials, Dieter, G.E., Mechanical Metallurgy, Third Edition,
AGARD SMP, AGARD-CP-569, pp. 17-1-17-10, 1996. McGraw-Hill, 1986.
Brock, D., Elementa>y Engineering Fracture Mechanics, Domas, P.A., Elevated Temperature Component Fatigue
Martinus Nijhoff Publishers, 1982. Robustness-An Holistic Approach, Thermal Mechanical
Broek, D., The Practical Use of Fracture Mechanics, Fatigue of Aircraft Engine Materials, AGARD SMP,
Klüver Academic Publishers, 1989. AGARD-CP-569, pp. 21-1-21-13, 1996.
Broichhausen, K., Effectiveness and Efficiency: The Donachie, M.J., Jr. and Sullivan, C.P., Properties and
Challenge of Aero-engine Development, XIII ISOABE Selection of Materials for Aircraft Gas Turbines,
Conference, Chattanooga, TN, 1997. Sawyer's Gas Turbine Engineering Handbook. Third
Burnett, D.S., Finite Element Analysis, Addison-Wesley, Edition. Vol. Ill, Sawyer J.W. and Japikse, D. (Eds.),
1987. Turbomachinery International Publications, pp. 3-1—3-
Cary, C. and Strudel, J.L., Apparent and Effective Creep 28, 1985.
Parameters in Single Crystals of Ni-based Superalloy-^. Dundas, R., Mechanical Design of the Gas Turbine,
Incubation Period, Acta Metallurgica, Vol. 25, pp. Sawyer's Gas Turbine Engineering Handbook. Third
161-111, 1977. Edition. Vol. 1, Ed. by Sawyer J.W. and Japikse, D.,
Cary, C, and Strudel, J.L., Apparent and Effective Creep Turbomachinery International Publications, pp. 10-1—10-
Parameters in Single Crystals of Ni-based Superalloy—II. 88, 1985.
Secondaiy Creep, Acta Metallurgica, Vol. 26, pp. Dyson, B.F. and Gibbons, T.B., Tertiary Creep in Ni-
859-870,"l978. based Superalloy, Acta Metallurgica, Vol. 35, pp.
Castillo, R. and Koul, A.K., Assessment of Creep 2355-2369, 1987."
Damage and Its Rejuvenation in Ni-based Superalloy, Edmunds, T.M., and Lawrence, R.A., Monitoring Engine
Metallurgical Transactions, A, Vol. 19A, pp. 2049-2066, Thermal Stresses, Engine Cyclic Durability by Analysis
1988. and Testing, AGARD PEP, AGARD-CP-368, pp. 2-
Chaboche, J.L., Lifetime Predictions and Cumulative 1-2-20, 1984.
Damage under High-Temperature Conditions, Low Cycle Ellyin, F., Fatigue Damage, Crack Growth and Life
Fatigue and Life Prediction, ASTM STP 770, pp. 81-104, Prediction, Chapman and Hall, 1997.
1982. Evans, W.J., Smith, M.E.F. and Williams, C.H.H., Disc
Chaboche, J.L., Constitutive Equations for Cyclic Fatigue Life Predictions for Gas Turbine Engines, Engine
Plasticity and Cyclic Viscoplasticity, International Journal Cyclic Durability by Analysis and Testing, AGARD PEP,
of Plasticity, Vol. 5, No. 3, pp. 247-302, 1989. AGARD-CP-368, pp. 11-1-11-13, 1984.
Chamis, C.C. and Johns, R.H., Computational Engine Fox, E.P. and Safie, F., Statistical Characterisation of Life
Structural Analysis, ASME Paper 86-GT-70, 1986. Drivers for a Probabilistic Design Analysis, AIAA Paper
Chulya, A. and Walker, K.P., A New Uniformly Valid 92-3414, 1992.
Asymptotic Integration Algorithm for Elasto-Plastic- Fuchs, H.O. and Stephens, R.I., Metal Fatigue in
Creep and Unified Viscoplastic Theories Including Engineering, John Wiley & Sons, 1980.
Continuum Damage, NASA, TM-102344, 1989. Gan, X., Kilic, M. and Owen, J.M., Flow and Heat
Coffin, L.F., Some Perspectives on Future Directions in Transfer Between Gas-Turbine Discs, Heat Transfer and
Low Cycle Fatigue, Low Cycle Fatigue, ASTM STP 942, Cooling in Gas Turbines, AGARD PEP,
pp. 5-14, 1988. AGARD-CP-527, pp. 25-1-25-11, 1995.
Coffin, L.F., Damage Monitoring, Life Prediction and Haiford, G.R., Low Cycle Thermal Fatigue, NASA
Life Extension, Low Cycle Fatigue and Elasto-Plastic TM-87225, 1986.
Behaviour of Materials, Ed. by K.-T. Rie, Elsevier Haiford, G.R., Thermal Fatigue Durability for Advanced
Applied Science, pp. 1-12, 1987. Propulsion Materials, NASA TM-102348", 1989.
Collins, J.A., Failure of Materials in Mechanical Design, Haiford, G.R., Evolution of Creep-Fatigue Life
John Wiley and Sons, 1993. Prediction Models, Creep-Fatigue Interaction at High
Cook, C.H., Spaeth, C.E., Hunter, D.T. and Hill, R.J., Temperature, ASME, AD-Vol. 21, pp. 43-57, 1991.
5-18

Haiford, G.R., Brief Summary of the Evolution of High- Aircraft Engine Materials, AGARD SMP,
Temperature Creep-Fatigue Life Prediction Models for AGARD-CP-569, pp. 8-1-8-8, 1996.
Crack Initiation, Computational Methods for Failure Koul, A.K., Wallace, W. and Thamburaj, R., Problems
Analysis and Life Prediction, NASA-CP-3230, pp. 121- and Possibilities for Life Extension in Gas Turbine
150, 1993. Components, Engine Cyclic Durability by Analysis and
Halford, G.F., McGaw, M.A., Bill, R.C. and Fanti, P.D., Testing, AGARD PEP, AGARD-CP-368, pp. 10-1-10-
Bithermal Fatigue: A Link Between Isothermal and 32, 1984.
Thermomechanical Fatigue, Low Cycle Fatigue, ASTM Koul, A.K., Thamburaj, R, Raizenne, M.D., Wallace, W.
STP 942, pp. 625-637, 1988. and DeMalherbe, M.C., Practical Experience with
Halford, G.R., Meyer, T.G., Nelson, R.S., Nissley, D.M. Damage Tolerance Based Life Extension of Turbine
and Swanson, G.A., Fatigue Life Prediction Modelling Engine Components, Damage Tolerance Concepts for
for Turbine Hot Section Materials, Toward Improved Critical Engine Components, AGARD SMP,
Durability in Advanced Aircraft Engine Hot Sections, AGARD-CP-393, pp. 23-1-23-22, 1985.
ASME, IGTI-Vol. 2, pp. 97-107, 1988. Koul, A.K., Immarigeon, J-P. and Wallace, W., Emerging
Halford, G.R., Saltsman, J.F., Verrilli, M.J. and Arya, V., Technologies for Life-Cycle Management of Turbine
Application of a Thermal Fatigue Life Prediction Model Engine Components, Canadian Aeronautics and Space
to High-Temperature Aerospace Alloys B1900+Hf and Journal, Vol. 34, No. 1, pp. 27-37, 1988.
Haynes 188, Advances in Fatigue Lifetime Predictive Koul, A.K., Bellinger, N.C. and Fahr, A., Damage-
Techniques, ASTM STP 1122, pp. 107-119, 1992. Tolerance-Based Life Prediction of Aeroengine
Heine, IE., Ruano, E. and DeLaneuville, R.E., TMF Life Compressor Spacers: I. A Deterministic Fracture
Prediction Approach for Turbine Blades, Thermal Mechanics Approach, International Journal of Fatigue,
Mechanical Fatigue of Aircraft Engine Materials, Vol. 12, No, 5, pp. 379-387, 1990.
AGARD SMP, AGARD-CP-569, pp. 20-1-20-6, 1996. Koul, A.K., Bellinger, N.C. and Gould, G., Damage-
Harrison, G.F. and Edmunds, T.H., The Role of Thermal Tolerance-Based Life Prediction of Aeroengine
and Stress Analyses in the Application of Damage Compressor Spacers: II. A Probabilistic Fracture
Tolerant Design, Damage Tolerance Concepts for Critical Mechanics Approach, International Journal of Fatigue,
Engine Components, AGARD SMP, AGARD-CP-393, Vol. 12, No, 5, pp. 388-396, 1990.
pp. 6-1-6-15, 1985. Kocks, U.F., Constitutive Behavior Based on Crystal
Harrison, G.F. And Homewood, T. The Application of the Plasticity, Unified Constitutive Equations for Creep and
Graham and Walles Creep Equation to Aeroengine Plasticity, Ed. by Miller, A.K., Elsevier Applied Science,
Superalloys, Journal of Strain Analysis, Vol. 29, No. 3, pp. 1-88, 1987.
pp. 177-184, 1994. Laflen, J.R. and Cook, T.S., Equivalent Damage - A
Harrison, G.F. and Shepherd, D.P., Lifing Philosophies Critical Assessment, Final Report, NASA CR-167874,
for Aero Engine Fracture Critical Parts, Aero Engine 1982.
Reliability, Integrity and Safety, The Royal Aeronautical Langdon, T.G. , Grain Boundary Sliding in as a
Society/The Institution of Mechanical Engineers, pp. Deformation Mechanism During Creep, Philosophical
5.1-5.12, 1991. Magazine, Vol. 22, pp. 689-700, 1970.
Harrison, G.F. and Winstone, M.R., Modelling and Lifing Larson, F.R. and Miller, J. Time-Temperature
of Structural Materials for Future Aeroengine Relationships for Rupture and Creep Stresses, ASME
Components, Advanced Performance Materials, Vol. 3, Transaction, Vol. 74, pp. 765, 1952.
No. 3/4, pp. 263-278, 1996. Lasalmonie, A., Pellerin, F. and Fournier, D.,
Harrison, G.F., Tranter, P.H. and Williams, S.J., Thermomechanical Fatigue in Gas Turbine Engines: The
Modelling of Thermomechanical Fatigue in Aero Engine Reason of a Concern, Thermal Mechanical Fatigue of
Turbine Blades, Thermal Mechanical Fatigue of Aircraft Aircraft Engine Materials, AGARD SMP,
Engine Materials, AGARD SMP, AGARD-CP-569, pp. AGARD-CP-569, pp. 1-1-1-5, 1996.
2-1-2-15, 1996. Lemaitre, J. and Chaboche, J.-L., Mechanics of Solid
Hertzberg, R.W., Deformation and Fracture Mechanics Materials, Cambridge University Press, 1990.
of Engineering Materials, i. Wiley, 1983. Liebowitz, H., Computational Fracture Mechanics,
Hessler, U. and Domes, B., LCF-Failure Analysis of an Advances in Fracture Research, Eds. K.Salama, et al,
Aero-Engine Turbine Wheel, Low Cycle Fatigue and Vol.3, pp. 1887-1921, 1989.
Elasto-Plastic Behaviour of Materials, Ed. by K.-T. Rie, Lin, Y.K., and Yang, J.N., On Statistical Moments of
Elsevier Applied Science, pp. 664-670, 1987. Fatigue Crack Propagation, Engineering Fracture
Hicks, M.A. and Pickard, A.C., Life Prediction in Turbine Mechanics, Vol. 18, pp. 243-256, 1983.
Engines and the Role of Small Cracks, Materials Science Malpertu, J.L. and Remy, L., Thermomechanical Fatigue
and Engineering, Vol. A103, pp. 43-48, 1988. Behavior of a Superalloy, Low Cycle Fatigue, ASTM
Kaufman, A. and Halford, G.F., Engine Cyclic Durability STP 942, pp. 657-671, 1988.
by Analysis and Material Testing, Engine Cyclic Manson, S.S., Thermal Stress and Low-Cycle Fatigue,
Durability by Analysis and Testing, AGARD PEP, McGraw-Hill, 1966.
AGARD-CP-368, pp. 12-1-12-12, 1984. Manson, S.S., Frecke, J.C. and Ensign, C.R., Application
Köstner, A., Chataigner, E. and Remy, L., Thermal of a Double Linear Damage Rule to Cumulative Fatigue,
Fatigue, A Useful Tool to Assess Low Cycle Fatigue Fatigue Crack Propagation, ASTM STP 415, p. 384,
Damage in Superalloys for Components Experiencing 1967.
Thermal Transient, Thermal Mechanical Fatigue of Manson, S.S., Future Directions for Low Cycle Fatigue,
 5-19

Low Cycle Fatigue, ASTM STP 942, pp. 15-39, 1988. Mechanics Concepts: A Historical Perspective, Fatigue
Manson, S.S. and Halford, G.R., Re-examination of and Fracture Mechanics: 28,h Volume, ASTM STP 1321,
Cumulative Fatigue Damage Analysis—An Engineering pp. 3-151,1997.
Perspective, NASA TM-87325, 1985. Newman, J.C, Jr., Booth, B.C., and Shivakumar, K.N. An
Marscher, W.D., Structural Design and Analysis of Elastic-Plastic Finite Element Analysis of the J-
Modern Turbomachinery Systems, Sawyer's Gas Turbine Resistance Curve Using a CTOD Criterion, Fracture
Engineering Handbook. Third Edition. Vol I., Sawyer Mechanics: Eighteen Symposium, ASTM STP-945, D.T.
J.W. and Japikse, D. (Eds.), Turbomachinery Reed and D. Reed, Eds., ASTM, pp. 665-685, 1988.
International Publications, pp. 7-1—7-56, 1985. Newman, J.C. Jr., Phillips, E.P., Swain, M.H. and
McClung, R.C., et ai, A Practical Methodology for Everett, R.A. Jr., Fatigue Mechanics: An Assessment of a
Elastic-Plastic Fatigue Crack Growth, Fracture Unified Approach to Life Prediction, Advances in Fatigue
Mechanics; 27'" Volume, ASTM, STP 1296, pp.317-337, Lifetime Predictive Techniques, ASTM STP 1122, pp. 5-
1997. 27, 1992.
McDowell, D., Basic Issues in the Mechanics of High Newlin, L„ Sutharshana, S., Ebbeler, D., Moore, N, Fox,
Cycle Metal Fatigue, International Journal of Fracture, E., Annis, C and Creager, M., Probabilistic Low Cycle
Vol. 80, No. 2-3, pp. 103-145, 1996. Fatigue Failure Analysis with Application to Liquid
McLean, B.F. and Dyson, B.F., Particle Coarsening, a0 Propellant Rocket Engines, AI A A Paper 90-1142-CP,
and Tertiary Creep, Acta Metallurgica, Vol. 31 pp. 1990.
17-27, 1983. Noor, A.K., Perspectives on Computational Structures
McGaw, M.A. and Saltsman, J.F., A High Temperature Technology, Computational Structures Technology, Ed.
Fatigue Life Prediction Computer Code Based on the by A.K. Noor and S.L. Venneri, ASME, pp. 133-140,
Total Strain Version of Strainrange Partitioning (SRP), 1995.
Creep-Fatigue Interaction at High Temperature, ASME, Noor, A.H., and Pilkey, W.O., State-of-the-Art Sun>ays
AD-Vol. 21, pp. 65-80, 1991. on Finite Element Technology, ASME, 1983.
McKnight, R.L., Structural Analysis Applications, Palmgren, A. Die Lebensdauer von Kugellagern,
Toward Improved Durability in Advanced Aircraft Verfahrenstechnik, Vol. 68, pp. 339-341, 1924.
Engine Hot Sections, ASME, IGTI-Vol. 2, pp. 83-95, Penny, R.K. and Marriott, D.L., Design for Creep,
1988. Chapman & Hall, 1995.
Merkle, J., Patterns and Perspectives in Applied Fracture Pickard, A.C., The Application of 3-Dimensional Finite
Mechanics, Fracture Mechanics; 26,h Volume, ASTM, Element Methods to Fracture Mechanics and Fatigue Life
STP 1256, pp. 3-40, 1995. Prediction, EM AS, 1986.
Meyers, M. A. and Chawla, K.K., Mechanical Pickard, A.C., Component Lifing, Materials Science and
Metallurgy, Prentice-Hall, 1984. Technology, Vol. 3, No. 9, pp. 743-749, 1987.
Miller, K.J. and Brown, M.W., Multiaxial Fatigue: A Pickard, A.C., Brown, C.W., and Hicks, M.A., The
Brief Revue, Advances in Fracture Research (Fracture Development of Advanced Specimen Testing and Analysis
84), Vol. 1, pp. 31-56, 1984. Techniques Applied to Fracture Mechanics Lifing of Gas
Millwater, H.R., Wu, Y.-T., Cardinal, J.W. and Chell, Turbine Components, International Conference on
G.G., Application of Advanced Probabilistic Fracture Advances in Life Prediction Methods, Ed. by D.A.
Mechanics to Life Evaluation of Turbine Rotor Blade Woodford and J.R. Whitehead, ASME, pp. 173-178,
Attachments, Journal of Engineering for Gas Turbines 1983.
and Power, ASME, Vol. 118, No. 2, pp. 394-398, 1996. Paris, P. C, Reflections on Progress in Fracture
Millwater, H.R., Chell, G.G. and Riha, D,S., A Mechanics Research, Fracture Mechanics: Twenty-Fourth
Probabilistic Creep Fatigue Computer Program for Volume, ASTM, STP 1207, pp. 5-17, 1994.
Structural Integrity Analysis, ASME Paper 96-GT-265, Priest, R.H., Beauchamp, D.J. and Ellison, E.G., Damage
1996. During Creep-Fatigue, International Conference on
Miner, M.A., Cumulative Damage in Fatigue, ASME Advances in Life Prediction Methods, Ed. by D.A.
Journal of Applied Mechanics, Vol. 67, pp. 159-164, Woodford and J.R. Whitehead, ASME, pp. 115-122,
1945. 1983.
Monico, R.D. and Chew, J.W., Modelling Thermal Romanoski, G.R., Antolovich, S.D. and Pelloux, R.M., A
Behaviour of Turbomachinery Discs and Casings, Heat Model for Life Predictions of Nickel-Base Superalloys in
Transfer and Cooling in Gas Turbines, AGARD PEP, High Temperature Low Cycle Fatigue, Low Cycle
AGARD-CP-527, pp. 24-1-24-10, 1992. Fatigue, ASTM STP 942, pp. 456-469, 1988.
Moreno, V. and Jordan, E.H., Prediction of Material Remy, L., Rezai-Aria, F., Danzer, R. and Hoffeiner, W.,
Thermomechanical Response with a Unified Viscoplastic Evaluation of Life Prediction Methods in High
Constitutive Model, International Journal of Plasticity, Temperature Fatigue, Low Cycle Fatigue, ASTM STP
Vol 2, No. 3, pp. 223-245, 1986. 942, pp. 1115-1132, 1988.
Murakami, Y (Ed.), Stress Intensity Factors Handbook, Riedel, H., Creep Crack Growth, Fracture Mechanics:
Pergamon Press, 1987. Perspectives and Directions (Twentieth Symposium),
Newman, J.C. Jr., Advances in Fatigue Life Prediction ASTM STP 1020, pp. 101-126, 1989.
Methodology for Metallic Materials, Theoretical Robinson, E.L., Effect of Temperature Variation on
Concepts and Numerical Analysis of Fatigue, Eds. A.F. Long-Time Rupture Strength of Steels, ASME
Blom and C.J. Beevers, pp. 301-325, 1992. Transactions, Vol. 74, pp. 777-801, 1952.
Newman, J.C, Jr., The Merging of Fatigue and Fracture Rooke, D.P., and Cartwright, D.J., Compendium of Stress
5-20

Intensity Factors, Her Majesty's Stationery Office, Gas Turbine Components, Materials Science and
London, 1976. Engineering, Vol. A103, pp. 49-61, 1988.
Schubert, F., Penkalla, H.J. and Frank. D., Thermal Vormwald, M., Heuler, P. and Seeger, T. A Fracture
Mechanical Fatigue of Model Blades Made from CC and Mechanics Based Model for Cumulative Damage
DS Superalloys, Thermal Mechanical Fatigue of Aircraft Assessment as Part of Fatigue Life Prediction, Advances
Engine Materials, AGARD SMP, AGARD-CP-569, pp. in Fatigue Lifetime Predictive Techniques, ASTM STP
22-1-22-9, 1996. 1122, pp. 28-43, 1992.
Sehitoglu, H., Thermo-Mechanical Fatigue Life Viswanathan, R,, Damage Mechanisms and Life
Prediction Methods, Advances in Fatigue Lifetime Assessment for High-Temperature Components, ASM
Predictive Techniques, ASTM STP 1122, pp. 47-76, International, 1989.
1992. Viswanathan, R., Low Cycle Fatigue Life Prediction of
Shih, C.F, deLorenzi, H.G. and German, M.D., Crack Fossil Plant Components, Low Cycle Fatigue and Elasto-
Extension Modelling with Singular Quadratic Plastic Behaviour of Materials-3, Ed. by K.-T. Rie,
Isoparametric Elements, International Journal of Fracture, Elsevier Applied Science, pp. 1-12, 1992.
Vol. 12, pp. 647-651,1976. Walker, K. The Effect of Stress Ratio During Crack
Slavik, D. and Cook, T.S., A Unified Constitutive Model Propagation and Fatigue for 2024 T3 an 7075 T6
for Superalloys, International Journal of Plasticity, Vol 6, Aluminium, Effects of Environment and Complex
No. 7, pp. 65L654, 1990. Loading History on Fatigue Life, ASTM, STP-462, pp. 1-
Sokolowski, D.E. and Ensign, C.R., Toward Improved ??, 1970.
Durability in Advanced Combustors and Walker, K.P., Research and Development Program for
Turbines—Progress in the Prediction of Nonlinear Structural Modeling with Advanced Time-
Thermomechanical Loads, ASME Paper 86-GT-172, Temperature Dependent Constitutive Equations, NASA,
1986. CR-165533, 1981.
Sawyer's Gas Turbine Engineering Handbook. Third Walker, K.P. and Freed, A.D., Asymptotic Integration
Edition. Vol I., Ed. by J.W. Sawyer and D. Japikse, Algorithms for Nonhomogeneous, Nonlinear, First Order,
Turbomachinery International Publications, 1985. Ordinary Differential Equations, NASA, TM-103793,
Saxena, A., Application of Fracture Mechanics in 1991.
Maintenance of High Temperature Equipment - An Webster, G.A., Ainsworth, R.A., High Temperature
Assessment of Critical Needs, Fatigue and Fracture Component Life Assessment, Chapman and Hall, 1994.
Mechanics: 28th Volume, ASTM, STP 1321, pp. 70-85, Wei, R., Life Prediction: A Case for Multidisciplinary
1997. Research, Fracture Mechanics; 27th Volume, ASTM, STP
Sih, G.C., Fracture Mechanics in Two Decades, Fracture 1296, pp. 3-24, 1997.
Mechanics: Perspectives and Directions (Twentieth Wu, X.J., Wallace, W., and Koul, A.K., A New Approach
Symposium), ASTM STP 1020, pp. 9-28, 1989. to Fatigue Threshold, Fatigue and Fracture of
Stouffer, D.C. and Salemme, G., The Impact of Engineering Materials and Structures, Vol. 18, No. 7/8,
Constitutive Modeling on Life Prediction, Fatigue and pp. 833-845. 1995.
Fracture of Aerospace Structural Materials, ASME, Wu, X.J. and Koul A.K., Modeling creep in complex
AD-36,pp. 163-176, 1993. engineering alloys, Creep and Stress Relaxation in
Sutharshana, S., Creager, M., Ebbeler, D. and Moore, N, Miniature Structures and Composites, Ed. by H.D.
A Probabilistic Fracture Mechanics Approach for Merchand, TMS Publications, pp. 1-19, 1996.
Structural Reliability Assessment of Space Flight Systems, Wu, X.J. and Koul, A.K., Grain Boundary Sliding in the
Advances in Fatigue Lifetime Predictive Techniques, Presence of Grain Boundary Precipites, Metallurgical
ASTM STP 1122, pp. 234-246, 1992. Transactions, A, 26A, pp. 905-914, 1995.
Sutharshana, S., Newlin, L., Ebbeler, D., Moore, O'Hara, Yamauchi, M., at at, Comparison of Creep-Fatigue Life
K., E. and Creager, M. Probabilistic High Cycle Fatigue Prediction by Life Fraction Rule and Strainrange
Failure Analysis with Application to Liquid Propellant Partitioning Methods, Low Cycle Fatigue, ASTM STP
Rocket Engines, AI A A Paper 90-1141-CP, 1990. 942, pp. 399-413, 1988.
Tada, H., Paris, P.C, and Irwin, G.R., The Stress Analysis Zienkiewicz, O.C., The Finite Element Method, McGraw-
of Cracks Handbook, Del Research Corporation, Hill, 1977.
Bethlehem, PA, 1985.
Tanaka, K., Mechanics and Micromechanics of Fatigue
Crack Propagation, Fracture Mechanics: Perspectives
and Directions (Twentieth Symposium), ASTM STP
1020, pp. 151-183, 1989.
Test Cases for Engine Life Assessment Technology,
AGARD PEP, AGARD-AR-308, 1992.
Thamburaj, R., Koul, A.K., Wallace, W. and
DeMalherbe, M.C., Life Extension Concepts for
Canadian Forces Aircraft Engine Components: J85-
CAN-40 Compressor and Turbine Material, National
Research Council Canada, National Aeronautical
Establishment Report, LTR-ST-1585, 1986.
Van Stone, R.H., Residual Life Prediction Methods for
 6-1

Chapter 6
Translation of Service Usage into Component Life Consumption
by
(G. Harrison)

Page

1. Introduction 6-3

2. Safety Requirement 6-3

6 3
3. Defining the Failure Point "
6 3
3.1. Life to First Crack "
3.2. 2/3 Dysfunction-Life and Damage Tolerance 6-3
6 3
3.2.1. Database Approach "
6 4
3.3. Damage Tolerance "

Damage Tolerance V LTFC 6-4

4.1. Materials Properties 6-4
4.2. Summary 6-5

5. Cumulative Damage Assessment 6-5

5.1. Calculation Processes 6-6
6 6
5.1.1. Metal Temperature Calculation "
5.1.2. Stress Calculation 6-7
5.2. Damage Assessment ""'
5.3. Component and Engine Testing 6-7
5.3.1. Sample Size 6-7

6. Li fing Algorithms ""'

6.1. Mechanical Stress Based Algorithms 6-8
6.2. Thermo-Mechanical Algorithms 6"!?
6 10
6.2.1. Cumulative Damage "
6.3. Damage Tolerance Crack Propagation Algorithms 6-13
6 13
6.3.1. Factors in Crack Growth "
6.3.2. Identification of Crack Growth Rate Controlling Parameters 6-13
6.3.3. Role of Residual Stresses in Crack Growth 6-13
6.4. Field Experience 6-13

7. Summary 6-14

8. Recommendations 6-14

9. References 6-14
6-2
 6-3

1/1000) will achieve the defined failure point. For

1. INTRODUCTION damage tolerance lifed components two types of flaw
This chapter discusses the techniques used to determine must be considered. For surface based flaws the safety
the structural component life with the safety levels criterion may be set to safety levels of 1 in 750. However,
imposed by the regulatory authorities. It describes how for buried flaws the criterion may be set such that not
materials algorithms may be developed and used, both to more than 1 in 10,000 components will exceed the
establish the declared component life and to evaluate the defined failure point. This is because buried flaws are
rate at which this life is consumed under service difficult to detect.
conditions.
DEFINING THE FAILURE POINT
Certification is a process that airborne equipment has to
pass through to demonstrate that it meets the safety 3.1. LIFE TO FIRST CRACK
regulations specified by the various airworthiness Within a general safe-life philosophy there are several
authorities. Typical examples of certification definitions associated with 'failure', where failure is the
requirements include: defined end-point. The Life-To-First-Crack definition
• Fan blade-off tests where an explosive mixture is uses a crack depth of 0.38 mm as the criterion for
used to cause the release of the whole of a single component failure. This means that all discs should be
airfoil and root fixing. The objective is to show that removed from service at the stage at which it is calculated
in the extremely unlikely event of a whole blade that a crack of 0.38 mm. depth would be present in the
being released, the fan casing would contain the weakest member of the total service population. Each
blade. disk is therefore treated as though it is the weakest disk in
• Cutting the shaft that takes the drive from the power the fleet.
turbine, to demonstrate that the control system can As stated above, a typical requirement is that with a
shut down quickly enough to prevent the loss of confidence level of 95%, not more than l-in-750
turbine blades, or a disk burst, during the resultant components will exceed the declared failure point,
over-speed. defined as a 0.38 mm deep crack.
Whether such tests are passed or failed is immediately Whilst this approach has successfully ensured that service
apparent. failures of engine disks are extremely rare, a natural
consequence is that most disks are retired from service
In the case of aero-engine fracture critical components after consuming just a small fraction of their available
exposed to widely varying thermo-mechanical loading life. Indeed, the average disk has consumed only 40% of
conditions, the failure modes progress in a complex its available Lifc-To-First-Crack, (LTFC).
manner. Two of the more important variables that
influence these processes are the material that an The depth of 0.38 mm was originally based on the
individual component is made from, and the way in effectiveness of crack detection techniques. Modern
which the engine is used. methods can easily exceed this, but the definition
provides a standard. It is emphasised that beyond 'first
At the end of the engine development process, each crack' the remaining (crack growth) life of the disc is a
fracture critical part is 'qualified' to the extent that it has significant portion of the cycles to dysfunction. In many
passed a set of tests, which are designed to demonstrate components, this portion may be typically 50% of the
that the regulatory requirements have been met. There declared safe life. However, minor cycles are relatively
may be some restrictions or limitations on the operational more damaging in propagation than in initiation. Hence,
use, but these should not be sufficient to prevent flying. in terms of the remaining additional missions, or engine
At this point an appropriate service life is authorised or flying hours that can be flown prior to disc burst, this can
recommended, depending on national practice, by the be less than 20% of the initiation life.
design authority for each safety critical structural
component. In many countries, the authorising body is the 3.2. 2/3 DYSFUNCTION-LIFE AND DAMAGE
engineering branch of the military service that operates TOLERANCE
the aircraft. When a component reaches the 'declared' life
In Europe, many newer components are designed so that
it is withdrawn from use.
the failure point is based on 2/3 of the dysfunction life.
The national qualification requirements are gradually This is a broader concept that allows failure criteria other
being harmonised. In respect of the manufacturers' than surface cracks of a defined size to be taken into
approaches towards their fulfilment, we have attempted to consideration. In general, lives declared using both
describe the broad concepts adopted and to illustrate methods are broadly similar, but the latter approach
typical techniques in general use. provides a determined safety level. For damage tolerant
designs in which the 2/3 dysfunction-life exceeds the life
2. SAFETY REQUIREMENT to a 0.38-mm deep crack, this approach provides the
Regardless of the details of the lifing methodology, it is opportunity for safe life extension.
essential that all declared lives be to a common level of 3.2.1. DATABASE APPROACH
safety. For many civil and military engines, service lives
of fracture critical components are specified such that In the database approach, a fracture mechanics method is
with a confidence level of 95%, not more than 1/750 (or used to combine both specimen and full-scale component
6-4

tests on different design features into a common database • In the absence of disc removal at the declared life to
for the material. If the respective stress intensities can be first crack (LTFC), the number of potentially
calculated for all relevant cases, the back-calculation of cracked discs will increase according to a log-
effective initial-defect sizes for all fatigue results together normal distribution as lives are extended beyond
with a statistical analysis of the size distribution, allows a 'first crack'. This is simply due to the nature of the
maximum probable-flaw size to be established. It is a crack initiation distribution as illustrated
simple process to use this effective initial-flaw size in a schematically in figure 1.
crack growth calculation to enable minimum component • The approach used is that of sampling without
lives to be established for each design feature. Declared replacement.
service lives are based on either the 0.38-mm life LTFC,
or the 2/3 dysfunction-life. Current US practice is to inspect at half of the calculated
mean crack-growth life (from an NDI sized defect to
In the above context, an assessment of a wide range of
dysfunction), and to retire components at a life associated
current aero-engine disc designs has revealed that in most
with a probability of cracking of 1 in 1,000. The major
cases, the application of the LTFC and the 2/3
exception is for some titanium disks, for which cryogenic
dysfunction-life concepts give very similar values for
spin-pit testing is undertaken to demonstrate that no
declared lives. In general, the constant safety margin of
defects greater than 0.006-in are present in the
the 2/3 dysfunction-life criterion is preferred. This is
component. In such cases, the NDI defect size can be
because the use of higher strength materials at increased
replaced by this much smaller size in fracture-mechanics
operating stresses has the potential effect of reducing the
crack-growth lifing-calculations.
critical crack size, associated with rapid crack growth, to
below the 0.38 mm depth of the engineering 'first-crack'. 4. DAMAGE TOLERANCE V LTFC
In European practice, where it can be shown that the 2/3
dysfunction-life exceeds the identified LTFC, fracture 4.1. MATERIALS PROPERTIES
mechanics crack propagation methods may be used to Consider the difference between the Life-To-First Crack
determine the available service life beyond 'first crack'. (LTFC) and burst distributions for a typical component or
For potential sub-surface locations, such as the disc mid- test piece. In figure l the mean life to a visible first crack
cob region, probabilistic crack growth methods are used is shown as 1.0. The life to burst is about 50% higher,
to account for the effects of defects of specific sizes, at therefore it has a relative mean value of 1.5 times the
specific locations. As discussed below, fracture mean life to first crack. This equates directly to spin-pit
mechanics methods may also be used uniquely from an testing experience.
established NDE size.
For any assumed crack size, the crack propagation life
3.3. DAMAGE TOLERANCE beyond this size would have the form AB, shown in
figure 2. However, the line AB specifically identifies the
In the US ENSIP (Ref. Mil Std 1793 (USAF), (1984))
minimum crack size for which, under high cyclic stresses,
approach for defect tolerant designs,
declared lives are based on quantifying
the available safe propagation life of
fatigue cracks. The initial crack size used — Life to First Crack
in fracture mechanics based lifing '—Life to Burst
calculations, is defined as the maximum 3 sigma = Usqrt(6) = L/2.45
defect size that is associated with a H
detection probability of 90% to a
confidence level of 95%. This relates
specifically to the equipment used in the Declare! Life
NDI inspection. The inspection interval
may be set to 1/2 of this dysfunction life,
which must be physically demonstrated in
full-scale engine tests. Importantly, this
approach allows advantage to be taken of
new inspection techniques, because
inspection capability is one of the major Life, L (log cycles)
limiting factors of this method. If
advantage is to be taken of available safe Figure 1 - Idealised population crack and burst distributions
life beyond 'first crack', the following
should be noted: the whole of the component life is consumed via crack
propagation. In practice, this is fully consistent with the
• In current materials, major cycles are largely 'effective-initial-flaw size' used in the UK fracture
responsible for crack-initiation, and minor cycles do mechanics database lifing methodology.
relatively more damage in crack-propagation than in
crack-initiation. Hence, typically, the mission Consider the fatigue curve, which shows the relative
damage exchange rate ratios may be up to 3 times significance of the crack initiation and propagation
greater during crack-propagation. periods. The line A-B is the tangent to the first part of the
 6-5

failure curve where, typically, the stress cycles are large 5. CUMULATIVE DAMAGE ASSESSMENT
and failure occurs in less than about 20,000 cycles. Np
represents the number of cycles in crack propagation, and The main reason for usage monitoring is to ensure that
engine components do not fail in service. In addition to
Ni those spent in crack initiation.
the direct airworthiness issue, the consequences of failure
Let us now consider the way in which declared lives are can be measured in several other ways including loss of
human life, loss of missions, and increased costs
associated with unplanned maintenance.
The main problems facing the design team centre on lack
of knowledge, about how the engine will really be used in
service. Experience shows that most military aircraft
undergo changes in the way that they are operated,
frequently without the conscious realisation of the
operator. The engineer tasked with estimating the service
life of each aero-engine component requires information
about the intended duty of the engine, the component
Cycles (log N)
material, component testing results, field experience, and
component management and retirement policy. Figure 3
Figure 2 - Life in crack initiation and propagation illustrates the stages involved in such life usage
assessments. With full knowledge, the designer can do an
calculated. If it is assumed that cracks are detectable at accurate job. Unfortunately, the design process is
0.38 mm depth, then from a small sample the declared invariably in an open loop situation. (Accelerated
life can be derived using appropriate statistical procedures simulated mission endurance testing (ASMET), is an
as below. attempt to close the loop, at least for 'hot parts'.
However, the difficulties of completely predicting service
In damage tolerance, with this starter-crack size and using conditions, and testing for them are such that risk cannot
mean crack-growth data, the crack propagation life is be completely removed.)
calculated. (In contrast to the minimum crack size that
can be detected by NDE, the damage tolerance approach Initially, the thrust to improve life usage predictions came
uses as its starting point the minimum crack size that will as a response to the management and logistic needs of in-
not be missed.). Typically, 50% of this calculated value is service fleets. Simple monitors were fitted to engines and
allowed as a service release life. If it is accepted that the data recorded and analysed by the manufacturer. For the
difference or scatter in crack growth rates, between the first time actual service usage data were available to the
strongest and weakest component, is typically a factor of operators and the designers to enable realistic life
four then the following applies: limitations to be imposed. Engine operators count engine-
life consumption in (engine flying) hours or (engine
• A component that has a crack size at the NDE limit, flight) cycles, where an engine flying hour just means an
that is not detected by inspection, has about a 1 in hour of engine flight, and an engine flight cycle means
750 chance of bursting in service use. simply one flight. In reality, the life consumed during an
• The mean risk of the population is approximately 1 engine run or flight is based on the effects of the
in 7500. continuously varying stresses, strains and temperatures
experienced at critical areas in the components.
4.2. SUMMARY
The values of these parameters obviously depend on the
Based on the figures illustrated, the damage tolerance actual mission profiles, engine intake conditions,
release life in cycles can be about 25% of the typical life- individual pilot reactions and several other parameters
to-first-crack (LTFC) release life. However, these and are used to assess the low-cycle fatigue, thermal
comments are based on release life in 'cycles'. For the fatigue, or creep service capability of individual
LTFC case, when the minor cycles incurred, during a components.
service mission, are converted to equivalent major
(reference) cycles, the exchange rates typically range The basic steps in cumulative damage assessment are:
from 2 to 5. This is based on analyses of mission usage Mission Specification;
tapes and is dependent on the severity of the mission Materials Characterisation;
flown. In contrast, for the same mission profiles, the
Stress and Temperature Modelling;
damage-tolerance exchange rates may typically be 3
Component Testing;
times more severe. Hence, when compared in terms of the
actual missions that can be flown, the damage tolerance Field Experience.
service release life of 25% of the LTFC release life in Component Management and Retirement
reference cycles converts into less than 10% of the LTFC
service missions. This corroborates and quantifies the Damage algorithms are developed in a form capable of
evidence of figure 2. It should therefore be noted that the application in real time life usage monitoring systems to
introduction of damage tolerance is a major step forward calculate the consumed life directly from measured
in terms of airworthiness and aircraft safety, but this is engine signals. The monitoring system must be able to
only achieved through increased 'cost of ownership'. respond to the rapid transitions of the input signals as
6-6

they appear during rapid aircraft

manoeuvring. Damage Accumulation
Using the measured time histories
of engine operating parameters
(such as spool speeds, intake and °I°I7I5I2I
gas path temperatures and
pressures) the algorithms calculate
the thermal and mechanical
boundary conditions for the engine
components. These boundary Input Signal from Engine Life Consumption
conditions are then used to
calculate the transient temperature
development within the parts and
subsequently the stresses or strains
at critical areas. These stress-
temperature histories are then used
to predict the life consumption
associated with each particular
V
Signal Checks, Filters, Temperature and Stress
flight measured in engine reference
Substitue Values Histories
cycle damage-related units. Both
on-board and ground based
processing systems have now been
developed.
measurements
In
are
the former,
processed
uy^^'V^]
continuously as they
developed, and in the latter
immediately after the end of a
are
V o
flight. The imposed damage is Simplified
summed over all engine runs to Thermo- Temperature
build up complete life dynamics Distribution ©
consumption records for all of Engine within Module
monitored parts of an engine.
Damage processes which are
monitored include low cycle
fatigue, thermal fatigue and creep,
as appropriate. Figure 3 - Overview of Life Usage Monitoring
Heat conduction between any temperature
5.1. CALCULATION PROCESSES differences in the component.
The life consumption calculation and monitoring process
is determined wholly from engine running history. Heat balancing for each of these points may be described
Generally, this is divided into three parts. With system by regular differential equations. The coefficients used in
power up, both the system and algorithms are initialised. the equations for the temperature model can be optimised
The main monitoring tasks immediately follow. in such a way that critical basic data used in the lifing
Computations are repeated every time step where a calculations are modelled as accurately as possible. In
typical time step-length may be 0.5 seconds. In the final practice, typical deviations from the full mainframe
part (at engine shut down) all the monitoring tasks are engine simulations are less than 10 K. The coefficients of
finalised and the results obtained are stored. Performance the equations depend on current engine operating
calculation consists of the determination of the parameters (e.g. spool speed) and the time step-length.
temperatures and pressures in the gas path and in the
cooling air paths. Calculations also include the The transient metal-temperature distribution of the
computation of torque, bending moments and normal component is calculated throughout each of the three
forces in shafts. These form the boundary conditions for phases of an engine 'cycle' (start, run and engine shut
the following steps. down). In phase 1, the initial temperatures for all relevant
points are determined. These values depend on the
5.1.1. METAL TEMPERATURE CALCULATION temperature distribution at the end of the last engine run
and the period for which the engine has been stopped. In
The temperature distribution of a component is phase 2, the temperatures for all points are updated at
established by sampling at a number of points. For each time step. They are calculated using the temperature
calculation of the metal temperature of a given point the distribution at the beginning of the specified time step and
following physical effects should be taken into account: the thermal boundary conditions operating at the specific
• Heat transfer to the temperature point from time. In phase 3, the temperatures corresponding to shut
surrounding gas and cooling air and vice versa down stress peaks are computed. Such metal temperature
distributions are required for both the calculation of
 6-7

thermal stresses, and for the temperature influence in done through a series of component tests, and statistical
damage assessment. assessment of the results. Hence, other basic LCF
programmes are used to provide the larger element of
5.1.2. STRESS CALCULATION proof by analysis, with the model tests providing
Stress modelling generally involves finite element stress confirmation and feedback.
analyses, determination of physical and thermal boundary In parallel with the engine-component structural-integrity
conditions, heat flow calculations and the step-by-step tests, the airborne and ground based monitoring
assessment of the stresses induced throughout a large computational system must be qualified. These must be
number of flights. Transient total stresses are calculated shown to give, within close limits, the same results with
for each monitored critical area by summing up the same data as the design models.
centrifugal stresses, thermal stresses, pressure-induced
stresses and any additional stresses. The centrifugal stress 5.3.1. SAMPLE SIZE
is related to the square of the spool speed. Thermal stress When a test sample is taken from of a population having
is derived from the current temperature distribution. either a normal or a log-normal failure distribution, its
Additional stresses include pressure-induced stress, the mean and variance are likely to be similar to that of the
effects of bolt clamping, residual stresses etc. If shafts are original population. As the sample size increases, so the
monitored, stresses due to torque and bending (resulting likelihood of a similarity increases. The confidence
from gyroscopic effects) may be included as well. interval (statistical correction factor) associated with any
The calculation of total stresses is repeated at every time stated confidence level (e.g. 95%) will get progressively
step during the main phase. In the initial and final phases, smaller as the sample size increases. This leads to the
the stress peaks which occur during engine start, and after conclusions that:
engine shut-down are also calculated. The coefficients • The sample must be of adequate size to obtain
used in the stress models are optimised in such a way that enough information to show statistically that the
given basis data are modelled as accurately as possible. sample itself is of a consistent and identifiable
Because in a cyclic damage regime the major cycle
distribution.
contributes most to accumulated damage, the accuracy is
• The sample can then be assumed to be
steered so that the maximum and minimum stresses are
representative of the original population.
calculated accurately, whereas for medium stress cycles,
typically, deviations of 2 % may be accepted. If damage 6. LIFING ALGORITHMS
assessment is based on fracture mechanics methods,
The algorithms discussed here relate to the damage
stress intensity factors or other fracture mechanics based
induced by the imposed stresses and hence apply in both
parameters are derived from the stresses.
the mainframe and the Reduced Order Algorithms (ROA)
5.2. DAMAGE ASSESSMENT that are implemented in the operational Engine Life
Monitoring System (ELMS). Differences between the two
The established stress-temperature histories are then sets of calculations relate to:
assessed with respect to the relevant damage mechanisms.
In the case of low cycle fatigue damage, stress cycles (i.e. Availability of data.
main cycle and sub-cycles) may be extracted using the Measurement accuracy and repeatability.
Rainflow process. The cycles obtained are converted into Data recording methods.
equivalent damage according Miner's damage hypothesis Computational speed.
via materials-specific LCF strain cycles. If fatigue is Computational memory limits.
assessed by fracture mechanics methods, the stress Operational requirements and limitations.
intensity cycles are assessed with respect to the relevant Logistics requirements.
crack growth law (da/dN curve). In both cases, damage
due to sub-cycles and main cycles is accumulated over The outcome ranges from simple engine based Total
the whole engine run. If creep is an important damage Accumulated Cycles (TAC), to individual component
mechanism then in every time step a creep damage tracking via complex thermo-mechanical algorithms. The
increment is also evaluated. derivation of the reduced order algorithms is described in
5.3. COMPONENT AND ENGINE TESTING
chapter 7.
As new component designs are developed, work is often Historically, such algorithms involve a number of
related to the potential advantages offered by new assumptions that include:
materials. This includes the testing of innovative design • Major stresses are speed related and are due to
techniques to offset any risks associated with the new centrifugal loads, such that the stresses are
material. Design validation may be from the qualification proportional to the square of the rotational speed, of
test programme. This may involve sub-scale component the component under consideration.
testing or as with ASMET, seek to prove the structural • The major rotating components are most likely to
integrity of the engine hot end parts through full-scale fail in fatigue due to the repetitive loading cycles
engine tests. Such tests are used to demonstrate that the induced under normal service usage.
engine can achieve a full service-life. However, these
• Induced stresses may consist of both steady and
tests are extremely costly and cannot be used to validate
alternating components. To ensure that failures do
statistically the declared service life. This can only be
6-8

not occur in service, the allowable

Start
alternating stress must be reduced as
the steady stress component increases.
Input data
This can be addressed via a 'Goodman' point
diagram or similar approach as
discussed later. Because of wide
variations in observed test results (even
with apparently identical samples) the
results must be assessed on a statistical
basis.
Fatigue damage is cumulative, in such a
way that the fraction of life consumed
at any given loading condition can be
added to the fraction of life consumed
at any other loading condition. This is
known as Miner's hypothesis and is
discussed later.

More advanced algorithms take into account

thermal loads caused by temperature
gradients across components, creep and stress
relaxation. It is essential that the life usage
computations should have mirrored service
practice to ensure acceptable levels of
accuracy and hence safety. Main differences
the more advanced usage monitors are that in
contrast to the earlier use of theoretical data,
sampled data or values calculated directly Pai( head ol Pair head ol
min stack with max stack with
from the recorded data can now be used. head ol max stack head ol min stack
Measured data may include air temperature
and pressure, altitude, engine rotor speed,
compressor delivery pressure and outlet
temperature. Extracted stress and strain
cycles and damage summation are all
Figure 4 - Rainflow algorithm
established using procedures similar to those
used in the design stage. However, the analysis models Where ar equals peak stress in the disc under imposed
used to calculate transient temperatures and stresses engine reference shaft speed conditions Nr(rev/sec), and
require substantial computing power and large storage a; equals stress in the disc corresponding to Nj (rev/sec).
capacity. Such calculations can only be effectively done
The above expression enables recorded histories of
on ground based installations. For on-board processing, it
rotational speeds to be used to determine the stress-time
is necessary to construct simplified heat transfer and
profiles experienced at critical locations in the disc
stress analysis routines that provide an acceptable
throughout any mission profile. In practice, however, the
compromise between computational time and accuracy.
fatigue damage incurred by the disc is not specifically
Hence, these must be correlated against mainframe
related to these stresses but to the cyclic thermo-
thermo-mechanical analyses and acceptable degrees of
mechanical stress-ranges that occur within the stress time
accuracy established. Formal qualification procedures
history.
must be used to verify that the results are robust and
directly comparable to those obtained from full analysis It is therefore essential to extract the significant stress
procedures. cycles from the stress histories of the sortie profiles. For
cycle extraction, a procedure such as the Rainflow
6.1. MECHANICAL STRESS BASED ALGORITHMS method is used. This allows the flight profile to be
In the simplest case of mechanical stresses only, these can analysed sequentially but requires only very small
be related directly to shaft rotational speed. Hence, the amounts of temporary data storage and hence is suitable
results from a stress analysis conducted for a specified for both ground station and on-board analysis. In practice
shaft speed can be used to determine the critical areas of it has been found that with efficient algorithms, data
the disc and the peak stresses operating in these regions. stores rarely contain more than four or five unpaired high
For all other loading conditions, the peak stresses in the or low turning values before the next set of minor cycles
disc can be related to the previous identified peak stress can be extracted. A flow diagram for the Rainflow cycle
or reference stress, a„ by an expression of the form: extraction process is shown in figure 4. The method takes
no account of rates of cycling or dwell effects and
assumes that interruption of large stress cycles to
<Ji ~ CXr (1) complete smaller cycles does not affect the damage
KNr. imposed.
 6-9

We may wish to assess the effects of a variety

of mean stress levels, in the absence of an all- Input (Mission)
inclusive fatigue database. To do so, an
estimate has to be made of the equivalent 'zero-
to-max' stress range that would impose an
amount of fatigue damage equivalent to that
caused by the identified high mean-stress- Performance Performance Performance
Algorithm Algorithm
~*v- (exact)
range test. Probably, the most common
empirical approach used to quantify the effects Frequency Oiltriöution

of mean stress level on fatigue is the Goodman

diagram.
In this procedure, mean stress is plotted along Temperature
¥
Temperature
Component
the horizontal axis and alternating stress range Algorithm Algorithm --«?* Temperature
(exact)
along the vertical axis. A line is then drawn to Frtquwcy OtslriDutlon
join the tensile strength of the material to the
stress amplitude of the push-pull test having fr
'N/ cycles to failure. It is assumed that all
Critical Area
combinations of mean stress and stress Stress Stress
amplitude that lie on this line will have a Algorithm Algorithm *9" Stress
(exact)
common life to failure. Hence, high r-ratio
stress-ranges can be expressed in terms of their
Goodman equivalent zero-to-max stress, AaE,
via the expression:
Damage
Model

CMAX ~ <7MIN
^<7E ' (2)
, (TMIN

CUTS

Although there has been little experimental Simulated Deviation Result Deviation
verification of this approach, it is widely used
in lifing calculations. Equivalent expressions
Figure 5 - Overview of the usage algorithms
can be constructed for strain based fatigue
results and an alternative formulation that conduction are calculated from the current operating
accounts for maximum stress and mean strain is the Smith conditions. The algorithms operate in such a way that the
Watson Parameter, >/(crmaxAeE) steady state temperatures are exact with transients being
Having identified the equivalent stress range. Miner's within about 10°C of their actual values.
hypothesis can be used to convert the damage imposed by The temperature transfer matrix describes heat transfer
such a cycle to its equivalent amount of major cycle between hot gas, cooling air and component surfaces,
damage. The Miner's expression has the form: heat conduction and radiation. Generally, in current
monitoring algorithms, time-steps, are fixed and
ÄÖS riRF
Y (3)
temperature changes per time-step can range between
about 10°C, after slam acceleration to 0.001°C in the
AOE \riF.F )
asymptotic phase. Although during a mission, transient
where nRF equals the number of reference cycles to failure temperatures can induce significant thermal stresses, the
at the reference stress, ArjR and nEF equals the number of major transient effects are imposed during the take-off
cycles to failure under the equivalent zero to max minor and landing stages in the flight profile. The most
cycle, AoE. The slope of the fatigue design curve is 'm'. sophisticated analysis procedures may involve a full life
analysis predicting component transient thermal and
6.2. THERMO-MECHANICAL ALGORITHMS mechanical stresses from flight recorded data. The
In practice, the simplified calculation of the engine coupled transient temperature and transient stress, finite
thcrmodynamic cycle frequently makes use of element analyses require extremely short time steps to
compressor and turbine performance maps for gas ensure adequate identification of both the instantaneous
properties and thermodynamic relationships. From these, transient thermal stresses and specifically of the turning
the temperature profiles of surrounding areas may be values associated with the changes in throttle setting.
indirectly monitored. When not all components have
The procedure is illustrated in figure 5 and involves:
reached ambient temperature between engine runs, a
time-bound estimate of the temperature profile is used. • The computation of gas stream temperatures,
During normal running, temperatures for the identified pressures and velocities from the known
time-step are derived from previous temperature characteristics of the engine;
distributions, and coefficients describing heat transfer and • The determination of appropriate gas-metal heat
6-10

transfer coefficients and transient temperatures since material mechanical-property data can be highly
throughout the component and throughout the temperature dependent. It is essential that the model is
flight; validated against the stress and temperature results from a
• The calculation of the thermal stress distribution full analytical solution. The verification should include
profiles throughout the flight from these calculated features not in the original conditions and must include
temperatures. transients between different engine states, and from non-
stabilised conditions. The combined thermo-mechanical
Where transient thermal stresses can make a significant stress equation is of the form:
contribution to the overall stress ranges, a large number
of computations must be solved rapidly and accurately. CTequiv =
°~0 + AN
' + 1 BiTt (8)
This requires significant computing power.
where cr0 represents any residual assembly stresses and
One long established approach to overcome this has been
via the use of a simplified thermo-mechanical stress the second and third terms are the mechanical and thermal
analysis model which is then matched with the full contributions to the total induced stress.
thermo-mechanical finite element computations.
Typically, for the centre regions of a large turbine disc,
6.2.1. CUMULATIVE DAMAGE
the characteristic response time of the disk can approach For single minor cycles, damage can be expressed in
20 minutes. In practice however, in excess of 20 major terms of equivalent reference cycles via the expression:
throttle movements may have occurred during this time.
A consequence of this is that most transient events A <J E (9)
commence from a non-stabilised state. The values of the
critical parameters involved can be uniquely established v ACT R)
by curve fitting to the results obtained for the detailed,
coupled, thermo-mechanical analysis. Hence, for each mission, the exchange rate ß, can be
expressed:
The assumption is made that temperatures always decay
exponentially towards their asymptotic values (if the f A
A aE^
defined conditions were held indefinitely). The n. (10)
/?=ZH,= Z
characteristic time constant of the decay depends on the \ ACT RJ
engine conditions. This can be modelled via a small
number of thermal masses (typically 3), with each having where nc equals the number of cycles extracted via the
4 or 5 dependable constants to control the temperature at Rainflow analysis of the mission profile, and, AnR equals
a particular location. Such models can be used to the equivalent number of reference cycles that have been
calculate transient temperatures and their associated consumed. Figures 6a and 6b provide a schematic
thermal stresses. illustration of the calculation involved in usage
Thus, for illustrative purposes in such an approach, given monitoring. The references to EUMS refer to the UK
the spool speed, N, the local gas-temperature, T, and gas Engine Usage Monitoring System, a tool used over many
pressure, P, for each individual mass: the asymptotic years on many engine types to provide data for all types
temperatures Ta, can be related to the gas stream of engine monitoring work.
temperatures. This is through a linear expression of the When using the basic low cycle fatigue algorithm as
form: described above it is common practice to apply the
concept of a 'cut-off stress range' below which the LCF
Tai(t) = KlT + Kf (4) fatigue damage of the disc is assumed to be infinite. This
The heat transfer coefficients Hj(t) can be obtained from: value is frequently set to a stress level at or close to the
endurance limit, and leads to a slightly more complex
0.7 \Kf
07
Hi(t) = KUNP/T ) (5) expression of the form:
and the temperatures of the masses T(t) from the
numerical solution of the differential equation: 1 ACTR- Acre
D = ■I7'I7' (ID
dTi riRF ■<TE- >(Tc
Hi(Tai-Ti) (6)
dt where AoR is the reference stress range, Aoc the
with an appropriate initial condition identified for the asymptotic stress range, AaE the total equivalent zero-to-
max stress range and nRF is the number of reference
start of the flight. These equations lead to a thermal stress
cycles to failure.
equation of the form:
aw = JLiKs,Ti(t) (7)

where 'i' is summed over the number of thermal masses

that significantly influence the selected location.
In addition to calculating the thermal stresses, it is
necessary to retain calculated corresponding temperatures
 6-11

Is
Z S M

6"

Ii

mM
ca „

eu •35 öl
$
re
e (9
o
Cc
£
b
5c
& n
5
Xi
9

<
z
o

2
i
a a -5 £ 3

t 11J1 § -S |
«/) 1* o.e «a n « F «a '

Figure 6a - Overview of usage monitoring procedures

6-12

Ü
£ a
INITIATION 1»
1 «
@"
nz 1 H ^ |
+ ill «*
<- n .2
o u =
z » el
E.2 =
« a —

8.
R
ä<_
z t»=
z t
+ 2
« '5-3 _• Ss 2
2 o § ell z
ET a
5 -- 5 _ £ .
> 8 «1 5
e i:
<& .
V ■»
c
V -I«
a
I « = 5 § e S I
« S -o ^ « -
o «2 .2
o 2
Til
I
z
a. 2.
Z
I <
| S11 i S > 1}
5 M
M J1 ä . e I
s fas Ü2
2Sw
=
D r« « > E c. . i ,5 g
S "2 11 £ M ' OS
2 Z L-J!lrJLwJ n .2:
O -a
Z3I 25 —
ET *,r,2, -? 111!
de
« « m*
•
si
E c
tu £ E
if g
i L 8
X
y .»
6 u
us u c
i—*

*F
u: £
o
0

«?
<r1"
PROPAGATION

UJ Q
UJ X UJ
J
3 cc - w y; 5 §<a§|
o 3 iu
<; £ fi Hi
w £: Ü3 ■ _
-J <■> J
< iu
-3 U- ü
fZ\
u ;
_ _ g CO CO
(J Q CC &
ffpD2

Figure 6b - Over\'iew of usage monitoring procedures

 6-13

6.3. DAMAGE TOLERANCE CRACK 6.3.3. ROLE OF RESIDUAL STRESSES IN CRACK

PROPAGATION ALGORITHMS GROWTH
As discussed earlier the basic lifing methodologies are There is considerable evidence to suggest that a major
based on a 'Life To First Crack' approach. An alternative, cycle that induces significant plastic strain at the tip of a
which is increasingly favoured in Europe for new designs crack, is likely to slow down crack growth during the
is the 'Life to 2/3 Dysfunction' approach. This section immediately following minor mission cycles. That is, the
describes the algorithms used in this latter approach. They position and magnitude of the maximum stress in the
depend heavily on modelling crack growth. In the missions to be flown is dominant in identifying both the
application of damage tolerance concepts, in contrast to boundary of the plastic zone, and the associated
crack-initiation lifing concepts, the size of the crack is compressive residual-stress field when the major cycle is
now a measure of life consumption. The prediction of completed. In such cases the following apply:
possible crack size considers the 'worst case' conditions.
• Where the loading conditions are high enough to
Practically, predicted crack sizes significantly exceed
cause local yielding at a stress concentration in the
those of any identified cracks. Crack size itself is of no
uncracked body, any subsequent crack growth will
practical use to the customer, except as a criterion for
only occur when the crack is fully open.
component removal. Therefore, based on the correlation
• The stress level at which this occurs depends on the
between accumulated cycles and crack size, the
magnitude of the local compressive residual stress
monitoring system can easily convert the 'crack size'
field created by a preceding overload. The effective
damage into reference cycle damage. This permits a
stress-intensity range driving the crack growth
single unit to be used as the measure of life consumption,
process is simply the peak elastic value associated
because within the monitoring system, safe-life and
with the maximum stress, minus the crack opening
damage-tolerance lifing concepts can be algorithmically
stress level.
combined.
• Where local plastic yielding can occur, due account
6.3.1. FACTORS IN CRACK GROWTH must be taken of the relative effects of the different
plastic zone sizes created under spin pit component
A large number of scientific papers have been written on
life determination and under service conditions.
fatigue crack-growth models. However, little attention
has been given to their application to real components in • The lifing methodology for crack growth must
service. Each factor affecting crack growth must be therefore be more complex than that for the crack
identified and statistically modelled to ensure that the initiation situation.
approach can be implemented without compromising
current levels of safety. The sequence of stages involved, In any fracture-mechanics based crack growth
together with the associated statistically adjusted damage methodology, it is essential to identify an appropriate
statistical model so that current safety levels are not
algorithms are illustrated in figures 6a and 6b. The
procedures are equally applicable to life extensions compromised.
beyond 'first engineering crack' and to the ENSIP damage 6.4. FIELD EXPERIENCE
tolerance concept.
This is the point at which the success or otherwise of the
6.3.2. IDENTIFICATION OF CRACK GROWTH design is evaluated in earnest. Unfortunately, several of
RATE CONTROLLING PARAMETERS the original design assumptions may no longer be
completely valid by this stage. Almost invariably, the
In practice, for laboratory specimens, crack initiation and
way in which the engine is actually used will be
growth processes under constant amplitude loading are
significantly different from the earlier assumptions and
reasonably well understood and they can generally be
the recipient may want to exploit to the full all of its new
quantified via the use of linear-elastic fracture-mechanics
or unique capabilities.
concepts. The transition to variable-amplitude loading
cases, such as occurs under service conditions At this stage, components may require a modified design
necessitates the identification of a scaling parameter. usage model. The first indication of this may come from
Although, in principle, the stress intensity factor could be the usage data collected from the engine monitoring
used, in practice for multi-axial loading and under both system. It may be seen that the rate of damage
global and crack tip yielding, this is a complex technical accumulation is not always that which was expected. As
step and accurate quantitative predictions is difficult. experience builds up, the cumulative damage picture
across the fleet will also build up and special cases that
The simplest and most conservative approach is to
cause very high or very low levels of damage may be
assume that for each cycle the full elastic load range,
identified. In some cases, this may lead to changes in
normal to the crack, contributes to the effective stress
pilot practice, or in others to changes in spares
intensity cycle providing the driving force for the crack
provisioning rates.
growth increment. However, this can lead to inaccurate
estimates of the available safe service life because the More importantly, the information that is being gathered
higher stresses imposed under laboratory test conditions provides a model of component usage based on the
may cause local yielding, and on unloading to greater indirect measurements and computations of the airborne
crack closure. This may result in more extensive system.
compressive residual stresses than are induced in a
service component. The results indicated by the monitoring system only work
6-14

according to the rules built into it. Any hidden flaws or G. F. HARRISON, J. E. GOODWIN, R. S. DENNIS:
errors in the original model calibration will not reveal Laboratory simulation of thermal fatigue, Int Conf
themselves unless a component fails. One way around Society of Environmental Engineers Fatigue Group VI
this is to remove a sample of components from service No 16 1-14 1976
and test them to destruction under controlled conditions. G. F. HARRISON: Deformation and fracture of some
This, however, is expensive. nickel based gas turbine alloys, PhD thesis, University of
London 1975.
7. SUMMARY G. F. HARRISON, M. G. COCKROFT: The effects of
cycle parameters on high temperature low cycle fatigue,
Few tasks in engineering are more critical than the 1974, AGARD CP 155, Paper 4 1-13.
prediction of the service lives of fracture critical F. HARRISON, G. P. TILLY: The state and cyclic creep
components and the complementary task of prediction of properties of three forms of a cast Nickel Alloy, Int Conf
component service-life consumption. Predictions of the on Creep and Fatigue in elevated temperature
state of stress and of strain are made difficult by intricate applications. 222.1-9.1 Mech Eng, Sheffield, 1974.
geometrical features and by uncertainties in transient F. COFFIN: Fatigue at high temperature, ASTM STP
temperature distributions. The effects of localised creep
520, 1972, 5-34.
and engine performance deterioration also cause this S. S. MANSON, G. R. HALFORD, M. H.
stress distribution to change significantly with time. In HIRSCHBERG: Creep fatigue analysis by strain-range
recent years proposed lifing models have become partitioning, NASA TM X-67838, 1971.
progressively more sophisticated, with more of the G. R. HALFORD, J. F. SALLSMAN, M. H.
identified lifing variables being incorporated. HIRSCHBERG: Ductility normalised strain range
It is clear that any model aimed at producing acceptable partitioning life relationships for creep-fatigue life
life predictions must examine the loading cycles in detail, predictions, NASA TM 73737, 1977.
and take into account variables such as the stress and T. KITAMURA, G. R. HALFORD: A non linear high
inelastic strain at each critical area of the disc. temperature fracture mechanics basis for strain range
Specifically with respect to fracture critical components, partitioning, NASA TM 4133, 1989.
safety cannot be compromised. Hence, to utilise the R. S. NELSON and J. F. SCHOENDORF: Creep fatigue
maximum available lives of such components it is life prediction for engine hot section materials (Isotropie)
imperative that individual component service-life- - fourth year progress review, NASA Report N89-12914,
consumption is tracked. This chapter has attempted to 1986.
illustrate the general principles involved in usage J. BRESSERS and M. M. BUCKLEY:
monitoring with simple illustrations of their application. Thermomechanical and isothermal fatigue of coated and
uncoated nickel-based single crystal superalloys, Rolls-
8. RECOMMENDATIONS Royce pic, private report, 1990.
G.F. HARRISON & D P SHEPHERD: Lifing
As more lifing variables are identified and taken into Philosophies for Aero Engine Critical Parts. Aero Engine
account, it is essential that sufficient material properties Reliability Integrity and Safety. Proceedings of the Royal
data is available to the designers. This may necessitate
Aeronautical Society, 1991.
increasing investment in materials databases, through P. WATSON & B.J. DABELL: Cycle Counting and
both the ENSIP and the European database approaches
Fatigue Damage. Jnl. of Soc. of Environmental
described above.
Engineers, Sept. 1976.
Normal variations in LTFC initially make a damage T.M. EDMUNDS & R.A. LAWRENCE: Monitoring
tolerance approach to design appear attractive. However, Engine Thermal Stresses. AGARD CP386 Engine Cycle
some advantages may accrue from selecting the approach Durability by Analysis and Testing, Netherlands 1984.
on a case-by-case basis. H. PFOERTNER & C. ROSS: Preparing Life Usage
Monitoring for the next Decade. Proceedings of the 18th
9. REFERENCES Symposium. Aircraft Integrated Monitoring Systems. 19-
21 September 1995. University of Stuttgart.
G. F. HARRISON, M. E. F. SMITH, J. NURSE:
J.BROEDE & H. PFOERTNER: Advanced Algorithm
Procedures for Aeroengine Component Life Usage
Design and Implementation in On-board Microprocessor
Prediction, AIDAA/AAAF/DGLR/RAeS 5th European
Systems for Engine Life Usage Monitoring. Proceedings
Propulsion Forum, Pisa, Italy 5-7 April 1995.
of the 15th Symposium. Aircraft Integrated Monitoring
G F. HARRISON, P.H. TRANTER, S.J. WILLIAMS:
Systems. 12-14 September 1989. University of Aachen.
Modelling of Thermomechanical Fatigue in Aero Engine J.BROEDE , H. PFOERTNER, K. RICHTER The
Turbine Blades. Importance of Testing for Successful Life Usage
G. F. HARRISON, A. D. BOYD-LEE: A Damage
Monitoring Systems. Proceedings of the 19th
Tolerant Lifing Methodology for High Strength Disk
Symposium. Aircraft Integrated Monitoring Systems. 4 -
Alloys. DRA/SMC/CR942045/1, October 1994.
7 May 1998. Garmisch - Partenkirchen, Germany
In turn the following references are made:
 7-1

Chapter 7
Lifing Procedures and Monitoring System
Verification and Validation
by
(P.-E. Mosser and P. Everitt)

Page

Engine Life Management 7-3

1.1. Introduction 7-3
1.2. Establishing the Authorised Life of Limited Life Components 7-3
1.3. Reducing the Margin Between Estimated and Actual Life Consumed 7-4
1.4. Main Specifications for an Engine Life Monitoring System 7-4
1.4.1. Lifing Methods Objectives 7-4
1.4.2. ELMS Specifications 7-4
1.4.3. Expected Benefit of an ELMS 7-5

2. Definition of Damage Units and Life Units 7-5

2.1. Squadron Based Definition 7-5
2.2. Mechanical Engineering Based Definition 7-5
2.3. Practical Choice 7-5
Reduced Damage Modelling for ELMS 7-6
3.1. Reduced Engine Description 7-6
3.2. Reduced Order Algorithm Initial Validation 7-7
3.2.1. Older Engines 7-7
3.2.2. Newer Engines 7-7
3.3. Long Term Validity of the Reduced Damage Model 7-9
3.3.1. In-Service Component Replacement 7-9
3.3.2. Aircraft and Engines Fly New Flight Profiles 7-10
3.3.3. Changes are Necessary in the Reduced Modelling 7-10
3.3.4. Engine Life Monitoring Hardware Changes 7-10
Engime Life Monitoring System 7-10
4.1. Integrity Requirement 7-10
4.2. System Architectures 7-11
4.3. System Elements to be Validated 7-12
4.4. Validation Process 7-12
4.4.1. Airborne Software 7-12
4.4.2. Airborne Hardware 7-12
4.4.3. Data Transfer Hardware 7-13
4.4.4. Data Transfer Procedures 7-13
4.4.5. Ground Hardware 7-13
4.4.6. Ground Software 7-13
4.5. Overall System 7-13
4.5.1. Laboratory Tests 7-13
4.5.2. Dynamic Tests 7-14
4.5.3. Service Environment Tests 7-14

5. Examples 7-14
5.1. Fighter Aircraft: Mirage 2000 and M53-P2 7-14
5.1.1. System Description 7-14
5.1.2. Validation Process 7-14
5.2. Helicopters: Agusta A129 and GEM1004 7-15
6. Conclusions and Recommendations 7-15
7. Summary 7-16
8. References 7-16
7-2
 7-3

The basic calculation methods and processes for predicting

1. ENGINE LIFE MANAGEMENT the life of a component and for establishing an appropriate
'safe life are covered in chapter 6. In this chapter, we
1.1. INTRODUCTION
consider the safe-life limit set through the methods of
The life control process for critical parts begins with the chapter 6 to be totally valid, and describe the derivation
definition of an appropriate safe or economic life for a and verification of a set of simplified models that are used
particular component. The design is then done using in-service to estimate life consumption. These models are
available engine models and material property databases. referred to as the Reduced Order Algorithms (ROA). They
The models and databases now available in industry are are used to perform the 'consumed life' calculations in a
extremely comprehensive, and almost unrecognisable from practical size of processor. We also describe the validation
those of the early days of the gas turbine. process that is necessary to demonstrate that the total 'in-
Checking and testing the various models against the real service' system provides a satisfactory representation of
world for accuracy and repeatability is known variously as the full design model and process.
verification and validation. These words refer to the
veracity or truth of the model or simulated process.
1.2. ESTABLISHING THE AUTHORISED LIFE OF
Unfortunately, use of the words verification and validation LIMITED LIFE COMPONENTS.
varies, and we need to define how they are used in this Where the consequences of a component failing in a life-
paper and elsewhere in this document. For our purposes: related mode arc critical, an in-service life limit is imposed
to control the risk of failure occurring. This 'authorised'
• Verification is checking the truth of an element of a life limit is imposed by the design or airworthiness
system. authority, depending on circumstance, and is set at a lower
• Validation is checking the truth of a complete system. level than the predicted safe-life because of the need to
These definitions are as used in ISO 9000, which
defines Quality Management Systems. Design to life
After manufacture, the life consumption of the
in-service component is then monitored against _w
the design safe-life using one of the approaches 0
described in this document. Most of the
approaches have been used extensively for many
Life limit
years and involve largely 'manual' processes,
(arbitrary
which do not pose any new validation or
unit)
verification questions. The most recent
Life consumption Residual life
approaches use a set of simplified models,
since new
known as reduced order algorithms. These are
implemented in a suitable recording and analysis "^- r-
system to calculate the life consumed for each
individual component. This technique presents Engine usage and /Engine component \
significant validation and verification issues that life monitoring V life database J
must be solved before the system can be used for
the executive control of critical component lives.
This chapter concentrates on the validation and Figure 1 - Designing for life and monitoring usage
verification processes for these systems.
A complete usage monitoring system contains provide a safety margin. This margin is designed to
the following functions: account for:
• Data Recording - acquisition of the necessary data for • Uncertainty in the design methods.
calculation of the component usage values. • Scatter in part dimensions and material properties.
• Damage Calculation - calculating the life used by • Ageing and damaging effects from the environment.
each component or critical area. • Variability in both engine response and usage.
• Life Management - Organising the life usage
information to support decisions on aircraft The evaluation of the margin between predicted life and
deployment, component retirement, engine removals authorised life is part of the lifing process. The method of
and engine and spares management. evaluation can vary:

The relationship between these three functions and • Depending on the particular experience of the
component life is shown in figure 1. manufacturer.
• Depending on the design year of the engine; The life
The damage calculation processing may be carried out on- of a recently developed engine type will rely on a
board or on the ground. In either case, there will be both certain amount of sophisticated mechanical analysis
airborne and ground based elements of the system. Figure plus some component and engine testing. Older
2 shows these arrangements, and figure 3 shows some engine types will be assessed through a smaller
possible data transfer methods. number of elementary analyses and more extensive
7-4

continuous testing of whole engines. described in chapter 6. The authorised life is declared at
the end of the lifing process by the design or airworthiness
Nevertheless, certification authorities for both military and authority, depending on national practice.
civil engines must approve all methods.
1.4.2. ELMS SPECIFICATIONS
1.3. REDUCING THE MARGIN BETWEEN The main objective of an ELMS is to compute the residual
ESTIMATED AND ACTUAL LIFE CONSUMED safe life of a part or a module after every flight. The
Improving the design methodology and manufacturing design models are too complex and too large to be run in
processes can reduce the life uncertainty due to the design real time even using a large mainframe computer.
methods and to the scatter in part dimensions and material Therefore, there is a need for simplified models that can
properties. A good design
may reduce the impact of On-board process On-ground process
ageing and of the
environment on the engine. Component Component
The flight profiles used in life data base life data base
life design can be chosen
so that they describe more
precisely the typical service Transfer:
I
Transfer:
flights. However, as engine identification/
id engine identification
'initial
mal '■'
life
described in chapter 9, the
variability of engine usage
and response induces a
very large scatter in the
Compute: Record:
conversion from number of On-board
engine component life flight and engines
flights or flight hours to consumption parameters

accumulated damage.
The most satisfactory way
to reduce the uncertainty in Transfer: Transfer:
estimating the effective engine identification engine identification
residual life flignt parameters
sorties performed by an
engine is to continuously
monitor and record enough
data to compute and record Compute:
the damaging effect of each engine component life
consumption and
successive flight. residual life

Figure 2 - Alternative architectures for data processing and transfer

1.4. MAIN SPECIFICATIONS FOR AN

ENGINE LIFE MONITORING SYSTEM
Let us first establish the differences between the
lifing procedures and methods and the Engine Life
Monitoring System (ELMS). The correlation
between the lifing and the monitoring computation
procedures is represented in figure 4. The first
column of processes represents the 'Perfect' model
and the second the reduced order algorithms.

1.4.1. LIFING METHODS OBJECTIVES

The aim of the lifing procedures in the engineering
department at any engine manufacturer is to detect
all critical areas of all the components of an engine
and to compute their life. To do so, they use finite
element models of engine components or
assemblies to perform transient thermo-
mechanical analyses for simulated flight profiles,
which cover the expected flight regimes of the
engine. The damage analysis procedures then
predict for each stressed location the critical
damaging mechanism and the corresponding life. Figure 3 - Data transfer methods
This analytical prediction is demonstrated through operate in an "on-board' computer or in a typical PC at an
testing of individual components or whole engines as operating squadron.
 7-5

Usually, such a system can be split into two main parts: overhauls, and aircraft deployment to remote sites.
• The computation of the consumed life for each Due to the scatter of the behaviour between any two
monitored engine or component. engines, these benefits can only be maximised through the
• The calculation, storage and management of residua! fleet-wide use of monitoring units, and ground teams
life for engines and components. trained to use such systems. Chapter 9 gives an indication
of the degree of benefit to be gained from various levels of
The two sub-systems must be able to exchange monitoring.
information. After the building or rebuilding of an engine
the reference identity, and life consumed for all the 2. DEFINITION OF DAMAGE UNITS AND LIFE
monitored components of the engine should be sent to the
UNITS
damage computation unit. At the end of each flight or
suitable number of flights, the consumed life for each 2.1. SQUADRON BASED DEFINITION
engine component is transmitted to the life database. The
maximum period before downloading the flight or damage There are two purposes to which usage monitoring is put
information to the ground station is important regardless of in service operation:
where the data processing is carried out. This is because it • The engineering officer at the base or squadron needs
affects the need for memory and the practicality of to know how much residual life is left so that he can
operating the ELMS in service. It is critically important decide on the operational capabilities of each aircraft,
during the design phase that the computational power and e.g. can it complete a proposed detachment.
memory needed for processing and storage is accurately • The logistics specialist at headquarters needs to know
determined, wherever the various functions are to be how much of the authorised life has been used at any
eventually implemented. time so that he can order the necessary spare parts. He
The reduced models used for life computation should be also needs to schedule engines into the repair shop for
simple enough to be performed either on-board or at the replacement of components whose life has been used
Squadron unit at the end of a flight or recording period. up.
They should incorporate the maximum possible physical
basis to make the demonstration of their validity easier. A variety of different units and counters are in use,
They should also be consistent with the engineering lifing sometimes simultaneously, to describe engine usage.
method used by the manufacturer of the engine. Depending on the aircraft operator one can find the
following counter types:
There is a range of complexity which can be incorporated
into these reduced order models depending on the • Clock time, total running time, after-burner running
availability of the input data and, particularly for older time, running time over X% of nominal, etc;
engines, on the availability of a full finite element • Number of flights, number of take-offs etc;
component analysis. A relatively modern engine would • Number of engine-starts, or after-burner light-ups.
have sophisticated models already available. In this case
the reduced order model should incorporate this and There are difficulties in translating such counter
should therefore provide an accurate measure of information into residual life. This has always applied to
component usage. Some older engines (50's and 60's) may military aircraft operations because of their inherent
not have been through a re-evaluation process that will variability, due to 'non-standard' routes and the need to
generate modem stress models, and in these cases, it may respond to the unexpected. These difficulties increasingly
be more appropriate to use simple algorithms based on the apply to civil operations because they, also, are becoming
existing stress models. Appropriate margins will have to more variable with the increasing use of flexible ratings.
be incorporated into these very simple algorithms, but they
will still give significant benefits compared to the most 2.2. MECHANICAL ENGINEERING BASED
basic life control methods based on flying hours. DEFINITION
The damage sustained by a part is not a uniquely defined
1.4.3. EXPECTED BENEFIT OF AN ELMS material property. In practice, it is defined through several
A number of benefits can be derived from a good life different mechanisms such as low cycle fatigue, thermo-
monitoring system: mechanical fatigue, creep, and oxidation together with a
combination rule as described in chapter 5. According to
• Life monitoring results in a better knowledge of
these models, a part will break when after a history of
effective flight parameters and their impact on
being stressed at different levels and different
component damage.
temperatures the value of the accumulated damage
• Overhaul of critical components can be undertaken variable reaches a predefined limit. By observing trends
closer to the life limit, with the same degree of
over time this information can provide a good indication
confidence in the safety of the process. of the right time to buy new parts.
• The ELMS provides practical help for predicting the
time remaining until overhaul is required and through 2.3. PRACTICAL CHOICE
trend analyses it allows a precise usage rate to be
computed for each engine. Practical consideration leads to the use of different units
for residual life evaluation and for the expected time
• Knowledge of individual engine life-consumption can
before overhaul. The principal considerations are given
be used for fleet-wide scheduling of inspections,
7-6

below:
/ Flight and engine , y Conventional design (light profiles
• However sophisticated the life parameters / Real flight profiles
/
management method may become, it
is necessary to keep some existing V
global simple usage measures Whole flow path
Performance modelling:
because they are the only systems to air flow temperature Some sections
rely upon if the high technology is
unserviceable.
• Normally the life limits of an engine v All engine,
component are expressed in terms of whole component volume
reference cycles defined by the Temperature modelling: Critical components
material temperature Critical life locations and
manufacturer. gauge locations
• In the engine life monitoring system,
the accumulated damage is
computed in the appropriate V Whole engine.
volume and surface stress,
engineering units, such as LCF non linear analysis
cycles, turbine creep units or other Stress analyses: Critical components, life locations
mechanical stress and stress components,
continuous damage variable. For local non linearity
homogeneity reasons, it will be
translated into the same reference
cycles as have been used for V
declaring the authorised life. Damage modelling: Full damage model
• The residual life is determined in life consumption Reduced damage model
residual life
flying hours or number of sorties,
depending of the aircraft operator,
using "temporary" exchange rates.
The Engine Life Monitoring System 1r
Design life
installed in each particular squadron / Residual life / Residual life
allows a local "temporary" exchange
rate, that can periodically, or Fig ure 4 - Comparison of the reduced model with the life desigil process
continuously, be reassessed instead
of the "normal" exchange rate based • The consumed life should be available after each
on overall experience.
flight. The damage computation may be made
3. REDUCED DAMAGE MODELLING FOR available in real time using an on-board computer or
ELMS very rapidly after the aircraft lands using a normal PC
type computer.
The development of a reduced damage model is usually a • When transferring damage or flight data from the
two-step process: engine to the ground the monitored components
• Development of the algorithms. This is usually should be unambiguously identified and associated
performed using a conventional computing system with the data. Engine configuration should be
such as mainframe, workstations or PC's. This system included within the database of the reduced engine
will be kept and maintained so that it can be used later description.
for modification or development of the reduced • Frequently, only one critical feature may exist in each
models and ground based life studies during engine module. If so, only this critical area need be
service life. monitored. For hot sections, there may be more than
• Implementation of the models into operational one critical feature, depending on whether the major
hardware. damaging mechanism is creep or fatigue. In this case,
enough features must be monitored to ensure the
This section will describe development of the algorithms. integrity of the modelling process. Figure 5 displays a
Implementation will be addressed later. typical representation of the monitored areas included
in the ELMS.
3.1. REDUCED ENGINE DESCRIPTION • The ELM model must take into account the engine
Models should be defined in conjunction with the operating conditions, and the mechanical properties of
decisions that are intended to be made as consequences of the materials and components. The analytical process
the outputs. Here the model monitors the life consumption to compute the life consumption from the engine
of the critical parts of the engine rotor. The decision will parameters is represented in figure 6.
be when to inspect or to overhaul the rotor.
Points 1 and 2 are mostly related to the hardware
Four basic requirements should be included in the organisation and the monitoring process and will be
monitoring system and the reduced engine model. These referred to in the second half of this chapter. Validation of
are: points three and four will be the main topics of the
following sections.
 7-7

3.2. REDUCED ORDER ALGORITHM

INITIAL VALIDATION

3.2.1. OLDER ENGINES

The reduced modelling appropriate for an engine
will depend on the design methods used to predict
its component lives. In practice, it will depend on
the date of development of the engine. For the
older engines designed using analytical
computations calibrated by testing, the ROA can
only be an application of the rules that used
engine-operating parameters. Usually, the ROA
will then be a direct computation of consumed life
from the maximum rotor velocity during each
O
flight.
For this single step approach, the validation
3 Thermodynamic airflow temperature section

procedure is to run the algorithms through a Venting air temperature

standard sortie that covers all of the expected flight Material temperature
conditions and exercises all of the logic in the
algorithm. All of the flight conditions in the sortie o Critical areas: fatigue, creep (temperature, stress)
are computed, both by the full design models and
by the reduced order algorithms. The results at Figure 5 - Reduced module model
each flight point are compared and assessed against incorporated into the reduced model is a simplified
the accuracy requirements. This gives a large sample derivation of the design material database. This is to
covering a wide range of conditions and effectively covers reduce the computer memory required by the ELMS.
the validation activity for the reduced order algorithms. Secondly, the thermo-mechanical history of a real engine
is much more erratic than the conventional flight profiles
3.2.2. NEWER ENGINES used in design. The life modelling and the material
The newer engines are hotter and more highly stressed database should accommodate:
than the older ones, and they are expected to achieve
higher levels of safety. They are designed using • Numerous small cycles of real flight;
comprehensive models that start with the instantaneous • Very rapid loading which occurs during rotor
operating state and predict successively the air temperature acceleration and deceleration;
within the engine, then the material temperature, then the • Constant loading can lead to some concerns in real
mechanical stress and finally the increase in accumulated time creep and fatigue modelling.
damage. Chapter six describes this process in more detail.
The reduced order algorithm will be divided into
modules that generally copy the structure of the life
design model of figure 4. Each module will / Internal \
I material j
incorporate reduced models, based to some extent on \ database J
the same physical realities and approaches as the full
models. The extent of this depends partly on the
philosophy of the engine manufacturer concerned.
Inevitably, some parameters in the reduced
algorithms will require optimisation during the Input: Damage modelling: Output:

algorithm development process. local temperature —► fatigue, creep, -> local damage
and stress history crack propagation increase
A brief description of the essentials for the reduced I J
model, including some important points for their
Figure 6 - Materi alsE)amage Modelling
calibration, will be given in the following sections.
Therefore, to be sure of the equivalence between the
3.2.2.1. MA TERIAL DAM A GE MODELLING design and the reduced models, all the verification
The active damaging mechanisms are determined during procedures of the reduced damage modelling are to be
the life design process for each critical location. The conducted for each material on a wide range of flight
damage mechanism is usually fatigue or creep, and some profiles. This will include real flights, and use the same
new algorithms incorporate crack propagation. Allowances sampling frequency as will be used in the life-monitoring
for crack initiation are more difficult. They depend on the device.
lifing philosophy, and the experience of the manufacturer,
and are unlikely to be included in a reduced model. 3.2.2.2. STRUCTURAL MECHANICS

Figure 6 shows the generic organisation of such a model. Centrifugal forces and temperature gradients cause the
main loads on a jet-engine rotor. For the discs and shafts
There are inherent differences between the design model the simplest equation, which can describe the generated
and the reduced model. Firstly, the material database stress at one location is of following type:
7-8

a = a0 + AxN + ZB'*T.
Where a0 is the assembly stress, N the rotational -£-
velocity and T-, is temperature at specific gauge ^
locations of the rotor. Typical gauge points are
shown in figure 7.
The reduced mechanical model is fitted to the
design model by adjusting the number of
temperature gauge points, their locations and the
model coefficients. The aim is to minimise the
difference between the stress levels resulting from
this model and those from the design model.
Whenever needed this basic equation can be altered Critical area: stress and temperature computation
to cope with:
—A_ Temperature gauge location
• Plastic loading;
• Overload detection; Figure 7 - Reduced thermo-mechanical model of a disk
• Modification of boundary conditions;
• Introduction of pressure loading.

A similar equation can be written for

torque loading.
The verification of the accuracy of
the structural model should be
performed on a specified variety of Calcute n

flights to ensure that all the possible -Calculs C

temperature situations have been

covered. In fact, the higher the stress
level and the more frequent the
occurrence of the situation leading to
this level, the more stringent is the
need for a precise reduced model.
Despite the apparent simplicity of the Temps (s)
equations, it is important to keep in
mind that this type of equation is a Figure 8 - Reduced model results compared to the full model
carefully constructed representation
of a very complex model. The temperature distributions It should be noted that the correct prediction of thermal
and centrifugal loads used during the model fitting must be stresses requires the accurate prediction of both low and
representative of real situations. If they are not, then the high temperature conditions.
reduced model can be exaggeratedly complex or even
inaccurate in the temperature range encountered in service. 3.2.2.4. THERMODYNAMIC MODELLING
The aim of this model is to provide the air temperature
3.2.2.3. MA TERIAL TEMPERA TURE MODELLING characteristics needed by the solid-component temperature
The locations where the material temperature model. Again, it is important to note that the correct
characteristics have to be modelled are: knowledge of thermodynamic state of an engine will be
available only after having investigated the full flight
• The critical points where life is calculated; envelope.
• The gauge points as needed by the stress calculations.
Most of the time the transient temperature calculation is a 3.2.2.5. CONNECTING THE REDUCED MODEL ELEMENTS
2-step process. At each instant first the steady state Some decisions concerning the structure of the whole
temperature is computed, and then the temperature change model should be taken before finally assembling the
rate. The inputs are the power setting, the air flow individual sub-models. Most of them relate to the interface
temperatures and the rotor speeds. Figure 8 shows the between the engine and the aircraft.
accuracy it is possible to achieve by this method for a
turbine disk location. The flight limits should be unambiguously defined:
Two important factors can affect the accuracy of the • When does a flight start and finish and when should
reduced thermal modelling: the damage calculation start and finish?
• Will damage be counted only between take-off and
• A bad choice of model parameters; touchdown?
• The full model does not become a 'perfect' • Will ground tests be evaluated, and how?
representation of the engine until almost the end of the • Will 'Touch-and-Go's' be considered as one flight
development programme.
 7-9

(one engine start-up and shutdown) or as many flights This would appear to involve an impracticable large-scale
as landings? comparison between full modelling and reduced
• Should the damage counting process differentiate a modelling. A method has been proposed (Reference 1)
hot start from a cold start? which consists of:
• Measuring the deviations of the individual sub-models
These decisions are specific to each project and should (fhermodynamic, thermal, mechanical, and material
take into account the hardware and software organisation life) from the 'perfect' design model on a set of
of the aircraft and engine. A particularly crucial point is standard missions;
the availability of electrical power for measurement and • Combining the partial results in a Monte Carlo
computation when the engine is shut down. Another is the simulation to build a deviation model for the whole
availability of signals from the aircraft. Finally, some
system;
physical situations cannot easily be modelled because they
• Using the result as a basis for having confidence in
are extremely variable. Examples are:
predictions for any new missions or design tasks that
• The decrease in rotor metal temperature after engine arise.
shutdown;
• The stress field across a rotor that has not completely The result provides the confidence level for the damage
cooled before the engine is started again. calculation, based on to the uncertainty of the individual
sub-models. Currently, this method has not been widely
Each damage computation model will then be refined and used, but it appears sufficiently promising to be mentioned
tuned in order to meet these operational requirements. here.

3.2.2.6. VALIDATION OF THE COMPLETE MODEL 3.3. LONG TERM VALIDITY OF THE REDUCED
One method of validation is to make a comparison DAMAGE MODEL
between the results from the full model used for life The information generated by the reduced damage model
determination and the results from the Reduced Order should be kept during the whole fleet life of maybe 30
Algorithms. The procedure consists of computing the years. It is thus necessary to keep it up-to-date, despite
damage for each critical area, for a given set of flight some inevitable modifications to the engine or the system
profiles including synthetic and real sorties, so that the around it. This section will identify the main situations
damage computation can be tested over the whole flight where a new validation process should be performed. If
envelope. Though all the models may have been the reduced model is revised and the ROA's changed then
individually calibrated, there can still be some the relevant parts of the validation process must be
unacceptable over or under-counting. This results from the repeated.
particular sensitivity of one section of the model to the
uncertainty of the previous section. When considering 3.3.1. IN-SERVICE COMPONENT REPLACEMENT
creep damage for a material used near the temperature When a component is replaced during servicing, by
limit, it is important to remember that a temperature another with the same life characteristics, this need only be
increase of 10°C could halve the creep life. Some recorded in the life database. The initial life of the
experience shows that this can be allowed for fairly easily, replacement component is 100% if it is new, or is the
because the effect is more significant in terms of previously recorded damage if it has been used before.
temperature gradients and stresses, than in terms of This situation only affects the life database management,
material properties. and not the monitoring system, except when the

Airborne Hardware Data Acquisition items

(Connections to existing system, new sensors etc.)
Processing items
(Data handling unit)
Software Implementation of the Requirements, (see note below)
Ground Hardware Data Transfer medium
(Hand held unit, smart card, cables etc.)
Computer Terminal
(Probably commercial off the shelf PC)
Software Implementation of the Requirements, (see note below)
Data Transfer Procedures.
Fieure 9 - Svstem elements to be validated
The reduced model is declared equivalent to the full model monitoring system may need to be reinitialised with fresh
when the mean values of computed damage are the same values for life-consumed. The need for this will depend on
and the maximum difference is below a defined limit. An the design of the particular system.
over-riding constraint is to ensure that the established If a component is redesigned, or a modification to other
material properties are not exceeded. components alters the behaviour of the engine, the rate of
An alternative method for the validation process is to damage accumulation for the monitored parts may change.
determine the statistical accuracy of the reduced algorithm. The reduced order algorithms may have to be altered, and
7-10

any changes will have to go through the verification and should apply and any effects on the performance of the
validation process. monitoring system must be evaluated against the criteria
given in paragraph 5. The data that has already been
Both cases present concerns over the configuration control
recorded must be kept and must be identified separately
of the engine standard, the monitoring system standard,
from 'post modification' data so that it is available for
and the life database. Control of the correct combination of
comparison purposes.
the first two is essential to ensure that the damage
accumulation calculations are correct, and control of the 4. ENGINE LIFE MONITORING SYSTEM
third is essential to ensure that all parts have the correct
total damage associated with them. This section covers the validation of the elements that
constitute a practical simplified system. The elements are:
3.3.2. AIRCRAFT AND ENGINES FLY NEW FLIGHT
Airborne system hardware including any new sensors.
PROFILES
Equipment for data transfer to the ground station.
This can occur when introducing the life monitoring Procedures for data transfer.
system on a fleet-wide basis, or when introducing the
Ground station hardware
system with a new operator. If the new engine usage is
Software for data acquisition and handling including
clearly different from that already experienced, it is
the accuracy of the Reduced Order Algorithms in
recommended that the accuracy of the reduced modelling
representing the results from the Full Algorithms.
for each new flight profile be checked. As the reduced
This software will always be split into two parts
modelling approximation is only valid for the situations
where it has been calibrated, a tuning of algorithms may because there needs to be some in the airborne unit
be necessary. Again, it is mandatory that the changes to and some in the Ground Station. The exact split will
the monitoring software be correctly managed. vary between projects.

If the changes apply to all engines of the type, it will be The equipment and processes used to handle the usage
necessary to evaluate the need for an alteration to the data after it has left the station^ase where it was generated
residual life of all existing parts. This is described in the are excluded.
next paragraph.
4.1. INTEGRITY REQUIREMENT
3.3.3. CHANGES ARE NECESSARY IN THE Gas-turbine component life-consumption is a progressive
REDUCED MODELLING and, normally, a long term (1000 hrs. plus) process.
This situation can occur every time a modification in the Therefore incorrect counting does not cause an immediate
understanding of engine operating conditions is introduced problem, and a usage monitoring system is not safety
in the design process and methods. For instance, the critical in real time in the same way as a flight control
following cases should be considered: system. Also temporary counting errors (for a few flights)
do not significantly affect the overall situation. However,
• Engine ageing causes component efficiency and systematic errors, which persist for a large proportion of
internal airflow to change and a new critical zone or a the time, are a cause for concern.
new damage mechanism is revealed.
• A new life design method results in a new predicted Monitoring system errors that cause undercounting have
life. possible safety consequences and procedures must be put
• The reduced model has been altered for any reason. in place to manage them. On the other hand, over-counting
errors do not result in safety consequences but do have
The validation at the end of the development of the new economic consequences. These errors will be satisfactorily
reduced model is a 2-step process: managed by the procedures used for undercounting.
• Establish the transformation rule, including spread The incorrect counting may be categorised as follows:
margins, between old and new damage predictions. • Inappropriate system design and implementation. This
• Re-evaluate the residual life of existing components could be due to either incorrect envelope
using their actual life, any knowledge about every identification, or changes in operational practice.
engine's life, and the new model. • System hard faults identified by BITE.
• Subtle system faults, BITE failure or drift, where
It is essential for the engine manufacturer to have an operation continues but the count is incorrect. An
extensive and continuously updated library of the effective example would be sensor calibration drift.
flight profiles performed by the whole fleet. This process
is particularly difficult to carry out. In particular, a good The first group can produce both temporary and systematic
compromise between safety margin and customer errors. The second group can only produce identified
acceptability can be difficult to reach if the residual lives 'temporary' errors that can be corrected and the third
are significantly reduced. group will usually produce 'systematic' errors. All three
groups can be managed by the combination of an initial
3.3.4. ENGINE LIFE MONITORING HARDWARE
validation programme and an ongoing service review
CHANGES
process.
These changes can affect the life database or the airborne
system. In this case, the normal rules for modifications The validation process, which will include all foreseeable
circumstances and confirm that systematic errors are very
 7-11

unlikely to occur during initial operations, will primarily will provide a routine system check, (say once per year)
cover the first group. Because operations change through and may be used to investigate suspicious results.
the service life of any engine application, regular reviews The following examples, in figure 10, provide some
of the life control process are carried out as part of current quantification of the change in probability of the fleet-
practice. This will still be necessary with a usage wide presence of a (0.38-mm) crack when undercounting
monitoring system. However, there will be information, occurs. They come from a study on retrofitting RAF
which is more detailed, available from the monitoring Tornadoes with engine monitoring systems. Critical
system to help the review. Any systematic incorrect components in the RBI99 engine were designed based on
counting which escapes these checks would certainly be the chance of a crack occurring being 'one in 750 with
missed by the present procedures. Therefore any other 95% confidence' described in Chapter 6. The table values
incorrect counting in this group will be due to temporary are considered to be pessimistic scenarios for undetected
causes and will not significantly compromise the cases of undercounting and none of the cases generates a
accumulated damage values. significant increase in the overall probability of a crack
The second group is the simplest to manage because the appearing.
fault is clearly identified. A clear and reliable mechanism The values in the table are for fleets of 50 and 200.
must be established to ensure that these events are Assume that 5% of the components in a fleet of 200
immediately reported to the data management staff so that undercount by 20% for 1/3 of their lives. It can be seen
all identified deficiencies are corrected. Some systems from line 4 in figure 10 that the probability of a crack
automatically use a pessimistic default value when a occurring at the design life rises from 1 to 1.04 in 750.
detectable fault occurs.
As a result of these considerations there is no reason, from
The third group of failures is probably the greatest threat the usage monitoring viewpoint, for any part of the
because it is a new type of problem for critical-part life- monitoring system to be safety critical (e.g. D0178B level
management, and associated with the introduction of A or B software). However, other considerations apply to
monitoring systems. This type of fault can be effectively the airborne elements of the system.
managed using the redundancy inherent in the information
The interaction with
% of cyclic % of life where % of fleet affected Fleet probability Fleet probabilitysafety critical control
damage or stress undercount occurs by undercount change factor change factor systems, which will
undercount generally result in the
(200 engines) (50 engines)
airborne software being
10 total 3 1.04 1.05 controlled to D0178B
10 total 5 1.07 1.08 level C, is probably the
most important.
20 total 1 1.04 1.05
However, it is possible to
20 33% 5 1.04 1.05 use Commercial Off The
1.07 1.08
Shelf (COTS) equipment
30 33% 5
for the ground based
Fig 10 - The effect of undercounting on the probability of a crack occurring elements of the system. It
is important to remember
management infrastructure, which is needed to make that most of the processes, which define the safe lives of
effective use of a monitoring system fit. The amount of critical components, are carried out on COTS computers,
additional capability depends on the existing techniques both hardware and software. Consideration of the whole
and equipment used to manage the fleet, and the system and built in redundancies makes this possible.
information management style chosen by the operator. The
ground 'units' can be PC's or mainframe computer 4.2. SYSTEM ARCHITECTURES
terminals or, at the other extreme, databases in hard copy Generally, the data processing for the health monitoring
form. functions is carried out on-board because the results are
When a system is operated in-service, there must be three required at the aircraft, to help the maintenance crew clear
elements for practical reasons: the vehicle for the next mission. On the other hand, the
calculation processes for usage monitoring need not be
• An 'on-board' unit which gathers the basic data. carried out on-board. This is because component life is
• A 'squadron' unit to provide information for first line used up slowly during normal operation, during which
maintenance. exceedences, by definition, do not occur. Any significant
• A 'headquarters' unit to provide information for exceedences will be recorded when each event occurs, and
spares ordering and operational planning. flagged to the ground crew. This can be used to trigger a
usage assessment.
Each of these elements can be used to review the results
and provide a check on the other elements and will provide The choice of system used on any particular project
an effective means to detect and allow for the third group depends on the specific requirements of that project and
of failures. In addition, wherever the processing is carried the degree of interaction between the 'usage' monitoring
out, it is desirable for the on-board unit to be able to record and 'health' monitoring functions. For example, an aircraft
1 to 2 flights of raw data for independent processing. This that is intended to operate for extended periods from
remote locations with minimal support will require more
7-12

capability in the airborne system. and discovering the differences between them. Full
validation covers the whole operational process from
Any system will require a ground station to act as the beginning to end, and should take into account the realities
interface with the operator's IT networks so that the of operating complex systems in adverse conditions.
information from the individual aircraft can be passed to Weather and factors such as the electro-magnetic
the support functions. Systems where the processing is environment are just two types of adverse condition that
carried out on-board are frequently able to record data for may be encountered.
several flights. For example, the Harrier GR Mk 5 can
record usage results for up to eight flights, and record bulk 4.4.1. AIRBORNE SOFTWARE
data up to the limit of the storage medium. If bulk data is
recorded, ground analysis can later use the data for The accuracy requirements will depend on the degree of
validation purposes, or investigate incidents in detail. This sophistication of the models, both full and reduced. If the
capability helps to maintain and demonstrate the integrity models are simple, as was common for engines of the 50's
of the total system and reduces the need for high levels of and 60's then appropriate safety margins will be needed.
control over individual items. It may also remove the need These can be incorporated within the reduced order
for Class A or B software. algorithms, applied to the resulting reference life units, or
applied to the authorised lives. For the newest engines, the
With modern commercial capabilities for data storage and reduced order algorithms are necessarily adapted to the
transfer, such as smart memory cards and Wide Area available memory and precision of the airborne computer.
Networks, the preferred method for usage monitoring is to
record the data on-board and carry out the analysis in a The validation process is to confirm that the implemented
ground station. This approach minimises the airborne system gives the same output as the full engineering model
element of the system and thereby significantly reduces when 'flown' through each of the full range of missions
the recurring costs of the validation and configuration expected in the service life of the engine. This activity will
control work required. It also allows the recorded data to include using a mixture of missions, some 'synthetic' and
be re-analysed later using revised techniques and material some 'real'. The synthetic missions are designed to ensure
information. that all the extremes of the flight envelope are covered
with a margin to allow the envelope to be exceeded
4.3. SYSTEM ELEMENTS TO BE VALIDATED without causing gross inaccuracies in the algorithms. They
should also cover the full range of input parameters and
The objective of the validation process is to demonstrate cycle sizes to demonstrate that the reduced models have
satisfactory operation of the system in all foreseeable the necessary robustness. The real missions are taken from
circumstances. recordings of typical flights, either by the actual aircraft or
The process should cover each of the items in the system by a similar one and are used to ensure that the algorithms
individually in a manner appropriate to that item. Having operate satisfactorily with the transient data characteristics
produced evidence that each element operates in of service operation.
accordance with the full specification requirements for that
Use of an appropriate approval process such as D0178B at
unit; further 'end to end' testing is required to demonstrate the appropriate level depending on the overall architecture
that the total system operates correctly in all foreseeable
of the application is recommended. For D0178B approval
circumstances.
would normally be at level C/D but there may be a need to
The process needs to consider both hardware and software clear it to a higher level if it is closely integrated with
aspects for the additional Airborne Units and for the other software (e.g. engine control) which itself requires a
Ground Station. The major aspects are shown in figure 9. higher level.
Note: The Reduced Order Algorithms may be 4.4.2. AIRBORNE HARDWARE
implemented in either the airborne or the ground units
depending on the architecture chosen. This affects the Airborne hardware is composed of specific measurement
software requirements for the two units in any specific devices, the link between the ECS and ELMS, and to the
case. If these algorithms are implemented in the ground ELM if this is a separate unit.
station then the airborne unit software is only required to The airborne hardware is a group of accessory units:
control the data acquisition process. If these algorithms are speed, pressure and temperature sensors, wiring, avionic
implemented in the airborne unit then the basic software in unit etc. Their functioning in the environment for the
the ground station is only required to handle the application is validated using the same specifications as
accumulated usage counts. However, it is recommended other similar units on the engine. This testing would
that the ground station should have the ability to process normally include aspects such as the following.
raw data recorded from service to confirm the satisfactory
operation of the system and to allow detailed analysis of Functionality;
any changes to the operating pattern. Temperature;
Vibration;
4.4. VALIDATION PROCESS Electromagnetic Compatibility (EMC);
Validation is the process of ensuring that the whole system Humidity;
works correctly or at least that any deficiencies are known. Salt Spray;
Figure 11 (also shown in chapter 6) illustrates part of the etc.
process: comparing the master and simplified algorithms,
The functionality and temperature requirements will be
 7-13

generated by the specific project; the others will usually be on the proposed ground station hardware with the
covered by specifications such as MIL-E-6041 for EMC proposed operating system to ensure that the combination
and MIL-STD-810 for Environmental Test Methods. functions satisfactorily.

4.4.3. DATA TRANSFER HARDWARE 4.5. OVERALL SYSTEM

This unit may be both a part of the airborne system and a Successful operation of the usage monitoring system
part of the ground station or it may be designated as a requires that all of the components and procedures operate
piece of Ground Support Equipment (GSE). The validation effectively together. Therefore, it is essential that a
process will depend on the system architecture, and the program of end to end tests be carried out on the complete
methods used should be appropriate to the status of each system after all the individual units have been qualified.
item.
4.4.4. DATA TRANSFER PROCEDURES Input (Mission)
The data transfer process will vary between
individual applications but will always be a
mixture of automatic data movement between
pieces of hardware and the manual activities Performance Performance
Performance
needed to make this data movement happen Algorithm Algorithm *?« (exact)
correctly. The validation of the data movements

¥
Frtqu«ncy Dialributiofl
can be carried out as part of the programme on
the overall system. The procedures which control
the manual activities and any techniques used to
Component
identify questionable data and to repair it or Temperature Temperature Temperature
Algorithm Algorithm --*?-
compensate for it should be subject to audit to (exact)
provide assurance that they are adequate and Frtqutncy Distribution

satisfactorily maintained.
This audit should take account of and cover the
following aspects: Stress
t
Stress ?x>
Critical Area
Stress
Algorithm Algorithm (exact)
• Identification of personnel carrying out the Fftqmncy DutribuBon

download under various circumstances; &

• Storage capacity of the airborne unit and x -,
transfer unit;
Damage Damage Damage ;
• The required frequency of downloads, and Model Model Model i
the number of aircraft and flights which can
be downloaded to the transfer unit, without
return to the ground station;
• The identification of, and compensation for,
system failures and data loss; c=o
• Programme for regular critical reviews of the Simulated Deviation Result Deviation
system outputs;
• Training programme for operating Figure 11 - Correlation between lifing and monitoring results
personnel.
There is no prescribed formula for these tests and they are
4.4.5. GROUND HARDWARE normally arranged to suit each specific project. However, a
typical practical approach is to carry out a sequence of
This will probably be commercially available equipment
such as personal computers, and therefore the 'validation' testing designed to cover the following stages.
activity is carried out during the selection process used to 1. Demonstrate that each function in the system operates
decide which unit to procure. Any specific testing should correctly over the full range of input data values
be based on the normal practice for ground support covering all possible flight conditions and that all the
equipment used to handle engine and aircraft data. logic 'loops' and 'switches' in the software behave as
intended.
4.4.6. GROUND SOFTWARE 2. Demonstrate that the system operates correctly in real
The operating system will probably be commercially time with typical dynamic inputs.
available software and therefore the selection process will 3. Demonstrate that the system operates correctly in the
include assessments of its integrity. Most operating real environment (EMI, weather, service operation
systems are reliable and the basic further constraint is to etc.)
ensure that it is mature. If the reduced order algorithms are
implemented in the ground station then they will be in a 4.5.1. LABORATORY TESTS
block of project specific software that should be subject to
testing to check that the algorithms have been The purpose of these tests is to demonstrate that the whole
implemented correctly. This testing should be carried out system functions over the full range of input parameters
7-14

and that all the logic in the software operates correctly. 5.1. FIGHTER AIRCRAFT: MIRAGE 2000 AND
They are intended to demonstrate item one of the above M53-P2
list.
5.1.1. SYSTEM DESCRIPTION
The testing requires a representative set-up of the complete
The M53-P2 is the engine of the Mirage 2000 multi-role
monitoring system, where artificially generated input
fighter, operated by various airforces. The engine control
signals can be fed into it and the resulting outputs
system uses digital technology. Since 1987, the French
monitored and checked against the expected values.
airforce has been equipped with a usage tracking and life
Ideally, this set-up should include any other units from the
monitoring system that has three components:
aircraft that interact with the monitoring function
(FADEC, cockpit instruments etc.). This can be carried out • The ELMS is an electronic unit placed apart from the
on a powered mock-up of the system, or as a ground test Engine Control System. It computes the residual life
on the complete vehicle. The sequence of input signal of the critical components on a flight by flight basis.
values and the expected outputs are defined by the group All of the flight and engine parameters are extracted
who produced the Reduced Order Algorithms. The values and transmitted by the ECS to the ELMS through a
should be chosen to cover all the possible operating specific connection.
conditions and to exercise all the features of the software • A ground station, on which the component lives of the
(All ioops' and 'switches'). The values should be chosen whole squadron are managed. This station will
to confirm that the boundaries of each loop and switch are periodically be fed with the residual life of surveyed
correct. For example, if speed ranges of less than 4000 engines.
rpm are to be ignored, then the test input values should • A man-operated transfer terminal that makes data
span and include this value. This controlled 'steady state' transfers between the ELMS and the ground station.
testing explores all the combinations but is not able to This device can used to display the residual life of an
investigate system performance with inputs that vary in a engine to the flight mechanic. It also initialises the
realistic manner. ELMS after overhaul and reconfiguration of the
engine. Its monitoring capacity is forty engines.
4.5.2. DYNAMIC TESTS
This testing can be partially done on suitable rigs and After having been initialised with the engine identification
bench engines, but can only be completed as part of a and the residual life of the monitored components, the
flight test programme because of the need for all the inputs damage calculator works independently of the ground
to be varying in a fully representative manner. The storage system. The lives of twenty critical parts are
duration of this test period depends to some extent on the individually monitored, for creep and low cycle fatigue.
accuracy of the correlation that is achieved, but should not
be less than five flight hours. The software is divided into four main sections:
• Thermodynamic, which calculates air temperatures
4.5.3. SERVICE ENVIRONMENT TESTS from the measured flight and engine parameters;
It is impossible for a practical flight-test programme to • Thermal, which computes temperatures of the solid
cover all environmental circumstances or to reproduce the parts;
normal service situation in terms of operating pattern and • Structural mechanics, which determines the local
personnel. Therefore, the results from the initial service stress from the temperature and the centrifugal
phase should be examined carefully over a significant loading;
period of flying, say 100 hrs. It should include several • Material mechanics, which derives the accumulated
aircraft. Initial operation is normally from one base only. damage at all the critical locations from the
As operation starts up at each significant new base, a temperature and stress values. This section integrates
similar validation exercise should be undertaken. the material property database.
The exercise should compare the usage calculated by the
system with what would be expected for an aircraft 5.1.2. VALIDATION PROCESS
carrying out the particular missions. This 'expected usage' The following examples will not describe each individual
will inevitably be a relatively crude estimate of reality. For step of the validation process but focus on some 'lessons
each project, the important criteria to help in this learned' during the ELMS development.
assessment must be defined. For example, the number of
'Touch and Goes' or, in a helicopter, the number of free- • During the development of the ROA a very strong
wheel disengagements may be the criterion. The purpose sensitivity between creep damage and temperature
of the exercise is to look for major mismatches between level was found. Thus, special attention was paid to
the counts recorded and what is expected from the the precision of the assembled software as measured
knowledge of the sortie; e.g. sorties with no counts or by the damage. Temperature and stress are not the
sorties with far too many counts. goals, they are the way to damage and life!
• Numerical approximations due to restricted on-board
5. EXAMPLES computer power and memory can lead to unacceptable
differences from the results achieved on the ground
The following examples are given to explain how two
computer. Extensive testing using various flight
ELMS have been designed to conform to a specific usage
profiles is necessary between the ground computer
specification. They will focus on validation process, ease
and the on-board system.
and cost of operating an engine life monitoring system.
 7-15

Flight testing also led to some discoveries: that exercise all the different routes within the
calculation process. This was to confirm that all
• Initially, the real flight-damage count was much aspects of the software requirements had been
higher than expected. This was because the effect of implemented properly so that the system operated
small cycles was over-emphasised. The material correctly in 'non-real time' and with a series of steady
property description and the fatigue model were state inputs.
modified, to correctly count small cycles. 3. Flight testing in an aircraft with a data recording
• Local temperatures differed from predicted levels in capability. This was to confirm that the system
certain flight conditions, and some parts suffered from operated correctly in the aircraft, in real time with
excessive creep. It was thus necessary to re-identify actual engine data inputs and that the other tasks of
the full-size model and to further develop the reduced the computing system did not compromise the engine
model and re-calibrate it. This type of difficulty can usage monitoring. The results calculated by the
arise whenever an engine is flying a new mission monitoring system were compared with those
type. calculated using a ground system fed with the data
from the flight test recorder (approx. 10 hrs of data).
To win acceptance by the operators the hardware and The ground system was programmed with the same
procedures, in particular for data transfer from and to the algorithms as the airborne system. This indicated
ground, must be as easy to use, and as secure as possible. agreement within 0.5% for the two sets of results.
In particular this means: 4. A small sample of the flight test data was re-analysed
• Any flight monitoring system faults must be displayed in an independent ground system with the algorithms
to the flight mechanics by a very clear and easy to implemented by a different software author. This was
access flag. to confirm that there were no errors introduced via
• For operational reasons, the on-board system should common software authorship of the flight software
be able to monitor several flights before the life and the validation software.
information is downloaded to the ground database. 5. Periods of service operation on several aircraft were
• The downloading device should be able to download examined to confirm that the correlation between the
and manage data from several aircraft, preferably a usage measured by the monitoring system and the
whole operational unit as defined by the operating known usage of the aircraft was as expected. This was
airforce, before the operator has to return to the to confirm that operation in service over a wide range
ground station. of environmental conditions did not cause any
significant errors in the calculated usage.
5.2. HELICOPTERS: AGUSTA A129 AND All these elements were applied to the low cycle fatigue
GEM1004 counting function because this has a clear effect on flight
This helicopter has an Integrated Multiplex computing safety. Turbine blade failure on a multi-engine aircraft
system that carries out several functions including aircraft does not represent a flight safety hazard and therefore only
and engine monitoring. The main requirement that led to elements 1, 2 and 5 were applied to the turbine creep
this configuration was the need to operate for extended monitoring function.
periods away from base, with minimal support. Because of The implementation of the creep monitoring function
this, the on-board system is designed to carry out all the revealed that for 'continuous accumulation' functions such
processing and to show the consumed life on the cockpit as this a monitoring system has to operate to very high
display screen. levels of precision. Because of the need to capture the
The engine element of this system calculates the low cycle short periods of very high damage-accumulation caused by
fatigue life consumed for all the critical rotating parts and exceedences the system has to calculate the damage rate
the creep life consumed for the gas-generator turbine over small intervals (less that 0.5 second). However, the
blades. It also monitors the usage of the rating structure damage rate at normal cruise conditions is very low
and reports the duration and level of any exceedance. (approx. 100,000 times lower). The system must be able to
register these low damage increments because this rate
The airborne processing is done in a 'universal' computing will apply for the vast majority of the engine operating
system that has flight critical functions. Because it was time.
certified to a high level of integrity, there was no need for
any hardware testing specifically for the engine 6. CONCLUSIONS AND RECOMMENDATIONS
monitoring functions.
• All usage monitoring systems should be validated
There were five elements in the validation programme for against the manufacturer's design model.
the usage monitoring functions. • Each element of the system should be verified against
1. Validation of the simplified algorithms against full the corresponding element of the master model.
stress analysis results. This was to confirm that the • The implementation must be matched to the users
simplified algorithms were a reasonable needs.
representation of the full cyclic damage analysis using • Allowances must be made for late completion of such
mainframe-computing systems. systems. The internal models cannot be completed
2. Validation of the implementation of the onboard units before there is adequate knowledge of the operational
by testing in the systems laboratory, using test cases environment, and may need adjustment after the
7-16

system has entered service.

• Careful system design can allow monitoring systems
to be designed to avoid the requirements of real-time
safety critical systems.
• The performance of monitoring systems should be
periodically reviewed.
• Consideration should be given to performing life
consumption and residual life calculations in ground-
based systems. This would facilitate effort, and reduce
the cost of implementing updates and corrections in
flight-worthy hardware and software.
7. SUMMARY
The most comprehensive method of life usage monitoring
applies a set of algorithms, to calculate the amount of
damage accumulated for each critical component in each
engine in the fleet. This chapter describes how the
'reduced order' algorithms suitable for this process are
generated and how both they and the system that
implements them are verified and validated.
There are three parts to a monitoring system: in-flight
systems, ground-based systems and data transfer systems
between the in-flight and the ground-based systems.
Validation is a multiple step process that includes ground-
based and in-flight testing.
Benefits of a monitoring system include:
• Better knowledge of flight parameters and their
impact on component damage;
• Overhaul of critical components closer to the actual
life limit with the same degree of confidence in the
safety of the process;
• Knowledge of individual engine life-consumption for
use with fleet-wide scheduling of inspections,
overhauls and aircraft deployment to remote sites.

8. REFERENCES
Reference 1 - The Importance of Testing for Successful
Life Usage Monitoring Systems. Juergen Broede, Hugo
Pfoertner, Klaus Richter. AIMS Conference 98.
Figure 11 was supplied by MTU.
 8-1

Chapter 8
Usage Survey and Mission Analysis
by
(J. Broede)

Page

1. Usage Tracking 8-3

1.1. Time Based Usage Tracking 8-3
1.1.1. Different Time Bases 8-3
1.1.2. Ground Running 8-3
1.2. Event Based Usage Tracking 8-3
1.3. Mission-Type Based Usage Tracking 8-4
1.4. Squadron Based Usage Tracking 8-4
1.5. Usage Tracking Based on Individual Components 8-5
1.5.1. On-Board or Ground-Based? 8-5
1.5.2. Bulk Data Storage 8-5

2. Cyclic Exchange Rate 8-6

2.1. Definition and Discussion 8-6
2.2. Procedure to Establish Cyclic Exchange Rates 8-7

3. Data Selection Criteria 8-7

3.1 Parameter Selection 8-7
3.2. Signal Quality 8-8
3.3. Plausibility Checks 8-8
3.4. Restoration of Faulty Data 8-9
3.5. Selection of Flight Points 8-9
3.5.1. Data Compression 8-9
3.5.2. Tagging of Data 8-10

4. Data Transfer 8-10

4.1. Functions and Requirements 8-10
4.2. Equipment and Media 8-10

5. Records and Accounting 8-10

5.1. Logistic Systems 8-11
5.2. Configuration 8-11
5.3. Database Systems 8-12
5.4. Database Maintenance Technology 8-13

6. Mission Characterisation 8-13

6.1. Mission Treatment for Analytical Purposes 8-14
6.2. 'Building Block'Techniques 8-14
6.3. Translation of Mission Elements 8-14
6.4. Generation of 'Synthetic' Mission Profiles 8-15
6.5. Summary 8-15

7. Conclusions and Recommendations 8-15

8. References 8-15
8-2
 8-3

maintenance actions can be accurately planned for long

1. USAGE TRACKING
periods.
The need for life usage tracking results from the fact that
A flight-time based tracking method tries to correlate the
fracture critical parts in aero engines have limited fatigue
consumed life directly with the flight time, assuming that
life, and the desire to operate the engines safely and all engines run roughly as long as the aircraft is airborne.
economically. In this context, Safe Engine Operation
Again, the consumed life is calculated from the time in
means ensuring engine integrity by not exceeding the life
effective flight hours (EFH), and an appropriate cyclic
limits during operational usage. Economic Engine
exchange rate. Now the cyclic exchange rate provides the
Operation means to approach the life limits as closely as relation between consumed life and flight time. All that
possible, in order not to waste life potential. The ideal is to
remains is to establish the cyclic exchange rate for all of
operate the engine precisely to the life limit, and then stop.
the monitored areas of the engine.
"Usage tracking' is a process, which monitors the usage of
The aircraft structure itself suffers from fatigue, and
each individual critical part in terms of life consumption.
requires usage tracking. Life consumption of the aircraft
For critical life limited parts, regulations require that each
structure is also often related to flight time. Therefore, it
individual part be traceable throughout its service life
may be an advantage that for both engine and aircraft
history, and that complete records of usage and supporting
monitoring the same input information can be used, thus
activities (inspections, repair) are maintained. For this
avoiding the need to capture two sets of data.
purpose, each critical part is identified by a unique serial
number. 1.1.2. GROUND RUNNING
Usage tracking would be easy if all missions flown, and The flight-time based usage tracking method does not
all the surrounding conditions were identical. In fact, this account for situations where engines undergo ground runs.
is not the case, particularly for military operations, where Ground runs may include, for example, pass-off tests,
the diversity of the mission profiles results in significant trouble-shooting and engine installation runs. In this case,
differences in life consumption. engine life is consumed, but cannot be accounted for as
Early maintenance policies were based on a 'hard-time' the flight time is zero. If engine ground runs are
philosophy, but the recent trend is more and more towards performed regularly, the cyclic exchange rates related to
an 'on-condition' philosophy. This makes it important to the flight time could be increased accordingly. Another
know what the 'condition' really is. method is to track engine life-consumption based on either
the engine operation time or the engine running time. For
There are a number of alternative strategies or methods for this, the engine running time, including all flights and
estimating life usage. The main methods used are: ground-runs, should be recorded. As in the previous
methods, a cyclic exchange rate must be established, to
• Time Based Usage Tracking;
provide the relationship between life consumption and
• Event Based Usage Tracking; engine ground-running time.
• Mission Type Based Usage Tracking;
• Squadron Based Usage Tracking; For the methods of determining cyclic exchange rates,
• Usage Tracking Based on 'real-time' measurement refer to chapter 6.
and calculation.
1.2. EVENT BASED USAGE TRACKING
As any synthesis more closely represents reality, so the The general idea behind event based usage tracking is that
results will become more accurate. damage is related to certain events. In particular, fatigue
damage is supposed to be directly correlated with stress or
1.1. TIME BASED USAGE TRACKING strain cycles. Let us assume that stress and strain cycles
The general idea behind time-based usage tracking is that are directly correlated to load cycles, (ignoring thermal
damage (life consumption) accumulates with time. Both transients and other effects) and that load cycles correlate
processes (time and damage) are irreversible, and so some directly to throttle modulation. Given this, a correlation
correlation should be possible. The simplest type of between throttle movements and life consumption might
correlation is the proportional one, saying that in a given be established. In practice, it may be more convenient and
time interval a certain fixed portion of life is consumed, accurate to measure spool speeds rather than throttle
and the relationship between time interval and life portion angle.
is constant. In this context, a throttle excursion can be considered as
an event, but in the same way, a sequence or recognisable
1.1.1. DIFFERENT TIME BASES
group of throttle movements can also be summarised as an
A calendar based tracking method assumes that the engine event. In the limit, a single flight also can be treated as an
is regularly used over the years and always with the same event. If it can be shown that the throttle movement
mission profiles and mission mixes. The consumed life is sequences are similar for all flights, it may be concluded
directly calculated from the number of years in service that life consumption is similar for all flights. If the life
multiplied by a 'cyclic exchange rate', which describes the consumption of a single flight is known (e.g. from engine
relation between consumed life and time. Once this cyclic design), then life usage tracking can be performed by
exchange rate has been established for a given engine multiplying the number-of-flights accumulated so far, by
component, life consumption can be easily tracked. The the life consumption of the single flight.
advantage of this method is that inspections and
8-4

As discussed above, for time based usage tracking one sequences can be decomposed into individual throttle
could argue that engine ground runs are not covered by excursions and an equivalent number of main excursions.
this method. They can be, however, included if instead of The corresponding life consumption per mission can then
the number of flights the number of engine runs is be determined.
counted. In this case the 'engine run' is considered to be
the governing event.
Experience has shown that flights might Tucano Information Management System
not be as similar as is presumed above. UNIT Ft Ä.«R»Se RATES DISPLAY

To account for differences between H<xr

throttle movement sequences, the whole Us«r W:

sequence can be decomposed into single SJ525

tine is Fleet Standard
throttle excursions. Generally, a flight (or
an engine run) contains one main $,6Xh
excursion from zero to max back to zero
and a number of sub-excursions (e.g. idle 0.C1S
- max - idle, idle - intermediate - idle,
max - intermediate - max). Now, one

■ ■ —■■
could simply count the number of
excursions, but this does not lead to a ■0.805 >
proper result, since excursions with
different ranges are known to cause
mwmmmmwMM
significantly different amounts of fatigue -SORTIE PROFILE CODE
damage.
The solution for this case is to weight the Figure 1 - Comparing squadron and fleet life consumption rates
excursions according to their contribution
to life consumption. Every range needs Mission-type based usage tracking means counting the
an individual weight factor. However, for practical number of each mission type, multiply them with their
applications it is more convenient to categorise the sub- respective life consumption per mission and add them up
excursions into major, minor and negligible. The main over all mission types. Figure 1 shows an example of
excursion is normally given a weight factor of one. mission based usage tracking.
Negligible excursions are zero-weighted, which means
This procedure can be easily extended to account for
that they need not be counted. Major and minor sub-
engine ground runs too. All that is necessary is to
excursions are weighted with factors between 0 and 1.
determine the life consumption of typical ground runs and
Finally, the procedure is as follows.
include them as 'special mission profiles'.
• Count the number of main throttle excursions, Mission type based usage tracking is not limited to the
including major and minor sub-excursions. event based method but can also be applied in
• Multiply the counts by their respective weighting combination with a time based one. In this case, the cyclic
factors. exchange rates need to be established separately for each
• Sum the weighted counts. mission type. This can be done either on the basis of well
defined synthetic profiles or recorded data from
This results in a number of equivalent main excursions. If representative missions.
the life consumption caused by the main excursion is
known, one can now calculate the life consumption of the One should be aware that the mission type based usage
flight with the analysed throttle sequence. tracking method is more complicated than the methods
where only times or numbers of flights need to be
Another method for determining the life consumption per recorded, since here as additional information the mission
flight is to record the necessary operating parameters of a type is required. This may cause difficulties if some
representative number of flights and calculate the typical missions cannot be allocated to the defined mission types.
values for all monitored critical areas. Scatter factors and
safety factors may be added as discussed elsewhere. This Furthermore, the logistic efforts are higher because more
method will probably be more accurate than the data must be handled and more care is required in order
decomposition of throttle sequences. not to mix it up.

1.3. MISSION-TYPE BASED USAGE TRACKING 1.4. SQUADRON BASED USAGE TRACKING
Generally, military aero engines fly a variety of different It has been established that aero engine life consumption
mission profiles (also known as mission types or sortie- varies between different squadrons or air bases, even if
patterns). The profiles are made up of a number of mission they fly the same mission types and same mission mixes.
elements such as take-off, climb, cruise high, cruise low, Influences include:
low-speed dash, and any others that may be appropriate. • Base locations relative to training areas;
Each mission profile is identified by mission codes or
• Differences in aircraft ground handling;
sortie codes. For each of the different mission profiles a
• Varying interpretations of handling instructions;
typical throttle sequence can be specified. These
 8-5

• The length and altitude of the runways. monitored is likely to be used. Such models are required
to permit rapid analysis on less powerful computers than
To account for these differences one can introduce could be used on the ground. They may also be required
different cyclic exchange rates or different figures for life for routine ground based analysis, because taking longer
usage per flight for the respective air bases. To do this than the flying time to analyse a flight is not an attractive
squadron identification must be added to the life usage proposition. If a 'reduced' or simplified model is used for
records. analysis it must be validated against the full design model,
which includes the full component and materials
As more variables such as mission types and squadron are databases.
recorded, the logistics and accounting procedures will
become more complicated. One may well question It has been suggested that life usage data may be
whether the improvement in life usage tracking accuracy transmitted to the ground, and this technique is used for
is worth all of the effort necessary to determine and trending by airlines such as Lufthansa. However, the
manage all the different rates and figures. volume of data required for life usage monitoring a fleet
of engines would be very high. If the data is to be
How cyclic exchange rates, or life usage per flight, figures analysed at a ground station the raw data must be recorded
are determined is discussed in detail in paragraph 2. and transferred from the aircraft. If the analysis is done
on-board then only the results need to be transferred on a
1.5. USAGE TRACKING BASED ON INDIVIDUAL
routine basis. Prudence would suggest that a bulk-data
COMPONENTS
storage-device should be available for trouble-shooting for
A number of problems associated with the usage tracking special investigations and for periodic re-validation of any
methods described above can be overcome if life 'reduced' models that may be in use.
consumption is calculated individually for each engine or
critical component. Because fatigue damage of a critical During the development of this report PC development
feature is related to its own stress-temperature history, has progressed rapidly and it is reported that modern
rather than what is happening in other engines, this ground stations can now analyse complete flights in about
approach is the only way of avoiding the premature 15 -20% of flight duration.
removal of components.
1.5.2. BULK DATA STORAGE
Algorithms which transform given stress-temperature When raw data is required for analysis, the required
sequences into consumed life are available elsewhere. The engine operating parameters must be recorded, and the
basic idea is to derive the stress-temperature histories at recorded data transferred to the processor system, which
the monitored areas from measured engine operating will perform the calculations. Data from one or more
parameters and convert them into life usage. This can be flights may be recorded on a single data carrier. For
done, and is, by on-board monitoring systems. This proper data evaluation, it is necessary that the recordings
approach covers all of the above differences between are complete, i.e. that they contain the whole period from
different mission profiles and squadrons, and additionally engine start to engine shutdown. If data from more than
addresses differences between the handling of engines in one engine run are stored together, the engine runs should
multi-engine aircraft, actual weather conditions, engine be clearly separated to avoid misinterpretation and faulty
airframe interactions, and pilots. results.
Of course, the accuracy of the result depends strongly on Signal selection should be specified according to the
the capabilities of the algorithms applied. They range from intended further use. As a minimum, all operating
the very simple to the very complex. Simple procedures parameters that are required in the subsequent evaluation
use equations, which calculate centrifugal stresses, process must be included. In many situations, it will be
(proportional to the square of the spool speed,) and keep beneficial if additional information is also available. When
temperatures constant. The most complex procedures take signal sources arc selected the accuracy and quality of the
into account transient metal temperature development in input signals should be considered. As with every
the components, centrifugal stresses, thermal stresses, measured quantity they deviate from the physical 'truth'.
assembly stresses, stresses due to gas pressure, and These deviations need to be considered, too. Details of
stresses induced by manoeuvre loads and other influences. these points are discussed elsewhere.
The algorithms themselves (irrespective of how
sophisticated they are) build a model, and as with every The bulk data can be evaluated with different intentions.
model they are only an approximation to reality. One intention could be the usage tracking of individual
Therefore, the accuracy of the applied algorithms must be engines. In this case, a clear tagging of the data, which at
analysed and the results taken into account when least uniquely identifies the engine to which the data
individual component life-usage-tracking methods are belong, is required.
assessed. Furthermore, it might be important to evaluate the flights
in the same sequential order as they were flown. This will
1.5.1. ON-BOARD OR GROUND-BASED?
be the case when damage increments depend on the
Individual life usage monitoring calculations may be current state of life consumption (e.g. when crack
performed either in 'real-time' on-board the aircraft or at propagation life is used). The order of the recordings
dedicated ground stations. If the analysis is done on-board should be tagged accordingly, either by a unique
then a 'reduced' model of the components being identification of the engine run number or by date and
8-6

time. feature determines the life of an engine, component, or

part. The life consumed at a critical area depends on the
If the intention is to establish cyclic exchange rates (e.g. operational usage of the individual engine. In general, life
for the use in time-based tracking methods), then it is consumption is the consequence of the variation of stress
sufficient to tag the recorded data only with the engine and temperature, and the only physically correct method
configuration (e.g. engine type and engine variant codes). for life consumption determination is the assessment of
This may be supplemented by information indicating the the stress-temperature histories at the critical areas. It is
mission types flown and the air base, if required. quite natural, that under variations in operational usage the
A pool of bulk data is very useful for the design, stress-temperature histories will also change, resulting in
improvement and verification of life monitoring changes of life consumption. However, this effect is not
algorithms. Experience shows that tests of newly proportional for all critical areas. In fact, some engine
developed or improved algorithms are much more manoeuvres selectively damage some critical areas, while
effective if recorded data are used, rather than synthetic other critical areas are virtually unaffected. For example,
data. In particular, it can be checked rigorously to ensure Quick Reaction Alert starts cause extreme life
that the control structure of the system (sensing engine consumption in bore regions due to the high thermal
start and engine shut down, switching between different stresses induced by the steep temperature gradients over
modes) works properly. The criteria for signal plausibility the disc. However, hot re-slams are typically detrimental
checks and the restoration models for substitution of faulty for rim areas due to the high thermal stresses induced by
data can be fine-tuned with real data, which is virtually the 'inverted' temperature gradients.
impossible if only synthetic data are available. These simple examples make it clear that the life limiting
area of a component does not exist, and cannot be
2. CYCLIC EXCHANGE RATE
identified from the design mission alone. Consequently,
2.1. DEFINITION AND DISCUSSION
one has to consider a number of critical areas as
candidates for becoming the life-limiting feature under
The Cyclic Exchange Rate (also called Mission Severity different patterns of operational usage. From the
Factor or simply ß-Factor) describes the relationship foregoing, it is clear that life usage monitoring should be
between life consumed (mainly fatigue life under cyclic applied to all of the potentially life limiting areas of an
loading) during operational usage and the correlated time engine part. This includes the calculation of cyclic
interval. As discussed above, the time-base can be exchange rates when these are used.
calendar time, engine flight time or engine running time.
Certain types of events (such as flights, single throttle To determine cyclic exchange rates on a sound basis, one
movements, sequence of throttle movements) are also in needs a pool of real (i.e. either directly evaluated or flight
use as a 'time-base'. For most military applications, the recorded) flight data. This pool should include a
engine flight time in hours (EFH) is used as the base, representative number of engines and a representative
whereas civil applications rely on the number of flights number of missions per engine. The spectrum of missions
(often called 'cycles'). Nevertheless, the general per engine should encompass all mission types flown in a
description applies to all types of time-base. typical mix. However, one should be aware that it is much
more important to have data available from a
The term 'cycle' takes a number of different meanings, representative number of engines than to have many data
including the following. from only a few engines. This can be concluded from the
• Flight; following consideration.
• Single throttle excursion from a certain beginning Life consumption is a cumulative process. This means that
level to a target level and back to the beginning; the consumed life at an individual critical area of a
• Sequence of throttle movements; particular part will be summed up over all flights of the
• Closed loop in the stress-strain history at a certain part's service time. Quite naturally, over this long period,
critical area; the critical areas will experience the whole variety of
• Reference cycle, meaning the most damaging loop in mission types and manoeuvres with high and low damage
the stress - strain history at a certain critical area rates, probably including the extremes. However, the
under the design reference mission. extremes will be rare, and provided the aircraft will be
operated typically, the deviations will be balanced out
In many cases, the correct meaning of the expression over the part's life. It can be concluded that the average
'cycle' can only be seen from the context. If the word value (i.e. cumulated life consumption divided by
'flight' is taken as the meaning of cycle then this also has cumulated time) will be sufficient to describe the life
many meanings including: consumption of an individual critical area of a particular
part. (Provided that the 'typical' operational usage will not
• Engine-on to engine-off, but only if the aircraft flys; change over the entire service period). This individual
• Take-off to landing; engine-specific average value can be interpreted as an
• Take-off to landing, excluding 'Touch and Go' 'individual' cyclic exchange rate.
manoeuvres;
• Whatever the operator says a flight is. In contrast, different engines will respond differently to
the same operational usage. The reasons are manifold.
The most highly stressed areas of a part are it's so-called Most important for our considerations is that these
'critical areas'. Generally, the most highly stressed area or differences do not balance out over the fleet's service
 8-7

period. This is shown in chapter 9. Thus, a more detailed • Calculate the feature specific individual cyclic
statistical treatment is necessary. The target should be to exchange rate, i.e. the average life consumption per
derive a 'worst case' or 'safe' cyclic exchange rate. unit time for each individual critical area for each
Therefore, the fleetwide statistical frequency distributions engine. To do this, sum up all life consumption
of the exchange rates of each critical area should be increments of the critical area in question and divide
established and a reasonable quantile (say 95% with 90% by the accumulated time (for the possible different
CL) determined. This quantile will be a figure somewhat time-bases refer to the discussion above). The
higher than the fleet average for each critical area. frequently used process of calculating the life
consumption per unit time and then averaging these
The factor between the 95% - quantile and the average of
figures cannot be recommended, because this process
this distribution can be considered as a dispersion factor
contradicts the physical nature of damage
for the cyclic exchange rate. Once this dispersion factor
accumulation.
has been established, a 'safe' cyclic exchange rate can be
• Establish a statistical frequency distribution of these
directly derived from the average over all individual cyclic
exchange rates. This may be of particular interest if feature-specific cyclic exchange rates for all engines.
Determine the 'safe' cyclic exchange rate as an
algorithms or operational procedures change and new
acceptable quantile (with a suitable confidence level)
cyclic exchange rates need to be predicted.
of this distribution. Repeat this step for all monitored
It might happen that the number of individually monitored areas.
aircraft is not large enough to derive the dispersion factors • If the number of engines within the investigated pool
individually for each critical area. In this case, it could be is not large enough to establish valid distribution
assumed that the dispersion factors are equal for all curves then calculate the average of individual cyclic
monitored areas. Based on this assumption, the statistical exchange rates (i.e. the average of the feature specific
distributions for all individual exchange rates can be averages). Then normalise the individual cyclic
normalised (preferably with their average values) and exchange rates with this average value.
combined to a single value. Assuming a number of 30 to • Combine the normalised figures of all monitored
50 monitored critical areas per engine and a pool of 10 areas to a common distribution curve.
individually monitored engines of a fleet, one will achieve • Determine the dispersion factor between desired
a statistical frequency distribution with 300 to 500 quantile and average of this common distribution.
samples. This will be sufficient to derive a common • Multiply the average of the individual cyclic
dispersion factor for the 95% - quantile with 90% CL. exchange rates (which can be interpreted as an
The 95% - quantile of the cyclic exchange rate distribution 'average' cyclic exchange rate) with the dispersion
determines a 'safe' cyclic exchange rate which for 95% of factor to yield the 'safe' cyclic exchange rate for each
the engines is greater than or equal to the actual individual monitored area.
life consumption, i.e. for these engine a conservative
approach. On the other hand, 5% of engines really do The desired quantiles may be determined by fitting the
consume more life than predicted by the 'safe' cyclic data to some commonly used distribution law (for
exchange rate. This results in some residual risk. This example Weibull-Distribution or Normal-Distribution).
residual risk can be quantified as the probability that life A fleetwide investigation of the RB 199 engine in the
consumed in excess of the released limit coincides with a GAF MRCA Tornado fleet has shown that the dispersion
'weak' part with an actual strength close to the -3 sigma factor for cyclic exchange rates typically takes values of
design line. One can easily see that this probability is very about 1.3.
small. Thus, the risk can be categorised as remote.
However, as a statistician once said 'Statistically speaking, 3. DATA SELECTION CRITERIA
if it can happen, it will happen. It is only a matter of time!'
3.1. PARAMETER SELECTION
2.2. PROCEDURE TO ESTABLISH CYCLIC
Life usage calculation and tracking will require a variety
EXCHANGE RATES
of input parameters. Which parameters are needed will
From the consideration above one can derive a depend strongly on the method used. If, for example, the
recommended procedure to establish 'safe' cyclic chosen usage tracking method is simply time-based, then
exchange rates, as below. only the flight time or engine running time needs to be
• Collect real (i.e. recorded) flight data from a recorded. In contrast, for highly sophisticated 'real-time'
statistically representative number of engines with a on-board monitoring systems a long list of input
sufficient number of flights per engine; parameters may be necessary. The requested data can be
• Calculate the life consumption increment of each categorised in 4 groups (the following list is considered
monitored critical area individually for each engine indicative and certainly not exhaustive).
and each flight. The calculation process should be • Engine parameters that include:
based on a thermo-mechanical model including all a) spool speeds, shaft speeds;
contributing transient temperatures and stresses. Life b) engine intake temperature;
consumption should be expressed in physical units c) compressor exit temperature;
(preferably in multiples of reference cycles per engine d) stator outlet temperature;
run). e) turbine blade temperature;
 f) exhaust gas temperature; when existing instrumentation does not cover the
g) engine intake pressure; operating envelope of the monitoring system with
h) compressor exit pressure; sufficient accuracy. However, there are examples where
i) jet pipe pressure; problems with poor quality sensor-signals have been
j) oil temperature; solved with appropriate algorithms in the monitoring
k) fuel flow; system software.
1) afterburner-select indicator;
m) pilot's lever demand. 3.2. SIGNAL QUALITY
The signal quality is mainly determined by the
• Aircraft parameters that include: characteristics of the sensor and signal acquisition chain.
a) air temperature; These characteristics include accuracy, resolution,
b) indicated airspeed; possible drift, and time delays in the response to transient
c) pressure altitude; inputs. From the monitoring point of view, the
d) angle of attack, angle of side slip; requirements for the sensor characteristic should be in
e) g-load, normal acceleration, yaw angle most cases no more stringent than from other clients. If
velocity; higher demands are considered, the necessity should be
f> weight-on-wheels indicator. carefully analysed. Typical requirements, normally
sufficient for life usage monitoring, are given in Table 1.
• Configuration data as:
Another important quantity is the signal sampling-rate
a) aircraft type, aircraft variant;
(also called update rate or iteration rate). The frequencies
b) engine type, engine variant;
need to be high enough to detect all modulations of the
c) aircraft and engine serial numbers;
signal and to avoid aliasing errors. On the other hand,
d) monitoring system hardware and software
sample frequencies at too high a rate should be avoided
versions.
because they do not provide additional or more accurate
information. The optimum sampling rate may depend on
• General information: the component that is to be monitored. Typical sampling
a) date and time; rates are shown in Table 2.
It is recognised that a strong correlation exists between Note: Aliasing error means an error introduced by a
several of the above parameters. In such cases, the
correlation can sometimes be used to derive other Monitored Component Sampling Rate
parameters thus making some direct measurements Compressor discs 1 ...2 Hz
unnecessary. A general rule for engine life monitoring Turbine discs 1 ...4 Hz
should be to use those signals already available for other Shafts 1 ... 2 Hz
purposes, e.g. for engine control, cockpit display or crash Turbine blades 2 ... 8 Hz
data recording. This rule is virtually mandatory when Table 2 - Typical Signal Sampling Rates
existing aircraft-engine configurations are to be equipped
with monitoring systems, but is well worthwhile for new sampling measurement system if the sampling rate is not
at least twice the highest frequency component of the
Signal Accuracy Resolution signal being measured. It prevents determination of the
Spool speed 0.1 % 0.05 % frequency of the signal, and can result in a measurement
Temperature IK 0.5 K offset.
Engine intake 4K 2K
Compressor exit In all cases the verification and validation processes
4K 2K
Stator outlet 4K described in chapter 7 will allow the system sensitivity to
2K
Exhaust gas 2K sampling rates, accuracy and resolution to be measured
IK
Turbine blade 4K before the first hardware systems are specified and built.
2K
Oil pressure 3.3. PLAUSIBILITY CHECKS
Engine intake 2kPa lkPa Monitoring input data are, as are all measured signals,
Compressor exit lOkPa 3kPa sensitive to disturbances. To ensure that lifing calculations
Jet pipe 5kPa 2kPa are reliably executed, a number of checks should be
Indicated airspeed 2kts lkts performed to detect possible implausibility of the input.
Pressure altitude 100 ft 50 ft Typical checks are range, rate, cross, and model checks.
G-load 0.01 K 0.005 g
Table 1 - Typical Signal Requirements Range check: for every signal, a plausible range is
established within which all measured data should fall.
developments. Experience has shown in many cases that it The plausible range is derived from physical limitations
is much easier to derive a required parameter from other (e.g. negative pressures are impossible) and operational
measured data with a suitable model algorithm than limitations (e.g. spool speeds beyond burst speed). Data
convincing customers and partners that additional sensors, outside the plausible range are declared as implausible.
and data conditioning devices are necessary. Particularly
in military applications, every piece of equipment not Rate check: for every signal, the change between two
required saves weight and cost. An exception may be successive samples is scrutinised. If this change exceeds a
plausible value, the data are declared as implausible. The
 8-9

plausible range for the data change is established based on replaced by substitute results which may be based on the
the engine and aircraft characteristics. (E.g. maximum and methods for time based usage tracking or event based
minimum changes of spool speeds are derived from usage tracking. For this purpose, it is worthwhile
engine acceleration and deceleration capabilities in the recording further engine operational data such as engine
considered aircraft-engine configuration). running time, engine flight time, number of engine runs or
engine flights. Depending on the complexity of the engine
Cross check: can be performed if for a certain signal more life usage monitoring system, substitute calculations for
than one source exists. In this case, the deviations between filling the monitoring gaps can be done automatically in
the different sources are considered. If the deviations the on-board processor or be performed in the ground
exceed a predefined range, the signal can be regarded as
station. Although these simple methods are crude, one can
implausible. However, it is also possible to employ a more
argue that the overall error in life usage tracking is
sophisticated logic to detect which of the sources produces acceptable as long as such actions are rare and remain the
the implausible input. exception. In any case, a substitute calculation - however
Model check: works in a similar way as the cross check. In crude it may be - is still better than leaving the gaps
this case the second source (or further sources) for the unfilled.
signal are model calculations which use other signals as
input. The models describe the correlation between 3.5. SELECTION OF FLIGHT POINTS
different signals. The models should be established on Techniques for engine-health, rather than life-usage
measured data, which are unambiguously plausible. monitoring, use selected flight points for their analyses.
Again, a range of acceptable deviations between the The emphasis is on stabilised flight and engine running
measured and modelled signals is to be defined. conditions, which are assumed to occur in certain phases
Deviations exceeding this range are an indicator that of the flight, and which are a prerequisite for performance
signals are implausible. If it is necessary to identify which trending analyses.
of the sources produces the implausible input, again a
more sophisticated logic is needed. For life usage analyses, these stabilised conditions are of
little interest. This is because life consumption (in
A strategy for handling implausible data must be decided particular fatigue life consumption) is mainly caused by
and built into any automated life usage monitoring changes in the engine running conditions, and those flight
equipment. All detected cases of implausibility should be points where the critical areas experience extreme stresses
reported to the user and appropriate warning messages and temperatures. For 'well defined' mission profiles - as
included in the output. they appear in commercial aviation - these phases of an
engine run are well known. They encompass:
3.4. RESTORATION OF FAULTY DATA
• Engine start;
If faulty data are detected (if data are declared as
implausible) then a process will be necessary to restore the • Engine run up for aircraft take off;
information. Depending on the extent of implausibility, a • Cruise climb;
variety of measures can be taken, ranging from simple • Holding pattern;
interpolation to sophisticated substitute calculation • Landing and thrust reverse;
techniques. • Engine shut down with the subsequent cooling phase.
If, for example, for a generally reliable signal only a few Commercial considerations about passenger comfort,
samples in the data stream are implausible and the satisfaction and repeat business ensures that all flying is
preceding and following samples plausible, then linear done as smoothly as possible. However, military mission
interpolation between the plausible values is sufficient. profiles look completely different. They are composed of
The error resulting for the life usage calculation is continuous sequences, often of relatively violent
negligible in such a case. manoeuvres. This makes it extremely difficult to predefine
Longer periods with an implausible signal or total loss of simple flight phases that produce information necessary
an input source will require substitute calculation of this and sufficient to accurately derive life consumption. One
signal. This can be done with a model using other signals should be aware that the different types of critical areas
as input. The same models as used for the model check react differently to various manoeuvres, so that flight
can be used, but consideration should be given to adding phases important for one area may be negligible for
some offset in order to push the life usage results to the another one.
conservative side. When establishing such models, the Consequently, the only technique that can be
effects of such substitutions on the life-usage results recommended for gathering flight data is continuous
should be carefully investigated. The results of this recording with a sufficiently high sampling rate. Pre-
investigation should be assessed to determine the definition of criteria to select relevant flight points is very
maximum possible errors. difficult and includes the risk of missing important
Input data may be even more disturbed. If more than one information with the possible consequence of non-
signal breaks down simultaneously, substitute calculation conservative life quotation.
could produce unacceptable errors or even might become
impossible. In such a case, life usage calculation should be 3.5.1. DATA COMPRESSION
interrupted or the lifing result declared as implausible. It must recognised that there are strong pressures to reduce
Implausible life usage results should be discarded and the amount of data to be handled and transferred. One
8-10

measure could be to reduce the signal sampling rate, but • Additional operational data;
there are limitations. Other options are data compression • Diagnostic information and warnings;
techniques. These can be categorised as 'Lossy' and 'Non- • Tagging information.
Lossy' or non-reversible and reversible. Non-lossy
techniques allow a complete restoration of the original Decisions must be made, about which tasks are to be
data during the decompression process, whereas the performed on-board and which in the ground stations, at
restoration can be incomplete when lossy techniques are an early stage of design and development. These decisions
used. In less agile aircraft, such as civil airliners some generally determine the interfaces and data transfer
trending systems record steady state information together requirements. Insufficient planning of the data transfer
with a duration field that shows the period for which the facilities can jeopardise the success of the whole system or
steady condition was maintained. - in less severe cases - cause additional cost.
Experience with standard non-lossy data compression Data transfer should be carried out with a high level of
techniques applied to recorded flight profiles of military automation. Manual support or even manual transcription
missions shows that compression factors of 10 or 12:1 are of data should be kept to a minimum. Provisions should be
typical. Higher compression rates can be achieved with made to ensure that data integrity is always maintained.
lossy techniques; but these cannot be recommended for
military mission profile data, since they bear the risk of 4.2. EQUIPMENT AND MEDIA
losing important information. Data transfer equipment is the means to perform
communication between aircraft installed parts and a
3.5.2. TAGGING OF DATA
ground station. Basic requirements are that data transfer
For life-usage monitoring, it is necessary that data are not must be quick, easy to operate, robust and user-friendly.
confused or mixed up. Thus monitoring data must be Depending on the architecture of the whole monitoring
uniquely assigned to an individual engine, engine system and the amount of data to be transmitted, the
component or engine part. To fulfil this requirement, transfer equipment can make use of removable memory
monitoring data are tagged at least with aircraft and engine devices (such as cassettes, tapes, discs and solid state
identifiers (e.g. serial numbers), date and time (or - modules) or data-bus links.
alternatively - with a unique engine related count of
engine runs). It is desirable for logistics reasons, that ground equipment
and data transfer devices are common for different types
Additional data should be added if they are demanded for of aircraft and engines. This requires a high degree of
special evaluation. These additional tagging data may flexibility in order to cover the entire range of possibly
include engine configuration, mission type and airbase different requirements and formats of the different aircraft
identifiers, as well as special comments advising about and engines. It is also difficult to achieve when
irregular conditions. procurement is on a project basis, and aircraft systems are
purchased from a single prime supplier through
4. DATA TRANSFER competition.
4.1. FUNCTIONS AND REQUIREMENTS In many cases, quick-look capabilities are required,
Engine monitoring systems are built up as distributed particularly for NO GO indications or warnings. They
systems, as shown in figure 2. They consist of components either can be realised as on-board displays or can be
located in the aircraft and at the engines, and ground provided as functions of the data transfer device. The main
equipment. Reliable and accurate communication between technical issues here are the availability of the data at a
these system parts is essential. Simple systems (which suitable point. If a rapid aircraft turn around is required,
only collect data on-board) require only unidirectional without the pilot leaving the cockpit, then it may be
data transfer, i.e. download of recorded data into the necessary to provide more than one access point for quick-
ground station for further processing. More sophisticated look displays.
applications need regular bi-directional communication.
Data transfer from ground systems to the on-board 5. RECORDS AND ACCOUNTING
components encompasses: Records and accounting - in this context - means the
activities necessary to record the life usage data of engine
• Initialisation of on-board equipment with
configuration and current usage data after new parts, engine components, modules, and complete engines
installation or repair; in a central or distributed register, and to keep these
records always up to date. Generally, the accounting
• Corrections in cases where substitute calculations
activities are ground-based tasks and are performed within
were necessary;
the ground equipment.
• Diagnostic queries;
• Updates to rules. To ensure correct accounting of lifing data, a well
organised parts management system is required. Such a
Data transfer from aircraft to the ground station consists system may range from a simple card index system to
of: complex computer-aided information management system,
which is able to communicate with airborne parts of the
• Recorded flight profiles; monitoring system directly via data transfer equipment.
• Life usage results;
The database management activities should be well
 8-11

integrated into the maintenance process and tailored to the

overall maintenance philosophy and the logistic
requirements. The following information can be provided:
• Monitoring of fleet-wide life usage (individual engine
and components, fleet average, frequency
distribution);
• Projecting of engine overhaul;
• Provisioning of spare parts and logistics support;
• Component life matching during engine builds, as
shown in figure 3;
• Correlation of maintenance history with operating
conditions (mission profiles, life consumption,
operating environment).
5.1. LOGISTIC SYSTEMS
Maintenance of any type of configuration data requires
logistics information. At the very least, this will
encompass keeping track of which engines are on which
Grourtd
aircraft. More sophisticated systems with full parts
tracking capability will require regular updating of part
numbers and serial numbers for all tracked components,
modules and engine parts.
Figure 2 - Typical data transfer methods
Updating of configuration data is necessary when:
• A new aircraft is
received or an old one . Assenfelv Actual E-u±ld
is moved to another
Assembly Actual Build
location;
• An engine is removed Parent Assembly Type PEGASUS EBGISB CHANGE Will
Top Level Sub Assembly ? Y
or replaced; Parent Assembly Part »umber WC105_
Parent Assembly Serial
• An engine is
disassembled or Sub Assemblies '
rebuilt; Assembly Type Assembly Part Jfamber Assembly Serial Position Life Left Param
• Parts of the engine G37 B938410 BD631 1 2225:15 HOURS
G38 B937205 BC860 5282 :05 HOURS
monitoring system
hardware are GALLERY 1683 KK 1 1101« 318:00 HO USLS
IGU KETG 1532 UK 1532012 631:55 B0 ULS
exchanged; KOI PEGK01109 0116 9C 318:00 nouns
• The engine monitoring K02 PSGH02109 021S9C 378:00 nouns
HD3 PBGH03109 S162C 318:00 HOURS
system software is HO« PSGK04109 04161C 378:00 HOURS
updated. l-IJiMJilL«AL,.!imi*UUl»lliU*AII,immMWI
<ÜCTl»ee>
As a minimum, manual Alt-5 EOR HELP I 19200 MSI

entry of logistics data must

Figure 3 - Component life matching during engine assembly
be possible. A more
effective means would be until the respective action is closed.
an electronic data transfer via appropriate media (e.g.
tapes, diskettes) or data links. Logistics data will be If separate systems have been established for aircraft and
required from different sources as well as transmitted to engine maintenance, then the engine monitoring ground
different destinations (e.g. engine manufacturer, overhaul station should not require duplicate entry of information.
facilities, other airbases). The engine monitoring ground station should be integrated
with other aircraft maintenance-data systems, if these
In many cases, a central database is employed. If this is already exist. Configuration data should be shared for both
the case, the ground station should be capable of both systems.
uploading and downloading information stored on the
central database. It may be additionally be required that 5.2. CONFIGURATION
certain other engine monitoring data can be transferred to
During the life of an aero engine, it will be removed from
the central database.
service several times for overhaul and repair. In the course
Engine removal and installation is logged in order to of these maintenance activities, parts are often replaced
ensure the validity of the engine configuration database. with spare parts, where the parts may be - although
Maintenance actions to remove and replace life-tracked interchangeable - of a different standard. The same is true
parts are logged, too, and it may be a requirement that the for complete modules from modular engines. In both
status of maintenance actions is tracked at a ground station cases, a high degree of interchangeability between
8-12

different standards is required. scenarios for which the aircraft were designed, and
different design philosophies. The ground storage systems
This interchangeabiliry must be reflected in a both store the data on a flight-by-flight basis.
configuration control function within the parts
management system. Its main task is to record the exact If the on-board and ground systems store the results in
configuration of engines. Less obviously, it should also different ways checks are needed. If in such a situation the
manage the configuration of the engine monitoring system on-board algorithms do not check the life usage results
itself, which also will undergo changes and improvements. prior to an update, then the necessary plausibility checks
should be performed at this point in the ground station.
Modifications lead, in many cases, to improvements of the
lifing behaviour of the affected parts. This improvement If life usage calculations performed in the ground station
means that either the cyclic life of the improved standard are based on recorded mission profiles, then the usage
is higher than of the original part, or that less life is increments are directly obtained. For storing accumulated
consumed during a given mission profile. Both these life consumption, the increments simply need to be
possibilities need to be adequately addressed in the summed up. Although storage of accumulated life with
monitoring system. The consequence is that the life usage updates overwriting the previous values uses the storage
tracking function must be informed about the physical capacity more efficiently, it seems worthwhile keeping the
engine configuration. In particular, if life consumption is life counts per certain dates, e.g. twice a year. This allows
monitored in an on-board system, the configuration for trend analysis, in particular for indications whether
information must be evaluated on-board to select correct there are changes in operation or other factors influencing
algorithms and constants. the life consumption. The diagrams in chapter 9 provide
an illustration of cumulative life usage data.
Additionally, compatibility checks are required. Checks
are required to ensure that the installed monitoring-system The only reason for having a ground station is to extract
hardware and software are able to correctly monitor the information from the database. This encompasses:
actual engine configuration. Any detected mismatch must
Calculation of cyclic exchange rates;
be flagged and corrective actions taken. Provisions are
also required to protect the life usage data against Substitution and restoration of lifing data;
incorrect updates, and to provide the data required for Trending;
substitute calculations. Conversion of cyclic data into time based data;
Reading across from monitored parts to parts not
5.3. DATABASE SYSTEMS directly monitored;
A database should be incorporated in the ground station. • Preparation of initialisation data for the on-board
This database should contain all of the detailed system.
information necessary for life usage tracking of each
engine belonging to the squadron or fleet. This Cyclic exchange rates provide the relationship between
information should include a complete description of the cyclic life consumption and a time-base, which describes
engine configuration, serial numbers of engine, modules the period within which the life is consumed. The time-
and life-limited parts as well as the life usage data for all base can be either the engine flight time or the engine
the parts. Additional information, such as accumulated running time. The cyclic exchange rate is the ratio of the
counts of engine runs and engine flights, engine running life consumed (in terms of cycles) divided by the
time and engine flight time or periods of special corresponding period (in hours). Cyclic exchange rates
conditions (e.g. afterburner or thrust reverse selected), can be provided for:
may be helpful. This may be used for further analysis, Individual engines;
trend interpretation and statistical evaluation. Certain mission profiles;
The content of the database should be regularly updated as Squadrons;
and when engine life-usage data of the aircraft become Air bases;
available. The preferred frequency is a daily update. If Fleetwide averages.
aircraft are to operating away from base then provision
must be made to collect and store data while they are Safety factors should be included for deviations and
away. When the aircraft return, the data must be correctly scatter in all cyclic exchange rates. As mentioned earlier
added into the database. fleet averages are probably not the safest fill-in factors to
Two different ways of storing life usage data in a database use.
are in use. Substitute calculations should be performed if lifing data
are corrupted or lost. It is mandatory that reliable backups
• Storage of the total number of cycles accumulated to
exist, so that data corruption or loss influences only short
this point in time;
periods. Substitution values for these periods can be
• Storage of the life usage increments for each engine-
calculated by using cyclic exchange rates and the length of
run.
each period.
The UK Harrier Engine Monitoring System (EMS) stores Trending of lifing data means both keeping track of
results on a flight-by-flight basis for up to eight flights, historical data and projecting trend lines into the future to
while the GE Tornado OLMOS system stores the anticipate the need for action. Trends are normally
cumulative total. This reflects the different operational provided for individual engines, but they can also be
 8-13

evaluated for certain mission profiles, air bases or fleet- or other ground computers using appropriate data transfer
wide. Trending data are often not required in terms of media.
cycles but in terms of time. Therefore, it is usual to A very important point - that should be given special
convert remaining life (in cycles) into remaining life in attention - is the 'human' interface of the database system.
engine hours using cyclic exchange rates. With the daily Not only the output formats but also the complete
engine usage in hours, one can forecast necessary handling should be user-friendly (i.e. easy, clear, quick,
activities for each engine or aircraft by calendar dates. robust). When planning and introducing a monitoring
Evaluation of trends produces information about changes system, one should be aware that acceptance or non-
in usage and isolated events that can be correlated with acceptance of the whole system mainly depends on the
reported problems and diagnosis. ease of handling, and on the user-friendliness of the
For engines where not all life-limited parts or critical areas 'human' interface.
are directly life monitored, it is possible to estimate the
life consumption of the parts concerned from the areas 5.4. DATABASE MAINTENANCE TECHNOLOGY
directly monitored. It is usual to apply read-across factors It is desirable that ground stations be sufficiently flexible
between the parts or areas concerned. But when doing so to handle communication with different engine monitoring
one should be aware that read-across factors may vary in systems. Database architecture and data management logic
the same manner as cyclic exchange rates, i.e. they depend should be general. Configuration management
on the individual engine, mission profile, environment and requirements for database system hardware and software
engine handling. should be established with this in mind. 'Object oriented'
concepts which tie the data format closely into the
Another ground station task is the preparation of
management code are being introduced in modern
initialisation data for the on-board part of the system.
systems, and assist the construction of such systems.
These initialisation data consist of identification and
configuration of the engine (as actually built), and the Database maintenance is necessary to ensure data
current state of life consumption for all monitored areas. accuracy and database integrity. Suitable means are trend
These data are to be transferred to the on-board system analyses and plausibility checks, recording of findings and
when a new (or repaired) engine has been installed or after corrective actions as well as regular (preferred daily)
the solution of problems with the engine monitoring backups. Plausibility checks for inputs into the database
system (e.g. data corruption or loss). should be performed in a similar manner as for on-board
inputs. One should be aware that errors could occur in all
The database system should be able to provide suitable
types of data at any point in the data path between original
outputs for maintenance and logistic purposes. Different source and the database. Particular attention should be
output formats are required for the different levels of
paid to manually entered data, which should be checked to
decision making. These are as follows. ensure conformance to the underlying data specification.
• Flight line (First line - at least a GO - NO GO Automatic data-validation checks can be built into the
decision must be possible.); 'business rules' at table level to ensure that a minimum
• Engine shop (Second line - life usage data for the level of consistency is enforced.
complete engine, modules and parts must be available The system should be flexible enough to allow for
to identify useable parts and the remaining life for changes of part related data (e.g. life limits) and the
each part. For planning of manpower and resources, introduction of new or modified (improved) design of
forecasts of engine removal and the amount of engine components and parts.
maintenance work are helpful.);
• Depot and overhaul facilities (Third line - detailed life 6. MISSION CHARACTERISATION
usage data are required, supported be parts history
information and reported problems.); Missions can be characterised in different ways. For
• Engine manufacturer (Fourth line - in addition to the engine life-usage, monitoring a description of the engine
detailed life usage data and parts histories, manoeuvres and their sequential appearance is important.
information about reliability and maintainability A number of engine manoeuvres can be directly related to
problems is important. The aims are improvements in aircraft manoeuvres. Indeed, it has been said that pilots fly
maintenance support, life extension and improved the aeroplane, and the engine goes where it is taken!
engine design.); Nevertheless, it should be noted that some aircraft
• Logistics (Statistical and individual life usage data are manoeuvres do not always require the same engine
required for deployment planning and stock keeping reaction. In the late 1970s, UK Engine Usage Monitoring
management.); System (EUMS) results showed that two pilots flying
• Engineering (Statistical and individual life usage data exactly the same route used very different amounts of
are required to evaluate existing life limits and design engine life. One used the throttle to climb over some hills,
in the event of a failure). and the other used the elevators. Typical manoeuvres
which appear in almost every mission include engine start-
Output is required in the form of formatted reports, data up, taxi, take-off, climb, cruise, approach and landing,
tabulations and plots, and may be displayed on screen or thrust reverse and engine shut-down. Where there is a
printed. Output from the ground station may also be possibility of different styles or habits developing, they
communicated to on-board parts of the monitoring system will exist, and will cause life to be consumed at different
rates. Combat and aerobatics manoeuvres are also typical
8-14

of military use and probably create even greater variations. given a flight syllabus as if they were flying the actual
The most accurate way of allowing for such variations is aircraft. Throughout the simulation, data describing
to fit a life usage monitoring system to every engine. the aircraft, engine flight data, and flight conditions
However, this is not always possible and it may be can be recorded. This approach is less costly and less
necessary to conduct mission analysis to generate typical accurate than the recording devices aboard actual
mission profiles. aircraft but it does produce greater fidelity than
simple pilot surveys. It also allows the effects of the
6.1. MISSION TREATMENT FOR ANALYTICAL expected dynamics of new aircraft and engine types to
PURPOSES be explored before they physically exist.
Mission analysis is done to predict engine life
consumption for mission profiles which are composed of 6.2. 'BUILDING BLOCK' TECHNIQUES
the above mentioned manoeuvres. Two differing aims The technique described here tries to build up a complete
exist: mission profile by combining a number of blocks, which
contain the profiles of the engine parameters
• To establish the profile of a particular mission and to corresponding to certain manoeuvres.
predict its particular life consumption;
• To create a 'typical' mission profile. In a first step, a representative number of mission profiles
should be analysed to identify both the manoeuvres that
A 'typical' mission profile may be used to predict life occur and the frequency at which they occur in each
consumption in a general way and to derive cyclic mission profile. Recordings for all identified manoeuvres
exchange rates or other life related information. should be obtained and separated into blocks. Attempts
Improvement of theoretical design missions and should then be made to determine the life usage related to
refinement of an engine design may also be possible. The particular manoeuvres or blocks, respectively. If the life
basic idea behind the creation of a typical mission is that it usage of a special type of manoeuvre is found from this
should be possible to represent real flight activity in a analysis, then an average profile of such a manoeuvre can
simplified way, with sufficient accuracy to maintain safety be created.
standards. Although the blocks are similar in the way that they
If this goal cannot be achieved in a single step, the process contain the engine parameters of a certain manoeuvre,
should be decomposed into smaller parts. The underlying they can be very different in the content of the manoeuvre.
assumption is that it should be easier to relate certain The types of manoeuvre may range from simple steady-
portions of the total life consumption to particular state conditions (e.g. cruise at some constant power
manoeuvres. setting) to highly complex sequences of throttle
movements (e.g. aerobatics).
The methods used to obtain the required input data are
various. Having created a typical profile for each manoeuvre, and
ascertained its frequency, one can construct a typical
• One method is to equip a single aircraft with a data- whole mission profile. Care should be taken to ensure that
recording device and systematically record a number the transitions from one manoeuvre to the next are
of mission profiles for mission types, which are smooth. This task should be much easier than the
supposed to represent the typical usage. manipulation of a whole mission profile. Assumptions in
• A second method (which is probably much more this process are:
representative) is to equip a number of aircraft with • That each manoeuvre is independent of those
data recording systems and operate them in the preceding and following it;
normal way. That is, not selecting them on purpose • Life usage is dependent only on the manoeuvre under
(but only randomly) for special mission types or consideration;
deployments. • Life usage is largely dependent on the transient
• A third method (if recording devices are not elements of engine usage.
available) is to ask the pilots. In this case, the pilots
are interviewed about type, extent and duration of the 6.3. TRANSLATION OF MISSION ELEMENTS
manoeuvres they fly, and typical sequences of throttle In the previous paragraph, it was assumed that recorded
settings as well as mission types and mission mixes. profiles were available for missions and manoeuvres. This
In the course of these activities a variety of is not always the case. In many situations, only a
representative missions are established which are description of the aircraft manoeuvres is given and the
simply described by pilot's lever sequences or thrust associated engine parameters must be derived, or deduced.
demand sequences. Historically, pilot interview has
not proven to be a particularly accurate method. The general process is that a profile of thrust demand is
However, it has provided insight into the qualitative calculated based on the aircraft performance. This requires
aspects of mission definition. the availability of an aircraft performance model, which
generally is developed by the aircraft manufacturer.
• A fourth method is via aircraft flight simulators. Depending on the type of manoeuvre (e.g. close
Simulators are used for training pilots and formation, tanking) a certain pattern of thrust modulation
maintaining their flight proficiency, in lieu of actual must be superimposed. In addition to the thrust demand
aircraft flight time. During this method, the pilots are
 8-15

profile, the intake conditions (e.g. described by altitude 7. CONCLUSIONS AND RECOMMENDATIONS
and airspeed) are requested, and are normally included in
the description of the aircraft manoeuvres. Based on a • Before introducing life-monitoring systems, all
model for engine intake performance, developed by the organisations should identify the problems that they
aircraft manufacturer or established in close co-operation wish to solve, and the benefits that they wish to
between aircraft and engine manufacturer, the engine obtain.
intake parameters (in terms of total temperature, total and • To maximise the benefits, the user must thoroughly
static pressure) can be deduced. prepare for a change in working practices before a
modern life usage monitoring system is brought into
In the next step, an engine performance model is service.
employed to calculate the profile of the pilot's lever • Priority must be given to the relevance of
position and the corresponding engine parameters information available at the ground station, and
necessary for life usage. robustness and simplicity of use.
Generally, the engine parameters are required as transient • Because fatigue damage of a critical area is related to
data (i.e. describing the transient response of the engine to its individual stress-temperature history, calculating
modulations of the pilot's lever). This means that the life consumption for individual engines or critical
engine performance model must be capable of providing components is the best way to avoid the premature
transient outputs. It is well recognised that accurate removal of components.
prediction of transient engine behaviour is much more • Pilot survey is the least expensive and least accurate
difficult than the prediction of a steady-state response. method of developing mission definitions.
Nevertheless, fine-tuning of the performance model Installation of in-flight data recorders in a random
should be carried out using recorded engine parameters. sample of the fleet aircraft to capture actual flight
data is more costly, but is the most accurate method.
6.4. GENERATION OF 'SYNTHETIC' MISSION
8. REFERENCES
PROFILES
With the tools and mission blocks described above, one SAE AIR 1872 - Guide to life usage monitoring and parts
can generate 'synthetic' mission profiles. These mission management for aircraft gas turbine engines
profiles can be composed as a (nearly) arbitrary sequence SAE ARP 1587 - Aircraft gas turbine engine monitoring
of all of the available blocks. Of course, some restrictions system guide
exist: every mission should begin with an engine start-up SAE AIR 4061 - Guidelines for integration of engine
sequence, followed by taxiing (of variable duration) and monitoring functions with on-board aircraft systems
take-off. In the same sense, every mission should be SAE AIR 4175 - A guide to the development of a ground
finished by an engine landing and shutdown sequence. station for engine condition monitoring to scan.
Any other technically necessary limitations in the
sequencing of manoeuvres should be observed in order to
obtain sensible profiles.
Another requirement for a synthetic mission profile should
be that it contains only plausible data. This means that
plausibility checks should always be passed. This includes
that the engine parameters do not exceed their predefined
range and that transitions between subsequent blocks are
smooth enough not to violate the predefined change rates.

6.5. SUMMARY
This chapter provides details of usage tracking, cyclic
exchange rate, data selection criteria, data transfer,
records, and accounting and mission characterisation. The
need for usage tracking is justified and the different
methods are discussed. The method of relating hourly
usage rates to 'design cycles' is described in detail. The
collection and storage of raw data in bulk form is
discussed and recommended. Input parameters are defined
in terms of sampling rate, accuracy, repeatability, and the
necessary plausibility checks are defined.
Operational and management aspects of usage monitoring
systems are described with emphasis on ease of use and
delivered value. The tasks of a ground station and its
configuration are described. Amongst these is mission
analysis, which is described in detail.
 9-1

Chapter 9
Usage Data from Operational Monitoring Systems
by
(M. Sapsard)

Page

1. Introduction 9-3

2. First Considerations 9-3

3. The Statistical Aspects of Monitoring 9-3

3.1. How Many Engines in a Fleet Should Be Monitored? 9-3
3.2. Using the Charts 9-5

4. Measured Results from Operational Aircraft 9-5

4.1. RAF Harrier 9-5
4.1.1. Visualisation of Results 9-5
4.1.2. Replacing Missing Data 9-7
4.2. GAF Tornado 9-8
4.2.1. Visualisation of Results 9-8
4.2.2. Statistical Modelling 9-8
4.2.3. Fleetwide Life Usage 9-9
4.3. FAF Mirage 2000 9-9

5. Individual Engine Monitoring 9-10

6. Summary 9-10

7. References 9-10
9-2
 9-3

may be needed.
1. INTRODUCTION • This question has yet to be answered by most
Once all of the research and development described operators and manufacturers. One practice is for the
elsewhere is finished, an aircraft, engine, and engine authorities and the operator to work in flying hours,
monitoring system will arrive in service. This equipment and for the results of the monitoring system to be
will be used in accordance with the manufacturers converted from 'Counts per flight' to 'Counts per
instructions and any flight safety requirements imposed by flying hour'. This is known as the 'Beta' factor by
the airworthiness authority. Whatever debate there may several European airforces and manufacturers. Once
have been earlier, and whatever scientific and statistical known different Beta factors are often applied to
uncertainty there may be about the following: components by airbase, engine position, engine or
aircraft modification state, or any other significant
• The materials; factor that is discovered.
• The algorithms used;
• The implementation; 3. THE STATISTICAL ASPECTS OF
• Cross correlation with the manufacturers design MONITORING
model;
When data is collected it can only have any statistical
validity if there is enough of it, and if tests on the data
the life usage counts recorded by the life usage monitors
show that it conforms well enough to one of the commonly
will provide 'REAL' life usage counts to the operator and
known distributions for the results of any further
the authorities. These numbers will be accepted 'warts and
mathematical analysis to be valid. Even if the raw data
all', and are the numbers that we discuss in this chapter.
does not conform well, it may be possible to transform it
In this chapter we discuss and show the following. into a distribution that does. An example of this is the log-
normal curve.
• The statistical basis for deciding how many aircraft,
engines or stress features to monitor. The following shows, in detail, how to calculate the
• Some real data collected from in-service aircraft answer for lifed components when their usage distribution
engines. approximates closely to a normal distribution. A graph that
• How the data has been interpreted. permits informed decisions to be made about sample size
is also derived. The worked example is followed by graphs
2. FIRST CONSIDERATIONS for lifed features which have a usage distribution that
approximates to the normal distribution.
When a working monitoring system is to be fitted to an
aircraft there are a number of questions that usually have Graphical displays allow non-statisticians to obtain an
to be answered. The following are probably the most insight into what is going on. However, it must be
frequently asked. remembered that they are only illustrations and that a
mathematical analysis and proof of conformity and hence
• Which features should be monitored?
validity is essential before statistical analysis is performed.
• How many engines on how many aircraft should be
monitored?
3.1. How MANY ENGINES IN A FLEET SHOULD
• How can we use the results?
BE MONITORED?
• How will we know what we are looking at?
• Should we continue to measure life in flying hours, It is sometimes assumed that if a fleet of several hundred
or change to using life usage counts as recorded by engines is in use then monitoring, for safety purposes at
the monitoring system? least, should be adequate if a small number are monitored.
It is well known that on production lines for mass-
The simple answers to these questions are:
• The first and most simple rule in an operational The sample and the whole population
environment, rather than a research environment, is
'Don't collect data unless you already know the
—"Sample
decisions you intend to make with it'. So if you do ■^"Whole population
not already know what you are looking for and why,
X <\ 'X* Is difference
then do not do it on operational aircraft. The only \ between means
exception may be when a trial is being deliberately
conducted. ^S
• This depends on the sample size and the variation
within it. Unfortunately, in statistical terms the size Life counts
of the fleet is a secondary factor in determining how
many aircraft or engines should carry monitoring Figure 1 - Population and sample relationship
equipment.
• This depends on the relationship between the produced car components, maybe only one in a hundred
operator and the manufacturer, the fleet size, and the components is measured. This is enough to ensure that
technical capabilities of the operator. very exacting standards are met. Unfortunately there is a
• In a deeply technical subject area, specialist help vast difference between the two situations. On the
9-4

production line most of the factors that cause variations there is some uncertainty about whether the average of the
have been identified and brought under tight control. sample is the same as that for the whole population. Figure
Whereas, the military aircraft, and more particularly, the 1 shows this difference as 'X'.
engine usage spectrum has not. In safety critical situations some allowance must be made
Let us assume that the measured parameters of each of the for the possibility that the mean of the sample is less than
above populations fit a standard distribution curve. the true whole population value. However, if the fleet is
Although the number of samples for each may be similar, large enough, as the sample size increases the smaller the
one is likely to have a very small variance or spread while uncertainty about the difference between the means will
the other has a large one. The following analysis is from become.
Sapsard MJ, Flanigan A, Pitwood IR (1996). When the sample variance and size are considered for a
Because the sample only covers part of the population fleet or population of various sizes we can construct the

Sample size for fleet of 1000

120 Acceptable error
(between sample
100 mean and
population mean)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Normalised sample variance (s/xbar)

Sample size for fleet of 100

— Acceptable error
(betw een sample
» * * « j { t I t t t ' * ' '=* mean and
population mean)

-1%
-5%

-20%

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Normalised sample variance (s/xbar)

Sample size for fleet of 10

■ Acceptable error
(between sample

^t^±i^^=t^*^^^** mean and

population mean)

-1%

- 30%

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Normalised sample variance (s/xbar)

Figure 2 - Sample size dependency on sample variance, for fleets of 1000, 100, and 10 engines
 9-5

curves in figure 2. These show what has to be done if the 3.2. USING THE CHARTS
operator wishes to ensure that the size of 'X' shown above
To decide how many engines in a fleet should be
is less than say 10% of the value of the mean.
monitored it is necessary to have enough sample data so
If the shape of the sample is close to the normal that the normalised sample variance can be determined in a
distribution, it may be assumed that the whole population statistically valid way. For guidance, typical values
has a similar distribution. We also know the whole observed on the UK Harrier and Tornado aircraft lie
population size - N. In this chart // is the mean of the between about 0.4 and 1.6 depending on the selected
whole population. From the sample, we can estimate the components. Most components have values between 0.4
range within which // falls, by using Student's t. This and 0.7.
states that: As an example, let us assume the following:
ts • We have a fleet of 100 aircraft;
JU-X + • The variance is 0.5;
Vr7 • We will accept an error between the mean of the
where t is a measure of the confidence level (a probability sample and the mean of the fleet of 10%.
between 0% and 100%) that can be given to the new From the middle chart we can see that about 72% of the
estimate of fU, and gives an estimate of the range of fleet should be monitored.
values within which JU must fall. A confidence level of
99% would necessitate a larger value of t and hence a
4. MEASURED RESULTS FROM
wider range of values within which fU could fall. This OPERATIONAL AIRCRAFT
would mean that there was only a 1% chance of // lying
4.1. RAF HARRIER
outside of the estimated range of probable values.
When the Harrier GR Mk 5 entered service in 1989 a
By considering the sampling requirements for a finite sample of Pegasus Mk 105 engine-component usage-data
population it can be shown that the sample size for a given from about the first 1700 flights of the first twelve aircraft
value of the difference between the means, as a proportion was collected. This data was collected at the rate of eight
p of the mean of the sample is given by: samples per second with real time on-board computation.
N The fleetwide values of /J for all components are shown,
in table 2, for confidence levels of 99%, 95% and 90% that
(tf-l)l^l +1 jU will lie inside the estimated range.
We now have a measure of the likelihood that fj. lies
By substitution, the graphs in figure 2 may be obtained for within the range indicated. In the worst case (largest
fleets of 10, 100, and 1000 components. range), which is also the safest assumption in terms of
The following assertions should be noted. airworthiness, the 99% confidence level should be
adopted, and fj = 5.17+/-2.82 as shown in table 1.
• The number to be sampled depends primarily on the
variance (or degree of scatter) of the sampled data When the above curves were applied to the variances
and the sample size, rather than on the fleet size. obtained from the Harrier GR Mk 5 Pegasus engine table 2
This means that the sample size required for a was obtained.
statistically valid sample could be larger than the
size of the fleet. Confidence Level % t X Range of fi
• Once the desired sample size for a particular fleet or 99 3.1 5.17 5.17+/-2.82
aircraft type has been determined, a doubling of 95 2.2 5.17 5.17+/-1.99
fleet size, with no change in the way in which the 90 1.6 5.17 5.17+/-1.63
aircraft are used, requires a 41% increase in the Table 1 - Range within which the fleet mean should lie
number of sampled aircraft.
• The charts derived here can, after appropriate tests This shows that for the high pressure turbine at least 72%
for applicability, be applied to lifed components in of the fleet should be monitored if an error between the
any mechanical or other system. sample and population means of less than 10% is to be
• The method can be modified and applied to any lifed achieved. In the case of this engine it was decided to fit a
component with a usage distribution that can be fleetwide engine monitoring system before it entered
mapped onto a normal distribution. An example is service.
the log normal distribution.
• At least 30 samples should exist for each statistical 4.1.1. VISUALISATION OF RESULTS
degree of freedom, for each permutation of variables
The figures 3 and 4 show what some of the Pegasus usage
such as squadron, operating base, sortie pattern, life-
distributions look like. The first is of high-pressure
usage affecting modification, time of year, etc.
compressor life usage across several engines. It can be
seen that two distinct distributions appear to exist. This is
because two squadrons were operated in the first year of
9-6

service. One was a

training squadron Difference
and the other an between sample 0% 5% 10% 20% 30%
operational unit. and population
The dual (twin means
peaked) Component % of Fleet to % of Fleet to % of Fleet to % of Fleet to % of Fleet to
distribution is be monitored be monitored be monitored be monitored be monitored
because the LPC 100% 93% 78% 61% 47%
aircraft were not LPS 100% 97% 89% 79% 68%
shared between LPT 100% 96% 86% 73% 61%
the squadrons, and HPC 100% 92% 74% 56% 42%
the usage was HPT 100% 91% 72% 54% 39%
different. This was CCOC 100% 97% 88% 76% 65%
clearly shown by THF 100% 94% 80% 64% 50%
the detailed Creep N/A N/A N/A N/A N/A
statistical analysis. Table 2 - RAF Pegasus engine: percentage of fleet to be monitored to limit the differences
The skew (long between the sample and fleet means to specified limits.
tail) to the right
appears to be engine in the Harrier aircraft have revealed approximately
typical of measured life usage distributions for fatigue Poisson distributions for individual engines and Normal
limited components. Typically, results for the Pegasus distributions for the whole fleet. From a general statistical

Pegasus HP Compressor Fatique Life Usage by Engine No

10 11 12 13 1* 15 te ;? 19 10 21 22 23 24
Usage Counts p«.' Fiighi

Figure 3 •••■ Pegasus hp compressor life counts

Pegasus Low Pressure Turbine Creep by A/C Taii No

QTaii 10
■Tail 9
m-!m. 3
QT»8 7
»t»s 8
£1*4 5
«T»i(«
Q1k!3
»TM;
;
STA 1 t
MKt Tft'i Mo

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24 0.26 Q.2S 0.3 0.32 0.34 0.36 0.38
Usage Counts per flight

Figure 4 - Pegasus Ip turbine blade creep life counts

 9-7

viewpoint the variation in observed

life usage counts is high. This is Cumulatwe HP Compressor Life Consumption
Pflgsiift •ngi'K an Ma"iof GR MlS
because the standard deviation is at 1 1 - -1
least 50% of the mean throughout. -»•11010 ■» 11026

To put this into context this ratio is -* 110U « 110S2

usually in the order of 20% to 30%.

It was also found that better
correlation was obtained if the data
was converted from 'counts per
flight' to 'counts per hour'.
Figure 4 shows low-pressure turbine
blade creep usage distribution
across the same group of aircraft.
This is not strictly relevant to Figure 5 - Variation between engines, in cumulative life consumption
critical parts usage, but illustrates
several points. Clearly different
rules apply here, and the normal [Cumulative HP Compressor Life Consumption |
distribution cannot be used unless
this distribution could in some way 1 1 1 1 1 1 1 1 1
»■■ T(«cw« if» uEJfl« Couna 0 *%K«a«4 ü'ff utjge Gcw
be mapped onto the normal curve. *"
«.go
In this case it is worth considering
how the Harrier aircraft is operated, 8 '"
and also looking at the cumulative *
200
usage curves for individual engines,
for a possible explanation of this
usage pattern.
The Harrier aircraft has a hovering
■ 00

a
«sä
itiStifl Hcpp
in *B
VID
&
|k#*

■□
n^tt Hunk«'
ma 130 <*D >DD <sa 100 120 j*o leo SID iaa i:a

capability, which dramatically

shortens turbine blade life if used Figure 6 - The effect of replacing lost data
extensively. All pilots must practice
flying in the hover to maintain their skills. This The difference in the rate of life consumption between the
manoeuvre is also demonstrated at air displays when the best and worst engines is about 20%. If this information is
aircraft 'bows' or 'nods' to the audience. These activities made available to the fleet manager, and included in fleet
help to produce the long tail to the right. Creep is planning considerations, it should be possible to achieve a
essentially a time and temperature phenomenon. Because number of advantages. These include the following.
the turbines are designed for acceptable life at the hot end
of the operational envelope, the cooling efficiency at Identification and rectification of engines with badly
normal operating conditions is very high and very little matched, or inefficient components.
creep damage occurs. Identification and rectification of aircraft with poor
intake efficiency.
Figure 4 shows how creep life usage can have very high
values. These very high values are all associated with air Identification of pilots, who consume more or less
show demonstrations. life than others. This may be used to advantage by
their trainers.
4.1.2. REPLACING MISSING DATA
Better selection of aircraft for tasks, which are
known to consume life more rapidly.
In figures 3 and 4, one aircraft is described as 'No Tail Overall fleet life management.
No'. This is because of inadequate labelling of collected
data. Although the data could not be applied to any In figure 6 the lower line shows the cumulative recorded
particular component, it was not discarded because it hp compressor life usage on a particular aircraft. The
added to the total knowledge about the fleet of horizontal parts of the line show where no data was
components. collected.
Figure 5 shows the cumulative life consumption for the hp The upper line shows the effect of replacing the missing
compressor for four different engines. Others have been data with the average usage for that component on that
removed for clarity. It can be seen that two lie very closely aircraft over the entire period. The effect is to account for
together, but that the other two are quite separate. This approximately 17% additional usage. The point of the
form of presentation is very useful for two reasons. observation is that although the visual difference between
• It removes the confusing effect of scatter (This will the lines may not appear to be very great, the real
be apparent from figures 7 and 8); difference in safety and operational terms could be a burst
disk.
• If two components show a distinctly separate line
the difference is probably statistically significant.
9-8

4.2. GAF TORNADO

The Tornado aircraft is operated by
four nations: Germany, Great
Britain, Italy and Saudi Arabia.
Each operating environment is
different and engines for each differ
in detail such as cooling hole design
in the turbine blades. These and
other differences lead to differences
in the life usage rates between each
group of engines. The GAF fleet is
equipped with the On-board Life
Usage Monitoring System
(OLMOS) engine monitoring
system, which has provided engine
life usage data from many
thousands of flights, J Broede and H
Pfoertner, (1997).
Figure 7 - Usage variations from GAF Tornado fleet
4.2.1. VISUALISATION OF
RESULTS

The overall results confirm the RAF Harrier experience.

Figure 7 indicates the scatter found in life usage for the
0- - A\
0.35 \
intermediate pressure compressor over 77 flights.
From this, the following have been deduced. 'to \
5 0.25
-a
\
• Clusters of similar usage values appear, and may be \
associated with a particular mission. £ 0.2 \
• Significant differences in usage exist between the ■a \
o Ma
right and left engines, due to engine handling £
0.1
procedures (namely, which engine is first switched
on and off, and the resulting warming up and o.o: - x.
cooling down periods).
2 2 4 6 B <0
• The correlation between flight time and life usage is
Life Usage per Flight
not as strong as might be expected.
• There is a sharp cut-off for low life usage values, Figure 8 - Statistical distribution of life usage
that is independent of flight time, due to the
minimum power requirement for take-off, and to the
start-up and shut-down stresses. , L

• Differences in engine response to the control system ,-r -.

o
and differences in settings (resulting possibly in =] n
£2
overshoots at the end of accelerations) have also CO
■

-,
been shown to contribute. Q
^ ■

sz
4.2.2. STATISTICAL MODELLING CD

Figure 8 shows the result of fitting a 3-parameter Weibull

-*-rff 111 rh-rn-w_.
distribution to the life usage data from figure 7. To do this
the flight time information has been discarded. Therefore, () 1 2
Relative Life Consumption
the distribution function describes the difference between
engines, but not different flights. Scatter in life Figure 9 - Relative life consumption - RB 199
consumption between different flights has already been
smoothed due to the accumulating nature of life consideration. However, to do this would cast severe
consumption. It was noted by the authors, J Broede and H doubts on the existing airworthiness of the components
Pfoertner, that the small number of high life-usage points concerned. Because air-worthiness is subject to formal
were not very well approximated by this distribution, and procedures and is therefore (hopefully) well proven, it
that this is one of the reasons for individual engine seems likely that future changes will be refinements of the
monitoring. If this is compared with the distribution for existing procedures.
the Harrier shown in figure 3 there is sufficient similarity Two examples of such improvements exist.
to suggest that these findings may be generally true.
• Within the GAF a thrust-rating system is used for
The only thing that may change this is an improvement in
take-off. This has resulted in the observation of an
the underlying algorithms, which brought new factors into
 9-9

increase in life-usage rates against time in service,

and consideration is now being given to including
the deterioration of seals and cooling air flows in 25
future life usage models.
20
• In some engines the transient thermal stresses are
now included in the dynamic models and much 15
more accurately modelled than in the past.
4.2.3. FLEETWIDE LIFE USAGE 10

The fleetwide normalised probability distribution function 5

for life usage across all RB 199 fracture critical cinCO
CL
components is shown in figure 9. This is a
Normal/Poisson distribution, with some skew to the higher 5 15 25 35 45 55 65 75 85 95
values. There is a considerable smoothing effect due to the
Life Usage (% of Released life)
varying life usage characteristics, but the lower cut-off
point of about 0.5 is clear.
Figure 10 - Life usage distribution for fleet manager
Figure 10 may be of more use to fleet managers. It shows
how the components in the fleet are consuming life.
Figure 10 shows the distribution of life usage in the fleet High Pressure Impeller - Rim
for an intermediate-pressure compressor disk. The
following observations have been made. MMU/EFH

2.00-
• The left-hand tail shows spares and components that ■
have recently been built into engines.
• The majority of parts are those which were built into ■ a
1,00
new engines - between the 30% and 80% life usage ^——^~^^. —-
bands. [
| • Squ#1 -*-Squ«2
• The right-hand tail shows components that are soon
to be withdrawn from service.
The high-end components are of greatest significance for
^ &&# &# £/ / /
safety and planning purposes. They represent fast/hot Figure 11 — HP compressor mean usage by airbase
running engines, and the repeated assignment of certain
aircraft to particularly demanding missions. Similar charts
can be derived for all components. This is often a task for
Low Pressure Turbine - Bore
the fleet management systems, and is discussed further in
chapter 8. MMU/EFH
2.00-
4.3. FAF MIRAGE 2000
The SNECMA M53P2 engine powers all Mirage 2000
aircraft. The Engine Life Monitoring System (ELMS) 1.00- " ■ "•.-"■•"---■ ■■"■

calculates and tracks the creep and fatigue damage for

about 20 critical locations. The life history is recorded and ■ Squ#1 -»-Squ>/2
tracked by component and engine module. If components
are moved to another engine, the individual component &#g#^#&#£4
data is also transferred, and stored in the ELMS. The
measured usage is therefore independent of engine Figure 12 - LP turbine bore mean usage by airbase
configuration, and characteristic of the squadron missions.
The current operational usage unit is the Engine Flying Mlx(MMU/EFH] High Pressure Impeller - Rim
/ Mln{MMU/EFH]
Hours (EFH). The damage related unit is the Mixed
5.00-
Mission Unit (MMU), which represents the damage ■ -■-"■

m- m '■
caused by a weighted mix of different flight profiles. 4,00-
a
Initially, this was thought to represent the mean flight
profile. However, it became apparent that usage scatter 3,00- / \ / ■
was important, and that no aircraft actually flew the mean
flight profile. The ratio MMU/EFH is a measure of the 2.00 ■
■
severity of the flight profile, and increases with throttle m Squ.01 -*-Squ.*2

excursions and turbine time-at-temperature.

e.* c& 4? <& <& 4" 4 4
The data in figures 11 to 14 was collected from two bases,
each with a different operational mission, over half yearly Figure 13 - HP compressor scatter by mission type
and annual periods from 1992 to 1997. For the example,
9-10

two critical features have been examined. These are the

MM(«MU/EFH) LOW Pressure Turbine - Bore
disk rim at the high-pressure compressor exit, and the bore 1 Mln(MMU(EFH|
of the low-pressure turbine. The first is subject to creep 5.00 -.
■ ■ Squ.*1 -»-Squ.#2
and fatigue, and the second to low cycle fatigue at low to
medium temperatures. Figures 11 and 12 show the long- 4,00-

term trend in the ratio MMU/EFH for each base. It can be

seen that the mean usage rate is relatively constant for 3.00-

both bases over the period. The differences are

predictable: one mission is a high-altitude fighter mission,
and the other is a low-altitude navigation based mission.
2.00-
^^^~~^~—S ■
The range of usage scatter for each mission type, over the J N# f / ^ # £ / ,# /
same intervals, is shown in figures 13 and 14 as the ratio
of maximum MMU/EFH to minimum MMU/EFH. Here it Figure 14 - LP turbine bore scatter by mission type
can be seen that the scatter varies considerably along the
time-line, from about 1.5 to 5. This result was
unexpected, and meant that the mean and range Average Exploitation of Released Life under
of usage were independent of each other, for GAF Usage Spectrum
this fleet. This means that monitoring the fleet
120%
mean is not sufficient to ensure that safety
standards are maintained. Care must be taken 100%
to ensure that the worst component remains in 0.17
a safe condition whatever this ratio may 80%
currently be. □ Uncertainly
60°.
A possible shortcoming of this technique is f) 08 ES Life Consumption
40"/ 83.00%
that the ratio of maximum MMU/EFH to
minimum MMU/EFH is more sensitive to a
20%
Iptfllj
1% change in the minimum value than it is to a 36v»
1% change in the maximum value. Although
this should be considered when using these A B
diagrams, the underlying message is clear. A - Fieetwide Monitoring
B - individual Monitoring
INDIVIDUAL ENGINE
Figure 15 - Fieetwide versus individual monitoring
MONITORING
may then lead to a decision to monitor all engines.
It has been shown elsewhere, Broede J (1995), that
• There are statistically significant variances between
individual monitoring of component life-usage can more
usage rates on nominally 'identical' engines in
than double the available life for a given part. There must
'identical' aircraft.
always be a point beyond which the marginal cost of
gaining another cycle from a part is greater than the • The variance or range of scatter varies significantly
benefit. However, it has been shown that the replacement with time.
cost may exceed the purchase price of a part by a factor of • That missing or lost data should, at least, have
10. Figure 15 indicates the additional life that may be average values substituted in their place.
attainable with a life monitoring system. This is based on • Individual engine monitoring potentially offers
GAF experience with the RBI99 engine. The uncertainty twice as much usable component life as fleet
region is a measure of the known inaccuracies of the sampling.
algorithms used. Clearly much more life may be extracted
with individual engine monitoring, than with fleet
7. REFERENCES
sampling techniques. In this example, a factor of two is Sapsard MJ, Flannigan A, Pitwood IR, 'Analysis of Usage
suggested. Data from the Pegasus MklOS in the Harrier GR Mk 5",
Aircraft Health and Usage Monitoring Systems - Selected
6. SUMMARY Papers AeroTech 95, IMechE Seminar Publication 1996.
J Broede and H Pfoertner 'OLMOS in GAF MRCA
The following have been shown. Tornado -10 Years of Experience with On-Board Usage
• How to estimate the number of engines to sample, in Monitoring' 33rd AIAA/ASME/SAE/ASEE Joint
a fleet. Propulsion Conference and Exhibit, July 6-9, 1997 /
• The number of engines to be monitored primarily Seattle, WA
depends on the sample size and sample variance, Broede, J., 'Design and Service Experience of Engine Life
rather than on the size of the fleet. Usage Monitoring Systems', 5th European Propulsion
Forum, Pisa, 1995.
• The usage variation is typically so high that most of
the fleet should be monitored to obtain a statistically
valid estimate of usage. Operational convenience
 10-1

Chapter 10
Conclusions and Recommendations

Page

1. Introduction l"-3

2. Highlights '0-3

3. Summary and Recommendations 10-3

10-2
 10-3

accumulation rates for different sortie patterns or different

1. INTRODUCTION operating environments; or they may use recordings from
Critical components, for which safe operating lives are a sample of operational sorties to estimate a safe rate.
defined, include the major rotating parts, the turbo- The 'damage tolerance' design approach can be used to
machinery disks and shafts, and structural casings achieve extended service lives of critical components,
subjected to high loads. In addition to the major typically beyond the minimum calculated life. The risk is
components, other components in the rotating assemblies, controlled by a sophisticated inspection programme.
such as spacers, cover-plates and seals, may also be However, it is not economical to apply this technique for
designated as critical. To avoid unacceptable risk of the whole component life because of the costs associated
catastrophic failure it is necessary to monitor the life with dismantling engines and carrying out the necessary
usage of critical components and retire them from service inspections.
before their allocated life has been exceeded.
The operation of most military aircraft types is highly
This report examines the methods that are currently in use variable. There is variability between mission types, and
to manage the operation of life-limited parts in gas variability between the ways that different pilots fly the
turbine engines. The purpose is to make same mission. Therefore, the only way to achieve the
recommendations on the methods which engine operators maximum life from the life-limited parts is to individually
can use to minimise the costs of this process. The report track the life of each part. This necessitates the use of
concentrates on the critical parts (disks and shafts) and
monitoring systems.
covers the needs and the various options for the
retrospective application of systems to monitor and The main obstacles to retrofitting monitoring systems to
manage the accumulation of component usage. It existing aircraft are the cost and complexity of the aircraft
considers engines designed in the period 1950 to 1990, changes, and the additional costs of the data collection
where these systems were not applied at the design stage. and management activities. These major cost elements
have to be balanced against the cost savings from
2. HIGHLIGHTS increased component lives.
There are different practices in different nations. The The best results will be obtained if the engine
discussions provided a forum that increased the manufacturer is involved in the production of the system.
understanding of all concerned, about the concepts and He must be involved in defining the algorithms used
practice elsewhere. because he has the best knowledge of the engine and the
manner in which it responds to variations in usage.
Because of the variability inherent in the operation of
most military aircraft, the maximum benefit in component Many engine designs in the 1960's and 1970's did not use
life comes from fitting monitoring devices to all aircraft, computer based finite element stress analysis that
preferably during manufacture. However, these benefits included transient thermal effects. These designs should
must be balanced against the costs of producing and be reanalysed using modem techniques, if the maximum
installing the units, as well as data management activities. benefits are to be realised. However, there arc still
significant gains to be made by using the original analysis
The relative merits of the 'safe life' and 'damage
results with a full-fleet fit of monitoring devices.
tolerance' approaches are discussed and explained.
The usage monitoring of critical components has a clear
The relative payback of fleetwide monitoring, versus no
relationship with aircraft safety. Therefore the validation
monitoring was estimated to be about 100% in realisation
process needs to be carefully thought out and the system
of potential life. should have an appropriate level of duplication. This
3. SUMMARY AND RECOMMENDATIONS should include, as a minimum, a ground system which
maintains a duplicate record of accumulated damage and
Several different approaches to the management of life which is able to provide a critical examination of damage
consumption are described and their relative merits accumulation rates.
evaluated. Configuration control is a vital element in any usage
The fundamental 'safe life' method employed with monitoring system. This applies not only to the
engines designed during the period under consideration monitoring system itself, but also to the engines being
was based on the standard measure of engine use, namely monitored and the data that comes from them.
flying hours. The procedure consisted of calculating, for
As soon as the control of the life of critical components is
each component, the minimum life capability and
separated from the lifing of the total engine (overhaul
applying a weighted average rate of damage accumulation
life), there is a significant increase in the individual
per flying hour. A life in hours, based on these two
component critical data that must be managed. This needs
values, was then released.
a suitable, high integrity, data handling process covering
There are several improvements on this basic process the transfer of data to and from each aircraft, and data
which aim to refine the damage accumulation rate management at all of the operating bases and at the
(exchange rate) used. These may use different damage logistic-control centre.
 Al-]

Appendix 1
US Military Engine Tracking and Operational Usage Methods
by
(P. Maletta - G.E. Aircraft Engines)
Al-2
 Al-3

Several US military engine applications incorporate on-board Since TAC accumulation is a strong function of the engine
engine monitoring systems to provide tracking data which is parameter monitored as well as the gates and methods that
critical to the successful Life Management of engine hardware distinguish the different types of cycles encountered, they are
in the field. These tracking systems can record various types not universally equivalent. One application's life limit quoted
of cycles based on the size of engine power excursions. Since in TACs cannot be compared with another's unless both count
many items (e.g. controls, electronic components, etc.) relate using the same methodology with the same gates. Therefore,
to time rather than cycles, the monitoring systems can also an engine's TAC accumulation rate is affected by both the
track engine operating time (EOT) and engine flight time missions flown and the counting system employed for
(EFH). For those components that are limited by time at high tracking. Furthermore, every counting system, except actual
temperature, dwell time at high temperature levels and time at on-board damage systems, must calibrate the life limits used
military (IRP) or augmentor power conditions are tracked. in the parts life tracking system to actual mission severity.
Engine maintenance plans should be based on a thorough For a given fleet of aircraft and engines, the use of TACs
understanding of the engine application's operational usage, provides a consistent measurement of tracking engine cyclic
and the appropriate limits set for each tracking parameter accumulation. It is through detailed and continual mission
based on analysis of this usage. An aircraft mounted flight assessment that these TAC accumulations arc calibrated to a
loads recorder (FLR) provides the most complete description specific engine application's life capability. The timely
of operational usage. Analysis of this data can identify analysis of engine field usage data and the subsequent
variations in base-to-base usage as well as fleet-to-fleet. determination of the resultant engine damage expressed in
The current system
of tracking engine Operational
cyclic usage in the Fteqirerrerts
USAF fleet is, for the
most part, based on
the use of Total
Accumulated Cycles Life Lirrits and Operations
Log sties
(TACs). Total Inspection SLffort Data
Accumulated Cycles FteqJrarents Analysis Cdlecticn
are simple gate based
counters used to
track individual
engine cyclic
Depot Nfeirtenance Inodert Parts l.fcarjR
accumulation. They Logistics Rspär Rarring Tracing Updates
represent a Barring Rccedres
somewhat arbitrary
unit of cyclic Mssion
damage, which Analysis
relates partial cyclic Lfedstes
accumulation to the
damage associated
with a start/stop &g're
cycle. fvörtenarte

The following Figure 1 - Operations Data Collection and Mission Analysis Updates
equation defines
TACs: TACs is central to the USAF "closed loop life" philosophy. It
assures that service limits and maintenance actions are based
TAC = LCF + FTC/4 + CIC/40 on the actual field usage of the engine fleet.
Where: The USAF accumulates data on fleet usage through a
LCF = Engine Start to IRP to Engine Stop Excursion formalised system that transfers recorded flight-loads data
FTC = Idle to IRP to Idle Excursion from the field to the engine contractor. Periodic fleet surveys
CIC = Cruise to IRP to Cruise Excursion are also conducted to obtain information from pilots. This is
IRP = Intermediate Rated Power used to complement the electronic data. The electronic data is
processed through the engine contractor's flight data analysis
Note that "cruise" is not rigorously defined and the implied system on a continuous basis. A fully integrated database
relationship of damage between the various types of cycles is allows for the rapid evaluation of mission statistics and the
not necessarily representative of any location within the comparison with the previous usage that was used to set the
engine. current life limits. A database usage report can provide
Al-4

extensive information on aircraft and engine usage, which is Once appropriate 'target values' are determined, the database
valuable to many end users. is then queried for possible candidate missions that best match
the cyclic content and time at power statistics. The candidate
The mission generation process produces a set of realistic
missions are reviewed to determine which one profile best
missions that best represent fleet usage. The goal of this effort
demonstrates the characteristics that the pilots described as
is to develop a complete definition of engine usage that can
being most typical of that sortie type. The 'Best Fit'
then serve as the basis for setting life limits. To attain this
candidates are the starting points from which the final field
goal, representative mission profiles have to be defined in
missions are derived. Damage studies show that the amount of
terms of power excursions, altitude and mach number
damage can be affected by where in the flight envelope
descriptions and combined with the appropriate mission and
engine cycles occur, and that the cause of stress varies for
ambient temperature mixes. Engine deterioration
each location throughout the engine. Hence, the goal of the
characteristics and transient
descriptions are also
considered.
E Engine Cycle Counting Description
Field usage surveys are
n LCF
required to develop an it\f
understanding of aircraft g
i Gate
and engine usage. This can
n
then be used during the
e Cruise
analysis of the electronic P Gate
flight data. Pilot interviews
a CIC
are very useful in r
understanding "typical" Idle
a Gate
flight scenarios as well as m FTC
the types of variations that e
can be expected in the t
electronic data. The field e
usage survey also provides r /
the opportunity to obtain \
information on ground runs Figure 2 - TAC cycle definitions
and testing as well as data
that may only be available
at the bases.
The basic mission
generation process, shown INPUT ANALYStS OUTPUT
in figure 3, is heavily
dependent on the
availability of Electronic
Flight Data (EFD) to -W- snö mission
ftvuttä C(C*
ifttertwwa

UMge Survey Da»

FUSHT specific
develop missions that STATISTICS «aKstto
characterise the fleet's field DATA BASE
/ lf
usage. This electronic data
is processed, plotted and Prof»« ptsWSftg
snrt Anaijsl* Software
analysed by the flight-data
analysis software. The | Pt«jw»serWat!v& Profis
resulting summary statistics ? for £äCK Wlssion Type

for each sortie are loaded Candidate iSisaKjn Pratt)»«

into a database. Statistics
collected include those
needed for mission
development as well as
those of interest to other Fkure 3 - TAC calculation
groups. After all of the
flights have been analysed and the appropriate statistics have mission generation process is to match all of the statistics for
been loaded into the database, the statistics are extracted for each mission category or sortie type. Matching statistics by
each of the sortie types that are to be modelled. Generally, mission category helps assure that the resultant model mission
statistics on pre and post flight time, EFH, EOT, IRP and A/B is representative of each category's fleet utilisation.
Time at Takeoff, Total Time at IRP and A/B, Number of A/B
In order to prepare the data needed to conduct the life
Lights, and a variety of cyclic statistics are obtained. The
analyses, the missions are smoothed to minimise the number
averages, distribution shapes and outliers are then evaluated.
of points in each sortie profile, and the cyclic content and
times are adjusted during this process. As a final check of the
 Al-5

resultant missions, their damage is compared to the damage of

the actual missions to assure that it is representative of fleet
experience. Finally, thermodynamic parameters are calculated
for each point in the model mission profile, which covers all
required deterioration levels and ambient conditions, and
detailed transient profiles are provided for each modelled
mission type.
The field limits for each component are then related to the
engine field tracking statistics to assure consistency in the
resulting field limits with the parts life tracking system data.
 A2-1

Appendix 2
Maintenance Policies and Procedures
by
(O. Davenport)

Page

1. Overview of Current Maintenance Strategies A2-3

A2 3
2. Canada - Canadian Forces "

3. France A2-4

4. Germany A2-4

A2
5. Greece - Hellenic Air Force ~^
A2 5
6. Italy - Aeronautica Militare Italiana (AM) "
A2 7
7. Netherlands "

8. Norway "

9. Portugal A2-8

10. Spain A2-9

11. Turkey A2"10

A2 12
12. UK - Royal Air Force "
A2 14
13. US Engines ~
A2 14
13.1 U.S. Air Force "
A2 14
13.2 U.S. Navy "
A2-2
 A2-3

1. OVERVIEW OF CURRENT MAINTENANCE STRATEGIES

This section provides information that is current at the publication date, and does not reflect advances that will surely
occur for the newest engines including the Fl 19-PW-100, Fl 19 derivatives, F120-GE-100 and the EJ200.

2. CANADA - CANADIAN FORCES

Canadian Forces (CF) engines are lifed based on the manufacturer's recommendations with a few exceptions. The CF
policy on equipment lifing does not specify detailed lifing approaches, but rather provides broad guidelines to the
equipment manager who is ultimately responsible for approving component lives. In general, the dynamic (rotating)
component philosophy is a safe life approach. The engine types have been divided into three main categories:
Transport/Patrol, Helicopter, and Jet.
ENGINE TYPE LIFING STRATEGY

TRANSPORT
T56 All CF T56 engine components are life limited by inspection wear limits. However, the
engine manufacturer has published LCF/Creep limits that are used by civilian operators.
(This is the main transport The USAF has indicated that it is in the process of implementing modified versions of
and patrol engine. The these lives. Therefore, the CF are re-evaluating their approach and have already
A310 and CF6-80 are implemented a 23,000-hour life on the Series II 2/3 turbine spacer based on failures and
maintained under civilian destructive component testing. This value will probably be lowered, based on further
regulation) testina and manufacturer/USAF recommendations.
HELICOPTER
PT6T All component lives are based on manufacturer safe life specifications (LCF limited).
Currently lives are tracked on operating hours, but investigation into using cycle counts
(mainly LCF) on the newer Bell 412 (CHI46) aircraft with its HUMS recording
capability, is taking place.
T58 Components are lifed, based on manufacturer safe life recommendations. All CF T58 are
in process of being upgraded to the -100 series and life limits will correspondingly
change to the new model values.
JET
J85-CAN40 Component lives were established based on the manufacturer's safe life limits and CF
Mission Severity Factors determined in consultation with the manufacturer. Recently, an
engine speed monitor has been installed on the CF Tutor aircraft. This is to more
accurately track cycle counts, and to establish more realistic conversion and seventy
factors. The CF also participates in a Component Improvement Program (CIP), where
the manufacturer provides lifing updates to the entire user community. Some damage
tolerance analysis has been completed for some components, and safe inspection
intervals determined.
NENEX The older NENE X has a combination of lifing approaches. The compressor impeller,
turbine disk, and turbine shaft are removed from service based on inspection criteria.
Additionally, the compressor impeller is safe life limited based on LCF cycles translated
to operating hours. The last complete review of the conversion was done in the early
'70s. Work is currently underway to install cycle counters to track actual usage of all
components and update conversion rates. Consideration is being given to using cycles
rather than operating hours for limits. A fracture failure analysis of the critical
components is underway to better characterise component condition against usage. Some
analysis of damage tolerance capability has been carried out for some components, and
safe inspection intervals determined.
F404 One of the newer CF engine fleets, the F404 is maintained under a modular concept with
a comprehensive recording system. This monitors and tracks individual cycles based on
manufacturer's algorithms and the manufacturer's safe life limits in cycles for LCF and
thermal cycles. A computerised configuration management system oversees all 26
critical components plus another 135 tracked components. The CF also participates in
the F404 CIP, which continually works with the manufacturer and other users to
improve lifing methods and limits.
A2-4

3. FRANCE
Engine Type Lifing strategy
JET
Larzac Component life is based on LCF cycles. Monitoring of engine life consumption is performed
through Engine Flying Hours and the declared flight profile. The conversion factor between
(Alpha jet) EFH and LCF cycles is calibrated for different flight profiles. For this purpose, some engines
were equipped with tape recorders to monitor flight parameters. The damage for each flight is
calculated using a ground-life-monitoring system. Life limited parts are scrapped once the life
limit has been reached.
Atar 9K50 Component life is based on LCF cycles. Monitoring of engine life consumption is done
through Engine Flying Hours. The conversion factor between EFH and LCF cycles is
(Mirage Fl) calibrated for different flight profiles using the crash recorder. The inspection frequency is
based on an EFH limit. Life limited parts are scrapped once the life limit has been reached.
M53-P2 Component life is based on LCF cycles. A life limit is set for each critical component. Each
engine is monitored by an on-board life monitoring system that calculates the fatigue, the creep
(Mirage 2000) damage and the residual life for about 20 critical part locations. The authorised life is released
after inspection of the fleet leaders to detect other possible damage mechanisms. The initial
life, the residua] life and the inspection frequency are expressed using the same flight damage
unit limit. The statistics of residual life and spares forecasts are established on-ground using
the data transferred from the on-board life monitoring system.

4. GERMANY
ENGINE TYPE LIFING STRATEGY
JET
J79- 17A Part's lives are based on LCF cycles. Life consumption is monitored using engine flying
hours (EFH). Cyclic exchange rates are provided by the engine manufacturer.
(Phantom) Fixed intervals (in terms of EFH) are defined for engine overhaul. During overhaul, parts are
inspected for cracks, wear, corrosion, erosion, FOD and damaged coatings. Damaged parts
are either repaired or retired. Parts with sufficient remaining life are re-used.
RB 199 Part's lives are based on LCF cycles that are established on analysis and tests predicting
minimum material properties. Lives are released in steps supported by a sampling
(Tornado) programme. For critical areas with high damage-tolerance capability, lives are extended into
the safe crack-propagation regime.
Technical Life Reviews are performed to check life assessment and assumptions for validity.
Life consumption is individually monitored with an On-board Life Usage Monitoring System
(OLMOS), using sophisticated thermo-mechanical engine models covering fatigue and creep
life consumption.
Life limited parts are inspected as they become available. Only parts with sufficient
remaining life are re-used.
RD33 Part's lives are based mainly on wear, corrosion and erosion. LCF is not life limiting. Lives
are established on analysis and accelerated mission testing. Life is released in terms of engine
(MIG 29)
running hours, where the appropriate scaling factor between test time and engine running
time was originally established by the manufacturer.
A national life-extension programme (in co-operation with the engine manufacturer) was
launched. With reduction of load and temperature, evaluation of German mission profiles
and on-condition monitoring and inspections, the released life in terms of engine running
hours could be significantly extended.
HELICOPTER
250 - C20B Part's lives are limited on either engine running hours or the number of engine starts,
whichever is reached first.
(BO 105)
Engine overhaul is performed at fixed intervals. During overhaul, parts are inspected for
cracks, wear, corrosion, erosion, FOD and damaged coatings. Parts with sufficient remaining
life are re-used.
T64-7 Part's lives are based on LCF cycles. Life consumption is monitored using engine flying
hours (EFH). Cyclic exchange rates were provided by the engine manufacturer. Based on
(CH 53)
recorded German mission profiles, updated exchange rates have been defined.
Fixed intervals (in terms of EFH) are defined for engine overhaul. During overhaul, parts are
 A2-5

inspected for cracks, wear, corrosion, erosion, FOD and damaged coatings. Damaged parts
are either repaired or retired. Parts with sufficient remaining life are re-used.
When cracks were detected at unusual locations, they were reported to the manufacturer and
led to new assessment of the parts and revision of the released lives.
T - 62T - 27 Component lives are limited by engine starts. The part's lives were originally specified by
the manufacturer, but then drastically reduced during service leading to a high number of
(APU in CH 53) life-expired parts.
In a national programme, updated life limits were established based on actual operating
conditions.
TRANSPORT
Tyne Mk 22 Part's lives are limited by LCF cycles. The manufacturer increased the initial lives, as the
result of a Life Development Programme.
(C 160 Transall) Cycles are counted as 1 cycle per landing, 1/2 cycle per touch and go, no cycles per engine
ground run. Average cyclic exchange rates have been established based on this counting
practice and German mission profiles.
Engine overhaul is performed at fixed intervals (in terms of EFH). During overhaul the parts
are inspected and - based on the findings - re-used or scrapped. Normally, engines are rebuilt
with remaining lives sufficient for the next inspection interval. In rare cases, a reduction of
the interval due to life limitations is accepted.
Tyne Mk 21 As for Tyne Mk 22, but cyclic exchange rates and inspection-intervals are different.
(Breguet Atlantique;
Sea Surveillance,
similar to transport)

5. GREECE - HELLENIC AIR FORCE

ENGINE TYPE LIFING STRATEGY

TRANSPORT
T56(C-130) These are life limited by engine operating time. Life usage counters are currently being
fitted in a number of aircraft (sample).

HELICOPTER
T-53 Life counters have been retrofitted into a number of engines and exchange rates, applicable
for the whole of the engines, are statistically calculated. Life limits are based on the
manufacturer's recommendations.
JET
J69 Life limited parts are retired on a flight time basis.
J85-4/13 The lives are based on the manufacturer's, the USN's, and USAF's recommendations. They
are in cycles. HAF mission severity factors for both the versions are statistically defined.
Counters have been retrofitted in a number (sample) of engines.
J79 Parts lives are based on the manufacturer and USAF's (given that USAF uses it)
recommendations and a conservative exchange rate is applied. Studies are currently being
undertaken to define the feasibility of retrofitting life usage counters.
TF-41 Flight hours are used to track engine-component life-consumption. The limits are based on
hours and are provided by the USN.

Fl 10 The lives are based on TAC's. The manufacturer's recommendation is used. Individual
engine tracking is adopted and some of the engine rotating-parts are retired as soon as the
life limit is reached. Others are inspected and cleared for a new period of use.

6. ITALY - AERONAUTICA MILITARE ITALIANA (AM)

In the past Turboprop, Helicopter and Jet Engine Manufacturer's recommendations were accepted, as engines were build
or assembled under a licence agreement. The Italian Air Force (IAF) Component Lifing Philosophy has evolved in the
last few years. This evolution began with the application of a sophisticated methodology for Data Recording, Monitoring
and Tracking (Maintenance Recorder). It became possible to track individual component's life consumption in cycles
and to analyse residual life. This can be performed via two methodologies, the first is the instrumentation of few aircraft
that will be used as fleet leaders, the second is the installation of a recording device on the full fleet. Specific information
A2-6

for engine type is provided in the following table.

ENGINE TYPE LIFING STRATEGY

TURBOPROP
T64 Originally, for all IAF Transport Aircraft Engines (T64, T56 & TYNE) belonging
to the old 'generation', life limited components were subject to periodic
(Aeritalia G222) inspections in term of operating hours and by inspection wear limits. This
T56 approach has been superseded by the introduction of a 'fixed' algorithm, given by
the engine manufacturers, which allows conversion of operating hours to cycles. A
(C130H) further improvement has been the 'variable' algorithm philosophy approach. This
TYNE derives a value directly from a table, originated by engine manufacturers. All
mission profiles flown in a given period of operation are recorded and analysed to
(Breguet Atlantique; obtain the Average Life Distribution Consumption Corrective Factors (Different
Sea Surveillance) factor for each spool). The resulting factors, which will be subject to further
regular updating, are used in the conversion algorithm. The IAF and Engine R&O
company are also involved in the T64 Component Improvement Program, in
which the engine manufacturer distributes lifing data updates to all operators.
The engine is maintained with an 'Overhaul at fixed intervals philosophy' (in
terms of EFH). Parts are re-used in accordance with the above-mentioned process.
HELICOPTER
Group 'A' component lives are based on engine manufacturer's safe life limits and
recommendations. Engine-component life consumption is recorded and parts are
replaced accordingly to the engine manufacturer's instructions.
T700-T6E Component lives for this engine, in the NH90 helicopter, will be set based on the
engine manufacturer's safe life specifications (LCF). Lives and cycle consumption
will be tracked via the EMA (Electronic Maintenance Assistant) that is part of the
EECU (Engine Electronic Control Unit). Maintenance actions will be driven by
the EMA.
JET
J79 For this engine, that belongs to the old 'generation', life limited components are
subject to periodic inspections in terms of operating hours. In J79 CIP the engine
manufacturer initially provided a 'fixed' algorithm that allowed the conversion of
operating hours to cycles. Subsequently, 'variable' algorithms that provided
Average Life Distribution Consumption Correction Factors were released. These
will be subject to further updating on a regular basis, and are used in the
conversion algorithm.
The engine is maintained with an 'Overhaul at fixed intervals philosophy' (in
terms of EFH). Parts are re-used in accordance with this process.
SPEY MK807 The SPEY fleet is currently maintained with an 'Overhaul at fixed intervals
philosophy' (in terms of EFH). In this programme, to calculate component LCF in
accordance with IAF Mission Severity Factors, two lead aircraft fitted with full
Flight Test Instrumentation (FTI) and On-board Data recording Unit, have flown
all Staff Requirement Missions in all foreseen profiles. The engine manufacturer
provides a comprehensive and traceable algorithm that is applied to the full fleet.
RB199Mkl01/103 RBI99 modular engine maintenance is supported by a dedicated system (MA.RE.-
Maintenance Recorder) that records engine function parameters (throttle
movement, duration, temperature, rotors speed, and thermal cycles) against all
critical plus several major components (more then 100 items). The system, which
equips the whole IAF fleet, has an On board Data Recording Unit and a Ground
Station. All recorded flight data are downloaded into the Ground Station, which
maintains the Engine configuration record and performs life consumption
calculations for the individual components installed on that aircraft/engine and
allocates subsequent cycle debits to the tracked components. The algorithms and
safe life limits contained in the ground station are based on engine manufacturer
data. The information retrieved from the system is also used in the life
improvement programmes.
PEGASUS Due to the uniqueness of this application (V/STOL A/C) and non-availability of
the on-board recording unit, the Italian Navy Pegasus fleet is operated under the
'Overhaul at fixed intervals philosophy' (in terms of EFH). The component lives
have been established in accordance with the engine manufacturer's
 A2-7

recommendations.
Parts are inspected at overhaul, and replaced or repaired and re-used if sufficient
life is still available.
EJ200 The EJ200 engine, like the RBI99 fleet, will be maintained in accordance with its
modular design. However, as the previous systems revealed operational
difficulties during extended deployment away from the Ground System, leading to
a loss of cycles, it will be supported with an enhanced system being developed and
applied to the whole EF2000 fleet. This system will elaborate and update the
recorded data in near real-time, after engine cut-off, via the On-Board Processing
Unit. This will avoid any data loss and subsequent analytical calculation in the
ground station.

7. NETHERLANDS

ENGINE TYPE LIFING STRATEGY

JET
PWA F100-200/220 Engine component lifing of this engine, installed in RNLAF F-16 fighter a/c, is
based on the manufacturer's Damage Tolerance (Retirement-for-Cause)
recommendations. Maintenance schedules and replacements are specified in terms
of factored engine cycles. A separate Load and Usage Monitoring programme is
performed. In this programme the high-pressure rotor speed N2 and the power
lever angle PLA are being monitored, combined with a variety of relevant aircraft
and flight parameters. The monitored data is transferred to the manufacturer that
can use this data to update the maintenance schedule.
HELICOPTER
Rolls-Royce Gem42 Engine hard time maintenance of the Gem42 engine, installed in the Royal
Netherlands Navy (RNLN) WHL Lynx naval helicopter is being applied on the
basis of the so-called Cyclic Life Control (CLC) concept. Since 1991, the RNLN
has applied CLC on a sample-monitoring basis, with 4 cycle counters measuring
engine usage on a rotation basis within the fleet of 22 helicopters. Recently,
RNLN has decided to switch to fleet-wide CLC. For this purpose a new fully
autonomous data-acquisition system has been developed, built, installed and
operated by the RNLN. This system combines engine usage monitoring with
structural load and usage monitoring. To date, the RNLN is the only Gem operator
to apply the Rolls-Royce designed CLC approach.

8. NORWAY

ENGINE TYPE LIFING STRATEGY

JET
J-85-GE-13 Component lives were established based on OEM, USAF safe life limits and RNoAF
Mission Severity Factor determined in consultation with the OEM/IEMP (International
Engine Management Program).
F100 The lives are based on TACs. The original engine manufacturer and USAF
recommendations are used. Individual engine tracking is adopted and some of the engines
rotating parts are retired as soon as the life limit is reached. Others are inspected and
cleared for a new period of use.
HELICOPTER
PT6T-3B All components lives are set based on OEM safe life specifications (LCF limited).
GNOME HI400 Component lives are established based on OEM recommendations. Lives are set based on
LCF cycles converted to operating hours. The last complete review of the conversion
factor was done by RNoAF in 1995 (1:1).
GEM 42 The lives of all accessories are on condition. The lives of rotating parts/components are
based on original engine manufacturer and are given in cycles or hours NRoAF use
hours.
TRANSPORT
A2-8

TURBOPROP
AIRCRAFT
RR/Allison T56 Engine Engine components are scrapped as a result of findings during scheduled inspections - no
LCF life limits are yet established. However, we are aware of the manufacturer's LCF
and creep limits published for commercial operators, and are currently evaluating the
USAF and USN approach in this matter. It will be decided whether the USAF and USN
policy should be applied, with or without adjustments to the RNoAF fleet.
RNoAF operates this engine in both the C-130 and P-3 aircraft, and has established a
overhaul interval of 5500 flight hr and 6000 flight hr respectively for these engine
installations.
P&WC PT6A-20 All components are lifed based on the OEM recommendations. For rotating parts, both
operating hours and cycles are monitored.
TURBOFAN
AIRCRAFT
AlliedSignalTFE-731 RNoAF operates three Dassault Falcon DA-20 aircraft, which have recently been re-
engined with the TFE 731-5BR.
All components are lifed based on the OEM recommendations. Engine performance and
health are being monitored through a 'jetcare' program, where essential engine
parameters are being recorded and analysed.

9. PORTUGAL
The Portuguese Air Force (PoAF) follows the recommendations of the engine manufacturer to monitor the aging of their
engines. For some, the PoAF has developed a 'health monitoring system' in co-operation with technical laboratories. The
'maintenance concept' is based on the manufacturer's recommendations and is supported by a set of adequate technical
publications.
ENGINE TYPE LIFING STRATEGY
JET
LARZAC LARZAC is a modular engine. The module lives are based on OEM (Turbomeca) safe
(Alpha Jet) life limits and are function of the mission profiles. The control is done by cycles and by
flight hours. For these particular engines PoAF has, in parallel, a health monitoring
system in co-operation with a technical laboratory. The results are accurate.
TF30 TF30 components lives were established based on OEM safe life limits. The PoAF has
(A-7P-Corsair) developed a program to track the rotating components and performs the rework of the HP
turbine at 900 hours and the high pressure compressor and turbine at 1500 hours.
F100 F100 is a modular engine. The engine condition is monitored by the 'Engine Monitoring
(F16) System' (EMS) which integrates an 'Advanced Compact Engine Tracking System'
(ACETS).
TFE 731 TFE 731 component lives are established based on OEM recommendations. The
(Falcon 50) condition is controlled by an 'Engine Tracking System.' The maintenance concept is in
accordance with civilian regulations.
TURBO-PROP
T56 All the T56 engines are life limited by inspection wear limits. For some, specific
(C130-Hercules) components are undertaken studies in order to determine if it is possible to rework them
(P3-Orion) to continue in service.
TPE331-251C PoAF follows the OEM recommendations and cycles and flight hours control the parts.
(Casa212) The hot section has a hot-end life inspection (HIS) that may lead to an extensive or
reduction of the OVHL time.
HELICOPTER
ARTOUSTE For this old engine, the components are removed from service based on inspection
(AL Ill-helicopter) criteria. No rework is performed.
MAKILA MAKILA is a modular engine. The modules lives are established based on OEM
(SA 330-PUMA (Turbomeca) safe life limits.
Helicopter)
 A2-9
10. SPAIN

ENGINE TYPE LIFING STRATEGY

JET
J-79 Maintenance according to USAF T.O. 2J79-83 and T.O. 2J79-86 manuals and
updates proposed by the Technical Integrated Group for Engineering and
Reporting Support (TIGERS) from General Electric.
There is a Component Replacement Program recommended by TIGERS group,
which consists of the systematic replacement through overhaul of a set of 23
elements. From all these elements, the replacement of the Outer Combustion
Cases (OCCs) is actually being performed, as it has been considered a very
critical element, due to the loss of some aircraft in other countries. In regard
with the rest of the elements, the critical elements will be replaced for Flight
Safety.

F404 The GE F404 is a modular engine, with fully interchangeable components, with
an 'on condition' maintenance philosophy. This is controlled solely by the
accrued fatigue induced by thermal stress, pressure, revolutions and operating
hours on those parts that are limited by Low Cycle Fatigue. The Lifing team in
GE has established fatigue limits for the parts. Each component is assigned a
fatigue limit (LCF), according to the its' failure probability and criticality.
Based on the studies of the manufacturer (GE), through finite element analysis,
endurance test, etc., the failure manner, appearance and propagation rate of the
crack and consequences of failure have been determined. With this information
all critical elements have their lives tracked (accrued fatigue). An automatic
system records the parameters causing the stress for each flight. These are
processed through an ADP (Automatic Data Processing) and accrued life
counts are assigned to each component. The algorithm is provided by GE. It
uses the average mission mix of each fleet.
SAF, currently, is in the process of applying a specific algorithm for each type
of mission and part, trying to optimise the life calculations. SAF has
accomplished simulations with this new algorithm that demonstrate in the case
of the Spanish fleet an average saving of 7% in the life consumed with respect
to the system currently used by other fleets.

F404 continued The life limits are checked periodically and revised in the light of experience,
removed parts analysis, new engineering studies, pressure failure studies,
specific trials, and through programs of Life Validation. Within the life
validation program are found the 'Fleet Leader', the engine analytical condition
inspection (ACI), the improvement components program (CIP), etc. Due to the
disparity in the life limits of each part, numerous engine disassemblies are
produced during the engine life.
SAF based on economic and operational concepts, has established the concept
of 'Window' grouping within a margin of LCF lives, to avoid unnecessary dis-
assemblies and to maximise engine availability at the lowest possible cost. This
system compels us to have sophisticated integrated logistical support, to adapt
the system to the continuous life limit modifications. The support system is a
function of the operative needs of the fleet, availability of human resources,
parts, facilities, economic, statistic, capacity of the difference maintenance
levels, industry response time.
SAF accomplishes simulations that help achieve the designed operational
objectives, in the most efficient way. These simulations are accomplished by
considering the future evolution of the fleet. This allows planning of the
necessary resources, and the prediction and avoidance of problems.
Turbo-Prop
T-56 I. Maintenance program forT.56-A-l4
Maintenance is 'ON CONDIITON' according to NAVAIR 01-75PAA-, 02B-
5DD manuals and OPNAVINST 4790.2 general instructions.
PERFORMANCES NA01-75PAA-6 is basically as follows: overhaul of
injectors every 4,800 hours; replacement of the turbine rotor at 7,500 hours
A2-I0

(35,000 hours for some P/N); withdrawal of the turbine stator at 35,000 hours
(7,500 hours for some P/N).
Engine inspection periods are:
28 days - special inspection of aircraft and engine.
112 days - special inspection of aircraft: Basically engine corrosion
224 days - special inspection of aircraft: Engine corrosion, lubrication, controls
and cleaning
150 hours - special inspection: Engine efficiency;
300 hours phased inspection of aircraft: Deep inspection of engines;
conditional inspections, etc.
Maintenance Program for T.56.-A-15
Maintenance according to USAF T.O. 2J-T-56-53 and T.O. 2J-T-56-56
manuals and O.T.E.-2J-T-.56-2 and O.T.E.2J-T-56-3 manufacturer (Allison)
manuals.
J85-GE13 Maintenance for J85-GE13 engines, which constitute the Air Force inventory,
complies with XX2F-5A-6WC-4 work cards and procedures issued by the
USAF.
These work cards specify a maintenance period of 300 hours, so every time the
engine accumulates this number of operating hours, it must go through the
workshops. Several maintenance activities will take place, including overhaul,
repairing or replacing different parts of the whole. The depth of inspection or
revision work increases such that the engine dismount level progressively
increases, until it covers a period of 2400 hours, when the engine is completely
dismantled. After this, a new maintenance cycle begins.
Most engine elements are 'on condition'. Depending on their condition when
dismantled during programmed inspections they will be replaced or repaired, in
conformity with limits specified by the applicable Technical Order issued by
the USAF. Certain 'fungible' elements as well as engine ignition plugs are
replaced when inspections are carried out.
Some of the components have an LCF limited operating life. Replacements
have been phased into the scheduled maintenance plan, where possible. Only
components for the impellers and turbine rotor (discs and spacers) are
excluded.
The severity of aircraft missions is recorded via operating hours counters.
Engine operating hours can be converted in fatigue cycles. Replacement of
parts that exceed the operating life periods is done at the programmed
inspections closest to the expiration time.

11. TURKEY
Turkish Air Force engines are lifed in accordance with the original engine designer and manufacturers'
recommendations. The dynamic component philosophy is mainly a safe life approach. There are some exceptions.
The Turkish Air Force manages aircraft engines through a weapons system approach. In this management style, the
policy does not specify detailed lifing approaches, but rather provides broad guidelines to the weapon system (engine)
manager who is ultimately responsible for approving component lives.
Specific information for each engine type is provided in the following table. The engine types have been divided into
four main categories: Turbofan, Turbojet, Turboprop, Turboshaft, and Auxiliary Power Units.
ENGINE TYPE LIFING STRATEGY
TURBOFAN ENGINES
F110-GE-100/129 Employed maintenance policy is ENSIP. Engine parts related to the ENSIP interval
arc inspected and cleared for a new period of use. Some of the rotating parts are
retired as soon as the life limit is reached.
Parts' lives are based on TAC, which is a combined measure of partial and full
LCF cycles accumulated by engine. For parts lifing, original equipment
manufacturer's (OEM) recommendations are used.
CFM56 Current lifing policy is strictly in line with OEM's publications.
TURBOJET ENGINES
J69 A study is being carried out by TuAF to define a new maintenance concept and
critical life limits according to inspection results of previous years.
 A2-11

Fixed intervals (1000 EFH) are defined for engine overhaul. During overhaul, parts
are inspected for FOD, wear, damaged coatings, erosion and corrosion. Damaged
parts are either replaced or repaired.
Used life of the life-limited parts is tracked by the engine flight hour.
J79 Fixed intervals (1200 EFH) are defined for engine overhaul. USAF Technical
Order uses number of sorties for overhaul period. A correlation study was
accomplished to define the engine flying hours limit. During overhaul, parts are
inspected for FOD, wear, damaged coatings, erosion and corrosion. Damaged parts
are either replaced or repaired.
Both USAF and OEM's recommendations are taken into consideration for the life
limit of the engine life limited parts. But; limits are subject to change based on NDI
results of overhaul.
TuAF also participates in the J79 TIGERS (Technical Integration Group for
Engineering and Repair Studies) which continually works with the manufacturer
(General Electric Aircraft Engine) and other users to provide lifing updates,
improve limits and share best practices.
J85 Fixed intervals (in terms of EFH) are defined for engine overhaul. TuAF defined
the interval and scope of the overhaul, in accordance with experience. This
maintenance concept is called 'Extended Depot Level Maintenance' in which base
level responsibilities are minimised and modular maintainability is also included.
During overhaul parts are inspected for FOD, wear damaged coatings, erosion and
corrosion. Damaged parts arc either replaced or repaired.
Part lives are based on USAF recommendation. Some counters (Engine Life
Monitoring, ELM) have been retrofitted in a number of engines to be able to
calculate an d compare the mission severity factors of different wings and bases.
Life limited parts are tracked by EFH and limits are totally defined by past NDI
history. OEM support is also used for the analysis of NDI results and risk
assessment.
TURBOPROP
T56 Engine overhaul is performed at fixed intervals (in terms of EFH). During overhaul
the parts are inspected for FOD, wear, damaged coatings, erosion and corrosion.
Based on findings, damaged parts are repaired for re-using or replaced. Part lives
are based on inspection wear limits, which are recommended by OEM.
TuAF also participates in the T56 CIP which continually works with the
manufacturer (Allison Engine Company) and other users to provide lifing updates,
improve limits and share best practices.
Tyne Mk 22 Engine overhaul is performed at fixed intervals (in terms of EFH). Although the
OEM's recommended life limits are based on cycle records, TuAF has defined the
conversion rate of EFH through a statistical study. During overhaul the parts are
inspected for FOD, wear, damaged coatings, erosion and corrosion. Based on
findings, damaged parts are repaired for re-using or replaced.
Parts have cyclic consumption are defined by the OEM and called Group 'A'
components. Group 'A' component' lives are based on OEM's safe life limits and
recommendations. Engine component life-consumption is recorded and parts are
replaced according to the OEM's instruction.
TuAF also participates in the technical support program with MTU-Germany and
other users to provide lifing updates, improve limits.

CT& An on-condition modular-engine-maintenance concept is used.

Part's lives for this engine are set based on engine manufacturer safe-life
specifications. Cycle consumption is tracking via electronic engine control unit.
TURBOSHAFT ENGINES
T700 An on-condition modular-engine-maintenance concept is used.
Part's lives for this engine are set based on engine manufacturer safe-life
specifications. Cycle consumption is tracked via electronic engine control unit.
A2-12

12. UK-ROYAL AIR FORCE

The authorised life for a RAF aero-engine type or mark, its modules and accessories is decided and controlled by the
appropriate Engineering Authority (EA) with advice from the manufacturers and the MOD(PE), and published through
the MOD(PE) Local Technical Committee (LTC) at the manufacturers works. Other than for ancillaries, these lives are
published in AP100E-01B. Lives for ancillaries are published in the Master Maintenance List (MML) for the aircraft. A
variety of strategies has been adopted to produce authorised lives, the most common being a safe life determined from
fatigue-based criteria, although the damage tolerant approach has been applied in a few cases.

ENGINE TYPE LIFING STRATEGY

JET
RBl99(AllMks) The RBI99 is a modular engine with 22 Group A components, which are
individually lifed in cycles based on the original manufacturers algorithms, and
(Tornado - all Mks) safe life limits in accordance with Tri-National Document TU346. These limits
can either increase or decrease depending on a series of on-going spin tests to
prove the LCF characteristics of each component. The rate of cyclic usage is
calculated for each engine Mk from a small sample of calibrated engines flown
through various sortie profiles.
The life of each Group A component is tracked with a computerised
management system that is capable of recalculating total life based on previous
usage. Additionally, specific component and fleet risk management is used as a
short term management tool for damage-tolerant fracture critical Group A
components. This allows component life to progress into the crack propagation
phase to permit identified items to exceed their release to service life to
minimise the effects of component life reductions.
AdourMkl5101 All Group A components are hard lifed in cycles in accordance with the Life
Management Plan. However, whereas several of the Group A components are
(Hawk - all Mks) subjected to some form of periodic inspection during their time in service,
(The RAF's Advanced others serve out their 'one shot' life without being inspected.
Flying/Tactical Training a/c)
AdourMkl5102 As per the Mkl5101, however, the Group A component lives and exchange
rates have been factored to compensate for the demanding aerobatic role of the
(Hawk for the RAF's Aerobatic aircraft.
Team, The Red Arrows)
AdourMk 10401 AspertheMkl5101, however, owing to the role of the aircraft and the fact that
the engine uses reheat (after-burner), the Group A component lives are subject
(Jaguar GR1, GR1A, T2 - low- to factored exchange rates. Further, given the nature of some GR1/1A out-of-
level, ground attack and area operations, the exchange rates are again factored and additional 'hot
reconnaissance a/c) section' inspections are carried out.
Pegasus The Pegasus has 36 Group A components which have individual lives based on
the original manufacturer's algorithms and safe life limits. Due to the specialist
(Harrier) nature of the engine, the authorised component lives are expressed in a
combination of fatigue counts, creep counts and thermal fatigue counts.
SURVEILLANCE
CFM56-2 Engine life is calculated in flight cycles linked to Group A cyclic lives;
although current lifing policy is strictly in line with CFM publications, the
(Sentry AEWMkl a/c) Company feels that the RAF are over recording and, therefore, wasting engine
life. Discussions are on-going with the Company to recover previously over-
recorded engine cyclic life. It is hoped that the results of a manual recording
exercise will allow the Company to provide the RAF with a reduced cyclic
recording penalty.
SpeyMk250/251 The engine/module has an overhaul life stated by the engine manufacturer.
Furthermore, all Group A component lives are based on the manufacturer's
(Nimrod) safe life specifications (LCF limited). Currently, life is tracked in operating
hours and factored via a cyclic exchange rate into usage cycles. There is
currently an on-going program to prove/revise the existing cyclic exchange
rate.
Avon The Avon is non-modular. All engine and accessory lives are set by Rolls-
Royce at a Local Technical Committee (LTC), accepted by the RAF and
(Canberra) published in the LTC Register and AP100E-01, although RAF policy is
 A2-13

ultimately decided by the EA. Engines are lifed by flying hours, with a
calendar back-stop. Internal engine components are separately lifed by flying
hours. Remaining components are 'on-condition'. An exercise to monitor the
cyclic exchange rates is in progress to confirm that they are correct.
TRANSPORT
ALF502R-5 ALF502 Group A life limits, based on cyclic consumption, are currently being
reviewed by DERA and Allied-Signal. The manufacturer's published cyclic
(BAe 146) lives are under discussion. The engine is operated under 'Power-by-the-hour'
arrangements to an on-condition policy.
Conway Mk301 The engine has an overhaul life stated by manufacturer. Furthermore, all Group
A component lives are based on the manufacturer's safe life specifications
(VClOallMks) (LCF limited). Currently, life is tracked in operating hours and factored via a
cyclic exchange rate into usage cycles. There is currently an ongoing program
to prove/revise the existing cyclic exchange rate.
RB211-524B4 The RB211 engine is ultimately controlled by Group A life. Component lives
are calculated by the manufacturer and published in the lifing manual. In-
(Tristar) service results are used by the manufacturer to reassess the cyclic lives.
T-56-A-15 Engine and accessory life policy is recommended by Huntings Airmotive, after
consultation with Allison Engine Company, at an LTC meeting and published
(C-130 Hercules) in the LTC Register. The EA then decides RAF policy, which is published in
AP100E-01. All engines and accessories are lifed in flying hours, with a
calendar back-stop for certain items. The remaining components are maintained
'on-condition'.
TFE731-3R Engine life is ultimately controlled by Group A life. Engine component cyclic
life is published by the manufacturer and monitored by the operator FRA
(BAe 125) Serco. The engine is operated under 'Power-by-the-hour' arrangements.
TRAINER
TPE331-12B Overhaul life is as recommended by the manufacturer and based on operating
hours. There is also an inspection of the 'hot section' at half life.
(Tucano)
Viper Mk301 Overhaul life is as recommended by the manufacturer, following life
assessment, and is based on operating hours.
(Dominie)
Allison 250-B17C Overhaul life is as recommended by the manufacturer and based on operating
(Islander) hours and starts. There is also a part life rework required at half life.
(Augusta 109)
HELICOPTER
T55-L-712F/714A Engine component lifing is based on the manufacturer's safe-life
recommendations. At present the T55-L-712F component lives are under
(Chinook) investigation by both the manufacturer and DERA Pyestock. DERA component
life recommendations, which are considered very conservative, have been
promulgated. Information gathered from the engine management system
(FADEC) has been used to support the DERA findings and an internal
component engine life assessment, the results of which are expected in January
1998.
Turmo III C4 Authorised engine life is as recommended by the manufacturer and based on
operating hours.
(Puma)
Gem 10001 Overhaul life is as recommended by the manufacturer and based on operating
hours.
(Lynx)
Gnome 122/3/4/5 Overhaul life is as recommended by the manufacturer, based on operating
hours and subject to a part life rework at half life.
(Sea King)
Artouste Mk2C Overhaul life is as recommended by the manufacturer, based on operating
(Alouette) hours.
Astazou (all Mks) Overhaul life is as recommended by the manufacturer, based on operating
(Gazelle) hours
(Jetstream)
A2-14

13. US ENGINES
13.1. U.S. AIR FORCE
The USAF has been employing ENSIP maintenance policies since 1979 on the F100-PW-100/200 engines. This is a
classical damage tolerance approach to life management. Critical rotating components, static structure, and pressure
vessels are retired at the minimum predicted LCF life. Present exceptions are some components of the F100-PW-100/200
engines, which make full use of the damage tolerance philosophy by employing 'retirement for cause'. All critical
hardware are inspected with eddy current, ultrasonic, or fluorescent penetrant inspection when they are procured from
the OEM and then re-inspected at pre-determined intervals set by damage tolerance analysis.
All older engines, which entered service prior to the establishment of ENSIP policies, are managed similarly to the Navy.
When problems such as LCF cracking, occur with these engines damage tolerance assessments are often used to
establish a risk based life management approach. This is to ensure supportability and fleet readiness while maintaining an
acceptable level of risk.
The USAF also procures transport and trainer aircraft that are FAA certified. For older engines, these are maintained just
as their civil counterparts. However, newly developed engines require damage tolerance analyses and a damage tolerant
life management approach for setting inspections prior to the FAA certified 'hard time' life limit.

13.2. U.S. NAVY

The US Navy maintenance strategy for critical rotating and pressure vessel components is to retire these parts at the
minimum predicted safe-life (LCF) limit. Parts are inspected for manufacturing defects, with fluorescent penetrant (FPI),
when they are purchased. Life limited parts are inspected during operational service when they are fully exposed at the
depot for other maintenance purposes. This is known as an opportunistic inspection and is conducted using FPI methods.
One part may be inspected many times within its service life while an identical part is never inspected. Whenever an
LCF crack is found, the part is immediately retired from service and forwarded to the manufacturer for further
investigation. A complete review of the lifing methodology, analysis and empirical data is performed to verify or
validate the existing life limit.
All engine LCF life limited parts are managed to a hard time limit. This is true for turbo-prop, turbo-shaft, turbo-jet and
turbo-fan engines. The majority of mature navy engine LCF limited parts are tracked by engine flight hours. Only the
most recently designed engines are fitted with high fidelity engine monitoring systems to track mechanical speed cycles
and other life limiting events.
Since 1993, the Navy life management organisation has initiated programs with most engine manufacturers and engine
models to update existing life limits. The updates involve:
• Lifing to the flight profiles recorded during missions;
• Enhancing thermal and stress models using state-of-the-art analysis tools;
• Incorporating fully characterised materials curves;
• Substantiating model inputs using instrumented engine test results.

ENGINE TYPE LIFING STRATEGY

Fighter
F402 The F402 turbofan has 36 flight critical (group A) parts which are individually lifed in
cycles based on the original manufacturers design and a composite mission representing
Harrier operations. There is an onboard cycle counter that records eight parameters, yet the
components are retired on flight hour limits. Twenty-four aircraft currently flying the -408
model are fitted with instrumented flight recorders to define the current mission profile.
When the new mission is defined, component exchange rates based on thermal transient
analysis will produce new life limits. A revised field management plan will be developed to
adjust spares needed to maintain the Harrier readiness rates. Additionally, fleet risk
management is used for damage tolerant parts allowing use of fracture mechanics life. The
use of propagation life minimises the impact of parts life reductions when usage values
increase and spares are inadequate. The F402 is managed with a Hot Section Inspection
interval.
F404 The F404 engine has 25 LCF flight critical parts that are tracked and managed to a cyclic
limit. The In-flight Engine Condition Monitoring system (IECMS) records the complete
mission as well as temperature, speed, and pressure counts. The Equivalent Low Cycle
Fatigue (ELCF) of each rotating part is tracked as a function of engine parameters. A
comprehensive Life Management Master Plan has been established to update life
management activities for both F404-400 and -402 engines over the last five years. Mission
updates have been completed in 1992 and 1997 with a complete reanalysis of part lives
 A2-15

between 1993 and 1997. The F404 is a modular engine that is maintained under the RCM
philosophy.
F405 The F405 turbofan has 10 flight critical (Group A) parts, which are individually lifed and
managed in cycles based on the original manufacturer's design and a composite mission
representing the T45 operations. These limits can be increased or decreased when new spin
pit or material test data becomes available. An update of the mission profile is currently
underway using improved Automated Data Recording (ADR) systems installed on 7 fleet
aircraft. When the new mission is defined, component exchange rates based on improved
thermal transient analyses will be upgraded. A comprehensive Life Management Master
Plan (LMMP) is being prepared that will detail the lifing methodology specific for this
engine. All aircraft are outfitted with an ADRS which records eight parameters used for
tracking life consumption. The engine is maintained under the RCM philosophy.
F414 The F414 turbofan has 20 critical Low Cycle Fatigue (LCF) and damage tolerant parts. The
design missions consist of 14 flight profiles, 2 ambient temperature conditions and 2 levels
of engine performance deterioration. Life analyses were performed for each of these
combinations of conditions and combined by weighted-average to obtain overall life
predictions. Current strategy upon completion of E&MD is to fly for the cyclic equivalent
of 4000 hours and introduce eddy current inspection at that point. Eddy current inspections
will be performed at fixed intervals until the LCF life limit is reached. A full mission
analysis update is scheduled for FY04. The F414 is a modular engine that is maintained
under the RCM philosophy.
FI10 The Fl 10 augmented turbofan has 22 flight critical parts, that are individually lifed and
tracked using the USAF TAC methodology. Mission profiles were updated in the early
1990's via instrumented flight recorders. All components life limits were reanalysed
between 1993 and 1997 using enhanced finite element modelling and heat transfer analyses.
Another mission analysis is in progress to assess life impacts from increased mission roles
and requirements. The F110 is a modular engine that is maintained under the RCM
philosophy.
TF34 The TF34 turbofan has 32 flight critical parts that arc managed to a 3 sigma standard
deviation life limit in hours. All critical parts are undergoing reanalysis with mission
updates and enhanced analytical modelling. The life limit updates were initiated in 1994.
The missions were updated in 1995 using in-flight data recorders and will be reviewed
again in the year 2000. The engine is managed under the RCM philosophy with a hot
section inspection interval.
J85 The J85 turbojet engine has two versions flying in the US Navy fleet. A nine stage
compressor in the -21 model and an eight stage compressor in the -4 model. The first is used
in F-5, and the later is used in the T-2 primary jet-trainer. All critical parts are lifed in cycles
and managed to flight hour limits. A life management plan has not been implemented for at
least 5 years, and mission analysis has not been updated within 10 years. Current life
reductions are proceeding and fleet logistics problems are anticipated. Evaluations of longer
aircraft use are currently being considered. A Life Management Master Plan will be
forthcoming.
Helicopter
T58 The T58 turbo-shaft engine has two versions in the US Navy fleet. The -402 model is used
in the H-3 and H-46 and the -16 model is used in the H-46E. The Navy is currently updating
part lives. The analysis should be complete by the end of CY98. The -16 task was started in
FY98 and should be competed by FY00. Part lives are based on flight hours. The engine is
maintained under the hot section inspection and overhaul philosophy. Mission updates are
being planned using in-flight data recorders. A field management plan for parts that have a
severe reduction in LCF life will be developed. Propagation life will be used to offset these
reductions with careful attention to fleet risk and flight safety.
T700 T700 turbo-shaft engine has 17 flight critical parts lifed in cycles based on the original
manufacturer's design and H-60 specification mission profiles. All engines were outfitted
with engine history recorders at production that track four parameters (LCF1, LCF2 and
Time at Temperature Index and Engine Operating Time) even though life limits are
maintained in flight hours. Analytical life updates were begun in 1993 and will be
completed in the next five three years. Mission profiles are being updated using in-flight
data recorders on fleet aircraft. The T700 is a modular engine maintained under the RCM
philosophy.
T64 The T64 turbo-shaft has five type/model/series in the fleet. The -413,-415,-416, -416A, and
the -419 are all used in various versions of the H-53. The critical part life updates for the -
416 has been completed. All critical parts except one are the same between all versions. Part
A2-16

lives are based on flight hours. The engine is maintained under the hot section inspection
and overhaul philosophy. In future years, the Navy will update material testing, mission
recordings, and instrumented engine tests.
T406 The T406 turbo-shaft engine program is managed under the 'power-by-the-hour' concept
based on commercial-engine maintenance-support experience. There are 21 flight critical
rotating parts which are individually lifed in LCF cycles and equated to mission hours based
on the original specification missions representing the V-22 Osprey operations. In addition
to the 21 flight critical parts, also 8 blades and vanes are life limited by either stress rupture
or hot corrosion. There are 15 other components tracked by operating time. The V-22
aircraft are equipped with on-board recording and monitoring system to gather engine usage
and mission profiles. Once production engines are fielded, engine usage and mission
profiles will be gathered and verified. These usage and profiles will be analysed and results
will contribute to establishing updated life limits.
Transport
T56 The T56 turboprop engine has 7 critical rotating life-limited parts. Two different series of
this engine are used in four different Navy aircraft platforms: P-3, C-130, E-2C/C+, and C-
2A. LCF and FM lives have been established for these parts and are managed using flight
hours based upon a cycle provided by the OEM. The mission analysis for all four platforms
has not been updated within 10 years. Recorded E-2C+ data will be utilised to update the
mission cycle to flight hour conversion factor. No life management plan has been developed
or followed within the last 10 years. A new life management plan is being developed.
Turbine spacer life limits are under review in light of new OEM materials data and life limit
reduction recommendations.
 A3-1

Appendix 3
Mechanics of Materials Failure
by
(W. Beres)

Page

1. Failure Modes of Gas Turbine Engine Components A3-3

2. Stress States in Gas Turbine Engine Components A3-3

2.1 Thermal Analysis A3-3
2.2. Stress Analysis A3-3
2.2.1. Stress Intensity Factor A3-3

3. Properties of Gas Turbine Engine Component Materials A3-5

3.1. Static Properties A3-5
3.1.1. Engineering Stress-Strain Curve A3-5
3.2. Flow Properties A3-6
3.2.1. True Stress-Strain Curve A3-6
3.2.2. Effect of Strain Rate on Flow Properties A3-6
3.2.3. Effect of Temperature on Flow Properties A3-7
3.3. Creep Properties A3-7
3.4. Cyclic Properties A3-7
3.4.1. Cyclic Stress-Strain Curve A3-8
3.4.2. High Cycle Fatigue (HCF) A3-8
3.4.3. Low Cycle Fatigue (LCF) A3-9
3.4.4. Thermo-Mechanical Fatigue (LCF) A3-9
3.5. Fracture Mechanics Properties A3-9
3.5.1. Fatigue Crack Growth Rate (FCGR) A3-9
3.5.2. Creep Crack Growth Rate A3-10
3.6. Influence of Mean Stress on Fatigue and Fracture A3-10
3.6.1. Influence of Mean Stress on HCF A3-10
3.6.2. Influence of Mean Stress on LCF A3-11
3.7. Influence of Multiaxial State of Stress on Fatigue and Fracture A3-11

4. Analytical Techniques for Damage Characterisation A3-12

4.1. Cumulative Fatigue Damage A3-12
4.2. Creep Damage A3-12
4.2.1. Larson-Miller Parameters A3-12
4.2.2. Monkman-Grant Type Relationship A3-13
4.2.3. Theta Projection Concept A3-13
4.3. Creep-Fatigue Interaction A3-13
4.3.1. Linear Damage Summation A3-13
4.3.2. Strain Range Partitioning A3-13
4.3.3. Ductility Exhaustion A3-14

5. References A3-14

Glossary G
A3-2
 A3-3

fracture mechanics methods.

1. FAILURE MODES OF GAS TURBINE

ENGINE COMPONENTS 2. STRESS STATES IN GAS TURBINE

The safe life approach has been criticised as being overly ENGINE COMPONENTS
conservative and costly on the ground that critical engine
components are usually discarded with significant amount 2.1. THERMAL ANALYSIS
of useful residual life. There are two major concerns when The temperature distribution for turbine discs depends on
a cycle-to-crack-initiation rejection criterion is used to life the engine operating conditions, and which phase of a
rotating parts. The first one is that, by implication, 99.9% flight or ground run the engine is in. For a stabilised high
of the components will be retired before any detectable temperature condition, the rim of the disc is hotter than
crack has formed. Possibly, these parts may be capable of the bore area, which creates compressive hoop stresses in
significantly longer service before they develop a the rim and tensile stresses in the disc bore area. This state
detectable crack (actually 0.8-mm in length). Secondly, of stress is reversed during engine deceleration when the
the components may be capable of tolerating crack sizes rim cools more quickly and a reverse temperature gradient
much greater than the 0.8-mm mentioned above. This is established. The rim is then subjected to high tensile
crack limit actually reflects the sensitivity and reliability hoop stresses, and the disc bore area to hoop compressive
of current ND1 methods rather than the mechanical stresses. Differences between the temperatures on each
tolerance of a particular part for cracks. Crack side of the disc create stress gradients across the disc
detectability by dye penetrant methods depends on crack thickness, which may also be taken into account.
width, whereas crack length is the critical factor for lifing
purposes. 2.2. STRESS ANALYSIS

In their most elementary forms alternative damage Generally, for the linear case, the state of stress at a given
tolerance based lifing procedures, which are also known point in the disc can be described as follows:
as Life-On-Condition, Retirement-for-Cause or simply
(7ij = B,(V2 + B2^T + B3T (A-l),
Fracture Mechanics (FM) lifing, make assumptions. They
assume that the fracture critical locations of a component where ais is the component of stress tensor, co is the
contain cracks of a size that lie just below the detection
rotational speed of the components, AT is the temperature
limit of the NDI technique used to inspect the component.
range, T is the iocal temperature and Bt, B2, B3 are
The crack is then assumed to grow during service in a
parameters depending on the material properties and
manner that can be predicted by linear elastic fracture
component geometry.
mechanics, or any other acceptable methods. This is
applied until a predetermined dysfunction limit is reached 2.2.1. STRESS INTENSITY FACTOR
beyond which the risk of failure due to rapid crack growth
becomes excessive (Koul at al., 1988). Under linear-elastic conditions, the general expressions
for the crack tip stresses have the form:
For damage tolerance based procedures to be successful,
supporting methodologies must be developed. These
methodologies include both deterministic and a.^j^/,/#) +A2gu(0) + A3hlJ(0)
probabilistic fracture mechanics based life prediction,
(A-2),
based on non-destructive inspection, mechanical testing
of test coupons and components, structural analysis, where the notation presented in 1 are used. K is the
mission profile analysis and condition monitoring of proportionality constant called the stress intensity factor,
components. In particular, extensive materials testing r and 6 are the radius and polar angle measured for the
involving room and elevated temperature crack growth crack tip, and crack plane respectively, A, are constants,
rate data must be performed for the application of new f,j(6), gij(&), hj/0) are dimensionless functions of #and
concepts. Therefore basic properties of materials and (/, j=l,2,3). If higher order terms are neglected (Ai=0,
databases of material degradations modes are summarised A}=0), the stress components ahead of the crack tip in
in this chapter. Mode I can be approximated by the expressions:
Not all components are candidates for life usage
management to damage tolerance limits. Two examples cr,, = cr =
K e , . 0 . 30
cos — 1 -sin —sin — (A-3)
are: 4lnr 2 . 2 2.
• Components made out of material with a short crack
propagation life; 0-n = <7y
K e , . 0 . 30'
, cos — 1 + sin —sin — (A-4)
• Components that cannot be inspected. 4lm- 2 2 2

Few, if any, mature engines were designed with damage K " . 0 0 30

*"„. = sm —c ;—cos— (A-5)
tolerance (FM) methodologies and some may be viable -Jim- 2 2 2
candidates for damage tolerance management. An
example of where this approach has been introduced after These equations describe the stress singularity around the
the engine has entered service, is the RBI99 engine in the crack because as r-0, the stresses approach infinity. Thus,
Tornado aircraft. A number of components, which have
proven crack resistance, have been reassessed using stress near the crack tip varies with 1/i/f, regardless of the
A3-4

configuration of the cracked body. It can also be shown C5

. .A » • » ' * f * f * * t
that crack opening displacement near the crack tip varies

with vi .The stress intensity factor (SIF), K, is usually

given a subscript denoting the mode of loading, i.e. Kh
1
K,h Kw. It was found that the SIF range is a main factor TK
governing the rate of growth of fatigue cracks. The SIF t
defines the amplitude of the crack tip singularity. This 0. *-| [<• 8.
means that stresses near the crack tip increase in

■?
proportion to K. In addition, the SIF defines completely
the stress conditions at the crack tip. If K is known, it is
possible to solve for all components of stress, strain and
displacement as a function of r and 6. This is one of the
most important concepts in linear elastic fracture *'
mechanics: having one single parameter that completely : \$
describes the conditions around a crack tip. A distribution
of an elastic stress ay in front of a crack is shown in figure "Crack *
2. Figure 1 - Stress Intensity Factor (SIF) at crack «p
To make SIF a useful parameter, it should be possible to
determine K from remote loads and the geometry of a
cracked component. Closed form solutions for SIF have
been derived for a number of simple configurations. For
example, for an infinite plate under tension treated in 2D,
figure 3, the stress intensity factor at the tip of the crack

(A-6),
*/
where <ry is the remote stress and a is the half of the crack
length.
This equation is modified when an edge crack in a half
plane is considered, figure 4:

K; = Fjcr 4m (A-7),

where F, is the boundary correction factor, which takes

into account the effect of free boundaries on the crack
opening. For example, for a short edge crack in a plate Fs
= 1.12.
When considering three dimensional cracks, the closed
form solution for a circular (penny shaped) crack fully
embedded in an infinite plate is:

K, =—<Jv^im (A-8)
simplest estimation of the size of the crack tip plastic zone
can be obtained by determining the distance rp from the
where cr is the remote stress, and a is the penny shaped crack tip for which the apparent elastic stress ay is larger
crack radius. than the yield stress, crys, figure 5:
As with the 2D case, this solution is modified when an
1 K2
edge crack is considered: Vy (for plane stress) (A-10)
2ncTys
2 /—
A-9) For conditions of plane strain where the triaxial stress
7T
fields suppress the plastic zone size, the plane strain
where F, is the boundary correction factor, and <p is the plastic zone size radius is smaller:
angle which describes the location of the point of interest
on the crack front circumference.
(for plane strain) (A-ll)
Linear-elastic fracture-mechanics analysis predicts 6^r cr
infinite stresses at the crack tip. In real materials,
however, stresses at the crack tip are finite because plastic The size of the plastic zone varies along the crack front,
deformation at the crack tip keeps the stress finite. The being largest at the two free crack surfaces and the
 A3-5

smallest at the mid-plane.

2.2.1.1. NUMERICAL ANALYSIS OF STRESS
INTENSITY FACTOR
t t t T i ' T f t
Finite element based stress analysis methods provide
displacement and stress distributions in a structure as a
primary solution to the problem. If a crack is present in
the structure being modelled, post-processing techniques
are generally required to extract fracture parameters from
these distributions. Three crucial parameters that are
extracted from two, and three-dimensional FE calculation
results are the stress intensity factor (SIF), the strain 2a
energy release rate (SERR) and the J-integral. The most
popular techniques for estimating these parameters are:
•
•
Virtual crack extension-based techniques;
Crack opening displacement-based techniques; TTTTTTTTT
• Stress or nodal-force approximation techniques; Figure 3 - Crack in infinite plate
• J-integral based techniques.

Because the finite element method is an approximate C3F

1
y*.
technique for solving boundary value problems in solid
mechanics, each method for extracting SIFs from
calculation results has advantages and disadvantages. The
choice of a particular method depends on the type of
problem analysed. Applications of FEM to fracture
mechanics were reviewed, among others, by Pickard
(1986), Liebowitz (1989), Banks-Sills (1991).

3. PROPERTIES OF GAS TURBINE ENGINE

COMPONENT MATERIALS <m
3.1. STATIC PROPERTIES

3.1.1. ENGINEERING STRESS-STRAIN CURVE

An engineering stress-strain curve is constructed from
load-elongation data, in figure 6. The initial portion of
i r
Figure 4 - Edge crack in a half plate
this curve OA is the elastic region in which Hooke's law
is valid. In this region, stress is proportional to strain.
Point A is the elastic limit, defined as the maximum stress
that can be applied without creating a permanent strain. a, . ■
The slope of the stress-strain curve in the region OA is the
modulus of elasticity or Young's modulus (E). Because the \ _■ Elastic Solution
modulus of elasticity is determined by the binding forces
between atoms, it is only slightly affected by metal heat V
i
treatment and cold work. However, the modulus of
elasticity decreases with increasing temperature. <V>
The limit of elastic behaviour of the specimen is
described by the yield strength, point B at figure 6,
defined as the stress which will produce a small amount
of permanent deformation, denoted by OC. When the load
further exceeds a value corresponding to the yield
strength, the specimen undergoes gross plastic
Crack
Ä
deformation, which is demonstrated by permanent r
deformation of the specimen after the load is released to
zero. 2a fp
Figure 5 - Plastic zone at crack tip
As the plastic deformation of the specimen increases, the
metal usually becomes stronger due to strain hardening - constant during test. From point D further on the diameter
so that the load required to extend the specimen increases of the specimen begins to decrease rapidly because the
with further straining. Eventually the load reaches the rate of strain hardening is overtaken by the loss of local
maximum value, point D. The diameter of the specimen bearing area due to necking and internal cavitation.
decreases as the length increases, since it is usually Because of this, the load required to continue deformation
assumed that total volume of the specimen remains drops off until the specimen fractures, Point E in figure 6.
A3-6

The point at which plastic deformation begins is hard to

define because for most materials there is a gradual Fracture strain
transition from elastic to plastic behaviour. Usually the
yield strength is defined as the stress required to produce D
a small specific amount of plastic deformation, so called n /
offset yield strength determined by the value a in figure 6. Biz ^E ' a
This offset is usually specified in US practice as a strain / t E
of 0.2% or 0.1% (0.002 or 0.001) or 0.1 and 0.5% for UK
practice.
7/ f !
CO
U_

/ / f O)
If the load is removed at Point B, the unloading follows Ig / / CD Q)
the path BC. The slope of this unloading curve is parallel / 1 •2 tB
to the elastic portion of the curve on loading. The /
/
i
t 8. M
permanent plastic deformation is the offset a in figure 6. c
CD
However, an elastic deformation is always present in the
tension specimen under load.
The tensile strength or ultimate tensile strength (UTS) is o C" Conventional Strain
the maximum load, Fmax, divided by the original cross- a
(Elongation)
sectional area of the specimen, A0: Figure 6 - Engineering stress strain curve
C _ Fmax
(A-12)
A
It should be remembered that although the tensile strength
is the value most often quoted from the results of a
tension test, it has little fundamental significance with ngineering stress-strain
regards to the strength of a metal. This is because the
cross al area of the specimen changes continuously as the E
test progresses.

3.2. FLOW PROPERTIES

3.2.1. TRUE STRESS-STRAIN CURVE
The true-stress a is expressed in terms of engineering D - Maximum load
stress 5 by: E - Fracture

i{e+l) (A-13)
4,
Strain
The derivation of this equation assumes constant volume
Figure 7 - Limit of elastic behaviour
and a homogeneous distribution of strain along the gauge
length of the tension specimen. The true-strain e may be dimensional form:
determined from the engineering or conventional strain e
by:
a a
ln(e+ l) (A-14) ■+a (A-16)
\a*j
Both equations are valid until the specimen necking starts.
Figure 7 compares the true-stress-true-strain curve with where £0=o</E ,a0 is the yield stress and E is the Young's
its corresponding engineering stress-strain curve. modulus of the material.

In the region of plastic deformation, the flow-curve for Relation (A-16) can be rewritten as:
metals may be expressed as a simple power law. This is: l
a (a
a = K£" (A-15) s = — + a\ — (A-17)
E [K
K is the strength coefficient and n is the strain-hardening
exponent. This equation produces a straight line on a Please note that although the strength coefficient K has
double logarithmic curve of the true-stress-true-strain. the same notation as the stress intensity factor described
Typical values for the strain hardening exponent n for elsewhere, it represents a different physical quantity.
metals and alloys range from 0.1 to 0.5. Perfectly plastic
solid has n=0, while a perfectly elastic solid has n=l. 3.2.2. EFFECT OF STRAIN RATE ON FLOW
PROPERTIES
The monotonic stress-strain curve can be approximated
by the Ramberg-Osgood relationship written in its non- A general relation between flow stress and strain rate for
constant strain and constant temperature can be expressed
 A3-7

as: primary, secondary and tertiary stages, respectively.

a = Ce" (A-18) (c) The third is a temperature function, which depends on

temperature, T, only.
where m is the strain rate sensitivity coefficient. This
In discussion of high-temperature material properties, the
relationship is illustrated at figure 7 of chapter 5, which is
temperature is often expressed as a homologous
drawn in a double logarithmic scale.
temperature. That is the ratio of the test temperature to
3.2.3. EFFECT OF TEMPERATURE ON FLOW the melting temperature on an absolute temperature scale.
PROPERTIES
For creep resistant super-alloys, creep becomes
significant at a homologous temperature greater than
The temperature dependence of flow stress at constant 70%.
strain and strain rate can be generally approximated in the
following manner: The most generally used temperature function is:

CT = Ce QIRT (A-19), /3 = A^cr" exp Q_ (A-25)

where Q is an activation energy for plastic flow, J/(g RT
mole), R is the universal gas constant, J/(K mole), and Tis where Q is the apparent activation energy, R is the
the testing temperature, K. universal gas constant and T is the absolute temperature.
Both Q and n are material dependent.
3.3. CREEP PROPERTIES
The component service temperatures are usually high It should be remembered that this set of equations is based
enough to cause ageing of the material, which results in on the phenomenological approach to creep modelling.
loss of creep strength and therefore its load bearing This means it is not based on the actual metallurgical
capacity. processes taking place in the creeping area. Rather it is
based on the observed physical behaviour of specimens
The shape of a creep curve is determined by several under test.
competing reactions, including strain hardening, softening
processes such as recovery, recrystalization, precipitate Of all the parameters pertaining to the creep curve, the
over ageing, damage processes such as cavitation and most important for engineering applications are er and tr.
cracking, and specimen necking. Of these factors, strain Specifically, their dependence on temperature and applied
hardening tends to decrease the creep rate whereas the stress are of extreme importance to the designer. This
other factors tend to increase the creep rate. dependence varies with the applicable creep mechanism.
Another useful first approximation of the creep strain rate, Damage accumulation during creep can be classified into
4, as a function of stress, a, time,;, and temperature, T, is three categories:
to present it as a product of the form: • Damage by loss of internal/external material
cross-section;
se = M<r)f2(t)f3(T) (A-20)
• Damage by degradation of the micro-structure;
The three functions are defined as below: • Damage by environmental attack.
(a) The first is a stress function, which depends on stress, Based on this classification, the continuum damage
<r, only. The most commonly used relations for stress mechanics, i.e., Kachanov-Rabotnov damage theory, can
function are: be used to describe the evolution of creep damage
coupled with the creep deformation. However, it does not
fl(cr)=A,a (A-21) recognise that inter-granular damage and intra-granular
damage occur by different mechanisms and therefore
(A-22) follow different laws. Recent work (Wu and Koul, 1996)
has shown that it is necessary to use a super-position
strain model, to describe the accumulation of intra-
\ granular damage (e.g. dislocation multiplication and
f](a)=A,exp (A-23) precipitate coarsening), and grain boundary damage (e.g.
V^oy cavitation and/or oxidation).

where At. A2. A3 are constants and a0, cr'o are the In all of the processes described above, diffusion (either
reference stresses. The first of these three equations is the grain boundary diffusion or volume diffusion) plays an
most extensively used in practice. important role. The diffusion itself can cause deformation
under certain conditions. Diffusional flow is generally
(b)The second is a time function: which depends on time, predominant at stress levels below those of the other
/, only It is expressed as a polynomial: mechanisms such as GBS and power-law creep. An
example of a deformation mechanism map for the above
f2(t) = at"3 + ßt + rti (A-24) mechanisms at steady states is shown in figure 8.
where a, ß, and y are material constants, which are 3.4. CYCLIC PROPERTIES
function of stress and temperature and which relate to
The latest approach for some aerospace components is to
A3-8

classify failures occurring below 105 cycles as LCF and LAND BASED TURBINE OPERATION
those occurring above 107 as HCF. However, many | AIRCRAFT TURBINE OPERATION
investigators now define the LCF range to be failure in
50,000-100,000 cycles or less. The area between 105 and _JP.EA^JH}EAJ?.siREtlGTH; IN738LC
DISLOCATION GLIDE
107 is a 'grey area' where attribution of LCF and HCF
should performed on a case-by-case basis. 10"
GRAIN BOUNDARY
3.4.1. CYCLIC STRESS-STRAIN CURVE t4> SLIDING
(DGB)
A cyclic stress-strain curve describes stress evolution
under strain-controlled cyclic loading. A stress-strain loop
under controlled constant strain cycling is presented in
INTERFACE REACTION
figure 9. During initial loading the stress-strain curve is CONTROLLED DIFFUSIONAL
CREEP
O-A-B. On reloading in tension a hysteresis loop
develops. The dimensions of the hysteresis loop are
described by its total width Ac, which is the total strain
range and A a, the total stress range. The total strain range,
Ac consists of the elastic strain component, Ac^AolE,
and the plastic strain component A%. Because plastic 0.2 0.4 0.6 0.8 1.0

deformation is not completely reversible, modification of

the metal structure occurs during cyclic loading and A DEFORMATION MECHANISM MAP
FOR IN 738 LC
results in the cyclic stress-strain response.
The cyclic stress-strain curve may be approximated by a Figure 8 - Deformation mechanisms. After Wu and
power curve similar to the curve for a monotonic stress- Koul (1995)
strain relationship:

Aa = K'(Aspy (A-26)

where K' is the cyclic strength coefficient and n ' is the

cyclic strain hardening coefficient. The equations for the
cyclic stress-strain curves are usually presented in the
form:

As A*, A (A-27)
2 2 2
l/ri
As _Ac 1 ~Aa (A-28)
2 2E 2 K'
The cyclic strain hardening exponent, n' has a value in
the range of 0.05 to 0.25 for metals and alloys. Cyclically
stabilised stress-strain curves provide an important means
of characterisation of a material cyclic response. Eq (A-
28) is similar to (A-17) except that the strain and stress
are replaced by the stress and strain amplitudes and the
material constants are denoted by the prime letters.

3.4.2. HIGH CYCLE FATIGUE (HCF)

The stress ratio is the algebraic ratio of two specified Figure 9 - Cyclic stress-strain loop
stress values in a stress cycle. Two commonly used stress
ratios are the ratio, A, of the alternating stress amplitude
to the mean stress
maximum and minimum stresses in one cycle:
A = ±-
a
»' (A-29) C™ =
ü
min ' ^max

and the ratio, R, of the minimum stress to the maximum (A-31)

stress:
and the stress range, Aa, is the algebraic difference
n min
between the maximum and minimum stresses in one
(A-30) cycle:

where the mean stress, om , is the algebraic average of the ACT = 0-mM ~ ömin (A-32)
 A3-9

High cycle fatigue life assessment is sometimes referred applicable to thermal fatigue, although thermal fatigue
to as the stress-life approach. The basic method for problems involve all of the complexities of mechanical
presenting HCF data is by means of S-N curves, which loading. In addition, all of the temperature-induced
plot the number of cycles to failure, Af vs. the stress range. problems are involved. Most low cycle fatigue problems
It should be stressed that the S-N curve does not at high temperatures involve thermo-mechanical fatigue
distinguish between crack initiation and crack propagation processes. Hot section components of gas turbine engines
phenomena. are subjected to cyclic temperature simultaneously with
cyclic stress. Analysis of these loads and consideration of
For some engineering materials, the S-N curve becomes the associated fatigue damage become very complex, and
almost flat at low stresses, indicating a threshold value of many simplifications have to be introduced. Historically,
stress below which failure will not occur for practical thermal fatigue was considered as isothermal low-cycle-
purposes. This threshold value of stress, ac, is defined as fatigue at the maximum temperature of the thermal cycle
the fatigue limit. to which a component was subjected. However, recent
advances in test systems and numerical calculation
3.4.3. Low CYCLE FATIGUE (LCF) methods have made it possible to conduct thermo-
Low-cycle fatigue life assessment is sometimes referred mechanical fatigue tests under well controlled conditions
to as the strain-life approach. This method is based on the and also to analyse numerically complex thermo-
observation that in many components the response of the mechanical cycles. The mechanical low cycle fatigue and
material in the failure critical locations is strain thermal low cycle fatigue processes are mathematically
dependent. In the strain-life approach the plastic strain is expressed by similar set of equations, however using LCF
directly measured and quantified. results to predict thermal LCF performance should be
LCF tests are often conducted in the same manner as HCF undertaken with substantial care. Generally, it is found
tests except that the strain range is held constant, and the that TMF loading can be more damaging than pure LCF
stress is allowed to vary. The variation of stress with loading for the same total strain range applied.
strain in LCF tests typically leads to a hysteresis loop Some of the reasons of the differences between thermally
such as the one shown in Figure 9. A tension-compression and mechanically induced LCF are as follows (Manson,
stress range Aais established corresponding to the strain 1966).
range Ac imposed on the specimen. The entire loop width
Plastic strains in thermal fatigue tend to become
corresponds to the total strain range A%, which can be
concentrated in the hottest region of the body, since the
decomposed into the elastic and plastic strain ranges. The
yield point is locally reduced in these locations.
height of the hysteresis loop is the stress range.
In thermal fatigue, there is often a localised region of
Fatigue cracks are mostly initiated at heterogeneous
strain developed by virtue of plastic flow during the
nucleation sites within the material. These could be pre-
compressive part of the strain cycle to produce a bulging
existing, or generated during the cyclic straining process.
at the hottest region. This is followed by a necking
Pre-existing nucleation sites can be associated with
tendency adjacent to the bulge, which is caused during the
inclusions, gas pores, or local soft locations in the
tensile part of the strain cycle upon cooling, by plastic
material microstructure. Fatigue cracks are generated
flow.
during cyclic straining at the areas where persistent slip
bands are being formed in the material. Cyclic variation in temperature may have an important
effect upon the material properties and it ability to resists
It has been found that persistent slip bands contain sharp
LCF failure.
peaks, extrusions, sharp troughs, and intrusions. These
result from incomplete slip reversals at the material There are interaction effects caused by superposition of
surface or matrix/inclusion interfaces. These surface or simultaneous variations in temperature and strain.
sub-surface notches serve as fatigue-crack nucleation
sites. These extrusions and inclusions represent the initial Rates at which the strain cycling is induced may have
stage in micro-crack formation. It should be recognised important effects, since the testing speeds in thermal
that although cracking begins at a surface, the material fatigue tests are often greatly different from the rates used
within these persistent bands and below the surface is also in mechanical LCF tests.
damaged and this will have an influence on the location of These are compelling reasons to use extreme caution in
the surface crack nucleation. In polycrystaline materials, the prediction of thermal LCF component behaviour from
persistent slip bands can be arrested at grain boundaries, mechanical LCF results and vice versa.
affecting crack nucleation at the junctions of slip bands
and grain boundaries. 3.5. FRACTURE MECHANICS PROPERTIES

3.4.4. THERMO-MECHANICAL FATIGUE (LCF) 3.5.1. FATIGUE CRACK GROWTH RATE (FCGR)
Any model aimed at producing acceptable engine hot- The stress range Aa- a,mx - erm,„ in a given location of a
section life-predictions must take into account the component can be related to the stress intensity factor
component loading cycles, and consider how the stress range, AK:
and inelastic strain at each area of the component vary
with time. AK = Km„-Kmm (A-33)

Low cycle fatigue results can thus be qualitatively where Kmax and Kmin are the stress intensity factors
A3-10

associated with the maximum, amax, and minimum, crmin, Another model is obtained by randomising the crack
stresses in a fatigue cycle. This relation, for a particular growth rate equation:
component, is frequently established using finite element
da
analysis. The growth rate of a fatigue crack as described ■q(a)X(t) (A-36)
by Paris or Forman's relation is related to AK and the load dt
ratio, R: where X(t) is the random process, and q(a) is the general,
deterministic relation between the crack size a and the
R_ A min
(A-34) crack growth rate. The best results were be obtained if the
Kr X(t) process was modelled as a random pulse train, (Lin
and Yang, 1983). Other aspects of the statistical
which is a ratio of maximal and minimal stress intensity
modelling of the FCGR can be found in Besuner (1987).
factors in a load cycle.
The Paris formula described by Eq. (6) of chapter 5, is 3.5.2. CREEP CRACK GROWTH RATE
sometimes written as: The most commonly used parameters for describing creep
crack growth rate ä are stress intensity factor K, and the
da_ AK \" (A-35) creep fracture mechanics term C*. The following relations
=C have been produced, mostly to correlate individual test
dN AK,
data:
where AKZ is the arbitrarily established reference stress
intensity factor range. In this equation, the constant C has a = AK" (A-37)
the unit of m/cycle.
It is necessary to point out that fatigue threshold
phenomena described in section 5.5.1 are mostly observed ä = D0C'* (A-38)
in load shedding FCGR tests using specimens containing
a long crack. In these cases, FCGRs are strongly where A, Do, m, and tp are material constants which may
influenced by crack closure associated with the load- be temperature and stress state dependent. The creep
shedding procedure (Wu, Wallace and Koul, 1995). The fracture mechanics term C* is the contour integral
characterising steady state creep. It is similar to J-integral,
measured AK,h value for a long crack in such a test does
which characterises the fatigue crack growth rate in
not represent an intrinsic property of a material. Short
elastic-plastic fracture. The exponent m has a value close
crack fatigue experiments show that cracks do grow at AK
to the value of n in Eq. (A-25), while the exponent (pis a
levels well below the long-crack, AK,h. Therefore, for the fraction close to unity (Webster and Ainsworth, 1994).
analysis of crack growth near the so-called threshold,
AKlh, it is recommended that short crack data, rather than 3.6. INFLUENCE OF MEAN STRESS ON FATIGUE
data obtained for long cracks, is used to assess component AND FRACTURE
lives
Even if the test variables that effect fatigue crack growth
3.6.1. INFLUENCE OF MEAN STRESS ON HCF
rates are stringently controlled, there still remains a The curves, which show the dependence of limiting range
significant variation in the results, which must be stress, amax-ami„, on mean stress are called Goodman
attributed to the material microstructure. That is, to diagrams. An example of which is shown in figure 10.
random distributions of lattice defects, impurity atoms, This relationship is established for a fixed number of
slip systems, crystal sizes, grain boundary parameters, and cycles or at the fatigue limit. It can be seen in figure 10,
macro defects such as porosity and cracks. All of these that as the mean stress becomes more tensile, that the
effects imply the random nature of material damage allowable stress range reduces. A conservative
processes and suggest that a probabilistic approach should approximation of a Goodman diagram may be obtained
be taken when analysing the fracture processes in modern by drawing straight lines from the fatigue limit for
materials. In the probabilistic approach, a random completely reversed cycles to the tensile strength.
variability of crack sizes is considered as a function of The other method of presenting the mean stress effect on
time. This approach requires knowledge of the crack size HCF is known as a Haig-Soderberg diagram in which the
probability distribution function, the probability alternating stress is plotted against the mean stress, figure
distribution function of the extreme loads, and the 11. These plots were historically the subjects of numerous
probability distribution function of the material empirical curve-fitting procedures. A straight line
parameters, such as fracture toughness. represents the Goodman method while the parabolic curve
A number of models have been developed to describe the was suggested by Gerber. If the design is based on the
observed variability in crack growth. One of these is the yield strength, as shown by a dashed line in figure 10,
model based on Markov chain theory described by than oa is substituted for au.
Bogdanoff and Kozin (1985). This model has shown the The equations for these three relationships can be
ability to characterise both constant and variable presented as follows:
amplitude fatigue-crack growth. However, this model
requires knowledge of the first and second-order moment Goodman's Linear Relation
statistics of time for a crack to reach various sizes.
 A3-11

Compression Tension
■£L + £k = l (A-39)

Gerber's Parabolic Relation tfmoi^^J // '

r^Y

Tension
■+ (A-40) / /"•ml» ! 1

Stress range
<JK> K^uJ %• / I i
Soderberg's Linear Relationship / Mean stress, om

Compression
SL + -S=- = 1 (A-41)
.i'p
/ /
/
where aa is the alternating stress amplitude, om is the /
mean stress, oN is the fatigue strength at N cycles and, ovp
is the yield point of the material. Figure 10 -- Goodman diagram. After Dieter (1986)

3.6.2. INFLUENCE OF MEAN STRESS ON LCF

Walker (1970) proposed a method of taking stress ratio
into account during crack propagation and fatigue in the
form of an equation:
da
~dN
=C 0-*)"Xj (A-42)
where R is the stress ratio, C and n are the appropriate
parameters of the Paris law, Eq(6) of chapter 5, and m is
the experimentally established Walker constant having a
value between 0 and 1.
This Walker method used for correction for non-zero
mean stress in the FCGR analysis is also applied, as a
way of deriving material LCF data for stress conditions
other than zero minimum stress (A *1), Eq (8) of chapter
Figure 11 - Haig-Soderberg diagram. After Collins
5.
(1993)
With respect to the Walker model, it should be
remembered that:
F = a
= — {£x-£2)2+{s2-eJ +{£i-sif}
• Not all Walker parameters are applicable to all ' 2 3
materials and to all stress ratios: each material and (A-43)
stress ratio is usually defined by a unique Walker
parameter. where eh c2, e3 are the principal components of strain, and
• The method is most effective in R ratio range of-1 yocl is the octahedral strain.
to 0.6 {A ratio from infinity to 0.25). The combined strain-state is consolidated in this equation
• It is a compromise in modelling behaviours at to provide eq< which is entered into the Coffin-Manson
various mean stresses, A or R ratios; there may be an formula, Eq (3).
uneven compromise at one or more sets of A -ratio
data. Similarly, it was proposed to use the following expression
• It can lead to unrealistic trends when very low and for fatigue failure under combined stress in low strain
very high stress ratios are involved. conditions - high cycle fatigue:
• The mean stress effect is more pronounced at room 1/2

temperature and drops off as temperature increases. fk.1 - <?al )2 + {°al ~ aal Y + (°"«3 " ^ Y
Therefore, the Walker exponent is a function of
temperature.
, \ v2cr
C2{crm}+<Tm2+(rm3)>-
• Significant compressive stress or negative mean K,
stress invalidates the method. (A-44)

3.7. INFLUENCE OF MULTIAXIAL STATE OF

where
STRESS ON FATIGUE AND FRACTURE oai is the alternating component of principal stress in
Combined fatigue under high-strain low-cycle conditions the ' 1' direction;
can be correlated with the octahedral shear strain: omi is the static component of principal stress in ' 1'
direction;
A3-12

• oe is the fatigue strength (fatigue limit) for

completely reversed stress;
• Kj is the fatigue notch factor, i.e the ratio of the
fatigue limit of unnotched specimens to the fatigue o
limit of notched specimens,
• C2 is the material constant taking into account the
influence of am on a,. At first approximation C2 =
0.5.

If fatigue properties are anisotropic, or become

anisotropic due to the effect of cyclic plasticity on
material properties, then the concept of a single
equivalent strain or stress is no longer valid. This is
because the direction of loading with respect to the l/r
direction of fatigue properties may become an important
factor.
Many questions remain to be answered with respect to the log t, = -C
validity of the above equations (Laflen and Cook, 1982). Figure 12 - Larsen-Miller time temperature grouping

4. ANALYTICAL TECHNIQUES FOR DAMAGE

CHARACTERISATION

4.1. CUMULATIVE FATIGUE DAMAGE

4.1.1.1. DOUBLE LINEAR DAMAGE RULE
It was proposed that cumulative damage estimates might
■ = 1
be improved by breaking the fatigue process down into a
crack initiation phase and a crack propagation phase and %WP)j (A-48)
applying the linear damage rule to each phase separately. In each case n is the number of cycles applied at the i1
Manson (1967) suggested that a crack propagation period andj'11 stress levels.
can be modelled as
The double linear damage rule gave good agreement with
Np=PN"f test data for two stress level tests for several materials.
(A-45)
where, 4.2. CREEP DAMAGE

Np is the number of cycles to propagate a crack to 4.2.1. LARSON-MlLLER PARAMETERS

failure after it has been initiated, A concept of a time-temperature grouping in the form
• Nf is the total number of cycles to failure, T(Ki + log t) was introduced by Larson and Miller
• P is the experimentally determined propagation (1952). For a given material a plot of stress vs. the above
coefficient, parameter resulted in a single plot regardless of the time-
• p is the experimentally determined propagation temperature combination employed to derive the
exponent. parameters, such as shown in figure 12. It can be shown
that the Larson-Miller parameter can be readily derived
In the original work, the coefficients P and p were from the stress and temperature dependence of the creep
experimentally established as 14 and 0.6, respectively. rate at time to rupture. The rate equations can generally be
written as:
Crack initiation period can just be obtained by subtracting
the number of cycles to propagate the crack from the total
number of cycles: (A-49)

N'=Nf-Np (A-46)
s=A,exp(-y) (A-50)
A linear damage rule is applied to each phase individually
to produce predictions of crack initiation and component
This parameter can be expressed in terms of time to a
failure as follows:
specific strain or time to rupture. This relation implies that
Fatigue cracks of critical size are initiated when the activation energy is dependent on stress. The Larson-
Miller parameter is usually written in the form:
Pt = f(a) = T(C + \ogwt) (A-51)
(A-47)
where P is the Larson-Miller parameter which is a
Fatigue crack are propagated to failure when: function of stress only, C is a constant and t is the time to
a specific strain or time to rupture. A plot of log tr vs 7/7"
 A3-13

results in straight lines whose intercept is a constant and basis. Its applicability is therefore material dependent.
whose slope is a function of stress as shown in figure 12. Contrary to experience, it also assumes that tensile and
compressive dwell periods are equally damaging. This
4.2.2. MONKMAN-GRANT TYPE RELATIONSHIP approach is sensitive to loading conditions and dwell time
The Monkman-Grant relationship is effectively a critical effects although it is consistent for varying temperature
strain criterion. It states that the strain accumulated during conditions. More importantly though, it has been shown
secondary creep is a constant at failure so that the product that significant interaction, which reduces the overall life
of the secondary creep rate and the rupture life tr is of the component, can occur.
constant: Specimen testing shows that a reduction in life can be
(A-52) obtained with a relatively small percentage of mixed
etr = C MC creep and fatigue. Loading histories and temperatures
where CMG is the Monkman-Grant constant. When should be evaluated to establish which of these effects are
secondary creep is dominant, constant creep ductility likely to occur. From a practical point of view, when low
independent of stress and temperature is predicted. combined creep-fatigue lives are calculated, detailed
specimen and component testing is performed to validate
4.2.3. THETA PROJECTION CONCEPT the calculations. Despite these limitations, the damage
A very general description of the creep curve under summation method is very popular because it is easy to
constant-stress conditions is given by the 8 projection use and requires only standard S-N curves and stress
concept in which creep strain, e, is considered to be the rupture curves.
sum of two competing processes according to the
4.3.2. STRAIN RANGE PARTITIONING
equation:
The strain range partitioning (SRP) approach involves
£=Oi[l- expf- 02 t)J + 03 [exp(0, t) -1] partitioning of the total inelastic strain range into four
(A-53) possible components depending on:
In this expression, 0|, 82, 63, and 04 are all experimentally • Whether the stresses are tensile or compressive;
determined constants which are functions of stress and • The type of inelastic strain accumulated (creep or
temperature. 9i and 62 define the primary or decaying time independent plasticity).
strain-rate component, and 93, and 04 describe the tertiary
or accelerating strain rate component. A wide range of Figure 13 shows a typical stress-strain hysteresis loop,
creep curve shapes can be modelled with various and figure 14 shows the four generic types of hysteresis
combinations of these constants. loops for the four types of strain range. The hysteresis
loop from a creep-fatigue test (i.e. LCF test with dwell
4.3. CREEP-FATIGUE INTERACTION time) is broken down into the component strains: Acpp,
Aecc> Aepc, Ascp. The terms Aepp and Aecc represent the
4.3.1. LINEAR DAMAGE SUMMATION pure reversed plastic and reversed creep ranges,
The most common approach is based on the linear respectively and the two other terms represent combined
superposition of fatigue and creep damage. In this creep and plastic strain ranges. For each type of strain
approach it is assumed that the damage fraction occurring range, the Coffin-Manson relationship, Eq. (A-3) can be
in fatigue and the fatigue damage fraction occurring in applied. For instance:
creep may simply be added. This approach combines the
damage summation of Miner's rule and Robinson's creep NPP = A(& £pp f (A-55)
damage rule as follows:
The fractional strain for each type of strain with respect to
N t the total inelastic strain is expressed, for example, as:
D' = 1— + T-<J (A-54)
Nf tr
'PP
Ff (A-56)
where N/N/ is the cyclic portion of the life fraction, in
which N is the number of cycles at a given strain range
and A^ is the pure fatigue life at that strain range. The time By adding up the fractional damage for each type of
dependent creep-fatigue fraction is i/tr, where t is the time strain, the total damage is estimated by the expression:
at a given stress and t, is the time to rupture at this stress.
The stress relaxation period is divided into time blocks F. F„ F,cp
■+■ •+■ ■+•
during which an average, constant value of stress prevails, N, N. JV" N pc N cp
and for each time block t/tr is computed and summed. D'
is the cumulative damage index, such that when D-\, (A-57)
failure is presumed to occur. where Npp, NCCI etc. represent the number of cycles to
Unfortunately, several material and test parameters may failure for each type of strain.
affect the distribution of D', and there is no satisfactory One of the major problems with this approach is the need
way of applying the linear damage rules at present. A to generate baseline data based on complex dwcll-period
serious drawback of this approach is that the life fraction tests. Extrapolation of predictions to long dwell periods
rule is purely phenomenological, having no mechanistic and small strain ranges also needs further verification. In
A3-14

order to apply Eq. (A-57) to life prediction for any Str<3SS

arbitrary cycle, the following information is needed:
Cree
Plastic flow ?
a) From a stable hysteresis loop of the stress-strain cycle,
the partitioned strain range components Aepp, Aecc, Aepc ' / //
(or Aecp) and the total inelastic strain range Ae are // / //
/ / //
obtained.
// //
b) The fractional strains Fpp, Fcc and Fcp are then B / /
calculated by use of the information obtained in step 1 A / C /D Strain
and by use of Eq (A-56). i I
//
i
/
c) The number of cycles to failure for each given type of
/// // /
/
strain (i.e the relationships A^ = A Aepp, etc) must be /
known from independent laboratory experiments. // //
Plastic flow
d) Now that the fractional strains Fpp, Fcc, etc. and the Creep
cyclic life for each type of strain, Npp, Ncc, etc. are known, Figure 13 - Stress-straiil hysteresis loop
the interaction damage rule Eq (A-57) is used to calculate
the cyclic life for the arbitrary cycle.

4.3.3. DUCTILITY EXHAUSTION

The ductility exhaustion approach is simply a strain-based Plastic
Creep
life prediction rule in which the fatigue damage and creep
damage are summed in terms of the fractional strain
damage for each category, as follows: *- i

J_ (A-58)
Nr DP Do Plastic Plastic
where Aep is the plastic strain-range component at half (b)
life, Dp is the fatigue ductility obtained from pure fatigue
tests, AEC is the true tensile creep-strain component, and
Dc is the lower-bound creep-rupture ductility of the Plastic
material. The first term in this equation denotes the Creep
fatigue-damage component, and the second term denotes
the creep-damage components. The ductility-exhaustion
approach is simple to use and has some mechanistic basis. -*■ t *- t
Selection of appropriate values for Dc and Aec is,
however, arbitrary and subject to errors.

5. REFERENCES
See Chapter 5.
Figure 14 - Four generic types of hysteresis loop
 G-l

Glossary
Term Meaning
Accelerated Mission Testing programme designed to expose selected types of design defect on one or more test-
Testing (AMT) bed engines, or models.
Accelerated Simulated As AMT.
Mission Endurance
Testing (ASMET)
Airworthiness Fitness to fly.
Availability The percentage of a fleet of equipment that is available for use.
Banding A method used for assigning usage damage to components. The bands are typically engine
speed
„i ranges. Also known—__.
—„-_. as gating.
_, __ .

Calibration Comparison and recording of the differences between a known standard and an instrument
that measures it.
Component Part of an engine. An item that forms part of an assembly-
Condition monitoring Monitoring the ability of an equipment or system to perform its function.
Creep Permanent deformation that occurs as a result of the prolonged application of stress. The
applied stress is below a level that would be expected to cause plastic deformation. Occurs
at temperatures that are high relative to the melting point of the material.
Creep, Mechanical Permanent elongation of a component when exposed at high temperatures to stresses that
would only cause elastic deformation at low temperature. As above
Creep, Metallurgical Very local yielding of material in stressed components at high temperatures. The effect is to
redistribute local stresses, without any apparent gross deformation of the component.
Cumulative damage The sum of damage which occurs over a period.
Damage Physical changes in a component, which cause it to deviate from its 'as new' condition.
Some forms of damage, such as fatigue, may not be visible to the naked eye.
Damage tolerance A design philosophy for safety critical parts. The basic assumption is that some form of
metallurgical damage exists in all new metallic components. It is assumed that a finite time
will be required for the damage to become a detectable defect. Design techniques that
assume that such damage exists, and allow the component to be used safely until the
damage increases to a pre-determined limit are known as 'damage tolerance' design
techniques.
Data Records of observations. Used in both the singular and the plural.
Duration A period (of time).
Engi In the context of this document, a gas turbine engine.
Engine flight time Length of period when engine is airborne.
Engine module A designated and interchangeable engine sub assembly. In many cases, it may have the
same traceability requirements as a complete engine.
Engine Life Monitoring The complete monitoring system including all airborne and ground based equipment.
System (ELMS)
Engine Life Monitoring An airborne unit which processes and records relevant engine and aircraft data.
Unit (ELU or EMU)
Engine qualification Formal tests that all new engine types must pass to demonstrate that the design and
testing construction meets the required airworthiness regulations and standards.
Engine running time Total recorded engine-running time. In some cases, this may not include test-bed and on-
wing maintenance running.
ENSIP Engine Structural Integrity Program. Used for proving the safety of US military engines.
Equipment An item designated as 'equipment' for ease of identification and logistical purposes.
Estimate A value that is calculated, or indirectly determined. Not measured.
Exchange rate A number used to convert flying hours into equivalent usage or damage counts. Also
known as Beta factor.
Fatigue, Low Cycle Damage that is related to stress cycles that lead to component failure in less than 105 or 107
stress cycles. Stress and temperature cycles are of high magnitude at relatively low
frequency causing large high plastic strain over a microscopic region. Normal design
practices and calculations consider this type of damage. It is the major life limitation on
many rotating components, shafts and combustion chamber outer casings.
Fatigue, High Cycle Damage that is related to stress cycles that lead to component failure in more than 107 stress
cycles, due primarily to vibration. Occurs at stress cycles of low magnitude at high
frequency causing elastic microscopic strain or very small amounts of plastic yielding. This
G-2

type of damage is usually avoided through the development testing process, and detail
design practices rather than by detailed design calculation. Many HCF failures are initiated
by other damage, such as LCF cracks.
Fatigue, Thermal Damage that is induced by thermally induced stresses.
Flight The period from the wheels leaving the ground, until they touch again. Many definitions
exist. In some cases, a number of "Touch and go" manoeuvres would be counted as a single
flight. This is an operational convenience. In component lifing terms, the accumulated
damage on such a flight may be much higher than that on a normal flight.
Fracture critical part A part which will physically break, causing catastrophic damage, after experiencing a
statistically described number and mix of missions. Such components are identified at
design time, and removed from service before failure occurs.
Functional failure When an equipment no longer performs as required, or as designed.
Information A system which stores data and provides an interrogation and analysis facility. Part of an
Management System engine life monitoring system.
(IMS)
Life count A value related to the number of design reference cycles that a component is designed to
endure before failure. It is estimated or measured by a usage monitoring system.
Life-critical part A part which limits the in-service life of an engine or module.
Life, dysfunction The life at which a part is judged to have failed. This is based on a combination of the
defect initiation life and the subsequent crack propagation life. 2/3 of the dysfunction-life
may be taken as a statistical adjustment to allow for variations in materials properties, and
applied as the service life-
Life expiry Consumption of all of the permitted life counts.
Life extension An increase in the permitted number of life counts. Fracture mechanics techniques or other
methods may be employed to show that the life increase is safe.
Life, Safe The formally declared number of life counts or flying hours that a component may
accumulate in use.
Life usage The life usage counts or hours accumulated by a component.
Lifed component A component that has a formally declared design life.
Lifing Assigning a life in hours or reference stress cycles to a component.
Lift-off The moment when the wheels of an aircraft leave the ground.
Maintenance, On- Maintenance actions that are taken only if signs appear, which show that a loss of
Condition operational functionality is likely to occur.
Maintenance The art or science of organising maintenance procedures and practice.
management
Maintenance, Preventive Maintenance done before any loss of performance, to protect future operational
functionality.
Maintenance, Reliability Process that determines what must be done to ensure that any asset continues to fulfil its
Centred intended function in its operating context.
Maintenance, Scheduled Maintenance done according to a time schedule, regardless of actual need.
Maintenance, Maintenance done between scheduled maintenance periods, to restore lost functionality.
Unscheduled
Mission A military operation. Also known as sortie
Mission mix A weighted mixture of mission profiles used to estimate the likely life usage in service.
Mission profile A definition of a sequence of flight manoeuvres for a particular type of mission.
Mission profile code An identifying code assigned to each of the types of mission flown by a particular aircraft
type. Also known as sortie-pattern code.
Mission severity factor An airframe fatigue damage indicator assigned to each mission profile code.
Model A physical, software or theoretical representation of some reality.
Module An engine assembly that is designated as a module by the manufacturer. Modules usually
have log cards.
Monitoring Measuring, recording and making decisions based on the recorded data-
Operational Relating to service use.
Partitioning Separating different types of failure mechanism, such as strain-range.
Parts life tracking Recording the life consumption of each lifed item on an individual basis.
Plausibility test A test which data must pass before it is accepted for use.
Reference cycle A stress-strain cycle for a stress feature. Derived from the mission mix, and used by the
designer as the basis for calculating damage and consumed life counts.
Reliability The probability of failure of an equipment or component, in a period. Normally expressed
as mean time between failures. Critical components are not supposed to fail in service.
 G-3

Therefore, their reliability cannot be proven and expressed in the same way.
Rainflow analysis A method of extracting usage cycles from a time history of turning points. The greatest
maximum is matched to the lowest minimum, considered as a cycle and extracted from the
time history. This process is repeated until all cycles have been identified.
Safety A measure of operational success, in terms of deaths or injuries per some operating period.
Safety critical A component, the failure of which, endangers the safety of the aircraft or people in the
component vicinity.
Snap-shot Collection of data at a defined point in the flight cycle.
Strain Deformation.
Strain, Elastic Temporary deformation that recovers when the load is removed.
Strain, In-elastic Permanent deformation.
Strain, Mechanical Overall deformation or elongation of a component, relative to its original length.
Strain, True Point or incremental deformation of material relative to the instantaneous incremental
length.
Stress A load applied to an area of material. Stresses are normally resolved into two components -
Normal and shear. They apply to all three axes
Stress feature A part of a component that may be the life limiting feature under some conditions. Some
components have several stress features which are the subject of intensive design effort.
Stress rupture Failure when a component has been statically loaded at an elevated temperature for a long
period
Stress, Normal The stress perpendicular to a surface of an elemental cube.
Stress, Shear The stresses lying in the plane of a surface of an elemental cube.
Thermal transient When the surface temperature of a component changes it takes some time for the
temperature distribution across the component to stabilise. The presence of thermal
gradients induces stresses. As the gradient changes, so do the stresses
Thermo-mechanical The materials data used to make life assessments, is isothermal data implying that the strain
behaviour is cycled at constant temperature. In reality, both the strain and temperature will vary
throughout the cycle and therefore consideration of the thermo-mechanical fatigue
behaviour is necessary. Large test programs are underway to understand the differences
between TMF and isothermal data. Early evidence suggests that in certain situations an
over-prediction of life is obtained and that in other areas an under-prediction is predicted
when using isothermal data.
Touchdown The moment when the wheels of an aircraft touch the ground.
Trend analysis A variation of rain-flow analysis, which seeks to reduce the complexity of the algorithm
with negligible effect on the accuracy of the result.
Type test A formal test-bed engine test that proves the basic safety of an engine prior to release into
service. _^_________
Usage monitoring Estimating and recording of the usage of a component. Frequently used to refer to life
monitoring
Verification Verification is proving the truth of an element of a system.
Validation Validation is proving the truth of a complete system.
 REPORT DOCUMENTATION PAGE

1. Recipient's Reference 2. Originator's References 3. Further Reference 4. Security Classification

of Document
RTO-TR-28 ISBN 92-837-1032-0 UNCLASSIFIED/
AC/323(AVT)TP/22 UNLIMITED

5. Originator Research and Technology Organization

North Atlantic Treaty Organization
BP 25, 7 rue Ancelle, F-92201 Neuilly-sur-Seine Cedex, France

6. Title
Recommended Practices for Monitoring Gas Turbine Engine Life Consumption

7. Presented at/sponsored by
the RTO Applied Vehicle Technology Panel (AVT).

8. Author(s)/Editor(s) 9. Date
Multiple April 2000

10. Author's/Editor's Address 11. Pages

Multiple 178

12. Distribution Statement There are no restrictions on the distribution of this document.
Information about the availability of this and other RTO
unclassified publications is given on the back cover.

13. Keywords/Descriptors
Equipment health monitoring Tolerances (mechanics)
Machinery health monitoring Gas turbine rotors
Aircraft maintenance Safety
Service life Turbine components
Life cycle costs Failure
Gas turbine engines Control equipment
Operations Standards
Military aircraft Structural integrity
Commercial aircraft Fracture mechanics
Aging (metallurgy) Fatigue (materials)
Damage Aging aircraft

14. Abstract
The Task Group analysed the use of life monitoring systems in modern engines (from 1990)
and in ageing fleets. The design and operational factors to be considered beforehand are
described.
Particular attention is paid to turbine disks. Regulatory requirements for safety standards are
considered. Civil military practices, maintenance policies and procedures, modes and mechanics
of service usage are covered as well as their influence on life consumption. Lifing procedures,
monitoring system verification and validation, operational management considerations and usage
monitoring approaches are dealt with.
 NORTH ATLANTIC TREATY ORGANIZATION

RESEARCH AND TECHNOLOGY ORGANIZATION

BP 25 • 7 RUE ANCELLE DIFFUSION DES PUBLICATIONS
F-92201 NEUILLY-SUR-SEINE CEDEX • FRANCE RTO NON CLASSIFIEES
Telecopie 0(1)55.61.22.99 • E-mail [email protected]
L'Organisation pour la recherche et la technologic de l'OTAN (RTO). detienl un slock limile de certaines de ses publications rf.ce.ntes, ainsi
que de celles de l'ancien AGARD (Groupe consultatif pour la recherche et les realisations aerospatiales de l'OTAN). Celles-ci pourront
eventuellement etre obtenues sous forme de copie papier. Pour de plus amples renseignements concernant l'achat de ces ouvrages,
adressez-vous par lettre ou par telecopie ä l'adresse indiquee ci dessus. Veuillez ne pas telephoner.
Des exemplaires supplementaires peuvent parfois etre obtenus aupres des centres nationaux de distribution indiques ci-dessous. Si vous
souhailez recevoir loutes les publicalions de la RTO, ou siniplement celles qui concement certains Panels, vous pouvez demander d'etre
inclus sur la liste d'envoi de l'un de ces centres.
Les publications de la RTO el de l'AGARD sont en vente aupres des agences de vente indiquees ci-dessous, sous fonne de photocopie on
de microfiche. Certains originaux peuvent egaleinenl etre obtenus aupres de CASI.
CENTRES DE DIFFUSION NATIONAUX
ALLEMAGNE FRANCE PAYS-BAS
Strcitkräfteaint / Abteilung HI O.N.E.R.A. (ISP) NDRCC
Fachinformationszentmm der 29, Avenue de la Division Leclerc DGM/DWOO
Bimderswchr, (FIZBw) BP 72, 92322 Chatillon Cedcx P.O. Box 20701
Friedrich-Ebert Allee 34 2500 ES Den Haag
D 53113 Bonn GRECE (Correspondant)
Hellenic Ministry of National POLOGNE
BELGIQUE Defence Chief of International Cooperation
Coordinatcur RTO VSL/RTO Defence Industry Research & Division
Etat-Major dc la Force Aerienne Technology General Directorate Research & Development Department
Quartier Reine Elisabeth Technological R&D Directorate 218 Nie|K)dleglosci Av.
Rue d'Evere, B 1140 Bmxelles D.Soutsou 40, GR 11521, Athens 00 911 Warsaw
CANADA HONGRIE PORTUGAL
Direcleur - Recherche et developpement Department for Scientific Estado Maior da Forca Aerea
Communications el gestion de Analysis SDFA Centre de Documentacäo
l'information DRDCGI 3 Instinite of Military Technology Alfraaide
Minislere de la Defense nationale Ministry of Defence P 2720 Amadora
Ottawa, Ontario K1A 0K2 H-1525 Budapest P O Box 26
REPliBLIQUE TCHEQUE
DANEMARK ISLANDE VTÜL a PVO Pralia /
Danish Defence Research Establishment Director of Aviation Air Force Research Institute Prague
Ryvanss Alle 1, P.O. Box 2715 c/o Flugrad Narodni informacni stiedisko
DK 2100 Copenhagen 0 Reykjavik obrannelio vyzkumu (NISCR)
ITALIE Mladoboleslavska ul., 197 06 Pralia 9
ESPAGNE
INT A (RTO/AGARD Publications) Centra documentazione ROYAUME-UNI
Carretera de Torrejön a Ajalvir, Pk.4 tecnico scientifica della Difesa Defence Research Information Centre
28850 Torrejön de Ardoz - Madrid Via Marsala 104 Kentigern House
00185 Roma 65 Brown Street
ETATS-UNIS Glasgow G2 8EX
NASA Center for AeroSpace LUXEMBOURG
Information (CASI) Voir Belgique TLRQUIE
Parkway Center NORVEGE Mill! Savunma Baskanligi (MSB)
7121 Standard Drive Norwegian Defence Research ARGE Dairesi Baskanligi (MSB)
Hanover, MD 21076 1320 Establishment 06650 Bakanliklar - Ankara
Ami: Biblioteket
P.O. Box 25, NO 2007 Kjeller

AGENCES DE VENTE
NASA Center for AeroSpace The British Library Document Canada Institute for Scientific and
Information (CASI) Supply Centre Technical Information (CISTI)
Parkway Center Boston Spa, Wetherby National Research Council
7121 Standard Drive West Yorkshire LS23 7BQ Document Delivery
Hanover, MD 21076 1320 Royaume-Uni Montreal Road, Building M-55
Etats-Unis Ottawa K1A 0S2, Canada
Les demandes de documents RTO ou AGARD doivent comporter la denomination "RTO" ou "AGARD" selon le cas, suivic du
numero de seric (par exemple AGARD-AG-315). Des informations analogues, telles que le title et la dale de publication sont
souhaitables. Des references bibliographiques completes ainsi que des resumes des publications RTO et AGARD figurent dans les
journaux suivants:
Scientific and Technical Aerospace Reports (STAR) Government Reports Announcements & Index (GRA&I)
STAR pent etre consulte en ligne an localisateur de publie par le National Technical Information Service
ressources uniformes (URL) suivant: Springfield
http://www.sti.nasa.gov/Pubs/star/Star.html Virginia 2216
STAR est edite par CASI dans le cadre du programme Etats-Unis
NASA d'information scientifique et technique (STI) (accessible egalement en mode interactif dans la base dc
STI Program Office, MS 157A doimees bibliographiques en ligne du NTIS, et sur CD-ROM)
NASA Lanaley Research Center
Hampton, Virginia 23681-0001
Etats-Unis
 NORTH ATLANTIC TREATY ORGANIZATION

RESEARCH AND TECHNOLOGY ORGANIZATION

BP 25 « 7 RUE ANCELLE DISTRIBUTION OF UNCLASSIFIED
F-92201 NEUILLY-SUR-SEINE CEDEX • FRANCE RTO PUBLICATIONS
Telefax 0(1)55.61.22.99 • E-mail [email protected]
NATO's Research and Technology Organization (RTO) holds limited quantities of some of its recent publications and those of the former
AGARD (Advisory Group for Aerospace Research & Development of NATO), and these may be available for purchase in hard copy form.
For more information, write or send a telefax to the address given above. Please do not telephone.
Further copies are sometimes available from the National Distribution Centres listed below. If you wish to receive all RTO publications, or
just those relating to one or more specific RTO Panels, they may be willing to include you (or your organisation) in their distribution.
RTO and AGARD publications may be purchased from the Sales Agencies listed below, in photocopy or microfiche form. Original copies
of some publications may be available from CASI.
NATIONAL DISTRIBUTION CENTRES

BELGIUM GREECE (Point of Contact) POLAND

Coordinates RTO - VSL/RTO Hellenic Ministry of National Chief of International Cooperation
Etat-Major de la Force Aerienne Defence Division
Quartier Reine Elisabeth Defence Industry Research & Research & Development
Rue d'Evere, B-1140 Bruxelles Technology General Directorate Department
Technological R&D Directorate 218 Niepodleglosci Av.
CANADA D.Soutsou 40, GR 11521, Athens 00-911 Warsaw
Director Research & Development
Communications & Information HUNGARY PORTUGAL
Management - DRDCIM 3 Department for Scientific Estado Maior da Forca Aerea
Dept of National Defence Analysis SDFA - Centro de Documentacäo
Ottawa, Ontario K1A 0K2 Institute of Military Technology Alfragide
Ministry of Defence P-2720 Amadora
CZECH REPUBLIC H-1525 Budapest P O Box 26
VTÜL a PVO Pralia / SPAIN
Air Force Research Institute Prague ICELAND INTA (RTO/AGARD Publications)
Närodm rnformacm stfedisko Director of Aviation Carretera de Torrejön a Ajalvir, Pk.4
obranneho vyzkumu (NISCR) c/o Flugrad 28850 Torrejön de Ardoz - Madrid
Mladoboleslavskä ul., 197 06 Pralia 9 Reykjavik
TURKEY
DENMARK ITALY Mill! Savunma Baskanligi (MSB)
Danish Defence Research Centro documentazione ARGE Dairesi Baskanligi (MSB)
Establishment. tecnico-scientifica della Difesa 06650 Bakanliklar - Ankara
Ryvangs Alle 1, P.O. Box 2715 Via Marsala 104
00185 Roma UNITED KINGDOM
DK-2100 Copenhagen 0 Defence Research Information
FRANCE LUXEMBOURG Centre
O.N.E.R.A. (ISP) See Belgium Kentigern House
29 Avenue de la Division Leclerc 65 Brown Street
BP 72, 92322 Chätillon Cedex NETHERLANDS Glasgow G2 8EX
NDRCC
GERMANY DGM/DWOO UNITED STATES
Streitkräfteamt / Abteilung III P.O. Box 20701 NASA Center for AeroSpace
Fachinformationszentrum der 2500 ES Den Haag Information (CASI)
Bunderswehr, (FIZBw) Parkway Center
Friedrich-Ebert-Allee 34 NORWAY 7121 Standard Drive
Norwegian Defence Research Hanover, MD 210764320
D 53113 Bonn
Establishment
Attn: Biblioteket
P.O. Box 25, NO-2007 Kjeller

SALES AGENCIES
NASA Center for AeroSpace The British Library Document Canada Institute for Scientific and
Information (CASI) Supply Centre Technical Information (CISTI)
Parkway Center Boston Spa, Wetlierby National Research Council
7121 Standard Drive West Yorkshire LS23 7BQ Document Delivery
Hanover, MD 21076-1320 United Kingdom Montreal Road, Building M-55
United States " Ottawa K1A 0S2, Canada
Requests for RTO or AGARD documents should include the word 'RTO' or 'AGARD', as appropriate, followed by the serial
number (for example AGARD-AG-315). Collateral information such as title and publication date is desirable. Full bibliographical
references and abstracts of RTO and AGARD publications are given in the following journals:
Scientific and Technical Aerospace Reports (STAR) Government Reports Announcements & Index (GRA&I)
STAR is available on-line at the following uniform published by the National Technical Information Service
resource locator: Springfield
http://www.sti.nasa.gov/Pubs/star/Star.html Virginia 22161
STAR is published by CASI for the NASA Scientific United States
and Technical Information (STI) Program (also available online in the NTIS Bibliographic
STI Program Office, MS 157A Database or on CD-ROM)
NASA Langley Research Center
Hampton, Virginia 23681-0001
United States
Printed by Canada Communication Group Inc.
(A St. Joseph Corporation Company)
45 Sacre-Cmir Blvd.. Hull (Quebec), Canada K1A 0S7