Standardization of Fretting Fatigue

Copyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
Downloaded/printed by
Universidad Del Valle (Universidad Del Valle) pursuant to License Agreement. No further reproductions authorized.
STP 1159
Standardization of Fretting
Fatigue Test Methods and
Equipment
M. Helmi Attia and R. B. Waterhouse, editors
ASTM Publication Code Number (PCN)

04-011590-30
As M
1916 Race Street
Philadelphia, PA 19103
Library of Congress Cataloging-in-Publication Data

Standardization of fretting fatigue test methods and equipment / M.
Helmi Attia and R. B. Waterhouse, editors.
(STP ; 1159)
Proceedings from a symposium held in San Antonio, Tex., Nov. 12-13, 1990.
"ASTM publication code number (PCN) 04-011590-30."
Includes bibliographical references and index.
ISBN 0-8031-1448-6
1. Materials--Fatigue--Testing--Standards--Congresses.
2. Fatigue
testing machines--Standards--Congresses.
I. Attia, M. Helmi
(Mahmoud Helmi)
II. Waterhouse, R. B. (Robert Barry), 1922III. Title: Fretting fatigue test methods and equipment.
IV. Series: ASTM special technical publication ; 1159.
TA418.38.$68
1992
92-17270
620.1' 126'0287--dc20
CIP
Copyright | 1992 AMERICAN SOCIETY FOR TESTING AND MATERIALS, Philadelphia, PA. All
rights reserved. This material may not be reproduced or copied, in whole or in part, in any printed,
mechanical, electronic, film, or other distribution and storage media, without the written consent of the
publisher.
Photocopy Rights
Authorization to photocopy items for internal or personal use, or the internal or personal use of specific
clients, is granted by the AMERICAN SOCIETY FOR TESTING AND MATERIALS for users registered
with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee i
of $2.50 per copy, plus $0.50 per page is paid directly to CCC, 27 Congress St., Salem, MA 01970; (508)
744-3350. For those organizations that have been granted a photocopy license by CCC, a separate system of payment has been arranged. The fee code for users of the Transactional Reporting Service is
0-8031-1448-6/92 $2.50 + .50.
Peer Review Policy

Each paper published in this volume was evaluated by three peer reviewers. The authors addressed all
of the reviewers' comments to the satisfaction of both the technical editor(s) and the ASTM Committee
on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors and the
technical editor(s), but also the work of these peer reviewers. The ASTM Committee on Publications
acknowledges with appreciation their dedication and contribution to time and effort on behalf of ASTM.
Printed in Baltimore, MD
July 1992
Foreword
In 1988, the ASTM Committee E-9 on Fatigue approved the formation of a Task Group on
Fretting Fatigue Testing to develop standards for the fretting fatigue test methods and equipment. This task group, chaired by one of the editors of this special publication (M. H. Attia)
has recognized the gravity of its responsibility and realized the need for an international cooperative effort to achieve its objective. As a first step towards this goal, the idea of organizing a
symposium on this subject matter was born.
This publication, Standardization of Fretting Fatigue Methods and Equipment, contains
papers presented at the Symposium of the same name in San Antonio, TX on 12-13 November 1990. The symposium was sponsored by ASTM Committee E-9 on Fatigue. Dr. M. Helmi
Attia, of Ontario Hydro Research Division, Toronto, Ontario, Canada and Dr. R. B. Waterhouse, of the University of Nottingham, Nottingham, UK, presided as symposium chairmen
and are the editors of the resulting publication.
The Cover
The photoelastic picture on the cover depicts the change in the stress field and the contact
pressure distribution at the fatigue specimen/fretting pad interface as a result of the change in
the height of the pad. The latter is usually chosen arbitrarily and as such, the variability in the
test results is not unexpected. It is hoped that the picture will capture the attention of those
involved with fretting fatigue testing to the necessity of standardizing the test specimens configuration, methods, and equipment.
The picture was obtained from the Fretting Laboratory, Mechanical Research Department,
Ontario Hydro Research Division.
Contents
Overview--M. H. ATTIA AND R. B. WATERHOUSE
A Historical Introduction to Fretting Fatigue--R. B. WATERHOUSE
OPENING PAPER
The Problems of Fretting Fatigue Testing--R. a. WATERHOUSE
13
FUNDAMENTAL ASPECTS OF FRETTING FATIGUE TESTING--CONCEPTUAL FRAMEWORK

Mechanisms of Fretting Fatigue and Their Impact on Test Methods
Development--o. w. HOEPPNER
23
Testing Methods in Fretting Fatigue: A Critical Appraisal--L. VINCENT,

Y. BERTHIER, AND M. GODET
33
Fretting and Contact Fatigue Studied with the Aid of Fretting M a p s - o. B. VINGSBO
49
Variables of Fretting Process: Are There 50 of T h e m ? - - J . DOBROMIRSKI
60
FUNDAMENTAL ASPECTS OF FRETTING FATIGUE TESTING--MECHANICS OF CONTACT

The Development of a Fretting Fatigue Experiment with Well-Defined
Characteristics--D. A. HILLS AND D. NOWELL
69
Determination and Control of Contact Pressure Distribution in Fretting F at i g u e- K. SATO
85
Fretting Fatigue Analysis of Strength Improvement Models with Grooving or

Knurling on a Contact Surface--T. HATTORI, M. NAKAMURA,
AND T. ISHIZUKA
101
Effect of Contact Pressure on Fretting Fatigue of High Strength Steel and

Titanium Alloy--K. NAKAZAWA,M. SUMITA, AND N. MARUYAMA
ll5
FRETTING FATIGUE T E S T I N G - - M E T H O D S AND EQUIPMENT
A Critical Review of Fretting Fatigue Investigations at the Royal Aerospace

Establishment--D. B. RAYAPROLU AND R. COOK
129
Fretting Fatigue in the Power Generation Industry: Experiments, Analysis, and

Integrity Assessment--T. C. LINDLEY AND K. J. NIX
153
Techniques for the Characterization of Fretting Fatigue Damage--c. RUIZ,

170
Z. P. WANG, AND P. H. WEBB
The Influence of Fretting Corrosion on Fatigue Strength of Nodular Cast Iron and
Steel under Constant Amplitude and Load Spectrum Tests--G. FISCHER,
V. GRUBISIC, AND O. BUXBAUM
Adaptation of a Servohydraulic Testing Machine to Investigate the Life of

Machine Components Operating under Fretting Conditions--J. LABEDZ
178
190
ENVIRONMENTAL AND SURFACE CONDITIONS
Improving Fretting Fatigue Strength at Elevated Temperatures by Shot Peening in

Steam Turbine Steel--Y. MUTOH, T. SATOH, AND E. TSUNODA
199
The Fretting Fatigue Properties of a Blade Steel in Air and Vapor Environments-D. YUNSHU, Z. BAOYU, AND L. WEILI
210
The Application of Electrochemical Techniques to Evaluate the Role of Corrosion

in Fretting Fatigue of a High Strength Low Alloy Steel--s. PRICEAND
217
D. E. TAYLOR
NONCONVENTIONAL MATERIALS AND TEST METHODS
ACSR Electrical Conductor Fretting Fatigue at Spacer Clamps--A. CARDON,

L. CLOUTIER, M. ST-LOUIS, AND A. LEBLOND
231
Fretting Fatigue of Carbon Fiber-Reinforced Epoxy Laminates--o. JACOBS,

K. SCHULTE~ AND K. FRIEDRICH
243
CLOSING PAPER
Fretting Fatigue Testing: Current Practice and Future Prospects for

Standardization--M. H. ATTIA
263
Author Index
277
Subject Index
279
STP1159-EB/Jul. 1992
Overview
Introduction
With the present state of knowledge, the fretting fatigue problem is commonly approached
empirically by testing the material/component under simulated conditions of contact and
environments. The extreme difficulty in performing fretting fatigue testing manifests itself not
only through the large number of process variables but also through their mutual interactions
and the self-induced changes in the tribological system. The discrepancy among published
data is, therefore, not surprising. The possibility and potential for improving the repeatability
of test data do, however, exist with proper and comprehensive understanding of the sources of
uncertainties.
Objectives
The main objectives of this symposium/publication are as follows:
1. Review the present state of knowledge and the current fretting fatigue testing practice.
2. Identify the areas of uncertainties in conducting fretting fatigue testing, including the
design of the test specimens, as well as the measurement and control aspects.
3. Identify the measures that should be taken to improve the repeatability of test results and
to minimize their dependence on the design of the test equipment.
4. Examine the future prospects for standardization, and identify the areas that warrant
further research.
This book will be useful to tribologists, physicists, and mechanical engineers who are
involved with fretting fatigue testing and those who are concerned with contact problems, particularly where fatigue and vibration are concerned, for example, in turbines, generators, aircraft, structures involving steel ropes, and so on. The paper presented by Hattori et al., for
example, shows how problems have been overcome in the design of steam turbines. Vincent
et al. and Vingsbo discussed the use of fretting maps for controlling the fretting fatigue damage
in practice. Other papers show the effectiveness of certain preventative measures such as surface treatment and cathodic protection in marine environments. The papers presented in this
publication cover the response of common-place materials, such as steel and aluminum, as
well as the less conventional materials such as composites.
Overview of the Papers of the Symposium

This special technical publication contains 20 papers written by renowned authorities in this
field. The opening keynote paper, presented by R. B. Waterhouse, provides a global overview
of the problems of fretting fatigue testing and presents the author's perspective and views on
the main issues that should be addressed in any attempt to standardize fretting fatigue testing.
In addition, a total of four invited keynote papers were also presented by Vingsbo, Hoeppner,
1
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
FRETTING FATIGUE TEST METHODS AND EQUIPMENT
Hills, and Vincent to simulate and set the stage for focused and fruitful discussion during the
symposium. The closing paper by Attia, the Chairman of the ASTM Task Group E9.08.02 on
Fretting Fatigue Testing, examines the future prospects for standardization in relation to the
current practice. The paper presents also the results of a survey in which the input was solicited
from 65 active researchers in various parts of the world.
This special technical publication reflects the trends and testing philosophy in ten different
countries and is, therefore, characterized by its international flavor. Apart from the opening
and closing position papers, the papers of this symposium are grouped in five sections:
FundamentaI Aspects of Fretting Fatigue Testing--Conceptual Framework

This section includes four papers that provide a conceptual framework for the mechanical
and physical interactions associated with the fretting fatigue process and testing. Following a
brief presentation of the historical evolution of the fretting fatigue testing, Hoeppner reviewed
the mechanism of fretting fatigue and the contributions that have been made in understanding
the crack nucleation and in characterizing the fretting fatigue damage. He underlined those
parameters that can be considered as mechanism controlling and presented the recent developments in micromechanical modeling. The paper concluded with the recommendation for
standards development and the identification of some areas that warrant further research.
Vincent, Berthier, and Godet applied their concept of "velocity accommodation" to the
fretting process and showed that the relative displacement and velocity difference between the
core of contacting solids are accommodated at different sites (the rubbing solids, their interface, or the surface screens) and according to different modes (elastic, rupture, shear, and rolling). Depending on the surface tensile stresses and whether adhesive welds break before crack
initiation, it was indicated that the material responds to fretting in three different ways: no
degradation, crack formation, and particle detachment. Since different material responses can
be observed during a single test, the authors stressed the importance of constructing "fretting
maps" to identify the material response to specific running conditions. To extend the velocity
accommodation and fretting maps concepts to fretting fatigue testing and to overcome the
classical problem of the dependence of the displacement amplitude on the body stress level,
the authors proposed a new "fretting-static fatigue" testing method. This method, which is
based on applying a constant body stress and controlling the slip amplitude independently,
requires a set of fretting maps to be produced for different loads, slip amplitudes, and number
of cycles. The authors proposed also a measure for the "severity" of the test, and outlined how
the design engineer can use these maps to identify and avoid fretting fatigue failures. In this
paper, some fundamental questions were raised, regarding the contact mechanics parameters
that govern crack initiation/propagation, and the significance of the drop in the fatigue
strength measured in fretting fatigue test machines. The latter issue was discussed in relation
to the formation/retainment of wear debris, and the effect of the machine stiffness.
The subject of fretting maps, which define the effect of the process parameters on the extent
of the stick, partial- and gross-slip regimes, was also discussed by Vingsbo. Using a simple
model of surface asperities in elastic contact with a' perfectly flat semi-infinite body under
cyclic loading, the author concluded that surface fatigue is promoted by fretting under mixed
stick-slip conditions, both in terms of cyclic stress concentrations and plastic deformation in
the contact zone. The author's view on establishing fretting maps for a given tribo-system to
control the fretting fatigue damage in practice is readily applicable to the design of a controlled
fretting fatigue testing system.
Perhaps the most difficult problem to be encountered in developing standards for a controlled and well-defined fretting fatigue test is handling the large number of process variables.
The popular list of variables, which was originally assembled by Collins in 1964, includes as
OVERVIEW
many as 50 variables! In reviewing the stress models, which were successfully used to predict
the fretting fatigue strength, Dobromirski argued that the vast majority of these variables,
which are not explicitly included in the stress models, can be treated as "secondary" variables
that influence the process through their effect on the "primary" variables. The latter is a short
list of three variables, namely, the coefficient of friction, the displacement amplitude, and the
contact pressure. The coefficient of friction was further singled out and identified as the main
primary variable. By re-examining a large sample of available fretting wear/fatigue data, from
this perspective, the author was able to use the coefficient of friction as a common denominator to explain the effect of various process parameters on the fretting fatigue test results.
Beyond the obvious benefit of reducing the list of variables to a manageable and practical number, Dobromirski's analysis should be taken one step further to alert all of us that the time has
come to treat the coefficient of friction as one of the parameters that should be measured in
fretting fatigue testing. It will be noted throughout this book that the emphasis on the critical
role of friction force is echoed by many others.
Fundamental Aspects of Fretting Fatigue Testing--Mechanics of Contact

This section includes four papers that deal with the theoretical aspects of the mechanics of
contact, and the application of numerical techniques; for example, finite-element and boundary-element methods to calculate the contact stresses. Experimental verification, using the
caustics method, is also presented. The authors maintained their focus on the main objectives
of this symposium and presented their analysis in terms of two important issues: the design of
the fretting pad/fatigue specimen and the method of applying the normal contact load.
The paper presented by Hills and Nowell is centered around the idea that specimen/pad
geometry should be amenable to a well-defined stress field and fracture mechanics analysis.
They highlighted the drawbacks associated with the flat-ended fretting pad; for example, the
singularities in the contact stress distributions and the difficulty in defining the slip-stick zones.
They recommended the adoption of a "cylindrical bridges against flat tensile specimens" configuration, since it allows changing the contact size while keeping constant normal load, as well
as controlling the normal and tangential contact forces independently. The paper deals with
some points of interest to those involved with the task of developing standards for fretting
fatigue tests, namely, the contact size threshold phenomenon and the nature of the distribution
of the coefficient of friction over the contact area.
Using the boundary element method, Sato studied the effects of clamping position (central
versus edge clamping) as well as the bridge height on the magnitude and the distribution of the
contact pressure at the specimen/bridge interface. The results of the plane-stress analysis of the
bending fatigue problem were validated experimentally, using the method of caustics. The
concept of"equivalent stress amplitude," as defined by Tresca's yield criterion, was proposed
by the author for estimating the fretting fatigue strength. From the S-N fretting fatigue test
results, it was established that the bridge height affects the fatigue life only under central clamping conditions (negative effect). The author was successful in interpreting these results in relation to the contact pressure amplitude, defined as half the difference between the compressive
and tensile contact pressures at the outer edge of the contact area. The paper was concluded
with the recommendation to use either central clamping when the bridge height-to-contact
length H/L ratio is unity, or to use edge clamping for fretting fatigue tests with other H/L ratios.
To improve the fretting fatigue strength, the author demonstrated a way of reducing the contact pressure amplitude through the machining of grooves in the fatigue specimen near the
end of the bridge.
The application of the boundary element method for calculating the contact pressure distribution and the concept of controlling it through grooving and surface knurling were also
discussed by Hattori, Nakamura, and Ishizuka. In this paper, the fretting fatigue limit was predicted using the fracture mechanics approach. These predictions were also verified experimentally. The paper addresses some interesting points in relation to the measurement and
modeling of the effective stiffness of the contact interface. The example given in the paper for
improving the fretting fatigue strength through optimization of the groove geometry (to cour~teract the negative notch effect with the positive effect associated with the rise in the threshold
stress intensity range factor) provides a methodology for designing the configuration of fretting
fatigue test specimens.
The effect of the average contact pressure on the fretting fatigue strength was further invesL
tigated experimentally by Nakazawa, Sumita, and Maruyama. The test results indicated that
the relationship between the fretting fatigue life and the contact pressure is influenced by the
stress amplitude. At low-stress amplitude (<20% of the 0.2% PS of high strength steel), this
relationship is nonmonotonic and passes through a minimum and then a maximum before
reaching a constant level. At high-stress amplitude (>40% of the 0.2% PS), the increase in the
contact pressure leads to a continuous drop in the fretting fatigue life. The authors reported
also the increase in the frictional stress amplitude with the increase in the contact pressure. For
the steel used, it has been indicated that the crack initiation sites shift from the middle portion
of the contact area to the outer edge as the contact pressure is increased. This observation is of
a particular importance to fracture mechanics analysts who usually assume that cracks initiate
at the contact edge.
Fretting Fatigue Testing--Methods and Equipment

In this section which includes five papers, the present state of the art in fretting fatigue testing
is reviewed, and the relative merits of various test methods are evaluated. A few recommendations were made regarding the adoption of commercial equipment, proven techniques and
experimental test rigs, as a starting point for standards development. Some interesting concepts and observations were also made, providing guidelines for conducting proper simulative
tests.
The fretting fatigue testing and research activity at the Royal Aerospace Establishment
(RAE) the U.K. was critically reviewed by Rayaprolu and Cook. Over the last 15 years, the
test methods and test variables at RAE were progressively changing to satisfy specific requirements and objectives. Four stages or test series were identified by the authors to reflect such a
change. The conventional fretting fatigue setup with a proving ring was used in the first test
series to investigate the effects of the pad span, contact load and body loading type on the
fatigue endurance. The second test series was motivated by the need for knowing the local
stresses induced by fretting in order to apply fracture mechanics models. Here, the frictional
force measurement was introduced. In the third stage, the experimental research effort was
directed towards identifying the separate effects of the contact, frictional, and body loads on
the fatigue process. Using a biaxial fatigue machine with phase linked actuators, a fourth series
of tests is being currently undertaken to examine the effect of cyclic load variations on the
cyclic frictional load, as well as crack initiation and propagation. The paper summarizes also
the work related to fracture mechanics modeling at RAE. Recommendations for standard test
setup, procedures, and future work were presented in the last two sections of the paper. To
improve the fracture mechanics prediction capability, the effect of the contact parameters on
crack initiation and growth, particularly with reference to initiation sites and angular and short
crack growth, was identified as an important area for further research. It is worth noting that
this recommendation is well founded by the observations made by Nakazawa et al.
The paper given by Lindley and Nix described the two fretting fatigue test methods used at
the National Power Technology and Environmental Centre in the United Kingdom. These
OVERVIEW
methods are similar to those used and recommended in the previous paper by Rayaprolu and
Cook, namely, the proving ring and the biaxial test rigs. The advantages of the latter system
were discussed in terms of controlling the contact load and the relative slip between the specimen and the pad, as well as applying variable amplitude loading. The paper describes also
alternative fretting pad geometries and emphasizes the requirements for frictional force measurement during the test. The two approaches of fretting fatigue analysis, the S-N curve and
the fracture mechanics modeling, were also reviewed.
For a proper simulative fretting fatigue testing, Ruiz, Wang, and Webb introduced the
fatigue-fretting damage parameter (FFDP), as a measure of the severity of fretting fatigue damage. This parameter is a function of the tangential stress along the line of contact, the interface
shear stress, and the relative slip and, therefore, includes the variables that control the initiation of fretting surface damage (wear) and the growth of the cracks. The main thrust of the
paper is centered around the importance of getting the three components of the FFDP right
in any test designed to reproduce the conditions prevailing in a real structural joint. The paper
discussed further the issue of controlling these variables in three types of tests: biaxial, tension/
compression, and 3-point bending tests. The authors pointed out the proper choice of the test
method, depending on the ductility of the material tested.
The paper presented by Fischer, Grubisic, and Buxbaum deals with a very important and
fundamental issue in fretting fatigue testing: the effect of load sequence. The experimental
study carried out by the authors on the fretting fatigue behavior of nodular cast iron under
constant amplitude and load spectrum (random sequence) throws the light on a few important
findings. First, the common test practice of constant stress amplitude produces more reduction
in the fretting fatigue strength because of higher slip amplitude and higher degree of "embedding." Second, the widely accepted notion of the negative effect of the contact pressure on the
fretting fatigue strength (under constant stress amplitude) cannot be extended to the case in
which the stress amplitude follows a random sequence. Third, the significant improvement in
the fretting fatigue strength with residual compressive stresses, for example, due to shot peening, was not observed in plain fatigue testing under spectrum load. Although these conclusions
cannot be generalized, at the moment, beyond the test conditions reported by the authors, they
demonstrate the importance of proper simulation of the loading conditions encountered in
practice and suggest the improved repeatability of the test results under random sequence
loading, even when the contact pressure and residual stresses are not precisely controlled and
defined.
Labedz's paper deals with the adaptation of commercially available servo-hydraulic testing
machines and the use o f a univeral test rig for fretting testing. The proposed test method is in
harmony with Dobromirski's concept of primary/secondary variables and considers only five
essential test variables. The author brings to our attention two test parameters that are usually
ignored in fretting wear/fatigue testing: the contact temperature and the residual stresses. The
effect of the latter was experimentally investigated to confirm its importance and to demonstrate the proposed test method.
Environmental and Surface Conditions

This section includes three papers that deal with the effect of surface residual stresses and
the environmental conditions (for example, temperature, vapor content, and corrosivity) on
the fretting fatigue test results. These papers point out the importance of monitoring and duplicating the environmental conditions and the state of stresses at the surface of the specimen.
Some experimental techniques, for example, X-ray diffraction, scanning electron microscopy
(SEM), atomic emission spectroscopy (AES), Mossbauer spectrometry, and electrochemical
techniques were described.
The effects of the compressive residual stresses and the environmental temperature on the
fretting fatigue test results were investigated by Mutoh, Satoh, and Tsumoda. Some considerations for testing and measuring frictional forces at elevated temperatures were discussed.
The paper examines also the relationship between the coefficient of friction and the stress
amplitude. It has been concluded that for the given test conditions, this relationship is unique
regardless of the temperature and the surface residual stresses. This behavior was attributed to
the insensitivity of the following mechanisms to surface and environmental conditions: oxidation (to temperature), and surface roughness and hardness (to shot peening).
The paper presented by Yunshu, Baoyu, and Weili focused on the effect of the environment
on the debris structure and its tribological properties. Using surface analysis techniques, the
authors concluded that if the environmental conditions promote the wear debris to act as an
effective solid lubricant, the fretting fatigue strength will be partially restored, as in the case of
blade steel fretted in vapor. They also concluded that the environmental effects become less
important in the presence of compressive stresses.
The paper presented by Price and Taylor is concerned with two issues: the synergistic effect
of the mechanical and electrochemical components of the fretting fatigue process and the
application of electrochemical techniques to separate and evaluate the role of corrosion in tests
run in aqueous environment. An experimental setup was developed to control the corrosivity
of the medium and to identify the electrochemical dissolution process through the use of
impressed cathodic protection. For the test conditions specified in the paper, the authors
concluded that the electrochemical processes have the greatest influence on the fatigue life of
high-strength low-alloy steel. The paper draws the attention to the requirement of assessing the
contribution of the corrosion action in fretting fatigue testing, and provides a method for
achieving that.
Nonconventional Materials and Test Methods
This section includes two papers that deal with nonconventional test configuration and
materials. The fretting fatigue testing system developed by Cardou, Cloutier, St. Louis, and
Leblond to test overhead electrical conductors is based on exciting the conductor at the span
midpoint, with a controlled cyclic deflexion. The concept of primary and secondary test variables was independently applied in this paper, and two test methods were followed, namely
the wire fracture time sequence and fracture location analysis.
In the paper presented by Jacobs, Friedrich, and Schulte, a special test setup was developed
to study the mechanism of fretting fatigue of carbon fiber reinforced expoxy (CFRE) laminates. In contrast to the observation made by Lindley and Nix, the authors found that the
fretting fatigue life of CFRE is significantly affected by the fretting pad material. This was contributed to the mechanism of interaction between fretting wear damage and fatigue, which is
also sensitive to the contact pressure and the hardness of the fretting pad material. The authors
established that the fretting fatigue mechanism of fiber reinforced polymers is characterized
by multiple matrix cracking along the fibers and, therefore, the available fracture mechanics
models are not applicable to these materials. A theoretical model for the "fretting fatigue load
versus number of cycles to failure" and the "specific pseudo-wear rate" was developed and
verified experimentally.
Acknowledgment
The editors are indeed grateful to the authors for their valuable and original contributions.
The effort of the reviewers in streamlining and improving the clarity of the presentation is
highly appreciated. Special thanks are due to Dr. R. Frishmuth, of the Vecto Gray Inc., HousCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
OVERVIEW
ton, TX, for his support and his instrumental role in forming the ASTM Fretting Fatigue Testing task group. A word of appreciation is also due to Messers. G. Clarke, D. B. Craig, and N.
S. D'Silva, of the Mechanical Engineering Department, Ontario Hydro Research Division, for
their support. The support of the Department of Materials Engineering and Materials Design,
University of Nottingham is indeed appreciated.
The editors would like to express their thanks to the officers and members of the ASTM
Committee E-9 on Fatigue for their support and also to the publication staffofASTM for their
patience and support that made this publication possible.
This publication is only one aspect of the symposium. The sessions and the discussions contribute greatly to the mission of the symposium. The effort of the co-chairmen of the sessions
is acknowledged and appreciated. The editors are thankful to the attendees of the symposium
for the interesting points and useful comments they made during the discussions that followed
the paper presentaion, and during the panel discussion session. Their enthusiasm to follow up
this symposium with similar conferences in the future is appreciated and well taken. The editors hope that those concerned with the subject of fretting fatigue will find this publication
useful and stimulating.
M. Helmi Attia
Ontario Hydro Research Division, Toronto,
Ontario, Canada; symposium chairman and
editor.
R. B. Waterhouse
Department of Materials, Engineeringand
Materials Design, University of Nottinghamsymposium chairman and editor.
R. B. W a t e r h o u s e ~
A Historical Introduction to Fretting Fatigue

REFERENCE: Waterhouse, R. B., "A Historical Introduction to Fretting Fatigue," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM STP ]15 9, M. Helmi Attia and
R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 89.
KEY WORDS: fretting fatigue, fatigue properties, historical perspective, crack propagation
Fretting was first reported by Eden et al. in 1911 [1 ] who found that brown oxide debris was
formed in the steel grips of their fatigue machine in contact with a steel specimen. It was not
until 1927 that Tomlinson [2] conducted the first investigation of the process and designed
two machines to produce small-amplitude rotational movement between two annuli in the
first case, and an annulus and a flat in the second. The movement was controlled by a long
lever system. Since the resultant debris on his steel specimens was the red iron oxide c~Fe20~,
which had arisen from chemical reaction with oxygen in the air, he coined the phrase "fretting
corrosion." He also established that the damage could be caused by movements with amplitudes as small as a few millionths of an inch ( ~ 125 nm) and the important fact that relative
movement had to occur, which he termed "slip."
The effect that fretting could have on fatigue properties was first investigated by WarlowDavies [3] in 1941, who produced fretting damage on the gage length of steel fatigue specimens
and found a subsequent reduction in fatigue strength caused by the pitting of the surface, of
between 13 and 17%. This was to be expected, but later investigations, particularly by McDowell [4] showed that the conjoint action of fretting and fatigue, which is the usual case in practice, was much more dangerous, producing strength reduction factors of 2 to 5 and even
greater. Fenner and Field [5] in 1958 demonstrated that fretting greatly accelerated the crack
initiation process. I published my first research paper in 1961 and showed that recrystallization
of the ferrite occurred in the fretted region when a bright drawn mild steel was subjected to
fretting fatigue [6]. The first major investigation was by Nishioka and Hirakawa who published a series of six detailed papers that were inspired by a problem encountered in the rolling
stock of the Shinkansen [ 7]. Subsequent experimental investigations have been based on their
valuable work. They also were the first people, together with Liu et al. [8], to attempt an analysis of fretting fatigue. This is an area that has seen great developments in the succeeding years
and forms a major part of this publication.
References
[ 1] Eden, E. M., Rose, W. N., and Cunningham, F. L., "Endurance of Metals," Proceedings of the Institute ( f Mechanical Engineers" Vol. 4, 1911, pp. 839-974.
[2] Tomlinson, G. A., "The Rusting of Steel Surfaces in Contact," Proceedings q/the Royal Society, A
Vol. 115, 1927, pp. 472-483.
Department of Materials Engineering and Materials Design, University of Nottingham, University
Park, Nottingham NG7 2RD, England.
8
Copyright
by ASTM Int'l
(all rightsASTM
reserved);International
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
www.astm.org
WATERHOUSE ON A HISTORICAL INTRODUCTION
[3] Warlow-Davies, F. J., "Fretting Corrosion and Fatigue Strength," Proceedings of the Institute on
Mechanical Engineers, Vol. 146, 1941, p. 32.
[4] McDowell, J. R., "Fretting Corrosion Tendencies of Several Combinations of Materials," Symposium on Fretting Corrosion, STP 144. American Society for Testing and Materials, Philadelphia,
1953, pp. 24-39.
[5] Fenner, A. J. and Field, J. E., "La Fatigue Dans les Conditions de Frottement," Rev. MOt., Vol. 55,
1958, pp. 475-485.
[6] Waterhouse, R. B., "Influence of Local Temperature Increases on the Fretting Corrosion of Mild
Steel," Journal of Iron and Steel Institute. Vol. 197, 1961, pp. 301-305.
[7] Nishioka, K. and Hirakawa, K., "Fundamental Investigations of Fretting Fatigue," Bulletin of the
Japan Society of Mechanical Engineers, Vol. 12, 1969, pp. 180-187,397-407, 408-414, 692-697;
Vol. 15, 1972, pp. 135-142.
[8] Liu, H. W., Corten, H. T., and Sinclair, G. M., "Fretting Fatigue Strength of Titanium Alloy RC
130B," Proceedings ofASTM, Vol. 57, 1957, pp. 623-641.
OpeningPaper
Robert B. Waterhouse I
The Problems of Fretting Fatigue Testing

REFERENCE: Waterhouse, R. B., "The Problems of Fretting Fatigue Testing," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM STP 1159, M. Helmi Attia and R.
B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 1319.
ABSTRACT: Fretting fatigue testing usually arises as the result of some failure which it is imperative to overcome either by a modified design or application of some surface treatment. In such
cases the test rig is usually designed to replicate the actual situation as closely as possible (e.g., a
press fit or riveted joint) with environmental conditions as near to those occurring in practice
(e.g., high temperature or a marine atmosphere). In laboratory testing the purpose is much wider,
and usually entails, for instance, the assessment of different materials for their susceptibility to
fretting damage, or the effect of variables such as clamping load, amplitude of slip and frequency,
in particular environmental conditions. The type of Fatigue test is very relevant (i.e., whether the
response is the same in rotating-bending, push-pull or torsion, and whether in the latter two
cases, a mean stress is applied). The nature of the contact is also material (i.e., whether it is flaton-flat or cylinder-on-flat, or even crossed cylinder). How is the clamping stress to be applied? If
by a proving ring, then the pressure can change according to whether the debris is trapped or can
escape; a dead weight method might be preferable. Surface finish and residual stress are factors
which must be taken into account. All these matters must be considered in devising a recommended testing procedure.
KEY WORDS: fretting fatigue, fatigue testing, fretting device, clamping pressure, slip amplitude, frequency
With fretting being the small amplitude oscillatory m o v e m e n t between two contacting surfaces, it is obvious that in machines and structures subjected to vibration its potential occurrence is to be c o m m o n l y expected. The production of wear debris, although a nuisance and
with possible long-term consequences, is not so serious as the initiation of fatigue cracks and
their subsequent propagation, where the m o v e m e n t arises from the cyclic stressing o f one o f
the c o m p o n e n t s or in the presence of a static tensile stress. S o m e of the failures initiated by
fretting fatigue have had tragic consequences (e.g., the loss of the C h i n o o k helicopter in the
North Sea in N o v e m b e r 1986) [I ]; others have had serious e c o n o m i c consequences [2], as in
the case o f the failure o f a power station generator rotor. M a n y failures reported in the literature have had less devastating consequences. These have involved the output shaft flange o f a
helicopter [3], turbine disc failure in a gas turbine aero-engine [4], failure of wire reinforcements in radial tires [5], steel ropes [6], the supporting joint of a railway line [7] and an artificial hip j o i n t [8] to n a m e a few. The investigation of such incidents has provided useful inform a t i o n for avoiding continuing failures, but frequently m o r e information is required which
can only be achieved by s o m e form o f testing.
Associate Reader, Department of Materials Engineering and Materials Design, University of Nottingham, Nottingham, NG7 2RD, United Kingdom.
13
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
14
The Object of Fretting Fatigue Testing

Where an immediate fretting problem has to be tackled, the usual experimental arrangement is designed to replicate as closely as possible the actual failure situation (e.g., a fly wheel
on a shaft where the author constructed a scale model of the system, a turbine blade/disc dovetail fixing [9], the rail failure mentioned above [10], or bridge suspension ropes [11]). One of
the great difficulties here is to ascertain and then reproduce the mechanical conditions of the
original joint, particularly the loading in the contact and the amplitude of movement. Sometimes these may be measured experimentally [ 12], or failing that, finite element analysis may
provide the answer. The environmental conditions should also be reproduced, e.g., temperature, humidity, seawater or salt spray, or industrial atmospheres. The purpose of the testing is
to examine the effects of such factors as modification to the original design, surface finishing
and surface treatments, such as shot-peening, surface rolling, anti-fret coatings, or even a
change of base material. Some manufacturers will try several or all possibilities together,
whereas, if time permits, it is more economical in the lbng run to pinpoint the optimum solution and hence to identify the source of the original problem. Of the options available,
improvement in design is the most satisfactory, but this may be disruptive and expensive.
The more general objective of fretting fatigue testing is the assessment of the susceptibility
of different materials to this type of failure. In the case of metallic alloys this may involve consideration of such factors as stacking fault energy heat-treatment (particularly in the case of
steels whether tempering or normalizing is to be preferred) whether cold worked or annealed,
forged or cast, and, of course, hardness. Further to this is the combination of different materials, since contacts of the same material have been thought inadvisable, as local welding is
more likely. Also important is the effect of mechanical variables, such as contact pressure,
amplitude of slip, frequency, the influence of random or sinusoidal loading, the effect of residual stresses and the nature of the surface finish. Finally, a systematic study of environmental
influences may be necessary, particularly the temperature, whether high or low, the composition of the gaseous or liquid environment and possible fluctuations thereof.
Practical Considerations
Specimen
The type of specimen is dictated to some extent by the choice of fretting contact, discussed
below. The most c o m m o n contact is flat-on-flat or cylinder-on-flat. This means that the specimen generally has to have a gauge length with parallel flats. This also means that the specimen
must itself be machined from plate or sheet material, or if the specimen is of circular cross
section, parallel flats must be machined on the gauge length. Typical examples are shown in
Fig. 1. The existence of corners means that there are stress concentrations present. In practice,
it is usually found that the fretting is much more severe and cracks initiate there rather than
POUSH
LONGITUDINALLY
SECTION A - A
FIG. l--Design of specirnen with machinedflats.

WATERHOUSE ON PROBLEMS OF FRETTING FATIGUE TESTING
15
at the corners of the specimen. Attention must be paid to the type of grips holding the specimen, since fretting is often experienced in them. This can usually be overcome by using screw
threads or applying an anti-fret lubricant or some form &insert.
Fretting Contact
Fretting contact is usually provided by some form of bridge. A pair of these is clamped on
to the gauge length by a proving ring, which is strain gauged to allow adjustment of the clamping pressure. In a push-pull machine, one end of the bridge can be located on the grips. In this
case the amplitude of slip is determined unequivocally, whereas, in the isolated bridge, it is
usually assumed that the slip is equally distributed between the two feet of the bridge. Provided
that loading is the same on each foot, experimental observation of the damage supports this
assumption. Figure 2 shows two possibilities. If the specimen is vertical, as in many servohydraulic machines, a dead-loading arrangement may be possible to apply the normal load.
In rotating-bending types, the proving ring must be carefully balanced to prevent vibration.
The feet of the bridge may be flat and sufficiently low in height to minimize elastic deformation
and allow relative slip to occur. They may also be chamfered or cylindrical. Cylinder/flat con-
SPiCIMEN
II
8~i~D~ ~
RSOWNGR~N~
"
GAUGE
LENGTH
(a)
ELEVATION
GRIP
BRIDGE
SPECIMEN
pLAN
~fC
BRIDGEA L2
SPECIMEN
(b)
FIG. 2--Arrangement of specimens and fretting bridges"(a) unlocated (b) located on grips.
16
tacts allow stress distribution to be calculated initially, but as soon as any wear occurs, the
contact situation is changed.
The disadvantage of the proving ring is that if debris is formed and retained in the contact,
since it is usually of greater volume than the metal from which it arises, the pressure will
increase. On the other hand, if the debris can escape, there will be a decrease in pressure. Thus,
pressure needs to be checked regularly. Choosing the material of the bridge the same as that of
the specimen eliminates a complicating variable and possibly gives the most severe type of
fretting damage. The length dimension of the bridge will determine the range of amplitude
slip, which is governed by extremes in the cyclic stressing of the specimen.
Normal Load
The effect of normal load on the fretting fatigue strength is of the form shown in Fig. 3. It
would seem advisable to choose a load which is within the horizontal part of the curve. Therefore, slight variations, due to the effects of debris, will not be significant. Too heavy a load may
result in digging in if the bridge has sharp corners. A recent survey of the published literature
by the author indicated that in steel specimens and flat-footed bridges the range of clamping
pressure was mainly between 20 and 1,90 MPa.
Amplitude of Slip
In most of the investigations the amplitude of slip has not been constant, but has depended
on the range of alternating stress, and hence alternating strain, in the specimen. This is the
usual situation in practical cases of fretting fatigue. The range of slip amplitude can be imposed
by suitable choice of the length dimen,;ion of the bridge related to the stress range under consideration. An elaborate arrangement was developed by Nishioka and Hirakawa to allow the
amplitude of slip to be kept constant whatever the stress in the specimen [14].
Most damaging range of slip is between 18 [15] and 25u [16], although Lindley and Nix
have suggested that coefficient of friction is a more important factor [17]. A typical plot of
300
(..9
z
i.iJ
rr"
I.-
200
la_
100
100
200
CLAMPING PRESSURE, MPa
FIG. 3--Fretting Jatigue stren~h versus clamping load.

17
0.8
0.6
_o
I.- 0.4
Z
iii
~ 0.2
5
10
15
AMPLITUDE OF SLIP, p.m
20
FIG. 4--Co~,~cienl (~/i/riction versus amplitude of slip.
coefficient of friction against amplitude of slip is shown in Fig. 4 as the transition from partial
to total slip.
Type of Testing
In a push-pull machine, a non-zero mean stress can be applied. There is evidence that mean
stress has a somewhat greater effect in fretting fatigue than in plain fatigue [ 18]. In plane bending it is usual for one end of the fretting device to be located in the grips. Since a parallel sided
plane specimen will have a varying bending moment along its length, it will be important to
locate the fretting contact at exactly the same point in comparable tests. A shaped specimen
would overcome this to some extent, provided that the apparent contact areas were kept reasonably constant. In addition, there has been little consideration of torsional fatigue testing in
fretting fatigue.
The question of frequency of the alternating stress should also be considered. Higher frequencies allow testing time to be cut down. However, possible heating of the specimen and
fretting contact have to be borne in mind, particularly as chemical interaction with the environment is an important feature of fretting.
Conclusion
In devising a standard test for fretting fatigue it would seem advisable to stipulate a particular
type of specimen and fatigue testing facility. An agreed type of fretting bridge is also necessary.
Since the coefficient of friction is an important factor, the fretting bridge should, if possible,
be strain gauged to allow for its measurement. The normal load applied should be related to
the yield pressure o f the material. The amplitude of slip should be limited, since at higher values the wear process predominates and the effect on fatigue is reduced, as illustrated in Fig. 5.
Surface finish and surface residual stress need to be stipulated or, at least, recorded. For testing
in normal atmosphere, the temperature and humidity should be recorded and, if possible, held
within certain limits.
The question remains whether the information gathered from such testing can be applied
to a particular practical situation. A recent paper suggests that results of the fatigue limit of
18

i
10
Stick
Slip
Gross Slip
-14
("'"--
~E
10
-16
I
!
!
!
t
I
I
I
!
I
I
!
I
!
!
I
E
z
10- i s
,~
iI
Mixed
1
=
and
I
10
30
100
//~"R
eoi't "r~
I sliding
I
I
I
!I
I
I
I
I .Illl,
I
!
I
l
I
I
!
l
h
300
107
~9
E
Z
10 s
.J
I,
105
1000
FIG. 5--Relation between wear and.fatigue strength as a function of amplitude qf slip (Srderstrom and
Vingsbo).
specimens determined in plane b e n d i n g were equal to those obtained in a shrink-fitted specim e n tested in rotating b e n d i n g [19].
References
[1] Department of Transport Aircraft Accident Report 2/88 H.M. Stationery Office 1989.
[2] Lindley, T. C., McIntyre, P., Snow, D. J., iand Wilson, J. D., "Fatigue and Environmental Cracking
in Turbo-generators," Proceedings, Sixth Thermal Generation Specialists Meeting, Madrid, 1981.
[3] Eckert, J. and Richter, R., "Fretting Corrosion of an Output Shaft Flange," Prakt. Met., Vol. 2 l,
1984, pp. 140-143.
[4] Lindblom, T. and Burman, G., "Fatigue Failure under Fretting Conditions," Proceedings, Conference on High Temperature Alloys for Gas Turbines, Liege Belgium 4-6 Oct. 1982, Riedel Publishing
Co., Dordrecht, 1982, pp. 673-684.
[5] Seitz, N. and Schmid, R., "Korrosion bei PKW-Stahlgurtelreifen," Kautschuk + Gurnrni-Kunstoffe,
Vol. 40, 1987, pp. 20-27.
[6] Hobbs, R. E. and Ghavani, K., "The Fatigue of Structural Wire Strands," Int. J. Fatigue, 1982, pp.
69-72.
[7] Okazaki, A., Urashima, C., Sugino, K., Matsumoto, H., and Hattori, M., "Upper Fillet Crack in
Bolted Joint of Rails and its Causes," Transactions t?fl.S.1.J., Vol. 23, 1983, pp. B22.
[8] Smethurst, E. and Waterhouse, R. B., "Causes of Failure in Total Hip Prostheses," J. Mat. Sci., Vol.
12, 1977, pp. 1781-1792.
[9[ Ruiz, C. and Chen, K. C., "Life Assessment of Dovetail Joints between Blades and Discs in AeroEngines," Fatigue of Engineering Materials and Structures, I.Mech.Eng., London, 1986, pp. 187194.
[10] Urashima, C., Sugino, K., Nishida, S-I., and Matsumoto, H., "Factors on the Upper Fillet Crack
Initiation and its Preventive Measures," Transactions ofl.S.I.Z, Vol. 23, 1983, pp. B23.
[11] Blakeborough, A. and Cullimore, M. S. G., "Fretting in the Fatigue of Wire Rope," Advances in
Fracture Research, 6th Int. Conf. on Fracture, New Delhi, India, 4-10 Dec. 1984, Pergamon, New
York, Vol. 3, 1984, pp. 2133-2141.
[12] Fisher, N. J. and Ingham, B., "Measurement of Tube-to-Support Dynamic Forces in Fretting-Wear
Rigs," Journal of Pressure Vessel Technology, Transactions of ASME, Vol. 111, 1989, pp. 385-393.
19
[13] Waterhouse, R. B., "The Role of Adhesions and Delaminations in the Fretting Wear of Metallic
Materials," Wear, Vol. 45, 1977, pp. 355-364.
[14] Nishioka, K. and Hirakawa, K., "Fundamental Investigations of Fretting Fatigue Pt.2. Fretting
Fatigue Testing Machine and Some Test Results," Bulletin ofJ.S.M.E., Vol. 12, 1969, pp. 180-187.
[15] Fenner, A. J. and Field, J. E., "La Fatigue dans les Conditions de Frottement," Rev. M~t., Vol. 55,
1958, pp. 475-485.
[16] Gaul, D. J. and Duquette, D. J., "The Effect of Fretting and Environment on Fatigue Crack Initiations and Early Propagations in a Quenched and Tempered 4130 Steel," Met. Trans. A., Vol. 11A,
1980, pp. 1555-1561.
[ 17] Nix, K. J. and Lindley, T. C., "The Influence of Relative Slip Range and Contact Materials on the
Fretting Fatigue Properties of 3.5 NiCrMoV Rotor Steel," Wear, Vol. 125, 1988, pp. 147-162.
[18] Fenner, A. J. and Field, J. E., "A Study of the Onset of Fatigue Damage Due to Fretting," Transactions, N.E. Coast Instn. Engrs. and Shipbldrs, Vol. 76, 1960, pp. 184-228.
[19] Gotoh, Y. and Ohuchida, H., "Effect of Corrosive Environment on Fretting Fatigue Under Plane
Bending," Journal of the Society of Materials Science, Japan, Vol. 38, 1989, pp. 816-822.
Fundamental Aspects of Fretting Fatigue

Testing--Conceptual Framework
D a v i d W. Hoeppner I
Mechanisms of Fretting-Fatigue and Their

Impact on Test Methods Development
REFERENCE: Hoeppner, D. W., "Mechanisms of Fretting Fatigue and Their Impact on Test
Methods Development," Standardization of Fretting Fatigue Test Methods and Equipment,
ASTM STP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing
and Materials, Philadelphia, 1992, pp. 23-31.
ABSTRACT: At the beginningof this century very little information was available related to the
phenomenon of fretting and fretting-fatigue. However, significant progress has been made in
developing an understanding of the mechanisms of fretting-fatigue in this century. Progress
toward developing a holistic view of the process of fretting fatigue is presented.
Contributionsthat have been made in understanding nucleation of fretting-fatiguedamage are
reviewed. As well, characterization of damage is discussed related to improving our understanding. Those parameters that are viewed as mechanism controlling, thus important to test methods
development, are reviewed. The progress made in micromechanical modeling is also reviewed.
The paper concludes with recommendations for the standards development group to consider
in relation to test methods and some suggestions for future research and development.
KEY WORDS: fretting-fatigue, fretting corrosion, fretting wear, mechanisms, modeling,
standards
Fretting-fatigue has been a nemesis to designers, manufacturers, and operators of equipment of all types since humans first used machines to their advantage. It has been in the 20th
century that our progress at identifying, characterizing, and designing for prevention, alleviation, and controlling fretting fatigue has become more formalized. Furthermore, we have
characterized both the mechanisms and parameters that are involved in the fretting process.
However, no standardized procedures, at least within ASTM auspices, have been developed
to aid engineers at either the prospective or retrospective design stage. This continues to be a
major technological challenge. This symposium, as well as previous ASTM symposia concerned with this subject [ 1-3], attempts to focus attention on this problem. As a small portion
of this focus, this paper discusses some aspects of mechanisms that have been revealed during the recent decades. Subsequently, the role of various parameters in influencing these
mechanisms is discussed. Finally, some conclusions and suggestions for future research are
presented.
Occurrence of Fretting Fatigue
Numerous publications have documented the occurrence of fretting or potential occurrence
of fretting in any mechanically fastened joint or in surfaces in contact under "small" relative
motion. If one or both of the contacting surfaces are under cyclic load in addition, then one
1Professor and Chair, Mechanical Engineering Department, University of Utah, Salt Lake City, UT
84112,
23
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
24
or both o f the m e m b e r s m a y experience fretting fatigue. Reference 4 d o c u m e n t s m a n y specific

c o m p o n e n t s where fretting fatigue occurs. S o m e are in the following systems:
9 Helicopters.
9 Fixed-wing aircraft.
9 Trains.
9 Ships.
9 Automobiles, trucks, buses.
9 F a r m machinery.
9 Engines.
9 Construction equipment.
9 Orthopaedic implants.
9 Artificial hearts.
9 Rocket m o t o r cases.
9 Wire rope.
9 Etc.
W. Barrois, a leader in design in the aircraft industry and North Atlantic Treaty Organization-Advisory G r o u p for Aerospace Research and Development-Structures and Materials
Panel ( N A T O - A G A R D - S M P ) , wrote the following in 1970 [5]:
Until about 1940 fatigue investigations included: some basic research into the physics of metals in
an effort to discover its mechanism; systematic rotating-bending tests on smooth conical or waisted
specimens in order to qualify each metal by identifying its fatigue limit; finally, check tests on actual
parts.
Between 1940 and 1945 it became obvious to most specialists that the rotating-bending test on
smooth specimens was not representative of the behaviour of actual components and that the fatigue
limit was not the only characteristic of interest. One improvement consisted in performing axial tension-compression tests on cylindrical specimens having a V-groove with a rounded root, the smooth
specimen being replaced by a waisted one, with a very large radius of curvature, so that the heat
induced by internal damping should no longer increase the temperature of the specimen and falsify
the test results. Furthermore, it was recognized that the purpose of fatigue testing is not to specify the
fatigue limit, which is often hypothetical, but to provide the entire stress versus number of loadings
curve from static strength, corresponding to one load application, up to a large number of cycles in
laboratory tests of short duration. This number being large in comparison with the cycles sustained
by the structure during its service life, testing had to be speeded up as much as is practicable without
distorting the results.
However, a factor o f great importance in many servicefatigue incidents was left out in the notchedspecimen test: the contact alteration by friction due to very small relative displacements of the various
parts o f an assembly during the loading and unloading cycles. This phenomenon, known as "fretting,"
consists in the welding o f asperities on the surfaces in contact and in the tearing-up o f these microwelds; it is responsible for the initiation offatigue cracks in assemblies and, in cases like the fitting o f
wheels on shafts or the bearing o f bolts in lugs, it may reduce theJatigue life to a tenth ~f what it would
otherwise be. 2
After 1970, n u m e r o u s publications and conference proceedings emerged that provided
additional focus on the challenge o f fretting-fatigue prevention, control, and estimation [62 Italics mine.
HOEPPNER ON MECHANISMS OF FRETTING FATIGUE
25
19]. The seminal book of Waterhouse [7] has provided much insight into fretting-fatigue.
Another book, edited by Waterhouse [i8], focused attention on the engineering challenge
related to anticipation of fretting-fatigue in a prospective design as well as the prevention, prediction, and control of fretting-fatigue. Although some of the mechanistic understanding of
fretting-fatigue has been applied to the repair of such damage, this is still a very great challenge.
The role of modeling the fretting damage is vital here and fracture mechanics has been coupled
with fretting-fatigue and studied for many years. This aspect will be briefly discussed in a later
section. The intensive efforts of the past 20 years have made the mechanisms by which frettingfatigue proceeds much clearer. It has been recognized for many years that a systems view is
needed to study and design for fretting fatigue prevention since there are so many parameters
involved. References [20-30] have presented the need for a systems view in studying fretting
fatigue. The work by Czichos reported in [15] and expanded in [30] is a significant help to all
engineers and scientists embarking on the journey to help develop fretting-fatigue standards.
The first step in the systems view is to understand why we do fretting-fatigue experiments.
Figure 1 presents a simplified view of the reasons. Basically, the designer is interested in either
the determination of a fretting-fatigue life reduction factor (sometimes part of a "joint design
allowable") or in evaluating a fretting protection system. Additional experimentation often is
done to verify a fretting-fatigue prevention scheme.
The systems view of fretting-fatigue is presented in Figs. 2 and 3. The efforts of numerous
investigators have indicated that designing to resist fretting fatigue is an extremely difficult
HISTORICAL
PERSPECTIVE
ON FRETTING STUDIES
i
Purpose
Daiage
Stimulate Fastening
and/or Contact
Conditions
i
Design Oriented
Design
Oriented
Design
AIl~
Simulate
Mechanistic
Verification Life ~Tlesting

/
I
Joint
and/or
Contact
Detail
I
Alleviation/
Prevention
System
Develop
Understanding ~
of Fundamentals
L Models
FrettingFatigue
Life
Reduction
Factor
Standardize
Test Methods
f:-- Allowables
Fretting-Fatigue
Develop
Alleviation/
Prevention System
Develop Inspection
and Monitoring
System
k.
FIG. 1--Reasons.[or conductingfrening-]atigue experimental studies.

26

A component is
subjected to cyclic loads
Normal pressure, P N
[
Contact
Point ]
Une /
,II
I
171ii.
AP
8 = Relative slip
amplitude
9
9
9
9
9
9
constant load amplitude

variable load amplitude
frequency
waveform
mean stress
residual stress
FREI"rlNG-FATIGUE OF A COMPONENT
Obtain fatigue response for conditions of interest (coi)
9
9
9
9
9
9 8 , PN "magnitude, frequency
9
9
9
9
material compatibility
friction
temperature
environment
stress state
geometric detail
material
surface condition
failure criteria
FIG. 2--A conceptual view of#retting-fatigue emphasizing the numerous parameters.
FRE'I-FING-FATIGUE]
=l Environment
v Chemistry /
temperature
Contact
surfacesI
Basic Material ~
Mechanical Deformation ]
[ Response to environment,
1
combined contact and
cyclic mechanical deformation
I Normalload
Chemical or
L,
electrochemical factors
Magnitude
amplitude
of relative
displacement
Surface stress
Frequency
T, environment
Material
compatibility j
Friction I
Potential
Current density
Passivity
Oxide
Time
Pitting
Dissolution
Embrittlement
Film formation
" t Cyclic Loading ] -
(Stress State
l
I Stress range
I Stress amplitude
]Frequency
I Sequence of loading (spectrum)
t,Time/Waveform effects
Product form, thickness,
geometry, inspectability
='
FIG. 3--A systemsframework previously proposedjbr planning experimental studies q[[?etting-[atigue.

27
task. Part of this difficulty results from a lack of understanding of the mechanisms [7,9,10]
and the lack of standardized fretting fatigue apparatus, definitions, and test methods. Thus,
this paper will focus additional attention on this significant challenge. To assist with this, a
clearer view of the mechanisms will be helpful.
M e c h a n i s m s of Fretting-Fatigue
The National Materials Advisory Board (NMAB) report [4] provides an extensive review
of fretting fatigue mechanisms and is recommended to all who have an interest. Numerous
other studies have been reported in the literature. Over the years, the understanding of the
mechanisms of fretting fatigue have been reduced, by the author and others, to the following
phases of fretting fatigue. Aspects of this approach are summarized in Ref31 as well as numerous other writings.
Surface or
Crack Formation/
Crack Propagation
Near Surface ~
Nucleation
~
Damage
II
Instability
III
IV
Furthermore, intensive studies by many investigators have shown that the mechanisms in
Phase I, undoubtedly the most criticalphase offretting-fatigue, and the most difficult to study,
result in "damage" that is summarized in Table 1. It has also been found that numerous
parameters influence the "damage" formation. The role of numerous parameters on frettingfatigue mechanisms is below summarized by Mann [32]:
Consequently, all parameters which may affect the generation of the service induced surface
"damage" must be reproduced in the laboratory as closely as possible, if there is to be any hope of
developing transfer functions relating the experiments to "reality."
These parameters are discussed below beginningwith the loading related parameters already mentioned. Along with each are comments indicating the effect of the variable on the fretting fatigue
mechanisms and the component life under conditions meeting the requirements for fretting fatigue.
Cyclic Load (Stress)--Provides energy input for both crack nucleation and subsequent propagation; generates slip between the faying surfaces with the amplitude depending on the level of cyclic
stress; increases the cyclic stress, reducing life; and gives a lower frequency which can reduce life by
providing more time per cycle for corrosion (fretting interactions).
TABLE 1--"Damage"produced underJkettingz/btigue conditions.

9
9
9
9
9
9
9
9
9
Pits
Oxide and Debris (Third Body)
Scratches--"Wear Tracks"
Metal Transfer
Extensive Surface Plasticity
Subsurface Cracking
Metal Transfer (More General--Material)
Cracks of Various Geometries at Various Angles to the Surface, Including Parallel
Fretting Craters
28
Normal Load (Stress)--Generates contact stresses in body of fatigue specimen; provides energy
input for production of most surface damage mechanisms; and increases normal stress, usually lowering life, although, if sufficient stress is applied such that slip amplitude is reduced or eliminated,
can increase life.
Slip Amplitude--Results from applied load; its presence undoubtedly plays a role in surface damage mechanisms; and it usually increases the slip amplitude, reducing fatigue life, but in some cases
higher slip amplitudes are associated with increased life due to the wearing away of nucleated
cracks.
Number of Fretting Cycles--As the number of fretting cycles increases, the amount of surface
damage produced increases; for some conditions a fretting fatigue damage threshold exists in terms
of the number of fretting cycles; continued fretting beyond the damage threshold produces no further
reduction of fatigue life; and this clearly relates to the energy threshold for crack nucleation alluded
to earlier and will definitely depend on the type of damage generated.
Geometry of Mating Components--This in part controls state of stress in components and, consequently, affects crack growth thresholds and crack growth rates; extremely important in retaining
debris between contacting surfaces, which in turn affects surface damage mechanisms (detrimental
to life by generating pitting and enhancing Mode I crack growth rates); and retained debris affects
friction forces and stress sta~e.
Elasticity--This affects both stress state and slip amplitudes.
Hardness--Generally, harder surfaces resist fretting damage, however, they also reduce the toughness of the surface layer which lowers the threshold energy requirement to nucleate fretting fatigue
cracks; this may or may not offset the advantage of the harder surface.
Microstructure--This can control near surface crack growth and crack nucleation via grain size,
hardness, etc.
Combinations--Similar metal contacts promote welding of asperities, enhancing debris production and reducing fatigue life; galvanic cell considerations are important; and friction and thus tangential surface stresses in fatigue component can be dramatically affected.
Surface Roughness--"Rough" surfaces can provide escape routes for debris (increasing life) or
also encourage gouging, scratching, and debris production which is detrimental to fatigue life.
Environment--Consideration must be given to temperature, humidity, and corrosive atmosphere; and fretting fatigue life normally longer in vacuum due to absence of oxygen (which otherwise
can form oxides harder than parent material).
It should be noted that under each parameter, e.g., cyclic stress, normal stress, slip amplitude,
materials, and environment, there are offsetting or opposing phenomena as the magnitudes or conditions of the controlling parameters vary. This clearly is due to the interactions and synergisms of
these variables and therefore the complexity of the subject. Consequently, it becomes of paramount
importance to view fretting fatigue deliberations from a systems context.
F r o m M a n n ' s analysis, and the m a n y works on mechanisms, this need for a systems framework is essential. Thus, to develop standards in fretting-fatigue, a systems framework and even
greater coordination o f activity will be required than in d e v e l o p m e n t of previous fatigue
related standards.
The recognition that a fretting-fatigue damage threshold exists has had an influence on the
evolution o f both mechanistic and engineering understanding. In addition, this d e v e l o p m e n t
has had a bearing on the research done to characterize the m a n y parameters that are involved.
However, a great deal of research still needs to be done to fully characterize the influence of
the m a n y parameters in the pre-threshold domain.
The discovery o f the threshold also has had an influence on the d e v e l o p m e n t of fracture
mechanics based models for fretting-fatigue, as briely discussed in the next section. Even
though significant progress has been m a d e since the damage threshold, m u c h controversy over
the controlling parameters still exists. This is dramatized by m a n y of the ideas in other papers
o f this symposium.
HOEPPNER
ON MECHANISMS
OF FRETTING
29
FATIGUE
Modeling
Subsequent to the presentation of the conceptual framework for a fretting fatigue damage
threshold in [9,10], numerous publications [33,34] focus on the utilization of fracture
mechanics based modeling. These have all focused around the development of a methodology
as shown in Fig. 4. The phases of this process that deal with crack propagation can be dealt
with to an extent. However, the phases on the left, i.e., fretting "damage" production and formation of cracks, and structurally dependent crack propagation are still very difficult to deal
with because &limited knowledge in those areas. Part of the difficulty in studying these areas
is the lack of standards concerned with fretting-fatigue.
Future Needs
Based upon the extensive progress made in understanding the mechanisms of fatigue and
characterizing the parameters that influence fretting fatigue, certain needs become evident.
First, clear definitions of all terminology are essential. Even though we have developed definitions of fretting (Terminology Relating to Erosion and Wear, ASTM G 40), fretting corrosion (Definition of Terms Relating to Corrosion and Corrosion Testing, ASTM G 15), fretting
wear (ASTM G 40), and fatigue (Definition of Terms Relating to Fatigue, ASTM E 1150),
further unification of these definitions focusing on fretting fatigue is essential. In addition, definitions of fretting fatigue, damage, induced cracks, damage threshold, etc. are essential to
improving the uniform reporting of information on fretting fatigue.
Second, a compilation of terminology specifically concerned with fretting fatigue would aid
our efforts a great deal. It also would be of help to develop a uniform format (protocol) for
reporting fretting fatigue experimental results.
I Modeling I
cl~Slress
I [ Cr_.~_~
StyrainlScITess
l i IMOdel
c
r Cyclic~Strain/
ICycl~tress IF r
I Environment I ~:~e~e I
L o---j
Environment
k ~176 J==
TM
Failure
by
Fracture
(Unstabl
orStable)e
I
I
Time(cycles)
K ~ - In~at~l~'y
tMode I
or
Nfor
Nff
Cycles
to
initiation
of Fretting
I1
Cyclesto
Mode !
Crack Formotion
Know
tf or
Cyclesof
Fr~ting
Fatigue Life
KI(th)
1 - AK = exp.
log
K[b
FIG. 4--Fatigue life estimationJbrJkeuing-fatigue.

- e
30
Third, more coordinated efforts, such as could be stimulated within ASTM Committee E 9
on Fatigue would be helpful. A round-robin activity could be developed ifa sufficient number
of participants become available.
Fourth, the development of a fretting fatigue experimental practice guide similar to A Tentative Guide for Fatigue Testing and the Statistical A nalysis of Fatigue Data, A S T M STP 91A;
Supplement to Manual on Fatigue Testing, ASTM-STP 91; and Handbook of Fatigue Testing,
ASTM-STP 566 would be extremely helpful. This effort is already underway in laboratories
at the University of Utah. It has been found that the diversity of fretting-fatigue apparati is
astounding. In addition, many fretting-fatigue investigators, even though well intentioned,
violate fatigue testing practices as elucidated in Practice for Conducting Constant Amplitude
Axial Fatigue Tests of Metallic Materials, A S T M E 466; Practicefor Verification of Constant
Amplitude Dynamic Loads in an Axial Load Fatigue Testing Machine, A S T M E 467; Recommended Practice for Constant-Amplitude Low-Cycle Fatigue Testing, A S T M E 606; and
A S T M E 1150; etc. This is a situation which could be improved upon in the future with such
a fretting fatigue experimental practice guide.
Statistically planned round robin activities will eventually be needed, but it is imagined that
a great deal must be done before this can take place. Nonetheless, the necessity to agree on an
experimental protocol, including apparatus, and conduct statistically planned fretting fatigue
experiments is essential to our further progress.
Finally, continued interaction under ASTM auspices to accelerate activity is vital. The interest and activity in fretting fatigue has obviously increased as attested to by the other papers in
this volume. Collectively, it is desirable that all researchers and engineers working in the field
keep this momentum going forward.
Conclusions
From the information presented, the following conclusions can be made:
1. Mechanisms studied have resulted in identification of various forms of damage that
occurs under fretting fatigue conditions.
2. The principal parameters involved in fretting fatigue have been identified qualitatively.
Additional research is necessary to characterize quantitative influences.
Acknowledgments
I wish to acknowledge the support of all my students over the past sixteen years. In addition,
my colleagues at Battelle Memorial Institute and Lockheed Aircraft Corporation have been a
great inspiration to all my efforts on fretting. The past support of the Office of Naval Research
and the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged. Bob Jeal, now technical director of Hawker de Havilland in Australia, has continued
to interact with me and question my ideas on fretting-fatigue as well as numerous other areas.
I am grateful to him for his interest and critical assessment of ideas. Paula Jorgensen typed the
manuscript and aided with the preparation of my presentation. I am indebted to her for her
efforts. My thanks also to Mark Thomsen who aided with the graphics.
References
[1] Symposium on Fretting Corrosion, STP 144, ASTM, Philadelphia, 1953.
[2] Materials Evaluation Under Fretting Conditions, STP 780, ASTM, Philadelphia, 1981.
[3] Selection and Use of Wear Testsfor Metals, STP 615, ASTM, Philadelphia, 1977.
31
[4] Control of Fretting Fatigue, National Research Council, NMAB, National Academy of Science,
Washington, DC, 1977.
[5] Barrois, W. G., "Manual on the Fatigue of Structures--Fundamental and Physical Aspects,"
NATO-AGARD Manual No. 8, NATO, AGARD Symposium, June 1970.
[6] Harris, W. J., "The Influence of Fretting on Fatigue," NATO, AGARD Symposium, Advisory
Report No. 8, April 1967, and No. 21, March 1970.
[7] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, New York, 1972.
[8] Devereaux, O. F., McEvily, A. J., Staehle, R. W. (editors), Corrosion Fatigue: Chemistry, Mechanics, and Microstructure, NACE-2 Conference, NACE (National Association of Corrosion Engineers), Houston, 1972.
[9] Hoeppner, D. W., Uhlig, H. H., "Fretting, Cavitation, and Rolling Contact Fatigue--Critical Introduction," Corrosion Fatigue: Chemistry, Mechanic's, and Microstructure, NACE-2 Conference,
NACE (National Association of Corrosion Engineers), Houston, 1972, p. 607.
[10] Waterhouse, R. B., "The Effect of Fretting Corrosion in Fatigue Crack Initiation," Corrosion
Fatigue: Chemistry, Mechanics, and Microstructure, NACE-2 Conference, NACE (National Association of Corrosion Engineers), Houston, 1972, pp. 608-616.
[11] Hoeppner, D. W., Goss, G. G., "Research on the Mechanism of Fretting Fatigue," Corrosion
[12] Salkind, M. J., Lucas, J. J., "Fretting Fatigue in Titanium Helicopter Components," Corrosion
[13] Lum, D. W., Crosby, J. J., "Fretting Resistant Coatings for Titanium Alloys," Corrosion Fatigue:
Chemistry, Mechanics, and Microstructure, NACE-2 Conference, NACE (National Association of
Corrosion Engineers), Houston, 1972, pp. 631-641.
[14] Starkey, W. L., "A New Fretting Fatigue Testing Machine," Corrosion Fatigue: Chemistry, Mechanics, and Microstructure, NACE-2 Conference, NACE (National Association of Corrosion Engineers), Houston, 1972, pp. 642-645.
[15] Fretting in Aircraft Systems, papers presented at the 39th meeting of the Structures and Materials
Panel, CP-161, NATO, AGARD Symposium, 1975.
[16] Barrois, W. G., Manual on the Fatigue of Structures--H. Causes and Prevention of Structural Damage 6. Fretting--Corrosion Damage in Aluminum Alloys, NATO, AGARD Symposium, Manual
No. 9, Nov. 1975.
[17] Barrois, W. G., Manual on the Fatigue of Structures--H. Causes and Prevention of Damage 7.
Mechanical Surface Damage, NATO, AGARD Symposium, Manual No. 10, ENG, June 1981.
[18] Waterhouse, R. B., editor, Fretting Fatigue, Applied Science Publishers, Ltd., Essex, England,
1981.
[19] Wallace, W., Hoeppner, D. W., AGARD Corrosion ttandbook, Vol. 1, Aircraft Corrosion. Causes
and Case Histories, AGARDograph No. 278, VI, NATO, AGARD Sympsoium, July 1985.
[20] Waterhouse, R. B., "Fretting Wear," Wear, Vol. 100, 1984, pp. 107-118.
[21 ] Hoeppner, D. W., "Fretting of Aircraft Control Surfaces," Specialists Meeting on Fretting in Aircraft
Systems, published in AGARD Conference Proceedings No. 161, AGARD, 7 Rue Ancelle 92200
Neuilly Sur Seine, France, 1974, pp. 9-13.
[22] Goss, G. L., Hoeppner, D. W., "Characterization of Fretting Fatigue Damage by SEM Analysis,"
Wear, Vol. 24, 1973, pp. 77-95.
[23] Goss, G. L., Hoeppner, D. W., "Normal Load Effects in Fretting Fatigue of Titanium and Aluminum Alloys," Wear, Vol. 27, 1974, pp. 153-159.
[24] Hoeppner, D. W., "Comments on 'Initiation and Propagation of Fretting Fatigue Cracks'" (letter
to the editor), Wear, Vol. 43, 1977, pp. 267-270.
[25] Hoeppner, D. W., "Environmental Effects in Fretting Fatigue," Fretting Fatigue, R. B. Waterhouse,
ed., Applied Science Publishers, Ltd., Essex, England, 1981, pp. 143-158.
[26] Hoeppner, D. W., "Material/Structure Degradation Due to Fretting and Fretting-lnitiatedFatigue,"
Canadian Aeronautics and Space Journal, (Third Quarter, 1981), Vol. 27, No. 3, pp. 213-221.
[27] Hoeppner, D. W., Gates, F. L., "Fretting Fatigue Considerations in Engineering Design," Wear,
Vol. 70, 1981, pp. 155-164.
[28] Hoeppner, D. W., Goss, G. L., "A Fretting Fatigue Damage Threshold Concept," Wear, Vol. 27,
1974, pp. 61-70.
[29] Reeves, R. K., Hoeppner, D. W., "Microstructural and Environmental Effects on Fretting Fatigue,"
Wear, Vol. 47, 1978, pp. 221-229.
[30] Czichos, H., Tribology--A Systems Approach to the Science and Technology of Friction, Lubrication, and Wear, Elsevier, Amsterdam, 1978.
32
[31] Hoeppner, D. W., "Parameters that lnput to Application of Damage Tolerance Concepts to Critical
Engine Components," in NATO-AGARD CP-393, Conference on Damage Tolerance Concepts for
Critical Engine Components, NATO-AGARD, 1985.
[32] Mann, D. S., The Design and Development (?['an Experimental Apparalus./br Fracture Mechanics
Based Fretting Fatigue Studies with Eleetrohydraulic Closed Loop Servo-Control of Axial Load,
Normal Load and Slip Amplitude," masters thesis, University of Toronto, Department of Mechanical Engineering, Toronto, Ontario, Canada, 1982.
[33] Hoeppner, D. W., Mann, D., Weekes, J., "Fracture Mechanics Based Modelling of the Corrosion
Fatigue Process," Specialist Meeting on Corrosion Fatigue NATO-AGARD, 52nd Meeting of the
Structures and Materials Panel, 04:05-10, Cesme, Turkey, 1981.
[34] Edwards P. R. ``The Appicatin f Fracture Mechanics t Predicting Fretting Fatigue in FrettingFatigue, Applied Science Publishers, Essex, England, 1981, pp. 67-99.
Leo Vincent, 1 Yves Berthier, 2 and Maurice Godet 2
Testing Methods in Fretting Fatigue"

A Critical Appraisal
REFERENCE: Vincent L. Berthier, Y., and Godet, M., "Testing Methods in Fretting Fatigue:
A Critical Appraisal," Standardization ofFretting Fatigue Test Methods"and Equipment, ASTM
STP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 33-48.
ABSTRACI': Fatigue damage in fretting after crack formation is only initiated under very specific conditions of both amplitude and load. Experience shows, that for a given load the velocity
difference between the core of the rubbing specimen and the friction pads is accommodated as
amplitude increases through the three following mechanisms:
1. Elastic displacements in homogeneous rubbing solids (or first bodies) is a non-destructive
process which will not alter specimen life.
2. Elastic displacements in cracked rubbing solids (or first bodies) can lead to fatigue failure.
This mechanism is dangerous as it can lead to failure and thus to life reduction.
3. Shear in debris beds (third-bodies), formed by particles detached from first-bodies.In most
instances, these beds protect the rubbing surfaces and prevent or at least retard crack formation and thus fatigue.
The significantdrop in life observed experimentally in modified fatigue machines is noted only
because the conditions which lead to Mechanism 2 above are often favored in this type of device.
An original static fretting fatigue approach is proposed.
KEY WORDS: fretting wear, fretting fatigue, friction log, mechanics, material, particle detachment, cracking, velocity accommodation mechanism, fretting maps
Fretting fatigue is commonly encountered in quasi-static loaded assemblies and has been
studied widely. It is one of the forms of surface damage which significantly limits machine
element's life. Waterhouse [ 1,2] distinguishes between fretting wear and fretting fatigue. Relations between these two types of damage are best approached through "interface" or "thirdbody" tribology [3,4]. Many parameters govern fretting [5] and laws drawn from non-intrinsic
test methods invoke corrosion effects, contact temperatures, etc. [4]. Extensive bibliographic
surveys have already been published and only general approaches are discussed here.
Stress analyses are discussed in the classical works quoted in reference [6]. Chivers and Gordelier [ 7] described fretting fatigue as the superposition of surface stresses due to the contact
and to the bulk fatigue created by the external loadings. Similarly, Nishioka and Hirakawa [8]
described stress fields in which the friction coefficient varies during the first cycles. These two
analyses do not consider the slip induced material degradations except through the modification of friction coefficient values.
Failure through fatigue crack propagation only occurs if one of the surface cracks reaches a
Professor, Drpartement Matrriaux-M~canique Physique, CNRS URA 447, Ecole Centrale de Lyon,
Collongue, BP 163, 69131 ECULLY Cedex, France.
2 Charg6 de Recherche and Professor, respectively, Laboratoire de M~canique des Contacts, CNRS
URA 856, 69621 VILLEURBANNE Cedex, France.
33
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
34
critical length related to the contact area [9-12]. Two propagation stages were identified. First,
crack propagation is influenced by both externally applied load and stresses due to the contact,
then by the externally applied loads only when the crack has reached a given length. Propagation rates were measured as a function of displacement, contact loads [13-16]. Short cracks
propagated under high velocity rates, even for stress intensity factors inferior to the threshold
value, and closing effects were used to describe specific behaviors and possible increases in
fatigue limits for high contact pressure levels [13].
Few studies listed surface degradations which act as initiation sites. Fretting fatigue tests
were run on aluminum or titanium alloys to analyze the degraded area from which cracks
nucleate [17,18]. Debris formed in these areas explained the lack of decrease in fretting fatigue
limits [ 19], brought about by fretting.
Fretting maps serve today to determine the actual fretting regime and to identify contact
kinematics conditions (adhesion, partial slip, gross slip) [20, 21]. Partial slip clearly appears as
the most detrimental mode for crack initiation. Critical values of the slip amplitude were
found for which fretting fatigue lifetime is minimal [14]. Designers need data on both fatigue
limit decrease, due to fretting, and scatter in life, needed in "safe crack growth" approaches,
to give estimations for time between revisions. The methodology proposed here is different
from that used in classical fretting fatigue testers. It was developed around work performed
near Lyon (France) during the last ten years by the laboratoire de "Mrcanique des Contacts"
de l'Institut National des Sciences Appliqures (INSA) and by the Laboratoire "MatrriauxMrcanique Physique" de l'Ecole Centrale de Lyon (ECL).
Fretting Wear
Fretting Wear Tests

The results discussed were obtained in tests run on several materials including iron, titanium
or aluminum based alloys with different microstructures. Tests were run on a modified fatigue
tester described elsewhere [22]. For each cycle, the tangential force (F) is recorded as a function of the imposed displacement (D). Both are plotted versus the number of cycles (N) in a
3D "tangential force F~ displacement D~ number of cycles N" graph named "friction log."
Here, the maximum displacement D is naturally twice the fretting amplitude (a). Third-body
action or crack formation are illustrated using these friction logs. Friction logs, which are characteristic of debris formation and third-body action (Fig. 1), are divided in four:
1. Elimination of the natural pollution surface screens.
2. Increase in specimen (or first body) interaction accompanied by an increase in friction
and by the corresponding first body modifications [23].
3. Metallic particle detachment or debris bed formation and gradual transition from a two
to a three-body contact.
4. Three-body contacts are characterized by: (a)/t continuous formation and ejection of
debris (steady-state conditions prevail); and (b) the change in morphology and composition of the debris during their dwell time in the contact. Metallic debris oxidize and
form the well known red powder when steels are tested.
The transition between two- and three-body contacts, which can be dangerous from the
crack formation point of view, is discussed elsewhere [3]. Figure 2 gives a classical friction log
for tests in which the cracking mode prevails. Three parts are noted:
1. Elimination of natural screens, as described above.
2. Increase in the tangential load which corresponds to no or partial slip conditions. FD
VINCENT ET AL. ON TESTING METHODS
35
+750N F
-750
10 s
N
FIG. 1--RepresentativeJkiction log when debris is forrned.

50N IF
$0 N
103
ld
I0 s
FIG. 2--Representative friction log when cracks are present.

cycles are closed or very slightly open (elliptic). During this stage, cracks initiate mainly
at the edges of the contact.
3. FD cycles open further. This corresponds to an accommodation of the displacement by
crack opening and localized slip.
Note that the friction log is made out of individual FD cycles which can take on different
shapes (Fig. 3):
1. Closed (cc) FD cycle, characteristic of a non-dissipative process associated with purely

elastic accommodation found in the "stick" zone. Machine and tangential contact stiffness is given, by the slope of the FD line.
2. Elliptic (ec) FD cycle, characteristic of a slightly dissipative process and generated in contacts in which partial slip is found or in cracked configuration with interfacial crack friction. Depending on operating conditions, the area of the ellipse varies in size.
3. Trapezoidal (tc) cycle, characteristic of gross slip, the near horizontal segments are dissipative, the near vertical segments correspond to the elastic displacements noted above
which are always present.
36
c.c.
e.c.
t.c.
FIG. 3--Friction~displacement loops in fretting.
Velocity Accommodation Sites and Modes

Figure 4 shows that a three-body contact [24] can be broken down into five basic elements:
the two rubbing solids or first-bodies (FB 1 and FB2); the interface or third-body bulk (TBb);
and the two screens (TBs 1 and TBs2), that separate that bulk from the first-bodies. The screens
and interface bulk form the third-body. These five basic elements are known as sites and are
numbered S~ to $5. The difference in velocity between points A and B is thus accommodated
along line AB. However, the velocity distribution between A and B is unknown outside of
thick film or hydrodynamic lubrication.
If, as seen above, the velocity is accommodated at different sites, it is also accommodated
according to different modes. Visualization has shown that accommodation can take place
within any of the five sites and, accordingly, to any of the four modes labeled, respectively, M~
to M4, and which correspond to the elastic, rupture, shear, and rolling modes.
Velocity Accommodation Mechanisms

All velocity accommodation mechanisms (VAM) combine a site and a mode, and are
labeled S, Mj. In hydrodynamic lubrication, for instance, accommodation occurs across the
bulk of the third-body or site $3 and the fluid is sheared according to mode M3. The velocity
accommodation mechanism is unique and identified as $3 M3.
The combination of five sites and four modes leads to 20 VAMs which are presented briefly
below. As first-bodies S~ and $5, and screens $2 and $4 can be inverted, only 12 mechanisms
need be illustrated. However, in real situations the two first-bodies and screens can be different,
and the full 20 mechanisms must be examined. Visualization techniques are used to identify
both velocity accommodation sites and modes.
FIG. 4 - - Velocity accommodation sites and modes.

37
Visualization tests are run with either transparent (glass or sapphire) or standard specimens.
Observations are performed through the specimens when transparent and along the side.
Microscopes are used where necessary and video films are taken. A very rigid test device was
built for the purpose. The magnitude of the displacements which can be accommodated varies
with the effective velocity accommodation mechanism.
Sites S1 and $5
Machine elements, test specimens and supports deform elastically (St M~); this VAM is commonly found in fretting, particularly under small to medium (20 to 50 ~m) amplitudes. Elastic
deformations accommodate larger amplitudes when, due to high contact stresses, cracks (S~
M2) are formed in the first-bodies (Fig. 5). This is also fairly common and was observed with
metals [25] ceramics [26] polymers [27] and other materials. Significant displacements are
commonly accommodated through plastic shear of superficial layers (S~ M3) (Fig. 6) and
exceptionally [28] small displacements are accommodated (Fig. 7) through first-body bulk roll
formation (St M4). Thus, all four modes of velocity accommodation are observed in sites S~
and $5.
Sites $2 and $4
Screens are so thin (10 -9 m) that it is difficult to visualize the accommodation (Fig. 8) which
occurs within them. However, such screens [29] are elastic ($2 Mr), tear [30] ($2 M2), are
believed to shear (Sz M3) [31], and rolls ($2 M4) have been observed [32]. All four modes exist
in sites $2 and S,.
FIG. 5--Sj M2 accommodation (through cracks).

38
FIG. 6--S~ Ms accommodation (through first-body shear).
Site $3
Third-bodies are thicker than screens (10 6 m) and their behavior is easier to observe [33].
They are elastic ($3 M0, tear ($3 M2) (Fig. 9a), shear ($3 M3) (Fig. 9b), and form rolls ($3 M,)
(Fig. 9c). Here, roll formation is c o m m o n and has been observed with ceramics (Fig. 9c), polymers, elastomers, solid lubricants, etc. The roll formation process is always the same. The nat-
FIG. 7--$1 M4 accommodation (through first-body rollformation).

39
FIG. 8--$2 M2 accommodation in screens.
FIG. 9--$3 M2 accommodation (through third-body cracks).

ural (debris) or artificial (solid lubricants) powder beds coalesce to form rollers which line up
perpendicular to the direction of motion. All four velocity accommodation modes exist in site
$3.
Multiplicity of Mechanisrns
All 20 VAMs exist in different applications. More than one velocity accommodation mechanism can, however, act simultaneously in a given contact. Third-body rupture ($3 M2), thirdbody bulk shear ($3 M3) and screen shear ($2 M3) act simultaneously in a chalk powder bed
[34]. Further, mechanisms can change with time. Competition between third-body ($3 M3)
40
and s c r e e n (S 2 M3) shear is often encountered at the start of a test. The parameters which trigger
one mechanism rather than the other in an otherwise apparent steady-state condition are not
yet identified and the change from one mechanism to another cannot, therefore, be predicted
[35].
Fretting Fatigue
Contact Mechanics
It is not possible to explain in a few paragraphs how and to what extent content mechanics
govern fretting fatigue. The subject is well documented [3,5, 7,9,36 to 39] and an excellent
review is given in Ref 7. A few suggestions must be made to avoid some of the classical pitfalls
encountered in fretting fatigue evaluation. This discussion focuses on the device which is commonly used in most fretting fatigue tests in which a tensile specimen held in an oscillating
traction/compression machine is loaded on either side by fixed pads. Sliding between the specimen and fixed pads is induced by the extension and contraction of the specimen. As in all
fatigue problems, crack initiation and propagation must be discussed separately.
Point 1: What are the contact mechanics parameters that govern crack initiation in.fretting
fatigue?--Contact mechanics [7,36] show that high tensile stresses which can cause crack initiation are generated at contact exit by friction. Crack initiation is difficult to identify experimentally, e.g., how long is a crack when it is initiated? Results show, however, [38] that for
ball bearing steels the number of cycles required for crack initiation decreases as friction
increases. In fretting fatigue tests the axial loads generate tensile stresses (Fig. 12) of their own
which are superimposed to the tensile contact stresses, and it is likely that the combination of
both effects leads to crack initiation.
Point 2." What are the contact mechanics parameters that govern crack propagation in fretting fatigue?---Crack propagation can be characterized theoretically by the stress intensity factor K~ and K~I [39]. In fretting many cracks are situated below the contact zone. They are,
therefore, often closed. K~ is nil, yet cracks are believed to propagate. Crack propagation analysis must take both K~ and K . into account.
FIG. 10--$3 M3 accommodation (through third-body shear).

41
FIG. 11--$3 M4 accommodation (through third-body rollformation).
r y
GXX~
vl
~ax 1
GXX
y~Y
y~F YY
[?xSylx,
Oyj
FIG. 12--Sur[ace stresses and displacements.
Point 3: How significant, J?om a contact mechanics point of view, is the drop in fatigue
strength measured in fretting fatigue test machines?----This is important for the designer.
Curves found in the literature show significant drops in endurance limits when classical traction tests are modified to run with fretting pads. This happens because, consciously or not,
everything is done in setting up these tests to avoid debris formation and, thus, eliminate the
protective debris which exist in most applications. Test conditions are, therefore, much more
severe than industrial conditions and it is unrealistic and overpenalizing to base a design on
values extrapolated from them.
42
Point 4: Does "contact stiffness'" modify fretting fatigue test results?--"Machine stiffness"
should be substituted for "contact stiffness." Normal stiffness is high in common fretting
fatigue devices. As it is usually equal to the test specimen compliance, normal deformations
are one or two microns. Tangential stiffness can be much lower as the entire specimen support
and support assembly flex under the tangential forces induced by pad/tensile specimen friction. When friction is high a significant part of the nominal displacement is taken up by elastic
deformation and interface sliding is low. If friction is low the elastic deformation is small and
most of the nominal displacement is taken up through interface sliding. It is common to
absorb from 15 to 20 t~m through elastic deformation (S~ M~) in a 50 um amplitude test. Thus,
changes in friction coefficient which occur during fretting tests induce changes in sliding distances. This point will be taken up later.
Material Response
As in all fatigue problems, fretting fatigue studies must not consider only crack propagation
but crack initiation and propagation, even though classical tests, unlike practical industrial
situations, favor rapid if not immediate initiation. Initiation must, therefore, be examined
first. Propagation of single and multiple short and long cracks are examined later.
During the first few cycles, the relative displacement is accommodated through gross slip
across the pollution screens (or sites 2 and 4) which cover all surfaces. All four modes can
participate in the slip. As the protective screens are destroyed during these first cycles, adhesion
occurs, slip is reduced and most of the velocity difference is accommodated elastically (S, M~)
in the bulk of the rubbing specimens and in their support. No surface degradation will result
from Sj or M~ accommodation, as long as the surface tensile strength, ax,, is low. On the other
hand, when axx reaches values close to the fatigue limit, cracks form in time and displacements
are accommodated through cracked first-bodies which exhibit lowered stiffness (S~ or M2).
Note that these cracks can grow to several millimeters even under pure fretting conditions.
If the adhesive welds break before individual cracks are formed, and often after strain hardening which leads to low toughness has occurred, particles are detached and gross slip takes
place across the bed of newly generated particles or third-bodies. The effective VAM is then $3
Ms or 4.
Hence, from a degradation point of view, the material responds in three different ways to
fretting tests:
1. Through specimen and support compliance, which is a conservative (or non-dissipative)
process with no degradations noted.
2. The tensile stresses are sufficiently high to initiate cracks. Depending on the crack condition (open or closed during the loading cycle), crack length, orientation, interfacial friction, crack number (multiple cracks), and external conditions (fretting amplitude and
contact friction), these cracks can propagate or not. The stiffness of the specimen assembly (specimen and supports) drops slightly and larger amplitudes are accommodated
without further damage.
3. Particles are detached early on. They form a protective bed which in most cases protects
the rubbing specimens from crack propagation.
The same material can, thus, respond in three different ways during the same test. Of the
three, the second is the most dangerous as it can lead to fatigue failure. Fretting maps or maps
which identify material response to specific running conditions must be drawn and corrective
steps taken if they show that the material responds essentially through crack propagation.

1000
43
'/ ,/
Z
i,a-
ii
i -I
'/?
,y
1I
ro
I
0
50
I
100
amplitude (+) 0 (I.tm)

FIG. 13--Running conditions fretting map (RCFM) obtained for a 2091 aluminum alloy [40].
Discussion
Fretting Maps
This section presents an overview of fretting-wear tests run on aluminum alloys. Fretting
maps [40] were drawn for different numbers of cycles, as the VAMs change during the test,
for different amplitudes (15 < a < 100 ~tm), loads (250 < P < 1000 N) and roughnesses (0.4
< Rt < 8 ~m), but for the same frequency (f = 1 Hz). Figure 14 shows representative FD
loops obtained for the three fretting regimes encountered, adhesion or stick, partial adhesion,
or stick and gross slip. The tests are divided in three stages. Two fretting maps are drawn. The
first, the running condition fretting map (RCFM), plots loads, P, versus nominal amplitude,
a, traced for a given frequency and a given roughness. Zones of stick, partial slip and gross
sliding are identified. The second, the material response to the running conditions fretting
maps (MRFM), plots stress or equivalent stress versus amplitude. The three zones identified
in these maps are no degradation, cracks, and particle detachment zones.
Thus two maps are drawn for each condition (or each stage) from which velocity accommodation sites and modes are identified. Further, the F D cycle type (ec, cc or tc) is shown in
Fig. 3. Elastic contributions of sites $2, $3, $4 are neglected.
Representative types of RCFM and M R F M , drawn schematically in Fig. 15, show that these
maps give both guides and trends. The frontiers between zones are not as clearly indicated as
the transition from one zone to the other is gradual. Finally, the time during which each map
is representative of the operating conditions (or stage) varies with the material tested and the
amplitude depends on the test device stiffness. Frequency and roughness effects are not discussed [25]. Remember, material behavior is limited by its "natural" limits, i.e., the elastic
limit, fracture toughness, etc.
The first stage corresponds to the first few cycles. It can last for less than one cycle for ceramics and more than one hundred cycles for resin coated materials. Gross sliding is observed,
friction is low, and the FD cycle is near trapezoidal. The velocity is accommodated mostly in
screens $2 and $4 formed by the polluted metal surfaces. The RCFM map is divided into small
stick and large slip zones. No first body damage is observed.
During the second stage which can, for metals, last up to a few hundred cycles, friction
increases following screen destruction and metal to metal adhesion. The RCFM stick zone
44

F
/
D
/
25 # m ~ /
~llck
k~a = z 25 p . r n j
~(ick
m i x e d stick and ~lip
a = 15 ~ m j
//
A
,''-7
L--1
ka
= ~ 50 p . m J
~[ip
Fn = 2 5 0 N
a = 2 5 l~rn~
~lip
FIG. 14--FD curves at 50 000 cycles for the three jketting regimes obtained for several conditions of
normal loads and displacements for a 2091 aluminum alloy [40].
increases, partial slip is encountered, and gross sliding is reduced. The no degradation (ND)
and the particle detachment (PD) zones appear on the MRFM. Even though surface or metallographic examinations show no traces in the ND zone, that zone must be divided in two to
include the incipient cracking which will develop into the crack zone of stage three. The velocity is always accommodated through S~ M~ in the ND zone, but site $3 (third-body) and possibly mode M~ are involved following particle detachment.
Stage three lasts to the end of the test as long as steady state conditions prevail. A small drop
in the partial stick zone of the RCFM, due to the presence of the third-body formed from
trapped detached particles, is noted. The M R F M also shows a zone in which cracks induced
V I N C E N T ET AL. ON T E S T I N G M E T H O D S
Stage 1
Slage 3
Stage2
L ~ ~k / PartlaI
L ~StJ lCl Gross

IN/tE(Slid|ng
I I ~S2,'~rl
45
L ISUck
T
Part,a,/
s3n3 .._
Running Condition Fretting I'laps
(s}
(s}
{st
De?v
OegraOatl
7DetachmentIdati~CraJ
|/ [
f Partlcle
,Detachment
NO
Degradation
NO
Ma6emal Response Fretting Maps
FIG. 15--Schematic view (?/the two types o[i/~etting mapsjor the three typicalfretting stages.
by fretting wear (not by fretting fatigue) are observed. That zone can increase in size to cover
the ND zone found in stage two, while earlier formed cracks continue to propagate.
The fretting maps drawn for different a l u m i n u m alloys (2024, 7075, 2031...) are similar as
far as particle detachment and debris flow which govern contact behavior are concerned
[40,41]. On the other hand, large differences are noted when crack initiation and propagation
(Fig. 16) are examined. For instance cracks of up to 2 m m were found only with the 7075
0 ~
95
40
'{
200
Stick
~-/7075
Partial
GROSS slldlng
slip
400_
14m.
FIG. 16--Crack lengths l for three aluminum alloys (2024, 2091, 7075) tested during 50 000 cycles
under a normal load of 500 N in a ball (R = 1 m) on flat contact for different amplitudes [40].
46
FRETTINGFATIGUE TEST METHODS AND EQUIPMENT
aluminum alloy after 5 105cycles. This suggests that fretting fatigue behavior can be related
to the fatigue threshold K,,h, or to other fatigue parameters such as da/dN, and that the effect
of original residual stresses, which are of no interest in the case of particle detachment, should
be included in initiation studies. These analyses have been applied to other bare and coated
materials; they require well controlled tests, particularly as far as amplitude and device stiffness
are concerned.
Ifa machine component operates within the MRFM crack zone, it will in time fail through
crack propagation. If, on the other hand, that same component operates in the MRFM particle
detachment zone through an increase in amplitude, all other conditions remaining
unchanged, it will, depending on its trapping possibilities, lose function through wear. The
information produced by fretting maps can help designers, even though they show trends
instead of producing limiting values.
Methodology
The arguments developed above were limited to fretting wear tests, i.e., tests in which the
only load is the contact load. Under these conditions, stick, partial and full slip conditions were
observed. Metallographic investigations identified the presence of cracks.
Fretting fatigue tests differ from fretting wear tests in that an axial (or body) load is added
to the contact load. However, the same methodology which is centered around running condition fretting maps (RCFM) and material response fretting maps (MRFM) can be applied to
these tests as they both produce the tensile skin stresses, ~rxx,which generate surface cracks. The
difficulty, however, in such tests is that in most devices the contact amplitude a, which has
been shown to be a governing parameter in material degradation depends on the elongation
of the test specimen, which itself varies with the axial or body load. The fretting fatigue Wrhler
curves which are plotted for different stress levels (or axial loads) are, therefore, plotted for
different amplitudes. This is a very serious limitation which leads to erroneous conclusions
when transposed to industrial components.
To avoid this problem, it is best to run systems in which the body and contact loads and
amplitude are controlled separately. This can be achieved in fretting-static fatigue tests in
which body stress is constant (R = Cr~i./amax = 1)andtheslipamplitudeiscontrolled. Fretting
maps are then plotted for different loads and amplitudes at different numbers of cycles.
Test severity must be characterized. Earlier results suggested that skin tension, Crxx,is a good
way to characterize severity. This offers significant advantages: 1) it yields a criterion ~rxx which
can be compared to the material fatigue limit ~rt,;and 2) it is capable of adding the contribution
to the original residual stresses to the skin tension axx. These residual stresses contribute significantly to the definition of the MRFM crack zone, even though they have little or no effect
in the particle detachment zone.
Both maps can be used by designers in three ways: 1) to identify the type of damage expected
for given materials operating under given conditions; 2) to favor one type of damage over
another by varying, for instance, component stiffness; and 3) to choose one palliative (coating)
or another.
Fatigue failures which are the most dangerous follow crack initiation. To avoid them the
designer should favor either the elastic accommodation mechanism (S~ M~) or, if this is not
possible, seek particle detachment ($3 M3). This last mechanism does open up the entire problem of debris trapping, which is a study in itself, and is influenced by many parameters [23].
Life predictions can be made from fretting maps when cracking is the failure mechanism.
Crack initiation duration and propagation rates in the contact zone are given by the data collected to plot the fretting maps at different times. Propagation to rupture can, outside of the
contact zone, be predicted by classical propagation laws and models.
47
Conclusions
The arguments developed above show that it is preferable to:
1. Identify as closely as possible the contact operating conditions (contact and body forces,
amplitude,..) of the industrial c o m p o n e n t s u n d e r study;
2. Situate them with respect to those tested in the fretting maps;
3. Identify failure mechanisms rather than try to obtain a factor supposedly representative
of the drop in fatigue limit to be included in c o m p o n e n t life predictions; and
4. Include materials on which practical or prototype information is available to verify the
correlation between laboratory tests a n d practice.
References
[ 1 ] Waterhouse, R. B., "Fretting Fatigue," Applied Science, London, 1981.
[2] Waterhouse, R. B., "Fretting Corrosion," Pergamon, Oxford, 1972.
[3] Berthier, Y., Colombia, C., Vincent, L and Godet, M., "Fretting wear mechanisms and their effects
on fretting fatigue," ASME/Journal of Tribology, Vol. 110, July 1988, pp. 517-524.
[4] Berthier, Y., Vincent, L. and Godet, M., "Fretting fatigue and fretting wear," Tribology International, 1989, pp. 235-242.
[5] Beard, L, "Aninvestigationintothemechanismsoffrettingfatigue,"Ph.D. dissertation, University
ofSalford, U.K. 1982.
[6] Johnson, K. L., Contact Mechanics, Cambridge University Press, 1985.
[ 7] Chivers, T. C. and Gordelier, S. C., "Fretting fatigue and contact conditions: a rational explanation
of palliative behavior," Proceedings, Instn. Mech. Engrs. Vol. 199, No. C4, pp. 325-337.
[8] Nishioka, K. and Hirakawa, K., "Fundamental investigation of fretting fatigue," Bulletin of
J.S.M.E., Vol. 12, No. 51, 1969, pp. 397-407.
[9] O'Connor, J. J., "The role of elastic stress analysis in the interpretation of fretting fatigue failures,"
Fretting Fatigue, Applied Science Publishers, 1981, pp. 23-66.
[10] Hills, D. A., Nowell, D., and O'Connor, J. J., "On the mechanics of fretting fatigue," Wear, Vol.
125, 1988, pp. 39-52.
[11] Hattori, T., Nakamura, M. and Watanabe, "Fretting fatigue analysis by using fracture mechanics,"
ASME Paper No. 84-WA/DE-10 1984.
[ 12] King, R. N. and Lindley, T. C., "Fretting fatigue in a 3~ Ni-Cr-Mo-V rotor steel" 4Proc. ICF5, 1980,
pp. 631.
[13] Sato, K., Fuji, H. and Kodama, S., "Crack propagation behaviorin fretting fatigue," Wear, Vol. 107,
1986, pp. 245-262.
[14] Gaul, D. J. and Duquette, D. J., "The effect of fretting and environment on fatigue crack initiation
and early propagation in a quenched and tempered 4130 steel," Metallurgical Transaction, Vol.
11A, 1980, pp. 1555-1561.
[15] Nix, K. J. and Lindley, T. C., "The influence of relative slip range and contact material on the fretting fatigue properties of 3.5 NiCrMoV rotor steel," Wear, Vol. 107, 1986, pp. 245-262.
[16] Lutynski, C., Simansky, G. and Mc Evily, A. J., "Fretting fatigue ofTi 6AI 4V alloy", ASTM/STP,
Vol. 780, 1982, pp. 150-164.
[ 17] Hoeppner, D. W. and Goss, G. L., "Metallographic analysis of fretting fatigue damage in Ti 6Al 4V
MA and 7075-T6 aluminium," Wear, Vol. 62, 19890, pp. 287-297.
[18] Bill, R. C., "Fretting wear and fretting fatigue--How are they related?", ASME, Journal of Lub.
Tech., Vol. 13, 1981, pp. 1-9.
[ 19] Wharton, M. H. and Waterhouse, R. B., "Environmental effects in the fretting fatigue ofTi 6AI 4V,"
Wear, Vol. 62, 1980, pp. 287-297.
[20] Vingsbo, O. and Soderberg, D., "On fretting maps," Wear, Vol. 126, 1988, pp. 131-147.
[21] Vingsbo, O., Odfalk, M, and Sben, N. E., "Fretting maps and fretting behavior of some FCC metal
alloys," Wear of Materials, 1989, pp. 275-282.
[22] Vincent, L., Berthier, Y. and Godet, M., "Fretting wear and fretting fatigue damage," Fatigue 87,
Vol. 1, 1987, pp. 567-575.
[23] Blanchard, P., Colombir, Ch., Fayeulle, S., Pellerin, V. and Vincent, L., "Material effects in fretting
wear," Metal Trans. (to be published).
[24] 8erthier, Y., "Experimental evidence for friction and wear modeling," Wear, Vol. 139, 1990, pp.
77-92.
48
[25] Berthier, Y., Colombi6, Ch., Lofficial, G., Vincent, L. and Godet, M., "First and third-body effects
in fretting--a source and sink problem," Mechanisms and Surface Distress, Ed., D. Dowson, C. M.
Taylor, M. Godet, D. Berthe, London: Butterworths, 1986.
[26] Kapelski, G., "Etude des propri6t6s tribologiques de c6ramiques thermom6caniques en fonction de
la temp6rature et pour diff6rents environnements," ThOse de Doctorat de l'UniversitO de Limoges,
1989.
[27] Vincent, L. and Dahmani, N., "Interface dynamics in polymer friction," Interlace Dynamics. D. D.
Dowson, C. M. Taylor, M. Godet, D. Berthe, Eds., Amsterdam, Elsevier Tribology Series, 1988, pp.
13-18.
[28] Colombi6, Ch., Berthier, Y., Floquet, A., Vincent, L. and Godet, M., "Fretting: load-carrying capacity of wear debris," Journal ofTribology, ASME, Vol. 106, No. 2, 1984, pp. 185-194.
[29] Tonck, A., Kapsa, P. and Sabot, J., "Mechanical behavior oftribochemical films under a cyclic tangential load in a ball/flat contact," Journal ofTribology, ASME, 1986.
[30] Berthier, Y., Brendle, M. and Godet, M., "Velocity accommodation in friction--STLE Transaction,. Vol. 32, No. 4, 1989, pp. 490-496.
[31] Briscoe, B. J., Scruton, B. and Willis, R. F., "Shear Strength of thin lubricant films," Proceedings,
Royal Society, London, A333, 1973, pp. 99-114.
[32] Israelachvli, J. N. and Tabor, D., "The shear properties of molecular films," Wear, Vol. 24, 1973,
pp. 386-390.
[33] Berthier, Y., Vincent, L. and Godet, M., "Velocity accommodation in fretting," Wear, Vol. 125,
1988, pp. 25-38.
[34] Gorier, M. and Berthier, Y., "Continuity and dry friction: an Osborne Reynolds approach," Fluid
Film Lubrication, Osborne Reynolds Centenary, D. Dowson, C. M. Taylor, M. Godet and D.
Berthe, Eds., Elsevier, Tribology Series 11, Amsterdam 1987.
[35] Berthier, Y., Brendle, M. and Godet, M., "Boundary conditions: adhesion in friction," Interface
Dynamics, D. Dowson, C. M. Taylor, M. Godet and D. Berthe, Eds., Amsterdam, Elsevier, 1988,
pp. 19-25.
[36] Johnson, K. L. and O'Connor, J. J., "Mechanics of fretting," Proceedings, Institute of Mechanical
Engineers, Vol. 178, Pt 3J, 1963-64, pp. 7-21.
[37] Endo, K. and Goto, II., "Initiation and propagation of fretting fatigue cracks," Wear, Vol. 38, 1976,
pp. 311-324.
[38] Kisu, H., Ura, A., Nakashima, A. and Moritaka, H., "On the surface cracking and crack growth due
to the oscillated tangential force without macro slip," Faculty of Engineering, Nagasaki University,
Vol. 15, No. 25, July 1985.
[39] Dubourg, M. C. and Villechaise, B., "Unilateral contact analysis of a crack with friction," Eur. J.
Mech., A/Solids, Vol. 8, No. 4, 1989, pp. 309-319.
[40] Pellerin, V., Zhou, Z. R. and Vincent, L., "Wear mechanisms in fretting ofaluminium alloys," 2nd
International Conference on Aluminium Alloys, Bijing, China, 1990.
[41] Zhou, Z. R. and Vincent, "Fretting maps ofaluminium alloys," (to be published).
Olof B. Vingsbo l
Fretting and Contact Fatigue Studied with the

Aid of Fretting Maps
REFERENCE: Vingsbo, O. B., "Fretting and Contact Fatigue Studied with the Aid of Fretting
Maps," Standardization of Fretting Fatigue Test Methods" and Equipment, ASTM STP 1159,
M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, ! 992, pp. 49-59.
ABSTRACT: Fretting fatigue refers to fatigue life shortening in bulk testing, induced by the
simultaneous application of surface fretting, that provokes nucleation and growth of surface
fatigue cracks. This paper suggests controlling fretting fatigue by controlling the conditions of
the fretting contact.
Fretting is known for its complexity of parameters and variables. The present work has focused
on a systematic representation of the surface effects of fretting, called fretting maps. By dynamic
recording of tangential force and displacement for convex circular contact between metallic
specimens, different fretting regimes are distinguished. The corresponding modes of surface
damage are identified by metallographic studies of the fretting scars. For increasing displacement
amplitude, three regimes can be identified.
1. The stick regime is characterized by very limited interracial slide and surface damage. A
tangential force (Fr) versus displacement (~) plot for a complete load cycle is a straight line,
indicating predominantly elastic deformation and no sliding losses.
2. The mixed stick-slip regime contact surface is subdivided into a central stick area. surrounded by an annular yield and slip area. For soft materials contact fatigue cracks are
observed around the stick-slip boundary. The FT(6) curve displays a narrow hysteresis loop,
indicating some plastic deformation and relaxation by sliding.
3. The gross slip regime (transition) is recognized by a sudden drop in the recorded Fr(6)
curve, corresponding to a transition from static to kinetic friction (the "point of incipient
gross slip"). All adhesive asperity contact bridges are broken during every half-cycle and
plastic deformation takes place in the bulk of the contact zone. The scars are characterized
by sliding wear grooves in the fretting direction. The Fr(6) hysteresis loop is deformed by
the incipient gross slip drop in FT.
Fretting test series were performed at varying test parameters, and fretting maps were constructed by pairwise combinations of the recorded critical parameter values for transition from
one regime to another. Generally, there is a gradual transition from the stick regime to the mixed
regime, whereas, the mixed-to-gross-slip transition is sharp.
The mixed stick-slip regime is characterized by high cyclic surface stresses at the boundary
between the stick zone and the slip annulus, promoting contact fatigue. Thus, conditions of
mixed stick-slip will aggravate bulk fretting fatigue. Similarly, the sliding conditions of the gross
slip regime are characteristic of fretting wear, during which surface fatigue cracks will be successively worn away and prevented from contributing to fretting fatigue. Once the relevant fretting
maps for a given tribosystem are established, it is, in principle, possible to control the fretting
process by choosing the active parameters so as to promote or suppress a certain type of fretting damage.
KEY WORDS: contact fatigue, fretting fatigue, fretting wear, microstructural studies, fcc metals
Professor, Department of Mechanical Engineering, University of Houston, Houston, TX 77204, and
Department of Materials Science, Uppsala University-School of Engineering, Uppsala, S-751 21 Uppsala,
Sweden.
49
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
50
Fretting Fatigue Related to Contact Surface Fatigue

The concept of fretting fatigue refers to a decrease in fatigue life or fatigue limit in bulk testing, induced by the simultaneous application of surface fretting. The generally accepted explanation for this effect is that the fretting contact provokes nucleation and growth of surface
fatigue cracks in addition to the contributions caused by the bulk load cycling. The present
paper deals with controlling fretting fatigue through fretting contact conditions.
The Elastic Model of Tribo Contacts

Engineering surfaces are not ideally plane. Even grinding and polishing to a good surface
finish will leave some waviness. Two unlubricated surfaces brought together will first reach an
equilibrium position when three protuberances of bulk dimensions are in contact. Depending
on load, sliding, etc., plastic deformation will later occur and increase the number as well as
the size of contact areas. For studies of the contact geometry when tribo contact is being established, however, a representative model is that of a spherical protuberance being pressed
against a half-space. The model is further simplified by assuming that the contacting surfaces
are perfectly smooth on a macroscale, and that the bodies are of the same isotropic and ideally
elastic material. According to Hertzian theory, the contact surface will be a circle, the radius
a of which is a function of the applied normal load FN, the radius of curvature of the sphere
R, and the elastic modulus E and Poisson's ratio ~ of the material, as expressed by
~
3
a =
3(1 - v2) 9 FN" R

TE
(1)
The normal pressure distribution over the contact surface is given as a function of the distance (r) from the center by
3FN
r:
p ( r ) -- 27ra 2
1 -- a-7
(2)
If a tangenital force F r is simultaneously applied, it will generate a tangential traction tr(r)

in the contact surface. A first assumption that no slip occurs anywhere in the contact surface
requires the traction distribution [1,2]
FT
tr(r) = 2 r a ~
-- r 2
(3)
It is obvious from Eq 3 that this assumption would imply a singularity in t(r) at the outer
boundary of the contact surface (r --- a). If, on the other hand, it is assumed that t(r) may not
exceed the friction stress distribution over the contact area, according to
(4)
tr(r) <-- u " p ( r )
it seems that slip will occur outside a circle of radius r = a ' < a, as suggested by [1,2]. a ' is
given by
3
a' = a
~FN
(5)
VINGSBO ON FRETTING MAPS
51
FIG. 1--Elastic modelfor surJace contact under normal and tangential loads.
Thus, for a tangential force Fr < #FN (~ is the static coefficient of friction), the contact surface is characterized by a central stick circle of radius a', surrounded by a slip annulus (Fig. 1).
It can be shown [2] that the tangential traction is given by
3~FN
tr(r) -- 2ra2
r2
(6)
a2
where a ' --< r _< a (in the slip annulus), and
tr(r) = 2rra 2
~5
as
a
a;~
(7)
where r _< a ' (witt, in the stick circle).

The elastic deformation in the sphere and the half-space will result in a tangential displacement 6 [2]:
3
6-
E2R
1 --
1 --~N ]
(8)
of the center of the sphere relative to a fixed reference point in the half-space, far from the
contact zone, k is a material constant, given as a function ofv by
k-
(1 + v ) ( 2 - v)
2
2
3(1 - v2)
(9)
In the transient phase of applying the tangential load, microslip starts at the rim of the contact circle and penetrates inwards under formation of the slip annulus. The corresponding displacement, a, is also a function of the normal load, FN, according to Eq 8. It is seen from Eq
52
FRETTINGFATIGUETESTMETHODSAND EQUIPMENT
5, that a ' approaches zero when the applied tangential force, Fr, approaches the friction force,
uFu, and
Fr = uFN
(10)
is the condition for incipient gross slip over the whole contact area, with the critical displacement 6~,c 6(u " FN); Eqs 5 to 8 are then defined for 0 ~ FT <-- ~F~.
The displacement is related to a surface stress, o, the character of which varies with location
around the contact circle as described in Fig. 2. Particularly, the normal tensile stresses at the
leading edge in the protrusion and the trailing edge in the mating half-space promote the nucleation of surface cracks during the friction build-up phase, until conditions of fully developed
static friction are reached. At incipient gross slip, however, the protrusion starts sliding, and
bulk wear will gradually remove the cracked surface layer. The tensile stress peaks now correspond to the lower, kinetic friction and are less promotive of further crack nucleation. In the
half-space the cracked initial contact area is left behind and will not contribute to further surface deterioration. Thus, in unidirectional sliding, the high-stress conditions of crack nucleation prevail only in the transient phase of friction build-up before gross slide is initiated, and
the contributions to bulk wear and fatigue can be neglected.
=
Cyclic Loading and Fretting

Under an oscillating instead of a unidirectional tangential force FT(t), the situation may be
totally different from the tribological point of view. If the displacement amplitude 4, corresponding to the tangential force amplitude T, is large compared to the extension of the contact
area, i.e., for 2x >> a, the reciprocating movement will have essentially the same tribological
bulk effects as unidirectional sliding over the same sliding distance, as described above. If2x _<
a, however, the mixed stick-slip conditions, characteristic of the friction build-up, may be met
and repeated with a stress reversal every half cycle, and the crack nucleation phase will predominate during the whole cycling process. These are the conditions for fretting in the strictest
sense, and it is obvious that surface cracks nucleated during one half-cycle will not be removed
by wear during successive cycles. On the contrary, crack growth combined with continuing
crack nucleation will feed a continuously aggravating situation of surface fatigue. When debris
are formed, because of the small sliding amplitude, only particles from the rim of the slip annulus can escape, whereas, most of the particles will be trapped in the fretting scar and contribute
to further aggravation by grooving, generating stress concentrations, etc.
It has recently been shown [3,4] that plastic deformation plays an important role in fretting,
even under loads and displacements too low to generate considerable yield in static loading.
Extensive dislocation glide can take place in the fretting scar zones, also in low-amplituded,
mixed stick-slip fretting. Thus, fatigue crack propagation mechanisms based on dislocation
glide can operate and contribute to the fretting contact fatigue process in ductile materials.
PROTRUSION
,""'~e
nsile
compr,
FT
c o m p ~
tensile
HALFSPACE
FIG. 2--Character of sur/ace stresses in the contact area.
53
In summary, surface fatigue is promoted by fretting under mixed stick-slip conditions, both
in terms of cyclic stress concentrations and plastic deformation in the contact zone.
Fretting Regimes
A more general description of the fretting process than that given by the above elastic model
has been worked out by Odfalk and Vingsbo [4]. The displacement 6 is made up of three components according to
6 = 6~ + 6p + &~
(11)
where ~, refers to elastic displacement between the protrusion and the half-space according to
Eq 8. (The effect on 6, of the stiffness of the testing machine is not taken into account.) 6p
corresponds to plastic deformation in the bulk of the contact zone and 6s to slip in the slip
annulus. The interrelation between the different components is schematically demonstrated
for a complete fretting cycle in Fig. 3, where the total 3 is assumed to be externally controlled
as a perfect sine function. The elastic loading and unloading phases are seen as straight parts
of the 6~curve. A plot of the tangential force versus the total displacement will form a hysteresis
loop, as shown in Fig. 4, the surface area of which corresponds to the energy dissipated during
a fretting cycle.
In terms of contact conditions, the contact area of Fig. 1 will be slightly modified. The stick
circle, in which the asperities are deformed elastically, is surrounded by a thin yield annulus,
characterized by plastic yield of the asperities, The yield annulus is, in turn, surrounded by the
FIG. 3--Displacement as a fimcdon (gtime during a complete tangential load cycle.
/
-T
I
-8(T)
8(T)
FIG. 4--Tangential force versus displacement hysteresis loop.

54
slip annulus in which the asperities are sheared to fracture in the same way as in the elastic
model. It is instructive to describe the distribution of deformation and slip over the contact
area as functions of the displacement amplitude, A, in terms of different fretting regimes as
follows.
Regime I, a' ~- a, (Stick)
For very low displacement amplitudes, the yield and slip annuli will have a negligible width,
and the contact conditions are dominated by elastic deformation of bulk material and asperities in the stick circle. An FT(6) plot will be a straight line, as shown in Fig. 5a, and essentially
no energy is dissipated. The surface damage is low, and has to accumulate over large numbers
of cycles to be measurable.
Regime II, 0 < a' < a, (Mixed Stick-Slip)
For gradually increased displacement and tangential force amplitudes, plastic deformation
will play an increasing role, both in the bulk within the stick circle and in asperities in the yield
annulus. At the same time, a slip in terms of shear fracture of asperities will appear in the slip
annulus. The F~(a) plot will gradually develop into a hysteresis loop (Fig. 5b), the surface area
of which represents frictional losses by plastic work and slip (cp Fig. 4). The situation is characterized by the mixed stick-slip fretting conditions demonstrated in Fig. 1. These are the conditions for surface contact fatigue, discussed in the previous section on Cyclic Loading and
Fretting. Wear effects are small.
Regime llI, a' = O, (Gross Slip)
For displacements corresponding to incipient slip according to Eq 10, the friction coefficient
will drop from its maximum value of fully developed static friction Us,a,cto the kinetic value
~,,,e,,c < m,~,ic.This will generally be manifested as a drop in the Fr(t) curve, but not in the 6(t)
curve, and the Fr(6) hysteresis loop will be deformed as shown in Fig. 5c. For A > 6,,c, gross
slip conditions will be reached every half cycle, resulting in an increasing amount of wear for
increasing A.
Fretting Maps
For a given material (or material combination), atmosphere and contact geometry, the fretting conditions are controlled by the testing parameters/variables: the (generally constant) norFT
FT
(a)
T FT
T,I! ,?
8
Ij
I--
A1
A2
(b)
(c)
FIG. 5--Fr(6) plots for different displacement amplitudes A. (a) Regime L (b) Regime II. (c) Regime
HI.
55
mal load FN, the cyclic tangential force Fr, the fretting frequencyf and the number of applied
cycles n. Together, these variables represent a multidimensional space in which the three
regimes are defined. A fretting map is a two-dimensional plane section through this space,
describing the regimes with special emphasis on the boundaries between the regimes. Schematic examples are given in Fig. 6.
The information needed to plot a fretting map is the experimental criteria for the different
regimes and the critical parameter values for transition from one regime to another. A prerequisite for collecting this information is a fretting testing machine that permits the independent
control of tangential force amplitude, displacement amplitude, and frequency, preferably in
connection with equipment for post-test studies of the fretting damage (such as a scanning
and/or transmission electron microscope). An apparatus for this purpose has been built at the
University of Houston and has been described in detail in a previous publication, together with
the appropriate test methods [5]. In practice, T and 2x are generally varied during a fretting
experiment with c o n s t a n t f n and FN, until the critical transition values are found. For most
materials the I/II transition is gradual, and the corresponding T, and/x~ values represent intervals rather than sharp boundaries (compare with the vaguely defined yield point of materials
without a yield drop). The II/III transition, however, is generally well defined, and most fretting maps consist of T2(f), T2(Fu), A2(f), and Az(FN)diagrams. A series of investigations of
number of fcc alloys has been carried through in which the effect of variations in materials
parameters on the regime boundaries were studied [5,6, 7].
Figure 7 is an example [ 7] of a fretting map in which the s
boundaries for three different materials are compared. The materials, pure copper, a copper-3%-Si alloy and an austenitic stainless steel (AISI 304), have successively higher hardness values, and it can be seen
m~~
"f
:'~'~
~:~!~Tl::::i:~:::!
" FN
FIG. 6--Schematic diagrams representing one A(I) and one T(FN)fretting map.
10
e-
0
0
(FN 2/3) (in N2/3)

FIG. 7--A2(FZun) fretting m a p recorded at a f r e q u e n c y f = 50 H z for three materials. [] copper. s copper-
3%-silicon. 9 austenitic stainless steel.

56
that the A2(FN) boundary falls with increasing hardness. In addition, the plot was made versus
F~ 3. The linearity proves that 2x2is proportional to F~ 3, as suggested by Odfalk [ 7].
An example o f a 2x2(J) fretting map [5] is reproduced in Fig. 8a. The displacement amplitude for the II/III transition has been recorded in the frequency interval 10 ~<f--< 800 Hz for
a Cu-3%-Si alloy and a stainless steel. Both curves show a A2 minimum for frequencies around
300 Hz. The area below each curve represents the mixed stick-slip regime II, i.e., contact
fatigue is the predominant damage mechanism. Fretting oscillations of this character, superimposed on a tribo system exposed to bulk fatigue, would result in fretting fatigue in the broad
sense. In regime III, above the curve, gross slip occurs every half-cycle, and sliding wear is a
predominant mechanism. The sliding distance per cycle increases with increasing A2, resulting
in an increased fretting wear. Fretting fatigue, however, will be suppressed by the continuous
elimination of contact fatigue cracks by the fretting wear. It is possible to design the system
with respect to fretting fatigue resistance in two different ways. For a given, fixed vibrational
amplitude A >_ A2m,,for each curve, fretting fatigue would be promoted in the lower and higher
frequency parts of the frequency interval, but could be suppressed by choosing frequencies
from the mid-part of the diagram, thereby, passing the boundary into regime III and trading
fretting fatigue for fretting wear (which might be less detrimental). If, on the contrary, the
Copper-Silicon
~'
~2 (p.m) 2
Sleel
10
100
1000
f (Hz)
FIG. 8a--zX2(/) fretting map recorded for copper-3%-silicon and stainless steel (FN = 3.4 N).
10
::k
._=
100
101
102
t0 3
f (in Hz)
FIG. 8b--~2Og.#etting map, recorded for the copper-3%-silicon alloy at three normal loads. [] 1.6 N. A
3.4 N. Il l l . 4 N.
57
vibrational amplitude is a controllable parameter, it is possible to achieve the same result by

increasing A, irrespective of frequency, into Regime III.
It is possible to speculate over the particular shape of the dx2(f) curves in Fig. 8a. The displacement 6 has a component 6~, connected to plastic deformation. The corresponding dislocation glide velocity is sensitive to the strain rate, which leads to a frequency dependence. An
increase in fwill imply a strain rate hardening (decrease in &), balanced by an associated heating and thermal softening (increase in dx). In the low-frequency interval, the strain rate hardening dominates, whereas, in the high-frequency interval, the thermal softening dominates.
This type of U-shaped curve is a characteristic result of the competition between two mechanisms of opposite effects. The fretting map also shows that this phenomenon is more pronounced in the Cu-Si alloy than in the steel, and that A2 is higher in the Cu-Si alloy than in the
steel within the whole investigated frequency range. Fig. 8b reproduces the same type of fretting map in which three &2(f) curves for the Cu-Si alloy have been plotted for different normal
loads. It can be seen that the hardening-softening effect is more pronounced and that the &2
boundary is higher for higher loads.
A scanning electron micrograph of a fretting scar is reproduced in Fig. 9. The specimen
material is a low-carbon structural steel, and the fretting conditions were chosen to represent
the mixed stick-slip characteristics of Regime II. It is clearly demonstrated how a slightly worn
slip annulus is surrounding the essentially unaffected stick circle. It has also been observed that
mixed stick-slip fretting can give rise to a fatigue crack around the whole yield annulus (Fig.
10). Light microscopy of cross sections through such scars revealed that the crack propagated
FIG. 9--Scanning electron micrograph of a typical mixed stick-slip fretting scar in a structural steel.
FIG, lO--Scanning electron micrograph of a conically propagating surface fatigue crack in a fretted
copper specimen.
58
FIG. 11--Light optical micrograph of a metallographic section through a fretting scar with a conically
propagating surfaceJatigue crack (Cu-Si).
conically into the contact zone, until a large fragment was released, leaving a conical pit at the
center of the scar (Fig. 1 1).
Conclusions
The concept of fretting maps is new and has not yet been subjected to widespread exposure.
The availability of appropriate testing equipment, however, has made it possible to work out
practical methods of recording maps and to evaluate their basic features. The influence of testing and material parameters on the location and shape of regime boundaries has been successfully studied, and is an issue of central interest in future fretting research. If large efforts
could be channeled into recording fretting maps for tribosystems, known to be vulnerable to
fretting damage, and that information were made accessible via relevant data basis, it would,
in principle, be possible for designers to control the fretting process by choosing the active
parameters to suppress fretting fatigue damage.
Acknowledgments
This work has had partial financial support from the Tribology Program of the National
Science Foundation under Contract MSM-8516963.
References
[1] Cattaneo, C., "Sul Contatto di Due Corpi Elastici," Accademia dei Lincei, Rendiconti, Series 6, Vol.
XXVII, 19.38, pp. 342-348,434-436,474-478.
[2] Mindlin, R. D., "Compliance of Elastic Bodies in Contact," Journal of Applied Mechanics, Vol. 71,
1949, pp. 259-268.
[3] Vingsbo, O. and Odfalk, M., "Conditions for Elastic Contzict in Fretting," Proceedings, International
Tribology Conference, Nagoya, 29 Oct.-1 Nov. 1990.
[4] Odfalk, M. and Vingsbo, O., "An Elastic-Plastic Model for Fretting Contact," submitted to the International Conference on Wear of Materials, Orlando, Fla., 7-11 April 1991.
[5] Odfalk, M. and Vingsbo, O., "Influence of Normal Force and Frequency in Fretting," Proceedings,
STLE/ASME Joint Tribology Conference, Ft. Lauderdale, Fla., 1989, STLE Preprint No. 89-TC-4E1, 1989.
[6] Vingsbo, O., Odfalk, M., and Shen, N., "Fretting Maps and Fretting Behavior of Some fcc Metal
Alloys, Wear of Materials, ASME, New York, 1989, pp. 275-282.
[ 7] Odfalk, M., "Fretting Studies of Some fcc Metal Alloys with the Aid of Fretting Maps," MSc Thesis,
University of Houston, Tex., Dec. 1988.
59
DISCUSSION
L. Vincent ~(written discussion)--Fretting maps appear to be a good basis for the prediction
of fretting fatigue behavior. The effect of frequency has been discussed through increasing temperature and material modification effects. At the lower ranges, the effect of frequency can be
described through the trapping of detached particles. For instance, any increase in frequency
favors debris ejection for steels, whereas higher ranges are required for titanium or aluminum
alloys due to ridges formed at the limit of the contact. Have you any comment on this frequency effect in the case of a small increase in temperature?
O. Vingsbo (author's closure)--Your comment that the frequency effects in fretting can be
influenced by the trapping and escape of wear particles is correct. We have not taken this secondary effect into account when finding the regime boundaries of the fretting maps.
Ecole Centrale de Lyon, B.P. 163, 69 131 Ecully Cedex, France.

Jan M. Dobromirsk?
Variables of Fretting Process Are There 50 of

Them?
REFERENCE: Dobromirski, J. M., "Variables of Fretting Process: Are There 50 of Them?"
Standardization of Fretting Fatigue Test Methods" and Equipment, ASTM STP 1159, M. Helmi
Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia,
1992, pp. 60-66.
ABSTRACT: It has been reported that up to 50 variables might influence the magnitude and
rate of the fretting process.
Predictions of the fatigue strength under fretting conditions were reported to be successful
when based on "stress models" (e.g., the fracture mechanics approach). These predictions use
only a few variables of the fretting process such as external load, contact pressure, coefficient of
friction, and the amplitude of relative slip.
A concept of "primary" and "secondary" sets of fretting variables has been proposed. The
"primary" set of variables includes the coefficient of friction, magnitude of slip and contact pressure acting at the fretting interface. The "secondary" set of variables affects the fretting process
through changes these variables cause in the "primary" set of variables.
The coefficient of friction has been identified as the main variable of the fretting process. Analysis of the literature shows that the effect which many other variables have on the fretting process
can be explained by the changes these variables cause in the value of the coefficient of friction.
KEY WORDS: fretting, fretting fatigue, coefficient of friction

Introduction
It has been reported that up to 50 factors might influence the magnitude and rate of the
fretting process [/]. Fretting literature amassed during the last 40 years shows that great effort
has been m a d e to assess the effect of a n u m b e r of variables on the fretting corrosion, fretting
wear or fretting fatigue. The list of such variables which have been studied in research laboratories includes:
9
9
9
9
9
9
9
9
9
9
9
9
Contact pressure
Tangential force
Coefficient of fraction
External load
Slip amplitude
Contact area
F r e q u e n c y of vibration
N u m b e r o f cycles
Hardness of contacting surfaces
Temperature
Thickness o f oxide layers
Surface roughness
Research Scientist, Research Laboratory, KODAK (Australasia) Pry. Ltd., P.O. Box 90, Coburg, Victoria 3058, Australia.
6O
Copyright
by ASTM Int'l
(all rightsASTM
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
www.astm.org
DOBROMIRSKI ON VARIABLES OF THE FRETTING PROCESS
9
9
9
9
9
9
9
9
9
9
9
9
9
9
61
Ambient atmosphere (air, nitrogen, etc.)

Atmospheric pressure
Humidity
Lubricants
Residual stresses
Work-hardening of surfaces
Materials of the contacting members
Morphology of the materials
Corrosion susceptibility
Microstructure (martensite, austenite, etc.)
External load ratio
Plain fatigue level
Size effect
Metallurgical compatibility
Literature reports vary, sometimes significantly, in assessment of the influence of these variables. The difference can be explained by the fact that no established, standardized testing procedure of the fretting process exists. Different materials, geometries, types of loads and equipments are used. This gives a large scatter of results, making it sometimes difficult to draw
quantitative conclusions regarding the changes and influence of the various factors involved.
Furthermore, experimental evidence suggests that there are strong synergistic links between
many of those variables, i.e., the change in one variable may cause a change in another variable. This creates extreme difficulties in obtaining experimental data on the influence of a single variable on the fretting process. To obtain such data a completely independent and simultaneous control of many experimental variables would have to be required.
Different Models of Fretting Fatigue Process

There have been different models of the fretting process mechanism proposed in the literature. However, none of them can be accepted as a full and complete explanation of the process. The predictions of some models [2] were in conflict with certain experimental observations. The reason for this lies in a high degree of complexity of the process and the strong
interactions between many of the factors considered as variables of the process. Different
methods of predicting fretting fatigue life or fretting fatigue strength were proposed. Simple
methods of drawing fretting fatigue Wohler's curve were suggested by Stepanov [3] and Filimonov [4]. These methods, however, assumed that the fretting fatigue curve was unaffected
by the conditions of the fretting process and, consequently, a very poor correlation of these
methods with experimental data was reported [5].
More successful attempts were based on "stress models" of the fretting fatigue process.
These stress models assumed that crack initiation and propagation in fretting fatigue are governed mainly by stress conditions existing within a junction. Nishioka and Hirakawa [6] formulated an expression for the fatigue limit based upon the initiation of fatigue cracks under
fretting conditions. This equation used the fatigue strength of a plain specimen, the magnitude
of relative slip, contact pressure, and the coefficient of friction within a junction as the input
variables.
Sato [ 7] reported good agreement between results of his experimental tests and predictions
of the fretting fatigue life. These predictions were based on an equation utilizing a concept of
the equivalent stress of Tresca. The equation used the following factors as input variables: contact pressure and its stress concentration factor; stress amplitude of the external load; and the
coefficient of friction.
62
Many attempts to predict fretting fatigue strength have been based on the fracture mechanics application. Rooke and Jones [8] showed how to calculate stress intensity factors for fretting conditions. Their method was used by Dekhovich [9], Edwards [10], Nix [11 ], Switek
[12] and Hattori [13] to predict fretting fatigue strength. Good agreement with the experimental results was reported in all of these cases. Typically, the following variables of the fretting process were included in the calculations: a plain fatigue strength (or elastic constants of
the material); contact pressure; magnitude of slip; and the coefficient of friction at the fretting
interface. Fracture mechanics theory analyzes conditions required to propagate a crack within
a loaded member. These conditions (stress intensity factors) are expressed in terms of a member geometry, location and size of a crack, and also in terms of the stress field acting in the
analyzed location. The stress field for fretting conditions is defined by an external load, normal
pressure and the friction acting at the fretting interface.
Primary and Secondary Sets of Variables in the Fretting Fatigue Process

The "stress models" (e.g. based upon the fracture mechanics theory) have been reported to
successfully predict the fretting fatigue strength under different testing conditions. Stress models typically assume that the coefficient of friction, contact pressure and the extent of slip area
at the fretting interface are the governing factors of the stress field within the fretting junction.
Other variables of the fretting process are not included directly in these models. It can then be
hypothesized that those other variables are somehow included in the total equation through
their effect on the magnitudes of normal pressure, slip and the coefficient of friction within a
junction. Consequently, this would suggest that there is a "primary" and "secondary" set of
fretting process variables, the latter one affecting the fretting process through changes in the
first, "primary" set of variables.
If such a hypothesis was true, then more focused research of the fretting fatigue process could
be possible. Theoretical modeling based upon a "stress model" approach (e.g. fracture
mechanics), and appropriate experiments, would study the effect of the "primary" set of variables on the fretting fatigue process. Extensive theoretical and experimental work would also
be needed to focus on the link between the "secondary" and the "primary" set of variables.
Review of the literature suggests that the coeff
of friction is the main variable of the
proposed "primary" group of variables in the fretting process. In fretting wear "the predominant wear mechanism is related to the magnitude of the friction coefficient" [14]. In fretting
fatigue the coefficient of friction plays a very significant role in creating conditions for initiation and propagation of fatigue cracks. An elastic stress analysis of conditions existing within
the fretting junction has shown the coefficient of friction to be a "scaling factor of the axial
component of stress induced by a tangential frictional force" [15]. This axial component of
stress strongly contributes to crack initiation and propagation in the fretting junction. Edwards
and Cook in their fracture mechanics analysis of the fretting fatigue [16] stated that " . . . the
damaging effect of fretting is ascribed almost entirely to frictional forces which give rise to an
extra damaging contribution to the stress intensity f a c t o r . . . " . Nix and Lindley [1 I], wrote
" . . . any fretting pad-specimen configuration giving similar values of frictional force and contact pressure at stresses close to the fatigue limit should exhibit the same fretting fatigue limit,
independent of relative slip range and contact pad material."
Effect of Fretting Process Variables on Coefficient of Friction

Existing literature provides ample evidence about the effect many variables have on the
severity of the fretting process. Data can also be found concerning the effect some of these
variables have on the coefficient of friction within the junction. Comparison of such data for
63
some of the variables is given below. Some of this data are taken from fretting wear studies,
some from fretting fatigue experiments. However, Bill [17] claimed that fretting wear and fretting fatigue are influenced in a similar manner by the controlled variations in experimental
conditions, i.e., conditions which tend to accelerate fretting wear also accelerate fretting
fatigue failures.
Crystal Structure ojSolids

Different rates of the fretting fatigue process for materials with a different crystal orientation
have been reported by Goss & Hoeppner [18]. They found that aluminium (f.c.c. structure)
undergoes more significant life reduction from fretting than Ti-6AI-4V (mixed h.c.p, and b.c.c.
structure).
Bethune and Waterhouse showed that friction is strongly linked with the intermolecular
(adhesive) interaction of the surfaces [19]. The relationship between the ability to form such
adhesive interactions and the crystal structure of solids was reported by Karapetian [20]. He
found that the f.c.c, structure has about twice as strong an ability to form adhesive interaction
between contacting surfaces as the h.c.p, structure, and one and a half times as much as the
b.c.c, structure. Thus, aluminium (f.c.c.) might be expected to develop higher adhesion (and
consequently friction) in the fretting junction than Ti-6AI-4V (mixed h.c.p, and b.c.c.)..The
higher coefficient of friction would in turn cause aluminium to be more susceptible to the fretting fatigue process.
Normal Pressure
Contradictory reports have been published about the effect of contact pressure on the fretting process. Some reports confirm the dependence of wear rate on the contact pressure
[21,22], while others claim the wear rate in fretting to be independent of the load applied
[23,24,25].
Rubenstein reported that the coefficient of friction is influenced by the degree of work hardening of the contacting surfaces [26]. For fully work-hardened metals the coefficient of friction
would be independent of the normal load, while for softer, or not fully work-hardened metals,
the dependence of the coefficient of friction on the normal pressure would be expected. Fretting fatigue strength (or fretting wear rate) is in turn influenced by the coefficient of friction.
Thus, for softer and harder materials (more or less work required respectively to fully workharden) a different effect of normal pressure may be expected. This might explain the independence of the fretting wear rate of the contact pressure found for hardened steel [24] and
dependence for mild steel [21].
Magnitude of Slip
The coefficient of friction was observed to reach maximum value (approx. 0.7) at about 10
to 20 microns of slip [27,28], while the fretting fatigue strength was reported to reach the lowest
value for about 10 to 20 microns of slip [29,30]. This striking similarity of both slip ranges
indicates the probability of a direct link between coefficient of friction and the resulting fretting
fatigue strength.
Number of Cycles
The coefficient of friction was reported to rapidly increase initially~ then to decrease uniformly to a stabilized value [28,31]. Similarly, the fretting wear rate was shown to rapidly rise
initially and then to slow down to a steady value [32].
64
Increase in Frequency
The coefficient of friction decreases with an increase in frequency of relative movement
[33,34]. Similarly, the fretting wear rate was found to decrease with increasing frequency up
to about 30 Hz [35]. This effect was attributed to the chemical factor of the oxidation rate.
The higher the frequency, the less time for chemical reaction to occur. No frequency effect was
observed when experiments were conducted in nitrogen.
Fretting fatigue strength increases with an increase in frequency of relative movement
[36]. Endo [33] reported similar results and explained that the frequency affects the oxidation
rate of the contact surfaces, which in turn affects the coefficient of friction and, through that,
frictional stress on the surface.
Roughness of Surjaces
The coefficient of friction was reported to be high for very smooth surfaces [37]. Similarly,
fretting damage was found to be high for smooth surfaces [23,38].
Increase in Oxide Thickness

The coefficient of friction was reported to decrease as the oxide thickness increased [34]. An
increase in oxide layer thickness is thought to have a beneficial effect on the fretting wear rate
due to separation of metal surfaces.
Temperature
Hamdy and Waterhouse reported that the fretting fatigue strength of a nickel-based alloy,
Inconel 718, was doubled by raising the temperature from 20 to 540~ [39]. This result was
attributed to the protective nature of the glaze oxides formed on the sliding interface. It was
found that this glaze oxide produced a low value for the coefficient of friction and this resulted
in low wear rate as well.
Conclusions
A concept of "primary" and "secondary" sets of fretting variables has been proposed. The
"primary" set of variables comprises of the coefficient of friction, magnitude of slip and contact pressure acting at the fretting interface. The "secondary" set of variables affects the fretting
process through changes these variables cause in the "primary" set of variables. The coefficient
of friction has been suggested as the main variable of the fretting process.
The effects that a number of the fretting process variables have on fretting fatigue strength
(or severity of fretting wear) and on the average coefficient of friction within the junction have
been presented. This shows that it is possible to understand, and explain, the effect of many
variables on the fretting fatigue strength through the influence of those variables on the coefficient of friction. Additionally, it suggests that the coefficient of friction within the junction
could be lowered through manipulation of other variables in the fretting process. In engineering practice, such a reduction in the value of the coefficient of friction is usually obtained
directly through the application of low friction coatings or through the introduction of lubricants. Additional methods of controlling the value of the coefficient of friction (and thus the
stress field) within the junction have been presented. This approach requires, however, further
experimental research focused strongly on the coefficient of friction in fretting.
65
References
[1] Collins, J. A. and Marco, S. M., "'The Effect of Stress Direction During Fretting on Subsequent
Fatigue Life," Proceedings, American Society for Testing and Materials, Philadelphia, Vol. 64, 1964,
pp. 547-560.
[2] Uhlig, H. H., Tierney, W. D. and McClellan, A., "Test Equipment for Evaluating Fretting Corrosion," Fretting Corrosion, STP 144, American Society for Testing and Materials, Philadelphia,
1952, pp. 71-81.
[3] Stepanov, V. N., "Specific Features of the Fatigue Curves Under Fretting Corrosion Conditions,"
Soviet Materials Science, Vol. 3, No. 3, 1967, pp. 203-205.
[4] Fitimonov, G. N., "The Selection of a Base Number of Loading Cycles When Testing for Fatigue
Under Fretting Conditions," Soviet Engineering Research, Vol. 1. No. 5, 1981, pp. 27-28.
[5] Waterhouse, R. B., "Fretting and Fatigue," Fretting Corrosion, Pergamon Press, Oxford, 1972, p.
161.
[6] Nishioka, K.and Hirakawa, K., "Fundamental Investigations of Fretting Fatigue (Part 5, The Effect
of Relative Slip Amplitude)," Bulletin (fthe Japanese Society ofMechanical Engineers. Vol. 12, No.
52, 1969, pp. 692-697.
[7] Sato, K., "Damage Formation During Fretting Fatigue," Wear, Vol. 125. 1988, pp. 163-174.
[8] Rooke, D. P. and Jones D. A., "Stress Intensity Factors in Fretting Fatigue," Journal of Strain Analysis, Vol. 14, No. I, 1979, pp. 1-6.
T9] Dekhovich, L. A. and Makhutov, N. A., "Use of Failure Mechanics in Evaluating Fretting Fatigue
Strength," Soviet Materials Science, Vol. 17. No. 3, 1981, pp. 280-283.
[10] Edwards, P. R. and Cook, R., "Fracture Mechanics Prediction of Fretting Fatigue Under Gaussian
Random Loading," Royal Aircraft Establishment, Technical Report 78086, 1978.
[11] Nix, K. J. and Lindley, T. C., "The Influence of Relative Slip Range and Contact Material on the
Fretting Fatigue Properties of 3.5 NiCrMoV Rotor Steel," Wear, Vol. 125, 1988, pp. 147-162.
[12] Switek, W., "Fretting Fatigue Strength of Mechanical Joints," Theoretical and Applied Fracture
Mechanics, Vol. 4, 1985, pp. 59-63.
[13] Hattori, T., Nakamura, M., Sakata, H. and Watanabe, T., "Fretting Fatigue Analysis Using Fracture
Mechanics," Japan Society qf Mechamcal Engineers, Imernational Journal, Vol. 31, Series 1, No.
1, 1988, pp. 100-107.
[14] Jahanmir, S., "The Relationship of Tangential Stress to Wear Particle Formation Mechanism,"
Wear, Vol. 103, No. 3, 1985, pp. 233-252.
[15] Dobromirski, J. and Smith, I. O., "A Stress Analysis of a Shaft With a Press-Fitted Hub Subjected
to Cyclic Axial Loading," International Journal qfMeehanieal Science. Vol. 28, No. 1, 1986, pp.
41-52.
[16] Edwards, P. R. and Cook, R., "Frictional Force Measurements on Fretted Specimens Under Variable Amplitude Loading," Royal Aircraft Establishment, Technical Report 78059, 1978.
[ 17] Bill, R. C., "Fretting Wear and Fretting Fatigue--How Are They Related," Journal of Lubrication
Teehnology, Vol. 105, No. 2, i983, pp. 230-238.
[18] Goss, G. L. and Hoeppner, D. W., "Normal Load Effects in Fretting Fatigue of Titanium and Aluminium Alloys," V~,ar, Vol. 27, No. 2, 1974, pp. 153-159.
[19] Bethune, B. and Waterhouse, R. B., "Adhesion Between Fretting Steel Surfaces," I42,ar, Vol. 8, No.
1, 1965, pp. 22-29.
[20] Karapetian, S. S. and Korosteli, Yu. I., "The Dependence of the Friction Coefficient on the Crystal
Structure of Sotids: Physical Principles," I42,ar, Vol. 85, 1983, pp. 133-141.
[21] Uhlig, H., "Mechanism of Fretting Corrosion," Journal (~[Applied Mechanics, Vol. 21, 1954, pp.
401-407.
[22] Feng, I. M. and Rightmire, B. G., "The Mechanism of Fretting," Lubrication Engineering, Vol. 9,
1953, pp. 134-136 and 158-161.
[23] Tomlinson, G. A., Thorpe, P. L. and Gough, H. J., "'An Investigation of the Fretting Corrosion on
Closely Fitting Surfaces," Proceedings of lhe Institution q/Mechanical Engineers, London, Vol. 141,
1939, pp. 223-249.
[24] Wright, K. H. R., "An Investigation of Fretting Corrosion," Proceedings of the Institution qf
Meehanical Engineers, London. Vol. 1B, 1952, pp. 556-574.
[25] Kusner, D., Pooh, C. and Hoeppner, D. W., "A New Machine For Studying Surface Damage Due
to Wear and Fretting," Materials Evaluation under Fretting Conditions, STP 780, American Society
for Testing and Materials, 1982, pp. 17-29.
[26] Rubenstein, C., "The Influence of Workhardening on the Coefficient of Friction," Wear, Vol. 3, No.
2, 1960, pp. 150-153.
66
[27] Nishioka, K. and Hirakawa, K., "Fundamental Investigations of Fretting Fatigue (Part 3, Some Phenomena and Mechanisms of Surface Cracks)," Bulletin qf the Japanese Society qf Mechanical Engineers, Vol. 12, No. 51, 1969, pp. 397-407.
[28] Endo, K., Goto, H. and Fukunaga, T., "Behaviors of Frictional Force in Fretting Fatigue," Bulletin
qfthe Japanese Society OfMechanical Engineers, Vol. 17, No. 108, 1974, pp. 647-654.
[29] Marsh, K. J., "Research into the Fatigue of Metals," The Australasian Engineer, August 1965. pp.
34-36.
[30] Gaul, D. J. and Duquette, D. J., "The Effect of Fretting and Environment on Fatigue Crack Initiation and Early Propagation in a Quenched and Tempered 4130 Steel," Metallurgical Transactions,
Vol. 11A, 1980, pp. 1555-1561.
[31] Suh, N. P. and Sin, H. C., "The Genesis of Friction," Wear, Vol. 69, 1981, pp. 91-114.
[32] Feng, I. M. and Rightmire, B. G., "An Experimental Study of Fretting," Proceedings"(fthe Institution o['Mechanical Engineers, London, Vol. 170, 1956, pp. 1055-1064.
[33] Endo, K., Goto, H. and Nakamura, T., "Effects of Cycle Frequency on Fretting Fatigue Life of Carbon Steel," Bulletin (fthe Japanese Society qfMechanical Engineers, Vol. 12, No. 54, 1969, pp.
1300-1308.
[34] Iwabuchi, A., Kayaba, T. and Kato, K., "Effect of Atmospheric Pressure on Friction and Wear of
0.45% C Steel in Fretting," Wear, Vol. 91, No. 3, 1983, pp. 289-305.
[35] Feng, I. M. and Uhlig, H. H., "Fretting Corrosion of Mild Steel in Air and in Nitrogen," Journal of
Applied Mechani~', Vol. 21, No. 4, 1954, pp. 395-400.
[36] Waterhouse, R. B., "The Effect of Fretting Corrosion in Fatigue Crack Initiation," Proceedings of
the International ConJerence on Corrosion Fatigue, National Association of Corrosion Engineers,
USA, 1971, pp. 608-616.
[37] Rabinowicz, E., Friction and Wear of Materials, John Wiley & Sons, New York, 1965.
[38] Rahm, E. A. and Wurster, H. J., "Fretting Corrosion in Aircraft and Aircraft Accessories," Lubrication Engineering, Vol. 7, 1951, pp. 22-40.
[39] Hamdy, M. M. and Waterhouse, R. B., "'The Fretting Fatigue Behavior of the Titanium Alloy 1MI
829 at Temperatures up to 600~ '' Fatigue q(Engineering Materials" and Structures, Vol. 5, No. 4,
1982, pp. 267-274.
Fundamental Aspects of Fretting Fatigue

Testing--Mechanics of Contact
D. A. H i l l s 1 a n d D. N o w e l l 1
The Development of a Fretting Fatigue

Experiment with Well-Defined Characteristics
REFERENCE: Hills, D. A. and Nowell, D., "The Development of a Fretting Fatigue Experiment
with Well-Defined Characteristics," Standardization of Fretting Fatigue Test Methods and
Equipment, ASTM STP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society
for Testing and Materials, Philadelphia, 1992, pp. 69-84.
ABSTRACT: It is shown that many of the unknowns present in a rotating-bending fretting

fatigue experiment, fitted with a clamped bridge arrangement, may be avoided by moving to an
entirely different geometry, based on a standard uniaxial tensile test specimen. A pair of cylinders
is pressed against this specimen, producing a Hertzian contact, and a shear force is developed by
impeding axial displacement. The apparatus is run in the partial-slip regime so that the relative
displacement within the contact may be calculated.
The well-defined geometry permits a rigorous analysis of contact pressure, shear traction distribution, interior stress fields, and crack tip stress intensity factors for any cracks present.
Explicit equations for these quantities are developed which enable the physical parameters
responsible for crack initiation, development, and early propagation to be readily obtained.
Refinements to the classicalcontact calculations include analysis of the effects of using elastically
dissimilar contacting elements, the influence of the thickness of the specimen, and the effect of
surface roughness on the contact pressure and the interior stress field. A size effect discovered
using this apparatus described is also reported.
KEY WORDS: fretting fatigue, Hertzian contact, stress intensity factors
A popular method of assessing fretting fatigue, particularly from the materials perspective,
is to use a standard rotating-bendingfatigue testing apparatus, and to clamp two bridge shaped
specimens to the waisted section. As the bending specimen rotates the surfaces experience
alternating compression and tension, so that particles beneath the feet of the bridge tend to
move cyclically towards and away from each other. This produces an alternating shear force
and hence induces fretting. Whilst this type of test is easy to conduct, with readily produced
specimens [1] it is not entirely satisfactory from the mechanics point of view. Normally, flat
ended feet are used on the bridge, with subsequent difficulties in estimating the contact pressure [2] and shear traction distributions. Although a solution for the flat-ended elastic indenter
is available [3], more recent versions of the experiment have used cylindrical feet [4]. However, this does not circumvent another basic drawback of the apparatus, viz that it is not possible to alter the amplitude of fretting displacement other than by changing the span of the
bridge and there are considerable difficulties in establishing the zones of slip and stick [5].
A completely different approach to the design of a fretting apparatus was made by Bramhall
and O'Connor [6, 7] and it is on the basis of their design that the present apparatus was built.
A general view of the apparatus is shown schematically in Fig. la and a photograph of the
actual equipment is reproduced in Fig. 1b. A standard tensile test specimen (A) with a long
waisted section, in this experiment of thickness 12.5 mm, is mounted in a conventional servoLecturer, Department of Engineering Science, University of Oxfor& Oxford, OX1 3PJ, United
Kingdom.
69
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
70
0-
FIG. 1--(a) Schematic experimental configuration. (b) Experimental apparatus.

hydraulic test machine and subjected to a fully reversing tension/compression cycle, inducing
a sinusoidal uniform stress of amplitude, a0. Two blocks (B) are pressed onto this specimen,
one on either side, by compliant springs (C). These are tightened at the beginning of the test
to give the required normal contact force, P, and the use of compliant springs means that there
will be a negligible fall off in load should wear occur; none was detected. As may be seen from
the photograph, the blocks used have their contacting faces machined to circular arcs and a
notionally Hertzian contact results. These blocks are mounted on small rigid bases connected
by diaphragm springs to a table (D). Thus, the cylinders are able to displace perpendicular to
the axis of the specimen, but are probibited from rotating. As the main specimen is loaded the
ensuing strains mean that there is a tendency for the cylinders to move in the direction of the
specimen's axis together with the corresponding points on the surface of the specimen, but
they are inhibited by stiffsprings (E) which take the form of beams. Thus, a shear force, Q, is
developed in phase with the bulk loading. By adjusting the stiffness of springs (E) the relative
amounts of bulk tension and shear force induced may be varied, Values of these quantities are
monitored continuously by strain gauges.
The width of the cylinders was chosen to be the same as that of the specimen; had they been
narrower or wider, a singularity in the contact pressure at the edge o f the contact would have
occurred [8]. The present configuration is imperfect insofar as the central region of the contact
suffers plane strain while the ends are in plane stress. This relaxation of transverse constraint
means that the contact patch is slightly wider at the ends, but the contact pressure falls slightly.
A two-dimensional geometry was chosen for two reasons. First, all subsequent analysis is
HILLS AND NOWELL ON FRETTING FATIGUE EXPERIMENT
"71
much easier to carry out and permits more accurate results than for three-dimensional problems. Secondly, one object of the initial set of experiments was to investigate a size effect. For
a given pair of materials a plane Hertzian contact gives rise to a peak contact pressure P0 proportional to k/P/R (where R is the relative radius of curvature at the contact) [9]. The contact
half width, a, is however proportional to x/PR. Hence by conducting tests with cylinders of
differing radii but adjusting the contact load, it is possible to conduct a range of tests in which
the peak contact pressure (and the magnitude & t h e internal stress field) is held constant, while
the value of a (and the size of the contact and its resulting stress field) is varied. Had we decided
to use barrelled or spherical contacts, the peak contact pressure would have been proportional
to 3,~/p/R and a characteristic contact dimension would have been proportional to 3x/p-R.
Thus, an unrealistically large range of contact loads would have been needed to achieve an
acceptable range of contact sizes. In view of these observations it was thought that the slight
reduction in contact pressure at the ends (to about ( 1 - v2) of the pressure over the central
portion, v, being Poisson's ratio) was acceptable. Very little problem with corner-initiated failure was experienced and a typical failed specimen is shown in Fig. 2.
An initial batch of five series of tests was carried out to investigate the influence of contact
size on fretting fatigue life using specimens and pads made from HE15-TF, an Al/4wt%Cu
alloy. The values of contact pressure, bulk tension, and range of calculated contact widths used
are given in Table 1. The radius &curvature of the fretting pads was varied between 12.5 mm
and 125 m m to achieve the required variation in contact width. Since the experiments were
conducted in the partial slip regime, very little wear took place and the change in contact width
due to wear was likely to be small. Tests were stopped at 107 cycles, if prior failure had not
occurred, and the results of the lives found are shown in Fig. 3. The reproducibility of the
FIG. 2--Failed /)ezlin,~ /hli~,,ue.ST~C'cimen.

72
--RETTING FATIGUE TEST METHODS AND EQUIPMENT
TABLE l - - E x p e r i m e n t a l parameters.
Series
Stress
Amplitude,
(to ( M N / m 2)
Peak Normal
Pressure, P0
( M N / m 2)
Q/P
fna
acrit (mm) b
1
2
3
4
5
92.7
92.7
92.7
77.2
61.8
157
143
143
143
120
0.45
0.24
0.45
0.45
0.45
0.75
0.75
0.75
0.75
0.75
0.28-0.38
0.54-0.72
0.18-0.27
0.36-0.54
0.57-0.71
a acn, is the contact half width at which transition from long to

short fatigue life took place.
b f, is an estimate for the coefficient of friction in the slip zones
Eq 34.
results gives us confidence in the reliability of the design of the apparatus and the concept of
using a well-defined stress field. O n e of the problems of using a flat-footed bridge specimen is
that the contact pressure is critically dependent on both m a c h i n i n g imperfections and interfacial friction. For perfect adhesion between the bridge and the specimen asymptotic analysis
predicts a singularity in pressure at the edge of the contact [8]. In practice this would be
12.0
12.0
~ z~ ~
10.0 -
Series
Series
10.0
o Series
+ Series
.-g.
q3
-6
~) 8 . 0 -
-d
~D
(D
8.0
>~
L)
6.0
6.0-
O
4.0
4.0
o3
_J
2.0
2.0
0.0
0.00
0.50
1.o0
1,50
Contact s e m i - w i d t h a (mm)
0.0
0.00
0.50
Series
8.0-
O
>,
o
~o
C)
6.0
4.0
09
,42.0
0.0
0.00
0.50
Contact
1.00
1.50
Contact s e m i - w i d t h a (ram)
12.0 i
10.0
%.00
s e m i - w i d t h o (ram)
Copyright by ASTM Int'l (all rights reserved); FIG

Mon Mar
28 00:05:25 EDTin
2016
"{--Vr~riation
fh/i~u~~/ i ~ with conlact ~ize.
73
retrieved by slippage, localized flaws, and possibly plasticity, b u t the additional number of
unknowns is apparent.
Although not central to this paper, it may be noted that above a certain contact size, act,,
typically about 0.4 mm, the life of the specimen was strictly finite, about a million cycles, while
below this it was infinite ( > 107 cycles). It was the desire to explain this phenomenon which
led us to analyze the contact problem in some detail as described in the following sections.
Contact Pressure
It was stated earlier that the contact was "notionally Hertzian." It is not precisely so because
one of the components has a strictly finite thickness in comparison with the contact half-width.
A detailed investigation of the influence of strip thickness was carried out [ 10] and it was found
that, providing the strip thickness exceeded about five times the contact half-width, the idealization as a Hcrtzian contact is a very good one. This certainly held true for the tests
described. For completeness the influence of a mismatch of materials on contact pressure was
also evaluated [11]. It was found that the influence on pressure distribution was negligible for
all combinations of materials.
The contact pressure distribution may therefore always be taken to be (Fig. 4)
p ( x ) = Po ~/1 -- ( x / a ) 2
Ixl ~ a
(1)
where P0, the peak contact pressure, is readily found from the standard results [9].
Shear Traction Distribution
An essential feature of our experiments was that they were all carried out in the partial slip
regime (i.e., gross sliding did not occur once a steady state had been reached). At first sight this
would appear to be the classical Mindlin-Cattaneo problem [12,13] in which a Hertzian contact is subjected to an oscillating shearing force in the plane of the contact. The main difference
in the present configuration is that as the shear force, Q, is applied a significant tension, ~0, is
developed in one of the bodies (Fig. 4) and this may be expected to modify the classical theory
significantly.
o#:
slick
zone e+c /
e-c
~//< " / / / / A
l
Cl
-a
1"//. /////,t
|
FIG. 4--Contact con[iguration.
74
First, for comparison, we record the Mindlin-Cattaneo solution. The shear traction distribution is given by
q(x) = --fPo ~/1 -- (x/a) 2 + q'(x)
Ix[ _< a
(2)
where
q'(x) = 0
(slip zones)
c ~ Ixl ~ a
q'(x) = Jl)o ( C ) ~/ l - (x/c) 2
(stick zone)
(3)
]xl --< c
(4)
where the stick zone size is determined from

2
(5)
a n d f i s the coefficient of friction.

We shall tentatively assume that the effect of the bulk tension is to displace the stick zone
from the center line by an amount, e. We may then retain Eq 2 and write in lieu of Eq 3:
q'(x) = 0
(slip zones)
Ix -- e[ _< c
(6)
Furthermore, the slip directions must be consistent in the slip zones, i.e.,
(0
sgn ~ ( u , -- u2 = --sgn(q(x))
I x - - el > c
(7)
where ui is the x-direction displacement of surface particles of body i (Fig. 4) and t is time.
Within the stick zones we must ensure that surface particles remain at the same separation or
strain, i.e.,
Out
Ox
Ou2
-0
Ox
Ix-el--<c
(8)
The surface displacements are related to the traction distribution by [14]
Ou, _
Ox
Ou2 _
Ox
2(1 -
(1 -- 2,)(1 + ,)
(1 - 2u)(l +
F~
p(x)
2) f ] , q(f)d~
ax - ~
2(1 -- ~,2) f ~ q ( f ) d f
~) p(x) + ~
~o
x--- ~ + -~(1 - ~2)
(9)
(10)
where in Eq 10 the effect of bulk tension has been superimposed. Substituting Eqs 9 and 10
into Eq 8 and replacing q(x) as defined by Eq 2 gives
fe+~ q'(f) d f
-~.x----~-
fpo~
OoTr
jx - ef < c
(11)
75
This is an integral equation in q(~-) which may be inverted [15] to give
q'(x)=fpo(c)
~1
- (e-~cX) 2
(12)
where
o-oa
e = --
(13)
4fpo
The requirement that the slip velocity be consistent with the shear traction (Eq 7) may be
checked a posteriori by considering the loading history. The above result is valid, provided
that
o"0
4fpo
~<1 -
\/1
(14)
lP
Save for Series 2 tests this inequality holds for all the tests conducted. When this inequality
does not hold, it is not possible to invert Eq 11 analytically and a numerical method must be
used. Details of the solution in this case are given in [16]. For the case when inequality (Eq 14)
is satisfied it may be seen that the sole effect of the bulk tension is to displace the stick zone,
within which the sliding shear traction is still abated by an elliptical distribution. The effects
of reversing the shearing traction and simultaneously reversing the bulk tension, as occurs
under the experimental conditions of cyclic loading, were also analyzed [16] and the solution
for one case is depicted in Fig. 5.
q
fPo
1.0
/\
I
(IS
I I
I~
.._...__ Xla
03
-O.5 I- J
[ __
\1
( ~ l"ll il X
0
-(~ max
"1.0
FIG. 51Shear traction distribution at maximum forward load ( Q ~ ) and during unloading. (Q,,o~/fP
= 0.5, c~olfpo = 0.55).
76
State of Stress
When the surface traction distribution has been established the next step is to determine the
stress and strain fields. This enables the surface displacements (and, hence, the frictional dissipation over the contact patch) and the probable trajectories of crack growth to be determined. As it has been established that a half-plane idealization is acceptable, it is straightforward to find the state of stress from that for a sliding Hertzian contact [17,18], These solutions
C o n t a c t patch
y/a
1.5
2.3
C o n t o u r interval:O.O9Po
3-2
-1'.5
-;
-d.~
Contact
y/a
a . . . .
-'-2
t~
x/a
patch
-1.5
~5
1,
xta
FIG. 6--(a) Contours of most positive principalstress (c~/po)for series I experiments. (b)Principal stress
directions for series 1 experiments.
77
assume that both the pressure and shear traction are elliptical. Hence, by using superposition
the influence of the stick zone is readily incorporated.
Figure 6 shows contours of the most positive principal stress and stress trajectories. The latter is a set of orthogonal curves whose tangents at any point lie in the principal directions, and
hence give an indication of the probable direction of crack growth. It will be seen that the zone
of greatest tension lies on the surface "behind" the contacting cylinder, i.e., at x / a = - 1,
y = 0. This is the region where cracks are observed to initiate. They also grow initially at about
45" to the free surface, which is to be anticipated as this is a plane of maximum shear.
When the surface shear traction distribution is not given by a combination of ellipses, i.e.,
when a region of reverse slip exists, it is necessary to integrate the Flamant solution for a line
force [9] numerically. If we now restrict attention to the point of maximum stress ( - a , 0 ) , the
in-plane stress Crxxmay be found from
~xx
2 ~a q(x)
; j_,x~
adX
(15)
and the effect of the pressure added separately. If q(x) is now written as a perturbation on the
sliding solution (Eq 2), Eq 15 may be rewritten as
~xx = 2j,bo -
2 Ic'' q'(x) d x
7r J - a X + a
(16)
The maximum tension found is shown in Fig. 7. It can be seen that the bulk stress ~r0increases
the magnitude of the stress component due to Q in addition to its direct effect.
Detailed knowledge of the complete stress field is useful in that the probable location of
crack initiation may be located (where the tension is a maximum and where localized plasticity
provides the right conditions for dislocation coalescence), but it is not helpful in predicting the
rate of growth when once a crack has developed. To proceed further we must postulate the
existence of a crack and determine the corresponding crack tip stress intensity factors.
O'xXC
- -
O
f---~-
f Po
2.0
10
1.8
1,6. ~
1.41.2- ~
1.O.
O.~
>
11o
2o
FIG. 7--Variation o/stress component at trailing edge (~xxq) with c~o.
78
Stress Intensity Factors

In fretting fatigue it is c o m m o n for a number of microcracks to develop initially and at first
the situation is quite complex. Soon, however, one of the cracks becomes dominant and a
satisfactory first approximation to this situation is to ignore the smaller micro-cracks. Let us
suppose that a single crack, initially plane, grows from the trailing edge of the contact at some
inclination to the surface (Fig. 8). We first determine the state of stress occurring in the crack's
absence and, in particular, determine the traction components of stress present along the line
of the crack, 5yy, ~xy,and which violate the requirement that the faces of the crack must remain
free of traction. There are several ways of cancelling the unsatisfied stresses present, but the
most efficient way would appear to be to distribute an array of displacement discontinuities,
or dislocations, along the line of the crack.
This technique is made possible by the closed form solution for the state of stress induced
by a dislocation present in a half-plane [19,20]. Even though the solution is readily written
down in terms of elementary functions, the expressions involved are lengthy and we shall here
simply write down the stresses induced by a dislocation at a depth h beneath the surface as
(Fig. 8)
l a y ~ l (x,y) - 7r(K + 1)
G~xv(x,y,h)
Oyx~(X,y,h)
where bx, by are the Burger's vector components of the dislocation, the functions G,jk(x,y,h)
are given in [21], and K = 3 - 4, in plane strain.
The next step is to transform the state of stress (left hand side of Eq 17) into local inclined
coordinates and rotate the Burger's components (right hand side ofEq 17) by the usual transformation rules to obtain a new matrix of coefficients, defined by
We can now distribute dislocations along the length of the crack and determine the resultant
normal and shear tractions, N(x), S(x), due to both the applied stresses 5;;, ~y, and the dis-
FIG. 8--Crack geomelrv.

79
locations. These are then set to zero giving
N(2) = a~" + J~{f~Gx.,(2,f:,h)B,.(h)dh+f~Gvv,(2,f:,h)B,(h)dh]=Orr(~

+ 1)
. . . .
(19)
S(2) = r~" + ~[;[oG~v(2,gh)B,.(h)dh+f~O,,,.y(2,.9,h)Bv(h)dh}=Orr(~

+ 1)
.
.
.
.
.
.
(20)
where the dislocation densities are defined by
/~,(h) = ~ (h)
i= x or y
(21)
and whose integrals represent the crack opening displacement and shear displacement, respectively. It is not possible to invert coupled integral Equations 19 and 20 analytically, but an
efficient quadrature which permits a numerical solution is to use Jacobi polynomials, as
described by Erdogan, Gupta, and Cook [22]. The discretized version of these equations is
then
#
2r
+ u,){(~xy~.(2~,/i,)4)d/~,) + Gm,(2k,h,)0~(/~,)} =
~r(K + 1) i=1 2 N + 1
"
u
71"(K -{- 1 j t=l
-g~i,;(2k)
(22)
2~r(1 + u,) {0x,.(2~f,)0~(/~,) + 0~,xv(2kf)0,,(/~,)} = -~.~i(xk)

2N + 1
. . . . .
(23)
where
217,
~-=
((2i--1)rr 1
1 + cos\ ~-N~ i /
i = 1,2 . . . . N
(25)
k = 1,2 . . . . N
(26)
and
22k
~-=
( 2krr I
1 + cos \ 2 ~ - + ~ J
Equations 22 and 23 are two sets of 2N simultaneous algebraic equations in the 2N

unknowns (hx(/4,), 0v(/l,), from which the crack tip stress intensity factors may be deduced
directly. They are given by [21]
K~ = 2 ~
u
4,j(l)
~'(K + 1)
j =
xory
(27)
where Kx is the shear mode stress intensity and Ky is the corresponding opening mode quantity.
The values of 4),(1) may be found from [23]
~b,(1) - 2N + 1
cot \ ( - ~ _
i~/sin
(2i - 1)Tr 4,(/~,)
(28)
80
The technique just described, though at first sight rather complex, is in reality a very efficient
way to find crack tip stress intensities. It is particularly suited to problems where ~i., ~ vary
rapidly, as they do in the neighborhood o f an elastic contact, and convergence is extremely
rapid, with Nless than 20 often proving adequate. In a series of papers this method was applied
to an increasingly complex set of problems, including the open crack [21], closed crack [24],
and kinked crack [25]. The last is of some interest to the fretting fatigue problem, where the
crack normally turns from stage 1 (mode II) to stage 2 (mode I) at a short distance beneath the
surface. The depth at which this occurs varies with test geometry. In our experiments it was
found to take place about 0.5 m m beneath the surface. Criteria for the turning of cracks from
shear to opening mode growth are not fully established, but would appear to relate to the maxi m u m release of crack tip strain energy, as well as being influenced by crystallographic factors.
In earlier work, we had examined the history of the crack tip stress intensity variation as the
applied load was taken through a complete cycle [26,27]. This included calculating the
"locked-in" K . stress intensity during the part of the cycle when the crack faces are pressed
together. This is a very involved computation, and we are now of the opinion that it is unnecessary, as most of the crack tip damage will occur when the crack is open (the m i n i m u m value
of K~ is always zero). The calculation detailed above will, therefore, suffice and a polar plot
showing the variation of crack tip stress intensities with angle for a crack emanating from the
edge of the contact is shown in Fig. 9 for a typical case. As may be seen, the maximum value
of K , occurs when the crack is inclined at about 45 ~ to the free surface, the maximum where
the crack is inclined under the contact being the greater of the two. The mode I stress intensity
is a m a x i m u m for an approximately normal crack and it may be noted that for cracks inclined
beneath the contact the value of K~ becomes negative, indicating that the crack tips are closed.
In previous analyses we have assumed that Coulomb friction obtains between the crack faces
[26,27], but it is clear that real crack faces are far from plane or smooth, so that the lightest
compression will cause locking of the faces and inhibit any mode II stress intensification. It
would, therefore, seem that the most likely plane on which a crack might form is one at 45 ~to
the surface and projecting beneath the contact. However, it is also true that this region is perilously close to a closure region, so that if the transition from mode II to mode I growth does
not occur sufficiently early self arrest will result.
O.B
0.6
0.4
0.2
0.0
0.2
0.4
I
. . . . . . . . IKIII
"-~/
\\
0.6
i
0.8
-~\\ Crack t~p closure

-
\\
\\
66 ~
~ ~ '~
~ , 46o
60 ~
-2o~
C r a c k angle e
29 ~
;o
FIG. 9--Polar plot ( f stress intensity Jactors ,[or cracks propagating,from the trailing edge of contact
(series 3 experiments, normalized with respect to contact half-width, a).
81
Consideration of Friction
The mechanics of the fretting test are characterized by the parameters a0, P, Q, f All save
the last are under the direct control of the investigator, who must accept whatever value results
from the surface finish, cleaning, and choice of materials used. One might presumably expect
the coefficient of friction to be constant over the contact area at the commencement of the test,
with a typical value of the order of 0.3. The current tests were conducted under a notionally
fixed imposed tangential displacement and this is sufficiently large to cause the shear force, Q,
to reach the limiting value f P in the early stages. Under sliding conditions the contacting surfaces are modified by plastic flow and the coefficient of friction increases, uniformly if the variation of pressure is assumed not to have any effect, until the point is reached when Q = JP.
F r o m this point onwards further increases in the coefficient of friction will result in the formation of a stick zone [28] and surface modification (with a consequent change in friction
coefficient) can only occur within the slip zone, where relative displacement of surface particles
occurs. After n cycles, the variation of the coefficient of friction may appear as shown in
Fig. 10. Let us assume, for simplicity, that the standard Mindlin-Cattaneo configuration of
stick and slip zones is present. Then, from Eq 5
2
= 1
(29)
f~p
where the subscript n is added to denote a parameter's value after n cycles of loading. The
average coefficient of friction, ~Cmay be defined by
.?=
f~
q(x)dx
(30)
f~p(x)dx
where
(31)
q(x) = J ( x ) p ( x )
r---.-
0
V//////////A
Y///////////A
-a
-C
-- x
Cn
fo
FIG. 10--Variation in coefficient o[[riction after n cycles.
82
FRETTING FATIGUE TEST M E T H O D S AND EQUIPMENT
Substituting Eq 1 into Eq 30 and carrying out the integration yields
--4- =
fo k/1 -- (x/a) 2 dx +
. f(x) ~/1 -- (x/a) 2 dx

+
f~ ~/1 -- (x/a)Zdx
(32)
If now Eq 32 is differentiated with respect to n and Eq 29 is used to eliminate c,, we find
,-f = f
2re r
1--f@p+Sin-'
1--f~
df.
(33)
The integration may now be carried out and by substituting info = Q/P (i.e., if there is initial
sliding) we arrive at
) ~ = f n - 2Qrep - 2sina + 21n tan
+-Q-a-
tanc~
(34)
where
(35)
COS~ ~
which is a closed form relationship between the true coefficient of friction, f,, in the slip zones
and the average coefficient of friction, f, which would be revealed by stopping the experiment
and measuring the limiting friction coefficient by increasing Q until sliding occurs. This relationship is plotted in Fig. 11.
Slip z o n e f r i c t i o n
c o e f f i c i e n f f.
0.2
1.0
Q/fnP
0-4
0-6
0.8
0.8
0.6.
0.4-
0.2-
0"0
o-o
o!2
o'.~
o16
o.B
Mean frlcfion coefficlenf f"

FIG.
l l--Relationship between mean [kielion coefficient-/'and slip zone value j~ ./or initially sliding
contacts.
83
For Series 2 experiments such a sliding test was conducted. The displacement imposed was
kept small so as to retain the same surfaces in registration and an average coefficient of friction
o f ) c= 0.42 was found. As Q/P = 0.25 we find from Eq 34 that the true coefficient of friction
in the slip regions was about 0.75. This is very similar to that found in subsidiary tests where
Q was monitored under continuous sliding conditions. In these tests the coefficient of friction
was found to rise very quickly during the first 50 cycles and to have reached a steady state
within about 200 cycles.
Refinements to the Analysis

Various refinements to the above results have been carried out. First, if it is desired to assess
the effects of fretting fatigue between dissimilar materials it must be noted that there will be a
small influence on the contact pressure, but a very significant change in the stick/slip zone
regime. When the cylinders are initially pressed onto the tensile test specimen slip zones of
opposite sign will form at the edges of the contact zone [1I]. Then, as the tangential load is
imposed, one slip zone will adhere before giving way to slip in the opposite sense. Shakedown
to a steady, fully reversing pattern of stick and slip zones will occur very quickly, within the
first two or three cycles of loading. Depending on the combination of loading parameters the
steady state arrangement of stick and slip zones may be similar to the Mindlin case.
Secondly, the question of surface finish needs to be addressed, as at the smaller .contact
widths there may be only a few asperities in contact; thus the stress field may be very different
from that predicted for perfectly smooth bodies [29].
Conclusions
A greater understanding of the process of fretting fatigue will only be achieved when the
techniques of stress analysis and fracture mechanics are applied to the problem. It is, therefore,
essential that any adopted standard test geometry should be amenable to such analysis. The
authors have described a fretting fatigue test which may be carried out on a standard uniaxial
hydraulic test machine, which permits independent regulation of the controlling parameters
within certain limits. It has been shown that the ensuing geometry may be analyzed, permitting precise and in some cases closed-form display of the contact pressure, shear traction distribution, stick and slip zone location, internal stress field, crack tip stress intensity factors, and
a reconciliation of the varying coefficient of friction with the mean measured value. Thus, an
experimental geometry and analysis techniques have been established which will permit an
enhanced understanding of the fretting fatigue process.
References
[1] Fenner, A. J. and Field, J. E., "A study of the onset of fatigue damage due to fretting," Proceedings,
North East Coast Institute of Engineers and Shipbuilders, Vol. 76, 1960, p. 183.
[2] Edwards, P. R., "The application of fracture mechanics to predicting fretting fatigue," Fretting
Fatigue, Ed. R. B. Waterhouse, Applied Science, London, 1981, pp. 67-98.
[3] Khadem, R. and O'Connor, J. J., "Adhesive or frietionless compression of an elastic rectangle
between two identical elastic half-spaces," International Journal of Engineering Science, Vol. 7,
1969, pp. 153-168.
[4] Milestone, W. D., "Fretting and fretting fatigue in metal to metal contacts," Proceedings, AIAA
Structural Dynamics and Materials Conference, Denver, Colorado, 1970, p. 86.
[5] Doeser, B., "The study of fretting fatigue using finite element analysis and electron microscopy,"
PhD thesis, Nottingham University, 198 I.
[6] Bramhall, R., "Studies in fretting fatigue," D Phil thesis, Oxford University, 1973.
[ 7] O'Connor, J. J., "'The role &elastic stress analysis in the interpretation of fretting fatigue," Fretting
Fatigue, Ed. R. B. Waterhouse, Applied Science, London, 1981, pp. 23-66.
84
[8] Dundurs, J. and Lee, M-S., "Stress concentrations at a sharp edge in contact problems," Journal of
Elasticity, Vol. 2, 1972, p. 109.
[9] Timoshenko, S. P. and Goodier, J. N., Theory of Elasticity, 3rd Ed., McGraw-Hill, 1970, pp. 409420.
[10] Nowell, D. and Hills, D. A., "Contact problems incorporating elastic layers," International Journal
of Solids and Structures, Vol. 24, No. 1, 1988, pp. 105-115.
[11] Nowell, D., Hills, D. A. and Sackfield, A., "Contact of dissimilar elastic cylinders under normal and
tangential loading," Journal of the Mechanics and Physics ofSohds, Vol. 36, No. 1, 1988, pp. 5975.
[12] Mindlin, R. D., "Compliance of elastic bodies in contact," Journal of'Applied Mechanics, Vol. 16,
1949, pp. 259-268.
[13] Cattaneo, C., "Sul Contatto di due corpi elastici: distribuzione locale degli sforzi," Reconditi del
Accademia Nazionale de Lincei, Vol. 27, 1938, pp. 342-348,434-436,474-478.
[14] Muskhelishvili, N. I,, "Some basic problems of the mathematical theory cf elastlcity, "Noordhoff~
Gronigen, 1953.
[15] Muskhelishvili, N. I., "Singular integral equations, boundary problems q/i/hnction theory and their
application to mathematical physics, "Noordhofl, Gronigen, 1953.
[16] Nowell, D. and Hills, D. A., "Mechanics of fretting fatigue tests," International Journal (~[MechanicalSciences, Vol. 29, No. 5, 1987, pp. 355-365.
[17] M'Ewan, E., "Stresses in elastic cylinders in contact along a generatrix," Philosophical Magazine,
Vol. 40, 1949, pp. 454-459.
[18] Sackfield, A. and Hills, D. A., "A note on the Hertz contact problem: a correlation of standard formula," Journal of Strain Analysis, Vol. 18, No. 3, 1983, pp. 195-197.
[19] Dundurs, J. and Mura, T., "Interaction between an edge dislocation and a circular inclusion," Journal of the Mechanics and Physics oJ'Solids, Vol. 12, 1964, pp. 177- 189.
[20] Dundurs, J. and Sendeckyj, G. P., "Behavior of an edge dislocation near a bimetallic interface,"
Journal o]Applied Physics, Vol. 36, 1965, pp. 3353-3354.
[21] Nowell, D. and Hills, D. A., "Open cracks at or near free edges," Journal ( f Strain Analysis, Vol. 22,
No. 3, 1987, pp. 177-185.
[22] Erdogan, F., Gupta, G. D., and Cook, T. S., "Numerical solution of singular integral equations,"
Methods of Analysis and Solutions of Crack Problems, Ed. G. C. Sih, Noordhoff, Leyden, pp. 368425.
[23] Krenk, S., "On the use ofthe interpolation, polynomial for solutions of singular integral equations,"
Quarterly of Applied Mathematics, Vol. 32, 1975, pp. 479-484.
[24] Hills, D. A. and Nowell, D., "Stress intensity calibrations for closed cracks," Journal of Strain Anal
ysis, Vol. 24, No. 1, 1989, pp. 37-43.
[25] Li Yingzhi and Hills, D. A., "Stress intensity factor solutions for kinked surface cracks," Journal of
Strain Analysis, Vol. 25, No. 1, 1990, pp. 21-27.
[26] His D. A. and Cmninu M. ``An anaysis f fretting fatigue cracks during ading phase International Journal of Solids and Structures, Vol. 21, No. 7, 1985, pp. 721-730.
[27] Sheppard, S., Hills, D. A. and Barber, J. R., "An analysis of fretting fatigue cracks, part 2--unloading
and reloading phases," International Journal of Solids and Structures, Vol. 22, No. 4, 1986, pp. 387396.
[28] Hills, D. A., Nowell, D. and Sackfield, A., "Surface fatigue considerations in fretting," Interface
Dynamics, Proc 14th Leeds-Lyon Symposium on Tribology, Elsevier, Amsterdam, 1988.
[29] Nowell, D. and Hills, D. A., "Hertzian contact of ground surfaces," Journal ofTribology, Vol. 111,
No. 1, 1989, pp. 175-179.
Kenkichi Sato 1
Determination and Control of Contact

Pressure Distribution in Fretting Fatigue
REFERENCE: Sato, K., "Determination and Control of Contact Pressure Distribution in Fretting Fatigue," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM STP
1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials,
Philadelphia, 1992, pp. 85-100.
ABSTRACT: Contact pressure is one of the most important factors in fretting fatigue. The contact pressure distribution on loading of applied cyclic stress is markedly different from that on
unloading. To elucidate the effect of contact pressure on fretting fatigue strength, the contact
pressure distribution should be exactly determined. In this paper, the contact pressure distribution was calculated by using a boundary element method and monitored by an experimental
method of caustics. Differences in the contact pressure distribution between the shapes of fretting
bridges whose height is 10, l 5, 20, and 30 mm at a constant span of 30 mm, and between two
types of clamping, which are center clamping and edge clamping at the back face of the bridge,
were analyzed. In order to examine the effect of the contact pressure distribution on the fretting
fatigue strength, fretting fatigue tests in plane bending were carried out. The S-N properties
obtained from fretting fatigue tests could be estimated analytically by equivalent stress
amplitude.
KEY WORDS: fretting fatigue, contact pressure, boundary element method, method of caustics,
S-N plots
Fretting fatigue failure occurs at mating parts subjected to microslip due to applied cyclic
stress or vibration. It has been shown that fretting fatigue strength is affected by m a n y factors,
n a m e l y contact pressure, cyclic stress, relative slip amplitude, material, environment, temperature, a n d so on. Contact pressure has a significant effect on fatigue strength due to fretting.
Therefore, it is valuable to understand the effect of contact pressure on fretting fatigue, from
the standpoints of design and maintenance, and also establishing a fretting fatigue test method.
The effect of contact pressure has been studied by m a n y researchers. Nishioka and Hirakawa
[1] showed that contact pressure, c o m b i n e d with the value of relative slip, affects fretting
fatigue strength. Goss and Hoeppner [2] showed from the experiments with Ti-6A1-4V and
7075-T6 that the effect of contact pressure depends on materials. Lindley and Nix [3] observed
that the effect of contact pressure on fretting fatigue strength at 10 s cycles of 3.5Ni-Cr-Mo-V
steel is very small. Field a n d Waters [4] studied the effect of clamping positions, at the center
and near one end of the fretting bridge, on the fretting fatigue strength, The author et al. [5]
observed the effect of contact pressure on crack propagation behavior in fretting fatigue of a
stainless steel. Wright and O ' C o n n o r [6] reported the variation of contact pressure distribution
on loading and unloading in tension, bending, and torsion by using a finite element method.
The author and W a t a n a b e [7] proposed a method for estimating fretting fatigue strength by
using the equivalent stress amplitude which was calculated by considering the variation in contact pressure.
1Associate Professor, Department of Mechanical Engineering, Chiba University, 1-33 Yayoi-cho,
Inage-Ku, Chiba, 263 Japan.
85
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
86
As mentioned above, contact pressure is an important factor in fretting fatigue. Fretting

fatigue properties obtained from fatigue tests are affected by the magnitude and distribution
of contact pressure. The distribution of contact pressure is changed by loading types, namely
tension-tension and bending-loading. Since a marked change in the distribution during a loading cycle occurs in bending, the plane bending fatigue loading was selected in this paper. The
magnitude and distribution of contact pressure were obtained by using a boundary element
method. The contact pressure distribution at the contact surfaces was calculated for center
clamping and edge clamping (i.e., clamped at the center point or at the edges of the bridge
back face). The variation of contact pressure with bending stress, bending direction, clamping
method, and shape and dimensions of the bridge is discussed. The distribution of contact pressure was monitored by an experimental method of caustics. Fretting fatigue tests were also
carried out to confirm their effects on fretting fatigue strength.
Boundary Element Analysis of Contact Stresses

Contact Omditions and Loading Conditions
Fretting fatigue in plane bending, as shown in Fig. 1, was studied analytically and experimentally. The specimen, which is compressed by fretting bridges, was subjected to bending by
a force (displacement) at the end. The effects of four types of fretting bridges and two types of
clamping on the contact pressure distributions were studied. The shape and dimensions of the
bridges are shown in Fig. 1. The fretting bridge height was varied from 10 to 30 mm at a constant span of 30 mm and a contact length of 5 mm.
Contact Stress Analysis
For the geometry being studied, the stress components under fretting fatigue in bending are
shown in Fig. 2. They arise from the applied bending stress (Sb) and nominal contact pressure
Specirne~
"~
Fretting
h :10 mm
15
20
30
.... ~,- Center d G m p i n g
D Edge damping
Copyright by ASTM
Int'l1(all
rights reserved);illuslration
Mon Mar 28 00:05:25
EDT 2016
FIG.
--Schematic
dloading
conditions" underfrettingfatigue
in plane bending.
SATO ON CONTACT PRESSURE DISTRIBUTION
87
11 IP~,~1,~ Fretting bridge

Y~~/Specimen
Tens.interfoce ~Sxt -"~E~'J
......
OI Syt~Tx y t____~ ,,;/~7T
Comp;interfoceTXYC-~Sxc---~ "x
ttttttt
FIG. 2--Stress components underfretting fatigue in bending.
(P0)- Contact stresses, composed of contact pressure and frictional shear stress, were calculated
by using a boundary element method (BEM) into which a penalty function was introduced.
The mesh divisions of the model are shown in Fig. 3. To simplify the calculations while retaining reasonable accuracy, only half of the bridge geometry was considered with a shortened
specimen length of I 0 m m from the outer edge. The loading condition in bending was replaced
9-
4-
h=]O
mm
Dorr~in 2
k 5
15
.T
lO
Copyright by ASTM Int'l (all rights reserved);

Mon Mar 28divisions
00:05:25 EDT
2016 mode/used m the BEM analysis.
FIG. 3--Mesh
o/tbe
88
FRETTINGFATIGUETEST METHODSAND EQUIPMENT
by an equivalent set of applied forces, composed of tension, compression, and shear forces
applied at the end of the model (Fig. 3).
The analysis was done under plane stress conditions although plane strain conditions are
predominant. This assumption was introduced to simplify the calculations and due to the fact
that fretting fatigue cracks initiated near the contact edge and propagate initially along a plane
inclined to the contact surface, which corresponds to a yielding plane governed by Tresca's
criterion. The contact stress distributions at the tensile interface and compressive interface in
bending were solved simultaneously after some iterations. A penalty number of the magnitude
of 107 was used. Material constants were a Young's modulus and Poisson's ratio of 69 GPa
and 0.33, respectively.
Calculated Contact Pressure Distribution
Figure 4 shows the contact pressure distribution calculated for a nominal contact pressure
of p0 = 50 MPa, a coefficient of f r i c t i o n f = 0.5, and no applied bending stress. The value for
the coefficient of friction of 0.5 was obtained from previous experiments using aluminum
alloys under no lubrication condition. The heavy lines correspond to the distributions under
center clamping and the light lines to those under edge clamping. The maximum pressure on
the contact interface occurred at the inner edge, x / a = -- 1, with center clamping (for all values
3
2
Center damping [
Edge clamping J
Sb: OMPa J
po=50MPa
f =0.5
I I----- I U
cs
1
0
-
0
X/a
x/a
3
h=2Omm
h=3Omm
u-
X/a
X/a
FIG. 4--Calculated contact pressure distribution with bridge heig'hl ~fh = 10, 15, 20, and 30 mm.
SATO ON CONTACTPRESSUREDISTRIBUTION
89
of 10 =<- h < 30 mm) and with edge clamping for bridge heights of 15, 20, and 30 ram. At a
bridge height h = 10 mm, the m a x i m u m contact pressure occurs at the inner edge, x / a =
1, with center clamping, and at the outer edge, x / a = 1, with edge clamping. Figure 4 shows
also that for h = 10 ram, and in the case of center clamping, a separation occurs at about 1
m m from the outer edge.
Figures 5a and 5b show contact pressure distribution for P0 = 50 MPa, f = 0.5, and Sb =
100 MPa under center clamping and edge clamping, respectively. Variations in the contact
pressure distribution with tensile and compressive interfaces are seen. With the exception of
thecaseofh = 10 m m and center clamping, the contact pressure on the compressive interfaces
was high at the outer edge, x / a = 1. This location corresponds to the initiation site for fretting
fatigue cracks. On the other hand, on the tensile interfaces, the local contact pressure at x / a
= 1 was very low. Indeed, in some cases separation occurred. With a bridge height of 10 m m
under center clamping, no significant difference in contact pressure distributions was observed
between the compressive and tensile interfaces. The difference in the contact pressure distribution between center clamping and edge clamping was only significant for h = 10 mm.
-
--Tensile interface
--- Compressive
interface
Sb=IOOMPa
po = 50MPa
3\
02
~"2
h =lOmm
C
-
o_
31
k
C]
-
3
h=3Omm
h=2Omm
t
0
x/a
(a)
~l
1
Center clamping
FIG. 5a--Calculated contact pressure distribution with bridge height 05h = l O, 15, 20, and 30 ram for
center clamping.
90
& - - Tensile interface

I---C~
/
interface
3~
Sb=100MPa
[
P,, = 50MPa
o?
~"2
/
h =10mm
t
/
OI ~
-I
21
,
0
x/a
h=
20mm
f =0.5
-
h =15
mm
ql
h=30mm
at
6
(b) Edge damping
FIG. 5b--Calculated contact pressure distribution with bridge height of h = 10, 15, 20, and30 mm for
edge clamping.
Variations in the contact pressure distribution, for applied bending stresses of 50, 100, and
150 MPa are shown in Figs. 6 to 9. Figure 6 shows the contact pressure distribution for the
bridge of h = 10 m m under the center clamping. While separation occurred always on the
outer edge of the tensile surface, it was only observed on the compressive interface with low
bending stresses Sb ~ 100 MPa. Figure 7 shows the case of h = 15 m m with center clamping.
As the bending stress increases, the concentration qf contact pressure at the outer edge
increases on the compressive interface. Figure 8 shows the case of h = 10 ram, with edge
clamping. The contact pressure on the compressive interface at the outer edge increases with
increasing bending stress, and that on the tensile interface decreases. The top and bottom diagrams of Fig. 9 show the contact pressure distributions under center clamping and under edge
clamping, respectively, for a bridge of 30 m m in height. There is little difference in the contact
pressure distributions at the same bending stress. With a square bridge (i.e., a bridge height of
30 mm) the effect of the types of clamping on the contact pressure distribution is insignificant,
in agreement with Saint-Venant's principle. However, the rectangular bridge (i.e., a low bridge
height such as 10 ram) has an end effect, and causes the contact pressure distribution to be
dependent on the clamping position.
SATOON CONTACTPRESSUREDISTRIBUTION
4I
!~Tensde interface
i___ Compressive
interface
3
Po= 50MPa
h =lOmm
of ~ k
f =0.5
31
~2
2t
0
-
91
\\
o_,]
3~
\\
_%=IOOMPa
-150MPq
x/o
X/o
X/O
FIG. 6--Calculated conlacl pressure distribution under bending stress o[S~ = 50, I00, and 150 MPa,
and /br center clamping using a bridge ()/h = I0 mm.
:I
4 --Tensile interface
I--- Compressive
_/
interface
3 I- Po=5OMPa
| f =0.5, h=lSmm
G~ ~~ M P c l
:I
k~b= 150 MPa
_Sb=lO0 MPG
o_
0
x/o
-1
x/o
FIG. 7--Calcu/aled contact pressure distribution under bending stress o[S~ = 50, 100, and 150 MPa,
and [br center clamping using a bridge (?/h = 15 ram.
Variations in Contact Pressure A m p l i t u d e During a Bending Cycle
From the calculated results, the contact pressure amplitude during a bending cycle, which
is half of the difference in the contact pressure between the compressive and tensile interfaces
at the outer edge, was obtained. Figure 10 shows the variations of the contact pressure amplitude with bridge height and with bending stress under center and edge clamping. Under center
clamping, the contact pressure amplitude increased with an increase in the bridge height, so
that at a bridge height of 30 m m the contact pressure amplitude reaches the maximum. With
edge clamping, the greatest contact pressure amplitude occurred with a bridge whose height
was 10 ram. The contact pressure amplitude decreased with the increase in bridge height, for
92
Tensile interface
---Compressive
interface
3
Po=50MPa
f =0.5, h=lOmm
_Sb=5OMPa
2
-
&-..
jl
1
I
/
I
I
_Sb=1OOMPa
I
J
__Sb:150MPa
2
/
0
X/a
x/a
0
x/a
FIG. 8--Calculated contact pressure distribution under bending stress qf Sh = 50, 100, and 150 MPa,
andjor edge clamping using a bridge qfh = 10 mm.
the case of high stress amplitude OfSb = 100 and 150 MPa. With both types of clamping (i.e.,
center and edge clamping) the variations of contact pressure amplitude were small if the bridge
height was more than 20 ram.
Figures 11 and 12 show the variations in the contact pressure amplitude with nominal con-
2~
1
Tens. interface
Comp. interface
Po=5OMPa
f =Q5, h=3Omm
50MFb.
= IOOMPa l
_b = 150MPa
3
!
__Sb=IOOMPa
Sb= 50MPa
_Sb= 150MPa
2
]
X/a
X/a
FIG. 9--Calculated contact pressure distribution under bending stress qf Sb = 50, 100, and 150 MPa,
and for center clamping (top) and edge clamping (bottom) using bridge of h = 30 mm.
iX~v
75
93
v,v,T Edge
clamping
[],[]l Center
c[omping
q~
"O
50
C1.
E
O
i/
ip3/
u~
Ul
25
E
Q.
u
C
O
(D
/ / ~z~
Po=5OMPo
ii//
v, I Sb= ]501vlPct
~,/ v,~ = IOOMPo
[]
v , ~l
= 50,MPo
10 15 20
30
Bridge height h, mm
FIG. l 0 1 Variations o/contact pressure amplitude with bridge height, h. and with bending stress, Sb.
act pressures of p0 = 50 and 25 M P a and coefficients of friction o f f = 0.5 and 0. In Fig. 11,
he contact pressure amplitude for the contact pressure of 25 M P a is lower than that of 50
VIPa, except for center clamping with h = 10 ram. In Fig. 12, the contact pressure amplitude
tecreases also with coefficient of friction, except for the bridge height of 30 mm.
~- 75
9~Q_
50
v,v Edge clamping

i,n Center damping
\ ~
m
m 25
~/
Sb=100M~I
"
b_
f =Q5
u
O
__.~.~I~
f,I
Po=50MPa
v,n po=25MPQ
0
II
lO 15 2'0
30
Bridge height h, rnm
FIG. 11- - Variations of contact pressure amplitude with bridge height, and with nominal contact pressure, Po.
94
n~ 75
Edge damping
=,D Center damping
v, v
(D
O_
50
E
O
~(D
25
/,'
12)_
//'
i
lO
i ''p
Sb= OOU
p = oum
,,, f : 0.5
v, n f = 0
15 20
30
Bridge height h, mm
FIG. 12--Variations of contact pressure amplitude with brtdge height, h, and with coefficient q[ifriclion, f
Equivalent Stress Amplitude

The author [7,8] has proposed a method for estimating fretting fatigue strength using the
equivalent stress amplitude as defined by Tresca's rule for yielding. The equivalent stress
amplitude for fretting fatigue in bending (Fig. 2) is composed of the bending stress amplitudes,
Sb,, the contact pressure amplitude, Sp=, and the frictional shearing stress amplitude, Tf,. The
equivalent stress amplitude, &qa, is given by
Seq a = (S2a - - 2SbaSpa -}- S2pa -[-
4T~.)u2
(1)
where
Sba =
( S x t - Sxc)/2,
Spa = (S~ - Sy~)/2, and

Tfa = (Txy,- T, yc)/2.
The fretting fatigue life, Nrret,ng, for the stress amplitude, Sb, can be obtained by replacing
Nrre,~ngfor the plain fatigue life, Np~am,appropriate to the bendi ng stress amplitude, Seqa. By this
method, it was shown that the value of Nrre,,,ngcan be estimated accurately in practical application [9]. This equivalent stress amplitude is applied for estimating the fretting fatigue properties (S-N plots) in a later section of this paper (see Figs. 18 and 19).
Experimental Procedure
Visualization of Contact Pressure Distribution by Experimental Method of Caustics

The contact pressure distribution can be demonstrated by the method of caustics. A schematic drawing for the creation of caustic images is shown in Fig. 13. The specimens and contact bridges used in the experiments were made of a PMMA plate of 5 mm thickness. The
.I ~J(,Bridge(PMMA)
I~~Specimen
r ~L~,.~ ~..j (PMMA)
:~,,,~
SS"
95
Screen J
Laser ray
FIG. 13--Schematic drawing for the creation qf caustic image.
shape and dimensions of the bridges were those of the model used in the BEM analysis. The
specimens were rectangular plates of 40 m m height and 100 mm width. A nominal contact
pressure of 10 MPa was applied because PMMA has a low Young's modulus of 3 GPa. A HeNe laser ray impinged around the contact interface as a convergent light beam. The transmitted light created images of a caustic on a screen (Fig. 13).
Figure 14 shows the caustic images obtained from the experiments. The caustic images were
created at the contact edge, provided there was no separation. For center clamping with bridges
of h = 10 ram, the caustic image was created only at the inner contact edge, and not at the
outer contact edge, due to separation. In other cases, a caustic image at the outer edge was
created. The change in the experimental caustic images agreed well with the trend in the
change in the contact pressure distribution obtained from BEM analysis.
Fretting Fatigue Tests

Fretting fatigue tests in bending (Figs. 15a and 15b) were carried out with specimens and
fretting bridges of A2024-T3 aluminum alloy. The width of the specimens was 23 ram. The
bridges were identical to the shape of the model used in the BEM analysis and were of 30 mm
depth.
The cyclic bending, which varied sinusoidally, was applied at a frequency of 40 Hz by a
vibrator. The amplitude of displacement was controlled, and the stress amplitude was calculated by using the formula for deflection &cantilevers. This change of displacement with stress
is valid because the stress applied by bending was lower than the proof stress of the material.
Fretting fatigue failure occurred near contact edge, x/a = 1. The fretting fatigue cracks propagated initially at an oblique angle to the contact surface, and then normal to the surface. Thus,
a feature called "tongue" was formed on the fracture surface.
Figures 16 and 17 show the S-N plots obtained from the fretting fatigue tests as well as the
S-N diagrams for plain fatigue [10]. From Fig. 16 for center clamping, it can be seen that the
bridge whose height is 10 m m gives longer fatigue lives. The difference in fatigue lives corresponds to the value of the contact pressure amplitude given in Fig. 10. In Fig. 17, however,
there is no significant difference in fatigue lives between different bridge heights. The results
show, therefore, that lower S-N data from fretting fatigue tests were obtained in plane bending,
with high bridges (h > 30 ram) under center clamping or with edge clamping irrespective of
bridge height.
Figures 18 and 19 show the S-N plots obtained by using the equivalent stress amplitude,
S~qa. The S-N plots for both types of clamping correspond to those obtained from the experiCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
96
FIG. 14--Caustic images obtained experimentally. Semi-circles show the caustic images created at contact edges.
ments. It can be seen that fretting fatigue strength is significantly affected by contact pressure
amplitude, Pa. It can also be seen that decreasing the contact pressure amplitude during stress
cycles improves fretting fatigue strength.
Fatigue lives estimated at low stress amplitude, however, were short compared with those
obtained from the experiments. This is due to fretting wear at the contact edge, which is
ignored in the calculations.
Conclusions
The contact pressure distribution on the contact interface between the specimen and fretting
element analysis. The
Copyright
by ASTM Int'l
(all rights
reserved);fatigue
Mon Mar 28
EDT 2016
bridges,
under
fretting
in 00:05:25
bending,
was calculated by boundary
97
(a) Center damping

FIG. 15a--Experimental setupjbrjkettingj~ttigue tests in bending using center clamping.
' <L_~J Repeated

Specimen 3-~ ~
displacement
straingage
Frettig bridge
Loadceit~ ~ {
Vise
%
i ;
. ~
h=lq15,20, 30
(b) Edgedamping
FIG. 15b--Experimental setupforjkettingjatigue tests in bending using edge clamping.
contact pressure distribution depends on the bridge height and clamping position. The contact
pressure at the outer edge, which is the initiation site of fretting fatigue cracks, was low under
center clamping and was high under edge clamping. Under center clamping with the low
bridge of 10 ram, separation at the outer edge occurred. On the other hand, under edge clamping, the low bridge gives rise to a high stress concentration at the outer edge. There was little
difference in the contact pressure distributions between the two types of the center and edge
clamping with bridges of 30 mm in height. The calculated contact pressure distributions were
verified by the method of caustics.
Copyright by
ASTM Int'l
(all rightsfatigue
reserved); Mon
Mar in
28 00:05:25
EDT 2016
From
fretting
tests
bending,
it was confirmed that the fretting fatigue lives
98
~300 I
0<> 9
A h=lOmm
Fretting
Ifatigue
-~ 200
[]
<>
89
30
, --,:.
121.
FO 150
~ 100
,,i,,,o
ffl
Plain fatigue
po= 50MPa
Center clamping
.~_ 50
I1)
. . . .
i I
i i I
i i I
10 5
10 6
I0~
Number of cycles to failure Nf, cycle
FIG. 16--Experimentally obtainedjkettingj~ttigue S-N plots, Jar center clamping
300
r A h=lOmm
Fretting /I O 15
%250
03
~b
OJ
~fatigue
"] []
L0
20
30
AO-~
-o200
Q_
-~ 150
0
(./1
z~O 0
~ 100
-i,.,o
CD
"1:3
r
9 -
Plain fatigue
po= 50 MPo
Edge clamping
50
0
Illl
I I I i l
lo
IIIll
lo

FIG. 1 7 - - E x p e r i m e n t a l l y obtained fretting fatigue S-N plots, for edge clamping.
~o300}
"/Plain
fatique
200
OaO
.-I--'
1:3
E
150
O
0
~ 100
Fretting
~ 5O fatigue
cn 0
15
20
"O
r
Po= 50 MPa
Center damping
3O
I I li
105
Ilil
I I ; I[
I0e
IG
Number of cycles to failure Nf, cyde

jar center clamping.
FIG. 18--Analytically obtainedJketlmgjhtigue S-N plols
300
"~rPlain
~250
@
..I,--'
fatigue
200
ID_
ch
I.h
@o
~ 100
cn
91:3
~-
50
0
Fretting
fatigue
I
I i J I
h=lOmm
[]
O
20
30
1.5
po= 50MPo
Edge clamping
I I I I~
llill

FIG. 19--Analytically obtainedj?etting #atigue S-N plots, [br edge clamping.
99
100
obtained under center clamping using a rectangular bridge (i.e., a bridge of 10 m m in height
and 30 m m in length) were longer than those under edge clamping. It was also found that no
difference in fretting fatigue lives can be seen under either center or edge clamping with the
square bridge (i.e., as with a bridge height of 30 mm). These S-N properties in fretting fatigue
were estimated by using the equivalent stress amplitude, S~qa.
From analytical and experimental results, it is proposed to adopt the following clamping
method in fretting fatigue tests:
(1) Edge clamping for the tests with the rectangular bridge.
(2) Center clamping for the tests with the square bridge.
With regard to the intention to alleviate damage due to fretting, it is effective to decrease
and control the magnitude of contact pressure amplitude during stress cycles. It is relevant to
this point that creating a groove near the end of the bridge, for example, is effective in improving fretting fatigue strength [9].
References
[ 1] Nishioka, K. and Hirakawa, K., Bulletin (?{Japan Society of Mechanical Engineers, Vol. 12, 1969,
pp. 692-697.
[2] Goss, G. L. and Hoeppner, D. W., Wear, Vol. 27, 1974, pp. 153-159.
[3] Lindley, T. C. and Nix, K. J. in MultiaxialFatigue, ASTMSTP853, American Society for Testing
[4] Field, J. E. and Water, D. M., Report 275, National Engineering Laboratory, 1967.
[5] Sato, K., Fujii, H. and Kodama, S., Wear, Vol. 107, 1986, pp. 245-262.
[6] Wright, G. P. and O'Connor, J. J., International Journal (?/'Engineeringand Science, Vol. 9, 1971,
pp. 555-570.
[7] Sato, K. and Watanabe, T. in Proceedings, 31st Japan Congress on Materials Research, 1988, pp.
23-28.
[8] Sato, K., Wear, Vol. 125, 1988, pp. 163-174.
[9] Sato, K. and Higashiyamazaki, M., Pre-print of the 39th Meeting of the Society of Material Science,
Japan, 1990, pp. 40-42.
[10] Takeuchi, K., "Fatigue of Nonferrous Metals," in Fatigue qfMetals, The Society of Material Science, Japan, Maruzen, 1964, pp. 114-134.
T o s h i o Hattori, ~ M a s a y u k i N a k a m u r a , 2 a n d Tatsuro l s h i z u k a 3
Fretting Fatigue Analysis of Strength

Improvement Models with Grooving or
Knurling on a Contact Surface
REFERENCE: Hattori, T., Nakamura, M., and Ishizuka, T., "Fretting Fatigue Analysis of
Strength Improvement Models with Grooving or Knurling on a Contact Surface," Standardiza-
tion of Fretting Fatigue Test Methods and Equipment, ASTM STP 1159, M. Helmi Attia and R.
B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 101114.
ABSTRACT: Previously the authors introduced a fretting fatigue evaluation method using stress
analysis and fracture mechanics analysis of the contacting structures. Using this method the fretting fatigue limit can be estimated by comparing the calculated stress intensity factor of the crack
which originates at the contact edge with the material's threshold stress intensity factor range.
In this paper the authors apply this fretting fatigue strength evaluation method to fretting
fatigue strength improvement models with grooving or knurlingon a contact surface. These estimated results are compared with the experimental results of a fretting fatigue test using grooving
or knurling pad type specimens. The strength improvement of 'these two models can be confirmed analytically and experimentally. Finally, the optimization of groove shapes such as the
radius and depth of a groove are discussed using the above evaluation method.
KEY WORDS: fretting fatigue, fatigue limit, fracture mechanics, stress intensity factor, non-
propagating crack, boundary element method, groove, knurling

Introduction
Fretting can occur when a pair of structural elements are in contact under a normal load,
while a cyclic stress and relative displacement are forced along the contact surface. This condition can be seen in bolted or riveted joints [1,2], in shrink-fitted shafts [3,4], in the blade
dovetail region of turbo machinery [5,6], etc. Under this fretting condition the fatigue strength
decreases to less than one-third of that under the non-fretting condition [ 7,8]. The strength is
reduced because of concentrations of contact stresses such as contact pressure and tangential
stress at the contact edge, where the fretting fatigue crack forms and propagates. Formerly, the
authors presented a new fretting fatigue strength estimation method using fracture mechanics
[7,9]. In this method the stress intensity factors were calculated from the stress or pressure
distributions near the crack which initiated at the contact edge. The fretting fatigue limits were
estimated by comparing these calculated stress intensity factors with the material's threshold
stress intensity factor range.
In this paper the authors apply this fretting fatigue strength estimation method to fretting
fatigue improvement models with grooves or knurlings on the contact surfaces. These results
are compared with the fretting fatigue test results of these models. Finally, the optimization of
groove shapes such as the radius and depth are discussed using these results.
~ReliabilityImprovement of Machine Elements and Products, T300, Kandatsu, Tsuchiura-city, Ibarakipref., Japan.
2Researcher, Mechanical Engineering Research Laboratory, Hitachi Ltd., Tsuchiura, Ibaraki, Japan.
3Senior Engineer, Hitachi Works, Hitachi Ltd., Hitachi, Ibaraki, Japan.
101
Copyright
by ASTM Int'l
(all rightsASTM
reserved);lntcrnational
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
www.astm.org
102
FRETTINGFATIGUETEST METHODS AND EQUIPMENT
Stress Analysis of a Contact Region with a Crack and Calculation of Stress Intensity Factors
Analytical Model and Procedure

Since the transmitting characteristic of the frictional force between contacting surfaces is
the most important factor in stress analysis of a contacted specimen, the boundary element
method program was used which was developed for such contact problems [101. With this
analysis the nodes in the contact region are fabricated on both the specimen and pad to enable
direct calculation of the relative displacement and contact pressure between these node pairs.
The analytical and experimental models used are shown in Fig. 1. In these models the uniform
contact pressure Po is applied to the outermost edge of the pad and uniform nominal stress ~0
is applied to the specimen. Cracks are assumed to form at the contact edge and grow perpendicularly to the contact surface. The boundary element models are shown in Fig. 2. These are
quarter sections of the analytical models. For the knurling pad type specimen, equivalent stiffness was used for the knurling region, 0.5 m m thick. The equivalent stiffness is determined by
comparing the experimental results of stress distribution near a contact edge with the calculated results.
As shown in Fig. 3 the experimental results of stress distributions (marked []) col ncided well
with the calculated results when the equivalent stiffness in the knurling region was E = 20.6
oi
/
.
Specimen
40
(a) Grooved type
........
AKnurling
(b) Knurled pad type

F I G . 1--Frelling.[aligue
slrenglh improvement specimens.
H A T T O R I ET AL. ON S T R E N G T H I M P R O V E M E N T M O D E L S
103
Pad
Groove
':
-~Crack-
Specimen
(a)
Grooved type
:
Pad
Knurlin 8
.....
r - f ~ -- ~
'.
""Crack
Specimen
(b)
Knurled pad type
FIG. 2--Boundary element models ~f each specimen.

G P a (shown by the dotted line). Thus, the knurling region equivalent stiffness was 20.6 GPa.
The frictional coefficient of the contact surfaces is determined as 0.7 on the basis of the
observed results of contact surface damage and relative displacement at the contact edge in
the fretting fatigue test [3,5]. The Mode I stress intensity values K are derived using the calculated stress distributions near a crack tip as
~,o = Ao/'J-~ + A, ~
+ A,rVT
(1)
104
FRETTINGFATIGUETESTMETHODSAND EQUIPMENT
1500
1000
v
P = 196 M P a
Specimen ' \ [~1-~ Oo= -+ 196MPa

Strain gage
Smooth pad
c-
500
..
D~~~_______.~__
t'D-----D-~ O_._____O
......
Knurled pad
(Knurling region E=20.6GPa)
O9
o'.5
,'.o
,'.5
2'.o
Distance from contact edge b (mm)

FIG. 3--Stress distributions near a contact edge.
and
K=
~/~A0
(2)
where constants A0, A ~, and A2 are calculated by fitting Eq 1 to the calculated results.
Calculated Results o f Stress Intensity Factor
The relationships between the calculated results of the Mode 1 stress intensity values, K,
using the above method and the crack length, a, under each load condition are shown in Figs.
4 and 5. These can be expressed approximately as
K = A + (oro + B ) " C
(3)
where A and B are constants, and C is a function of crack length, a. These can be expressed as
shown in Figs. 6 and 7 for the grooved type and knurling pad type, respectively.
Fretting Fatigue Threshold Analysis

Using these calculated Mode I stress intensity values, K, and the threshold stress intensity
factor range, AK~h, we can estimate the propagation or non-propagation behavior of a small
crack which initiates at the contact edge. Generally, AK~his influenced by two factors: (1) the
stress intensity factor ratio, R, such that
R = K.,,~,JKm.~
(4)
HATTORI
ET AL
ON STRENGTH
IMPROVEMENT
MODELS
105
n
v
I.-
. ..:::::::::::
.::i~ii!i!!!!!!!! ,g5
.:.:...:-:.::.::.
' "' - " -
o 20
ZZ
..
a=l.Omm/
~ 10
E
a=O.3mm--"
.5'0 G) 0~.~
~ ~ t r e l s 0 S t o (
20
260
260
u '50
FIG. 4--Re/alionship between slress intensil)'Jactor K and nominal slress cro (grooved type).
KnUrling::::::::i
:::~~
.
a=1.0mm
._E 10
.... : :
/ ~ ' m a - - - - U . D m~
-50
m~
,
-10[
~50~10
mm
0~'~150
200
Stress oo (MPa)
IK= -8.84+(~'o+53"9) C I
FIG. 5--Relationship between stress intensityjactor K and nominal stress ao (knurled pad type).
106
FRETTINGFATIGUETESTMETHODSANDEQUIPMENT
0.1
d=l~m\
0.05
o'.5
11o
Crack length.a (mm)

FIG. 6--Relalionship between constants C and crack length a (groow.d type).
where Kmi~ and Km=~are the m i n i m u m and m a x i m u m M o d e I stress intensity values; and (2)
the crack length, a, especially in the short crack region.
Here, we consider the effect of the stress intensity factor ratio, R, and the crack length, a, on
dxK,h. We then d e t e r m i n e the relationships between dxK~hand the stress intensity factor ratio,
R, using various experimental results derived [11] as
AKth(R} = &K~h{R:o) " (1 -- R) ~ (when R < 0)
and
dxK,h(R) = &K,~(R:o~ " (1 -- R) 1~176
R~/~o~(when 0 ~< R < 1)
(5)
To predict/XK~h in the short crack region, El Haddad et al. [12] derived the equation
a +a ao
/XK~h(~) = 3,Kth(a=~o) 9
(6)
In their equation, the critical crack length, ao, is determined by

2
ao
Ao-.,o / 7r
(7)
107
HATTORI ET AL. ON STRENGTH IMPROVEMENTMODELS
0.1
Knuri:ingii
0.05
o15
1.o
C r a c k l e n g t h a (mm)
FIG. 7--Relationship between constants C and crack length a (knurled pad type).
where A~w,,is the fatigue limit of a plain specimen and ~K,h~....~is the threshold stress intensity
factor range of the long crack. The value, a0 is assumed to be constant for all R. Using Eqs 5
a n d 6, the threshold stress intensity thctor range, AK,h, considering both R and a, is derived as
AK, h = AK, h(R.O. . . . )" (1 -- R) ~ -
@a@ao
(when R < 0)
and
AK,h = AKn,IR=o. . . . )"
(l
--
R ) (0"5+l~
R)/10) ,
a
(when 0 ~ R < 1)
~3+a0
(8)
Using Eq 3, the Mode I stress intensity factor range, LXK,and the stress intensity factor ratio,
R, can be calculated as
2xK = Act0 9 C
n =
A +(~min+B)- C
A + (c~.... + B) 9 C
(9)
(10)
The comparisons of AK with 'XK,h, for a grooved type model with a 1 m m groove depth, are
shown in Figs. 8, 9 and 10. In each figure the stress amplitudes, era, are 137 MPa, 147 MPa and
108
n
3O
<1
t~
.~ 20
..-'".f''" ^If~ tq,- ~O'a= | 3 7 M P a

,'/"
LkKth
Crack length a (ram)
an
FIG. 8--Relationshq) betweo? A K a~d ~K'~h(grooved type, d ~ I ram. ~ = I57 MPa).
30
<I
t~
20
,I
5. . . .
AKth
10
t"
U~
0.5
" - ~i0
Crack length a (rnm)

FIG. 9--Relationship beLwee/~ ~ K a n d ~K,,~ (gtr
type, d = 1 ram, a~ = 147 MPa).
157 MPa, respectively. In the case of Fig. 8 (~,, = 137 MPa) the small carck formed by the
fretting damage stops at 20 ~rn. For Fig. ~13(a~ = 157 MPa) the crack formed by the fretting
damage grows s|owly unt;,l the specimen breaks. Figure 9 shows the critical condition where
the crack stops at 45 urn. h is important to note that this is the fretting fatigue limit. In Figs.
11 and 12 the fretting fatigue limit conditions for each grooved type with a 2 mm groove depth
and knurled pad type are shown.
HATTORI ET AL. ON STRENGTH IMPROVEMENT MODELS
30
<1
::::::::::::::::::::::::::
AK .... -''"
i-
20
1"
.. " " "
"" "
/ ~ - ~ G a = 157 M P a
UJ-I / I ' -
/ V L, ~m=O
.
f
f
r-
AKth
10
ul
I
I.'o
o.s
C r a c k l e n g t h a (ram)
F I G . l O~Re[ationship between ,~K and &Kfh (grooved o'pe, d = 1 ram, c~ = 157 MPa).
30
.::i!iiii!iiiii!R5 ~!
&K
_-
<1
J
t'-
20
//
f'
10"a'201MPa
-O'm=0
""
0
2
o
10
AKth
r"
.E
0
Crack length a (ram)

F I G . i l--Relationship belween ~ K and AK~t, (grooved O,pe d = 2 ram, ~ = 201 MPa).
109
110
FRETTING FATIGUE TEST METHQDS AND EQUIPMENT
30t|
AK,,.
/ .-..r
/ " .:ifiii!!iJ~Knufling
/
~..~...
(-
=~SMPa
AKth
10
if)
r-
015
1.'0
Crack length a (mm)
O9
F I G . 1 2 - - R e l a t i o n s h i p bet ween ,SK and AKrh (knurled p a d type, d = 1 mm, a. = 265 MPa).
i iiiiiiiiiiiii!ii:i: "
200
~"
98MPa
<1
d~
(-
100
4.a
oO
Distance from groove bottom r (ram)
F I G . 1 3 - - S t r e s s distribution near the groove bottom.
HATTORI ET AL. ON STRENGTH IMPROVEMENT MODELS
111
Optimization of Fretting Fatigue Strength Improvement Methods

From the estimated results of the fretting fatigue limits for a grooved type (see Figs. 9 and
11) we can see that fretting fatigue strength improves with increased groove depths. However,
the fatigue strength of the groove bottom decreases with increased groove depths. We must
consider both the fretting fatigue limit and groove bottom fatigue limit for optimization of the
groove shape. Stress distribution near a groove's bottom is shown in Fig. 13. Here, the stress
concentration factor is 2.18. The fatigue limit of the groove bottom can be estimated from
these stress concentration conditions. Estimated results of the fretting fatigue limit and groove
bottom fatigue limit for grooved type models are shown in Fig. 14. From these estimated
results it was concluded that the optimum groove depth is about 1.5 ram.
Fretting Fatigue Test

The fretting fatigue test apparatus is shown in Fig. 15. The contact pressure, P0 = 196 MPa,
is achieved by tightening the four screws and measured by strain gauges mounted at the specimen's center (strain gauge A in Fig. 15). The fluctuating axial stress G~,is achieved using a
closed loop servo controlled electro-hydraulic test machine with a load capacity of _+ 100 KN.
The test specimen and fretting pads are made, respectively, from Ni-Mo-V steel and c~/rbon
steel. The frictional coefficient of the contact surface is estimated by measuring the strain hysteresis near the contact edge by strain gauge B in Fig. 15. Using this method the frictional coefficient of the fretting damaged surface can be obtained as 0.7.
Experimental Results
For the knurled pad type test, the wear loss of contact apexes and reduction of contact pressure were assumed, so the contact pressure change during the fretting fatigue test was measured
" :+:+:::,::::-o
Fretting fatigue limit
o.. 200
b~
Groove bottom fatigue limit
g
.
0
Groove depth d (mm)
FIG. 14--Fatigue strength comparison qf grooved-type models.
FRETTIFATI
NGGUE
TESTMETHODS
ANDEQUIPMENT
112
Specimen
~ ~Pad Straingage[]
10
~- i:I
__....
i__i
~'",-~--~~ Screw~-~'l ' .....
Straingage~A~ ~ ~Pressplate ~U
FIG. 15--FrettingJatigue test apparatus.
0
13.
v
Time
-100
13.
OJ
:3
U~
G)
-200
MPa~"--- "-"-
-186MPa
-300
N=2X105
FIG. 16--Change ()/contact pressure during fretting Jatigue test.
HATTORIETAL. ONSTRENGTHIMPROVEMENTMODELS
113
as shown in Fig. 16. Thus, there is little reduction of contact pressure. The fretting fatigue
strengths of each strength improvement model are shown in Fig. 17. The fretting fatigue limit
of grooved type specimens (184 MPa with groove depth of 1 ram, 208 MPa with groove depth
of 2 ram) and knurled pad type specimen (216 MPa) increases significantly in comparison with
the plain fretting fatigue limit ( 120 MPa) [ 7]. These experimental results of fretting fatigue
limits coincided well with the estimated results shown in Figs. 9, 11 and 12. Fatigue fracture
conditions of grooved type specimens, near fatigue limit stress levels, are shown in Fig. 18. For
a groove depth of d = 1 m m the fracture occurred at a contact edge and for a groove depth of
d = 2 m m the fracture occurred at the groove bottom. These conditions coincided well with
the estimated results shown in Fig. 14. However, for high stress amplitude, the fracture
occurred at the groove bottom regardless of the groove depth. This is shown in Fig. 17 with
the 9 mark for a groove depth of 1 mm.
300
..~'~-~.~.~
8.
"''-.
Z&~z:~...~
~ o
Knurled pad type
"~'~A
200
"\\
"13
, N
O"
O-
"-.../
E
u3
U)
(R=Smm, cl=2mm) ./" ~

.
~'~\
(R=5mm, d=lmm)
"-...
Grooved type
"~
Plane fretting
100
uO
105
10 6
10 7
Number of cycles to failure Nf

FIG. 17--Fretting fatigue test resin]Is".
FIG. 1B--Fatigue fracture conditions of grooved type specimens.

10 8
114
Conclusions
The fretting fatigue strength of strength improved models is estimated using fracture
mechanics analysis of small cracks formed at the contact edge. By comparing these estimated
results with experimental results the following conclusions are drawn.
1. The fretting fatigue limit improves about 70-80% by making a groove at the contact
edge.
2. By increasing the groove depth the fatigue crack forms at the groove bottom. This transitional point is the o p t i m u m groove shape (in our case groove depth o f d = 1.5 m m for
a groove radius o f R = 5 mm).
3. Stress analysis and fracture mechanics analysis o f a knurled pad type model can be performed easily by using equivalent stiffness for the knurling region.
4. The fretting fatigue limit improves more than 80% by knurling the pad surface.
References
[1] Gassner, E., "The value of surface-protective media against fretting corrosion on the basis of fatigue
strength tests," Laboratorium fur Betrieb.~Jesligkeit TM 19/67, 1967.
[2] Buch, A., "Fatigue and fretting of pin-lug joints with and without interference fit," Wear, 43, 1977,
pp. 9.
[3] Hattori, T., Kawai, S., Okamoto, N. and Sonobe, T., "Torsional fatigue strength of a shrink fitted
shaft," Bulletin oftheJSME, 24, 197, 1981, pp. 1893.
[4] Cornelius, E. A, and Contag, D., "Die Festigkeits-minderung von Wellen unter dem Einfluj3 von
Wellen-Naben-Verbindungendurch Lotung, Nut und Paflfeder, Kerbverzahnungen und Keilprofile
bei wechselnder Drehung," Konstruktion, 14, 9, 1962, pp. 337.
[5] Hattori, T., Sakata, S. and Ohnishi, H., "Slipping behavior and fretting fatigue in the disk/blade
dovetail region," Proceeding,s, 1983 Tokyo Int. Gas Turbine Cong., 1984, pp. 945.
[6] Johnson, R. L. and Bill, R. C., "'Fretting in aircraft turbine engines," NASA TM X-71606. 1974.
[7] Hattori, T., Nakamura, M. and Watanabe, T., "Fretting fatigue analysis by using fracture mechanics," ASME Paper No. 84- WA/DE- 10, 1984.
[8] King, R. N. and Lindley, T, C., "Fretting fatigue in a 3~ Ni-Cr-Mo-V rotor steel," Proc. ICF 5, 1980,
pp. 631.
[9] Sakata, H., Hattori, T. and Hatsuda, T., "'An Application of Fracture Mechanics to Fretting Fatigue
Analysis," Role (?[FractureMechanics in Modern Technology, G. C. Sih, H. Nishitani and T. lshihara, Ed., Elsevier Science Publishers B. V., 1987, pp, 303.
[10] Ezawa, Y. and Okamoto, N., "Singularity modeling in two and three-dimensional stress intensity
factor computation using the boundary element method," Proceedings, 7th Int. Conf. on Boundary
Elements VII, Como, Italy, 1985, pp. 7-3.
[ 11] Usami, S., "'Applications of threshold cyclic-plastic-zone-size criterion to some fatigue limit problems," Proceedings, Int. Conf. on Fatigue Thresholds, Stockholm, 1981, pp. 205.
[12] El Haddad, M. H., Smith, K. N. and Topper, T. H., "Fatigue crack propagation of short cracks,"
Transactions, ASME, Vol. 101, 1979, pp. 42.
K o z o N a k a z a w a , 1M a s a e S u m i t a , ~a n d N o r i o M a r u y a m a I
Effect of Contact Pressure on Fretting Fatigue

of High Strength Steel and Titanium Alloy
REFERENCE: Nakazawa, K., Sumita, M., and Maruyama, N., "Effect of Contact Pressure on
Fretting Fatigue of High Strength Steel and Titanium Alloy," Standardization of Fretting
Fatigue Test Methods" and Equipment, ASTM STP 1159, M. Helmi Attia and R. B. Waterhouse,
Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 115-125.
ABSTRACT: The effect of contact pressure on fretting fatigue behavior has been studied using
a high strength steel and a Ti-6AI-4V alloy. In steel, at higher stress amplitude, the fretting fatigue
life decreased monotonously with increasing contact pressure. At lower stress amplitude, it
exhibited a minimum at a low contact pressure and a maximum at an intermediate contact pressure, then decreased again and became constant at high contact pressures. The fretting fatigue
strength at l0 7 cycles was high at an intermediate contact pressure. In titanium alloy, the fretting
fatigue life showed a similar contact pressure dependence. The frictional force increased monotonously with increasing contact pressure. The initiation site of the main crack depended on the
contact pressure and had a close relation to the width of stick region at the fretted area. The
contact pressure dependence was discussed in terms of stress concentration at the fretted area.
KEY WORDS: fretting fatigue, contact pressure, frictional force, crack initiation site, stick
region, stress concentration, high strength steel, titanium alloy
Fretting damage is known to have a deleterious effect on fatigue behaviors of steels and other
alloys. Fretting fatigue behaviors have been investigated so far from the point of view making
clear their mechanism. M a n y factors control the fretting fatigue. The effect of contact pressure
on fretting fatigue has been studied by several researchers. Most of these studies have shown
that the fretting fatigue life decreased with an increase in contact pressure [ 1- 7]. However, one
author reported [8] that the fretting fatigue life exhibited a m i n i m u m at a certain contact pressure. Thus, the effect of contact pressure is not yet fully understood. In this paper, fretting
fatigue behavior of a high strength steel and a Ti-6AI-4V alloy was studied at various contact
pressures,
Specimen Preparation
Materials used were a high strength steel of 0.18C-0.32Si- 1.26Mn- 1.04Ni-0.60Cr-0.49Mo0.26Cu-0.059AI-0.001P-0.003S (in weight %) and a t i t a n i u m alloy of 6.34A1-4.11V-0.14Fe0.200-0.007N-0,008C. The steel was q u e n c h e d and tempered in the following sequence:
heated to 1173 K for 2 h then air cooled; heated to 1153 K for 1 h then water cooled; tempered
at 838 K for 1 h then water cooled. The t i t a n i u m alloy was solution treated at 1213 K for 2 h,
then water cooled, and aged at 813 K for 5 h, then air cooled. The mechanical properties of
the steel and t i t a n i u m alloy along the rolling direction are shown in Table 1.
Senior Researcher, Head of Research Laboratory, and Researcher, respectiveIy, National Research
Institute for Metals, 1-2-I Sengen, Tsukuba, 305 Japan.
115
Copyright
by ASTM Int'l
(all rightsASTM
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
www.astm.org
116
TABLE l--Mechanical properties of steel and titanium alloy.

Material
0.2% P.S.
(MPa)
U.T.$.
(MPa)
El.
(%)
R.A.
(%)
Steel
Titanium alloy
920
1 010
1 010
1 100
15
15
70
30
Testing Procedure
The fretting fatigue device is shown schematically in Fig. 1. Dimensions of fatigue specimen
and fretting pad are shown in Fig. 2. Bridge-type fretting pads (span length 20 mm) of the same
materials as the fatigue specimens were used. The gage parts of the fatigue specimens and the
fretting pads were polished with 0-grade emery paper, then degreased with acetone.
~-FatigueSpecimen
FrettingPod
-~-
Normal L o ~
Cyclic Load
FIG. 1--Schematic representation offretting fatigue test.
FatigueSpecimen
10e
I
6
FrettingPad ~ 2
FIG. 2--Dimensions ~ffatigue specimen and fretting pad.

NAKAZAWA ET AL. ON CONTACT PRESSURE
1 17
The fretting fatigue tests were performed on a 100 kN capacity closed loop electrohydraulic
fatigue testing machine. A constant normal pad load was applied by a small hydraulic actuator
to which oil pressure was supplied from the main oil pressure source of the fatigue testing
machine. The contact pressure, calculated by dividing a normal force by the apparent contact
area of the fretting pad, was maintained below 160 MPa. The tests were carried out using a
sinusoidal wave at a frequency of 20 Hz, under tension-tension mode at a stress ratio of 0.1 in
laboratory air of 40 to 70% relative humidity at room temperature. Relative slip amplitude
between the fatigue specimen and the outer edge of fretting pad was measured using a calibrated extension meter at various stress amplitudes and contact pressures. The relative slip
amplitude agreed approximately with the value calculated by the equation, 6 = L cr~/2E,
where 6, L, ~o and E are relative slip amplitude, span length of pad, stress amplitude and
Young's modulus of fatigue specimen, respectively. 6 / ~ in steel specimens was about
0.05 um/MPa, and that in titanium alloy specimens was about 0.1 um/MPa. However, the
relative slip amplitude was influenced slightly by contact pressures, because the pad was
deformed elastically by the frictional force. The deviation of the relative slip amplitude from
the above values was less than _+ 10% for the contact pressure range investigated in the present
study. Frictional force between the fatigue specimen and the pad was measured using strain
gages bonded to the side of the central part of the pad. The plain fatigue life data were obtained
with hourglass-shaped fatigue specimens (stress concentration factor Kt = 1.08 for steel specimens and 1.04 for titanium alloy specimens).
Results
Fretting Fatigue L f e
The effect of contact pressure on fretting fatigue life at stress amplitudes of 180, 250 and 350
MPa for the steel is shown in Fig. 3. Plain fatigue lives, those at a contact pressure of 0, are
beyond 1 X 1 0 7 cycles for three stress amplitudes. The relative slip amplitudes at stress amplitudes of 180, 250 and 350 MPa are about 9.0, 12.5 and 17.5 urn, respectively. In specimens at
a stress amplitude of 350 MPa, the fretting fatigue life decreases drastically below 1 X 105
cycles at a contact pressure of 15 MPa. With the further increase in contact pressure, it
decreases gradually until it becomes constant at contact pressures of more than 80 MPa. However, in specimens at a stress amplitude of 250 MPa, the fretting fatigue life reaches a minimum, about 1.4 X 105 cycles at a contact pressure of about 25 MPa. With the increase in
contact pressure, it increases and reaches a maximum, about 2 X 105 cycles at a contact pressure of about 55 M Pa, then decreases gradually again to become constant at contact pressures
of more than 80 MPa. In specimens at a stress amplitude of 180 MPa, the fretting fatigue life
decreases below 1 X 1 0 6 cycles at contact pressures of 15 to 35 MPa, then increases sharply
with increasing contact pressure. At contact pressures of 55 to 80 MPa, it is beyond 1 X 107
cycles. However, it again decreases below 1 X 10 6 cycles at contact pressures beyond 90 MPa.
S-N curves of fretting fatigue at contact pressures of 25, 80, and 120 MPa for the steel are
shown in Fig. 4. They depend on the contact pressure as predicted from Fig. 3. The fretting
fatigue strength at 107 cycles is higher at a contact pressure of 80 MPa than at 25 and 120 MPa.
The effect of contact pressure on fretting fatigue life at a stress amplitude of 150 MPa for the
titanium alloy is shown in Fig. 5. The relative slip amplitude was about 15 um. The fretting
fatigue life takes a m i n i m u m at a contact pressure of about 20 MPa. The contact pressure
dependence of fretting fatigue life is similar to that of the steel specimens at a stress amplitude
of 250 MPa as shown in Fig. 3. S-N curves at contact pressures of 20 and 50 MPa for the
titanium alloy are shown in Fig. 6. At the lower stress amplitude, the fretting fatigue life is
shorter at a contact pressure of 20 MPa than at 50 MPa. The effect of contact pressure on
S-N curves is also similar to that in the steel specimens.
118

I
,o,k
Stress
ul
uI
/
/
",',
',',
1~
n
D
[]
.o-R~z~
10 4
amp.
180 MPa
A 350 MPa
;;\
o 250 MPa
106
Steel
I
20
40 60 80 100 120 140 16(
Contact Pressure
/
MPa
3 - - L ~ ' c t (?/contact pressure on /rettingJatigue /t72)for steel.
FIG.
500
Steel
o
n 400
:E
30O
.~
Fretting fatigue ",.,'~
.~
" Z X ~
~.2oo E
100
if)
Contact pressure
- -L~- - 25 MPa
80 MPa
- . - o - . - 120 MPa
I
10 4
10 5
10 6
10 7
Cycles
FIG.
Plain fatigue
to
Failure
4--Effect ~)Jcontact pressure on fretting fatigue
S-N
curves for steel.
NAKAZAWA
ET
107
AL.
ON
CONTACT
PRESSURE
119
Titanium alloy
Stress amplitude
150 MPa
--IO
10
20
40
Contact
60 80 100 120 140 160

Pressure / MPa
FIG. 5--Effect o/contact pressure on /ketting,/atigue l~/ejbr titanium alloy.

600
Titanium alloy
gsoo
fatigue
" 400
|
"~
"._.~
n
Frettin
fatigue ~ ~
300
s
.<
200 (/1
(/1
~100-
03
Contact pressure
zs 20MPa
o 50 MPa
104
105
106
Cycles to Failure
107
FIG. 6--Effect qf contact pressure on J?ettingjatigue S-N curves for titanium alh)y.
Frictional Force
Frictional force between the fatigue specimen and the pad varied with the number of cycles.
The degree of variation was small at high contact pressures but somewhat larger at low contact
pressures. In fretting fatigue, the crack initiation and the acceleration of the crack growth by
fretting usually occur after 104 to l0 s cycles [2,5,9-1l]. Hence, a frictional force determined
around 104 cycles was used.
120
FRETTING
FATIGUE
TEST
METHODS
AND EQUIPMENT
Relations between frictional stress amplitude and contact pressure for the steel specimens
at stress amplitudes of 180, 250 and 350 MPa are shown in Fig. 7. The frictional stress amplitude is defined as a shear stress amplitude acting on the contact area whose value is calculated
by dividing frictional force amplitude by the apparent contact area. At low contact pressures,
the frictional stress amplitudes increase almost linearly with increasing contact pressure. At
high contact pressures, the rate of increase drops and the frictional stress amplitude appears
saturated irrespective of stress amplitude, and the higher the stress amplitude, the larger the
frictional stress amplitude. For the titanium alloy, the relation of the frictional stress amplitude
and the contact pressure was similar to the steel specimens.
The coefficient of friction # is defined by the relation, u = f/P, w h e r e f a n d p are frictional
stress amplitude and contact pressure, respectively. The coefficients of friction were large at
low contact pressures, about 1.0 at a stress amplitude of 350 MPa, 0.9 at 250 MPa, and 0.7 at
180 MPa. However, they decreased with increasing contact pressure in accordance with the
saturation in frictional stress amplitude. In addition, at high contact pressures, the higher the
stress amplitude, the larger the coefficients of friction.
Crack Initiation
Initiation sites of the main cracks responsible for the failures for the steel specimens fractured at a stress amplitude of 250 MPa (Fig. 3) are shown in Fig. 8. Their frequencies, at sections of fretted surface divided into five equal parts along the cyclic stress axis, are shown for
two contact pressure ranges. Most of the crack initiations at contact pressures of 55 to 160 MPa
occur near the outer edge of the contact area. However, at contact pressures of 15 to 35 MPa
where the fretting fatigue life took a m i n i m u m , the crack initiations occur most frequently at
the middle portion of the contact area. In specimens of a stress amplitude of 180 MPa and
contact pressures of 15 to 35 MPa and in those of a stress amplitude of 350 M Pa and contact
pressures below 55 MPa, the crack initiation also occurred at the middle portion. At the higher
contact pressures, it occurred near the outer edge of the contact area at both stress amplitudes.
In the titanium alloy specimens, the effect of contact pressure on crack initiation was the same
as that in the steel specimens.
Figure 9 shows scanning electron micrographs of fretted surfaces near the initiation sites of
fracture in the steel specimens fractured at a stress amplitude of 250 MPa. At a contact pressure
of 80 MPa, the inner area of the fretted surface remains as polished. This area is the so-called
g
~100
~ 8o
n
6o
Stress amp.
n 180 MPa
o 250 MPa
zx 350 MPa
Steel
20
o
~ o J i
~- o 20
40 60 80 100 120 140 160

Contact Pressure I MPo
F I G . 7 - - R e / a l i o n belweenJHcziona] stress amplilude and contact pressure for steel.

1 21
3
Fretting Pod
Cyclic Lood
<
2ram
~,
Specimen
(a) p = 55-160MP,
(b) p=lS-35MPo
l~tiation Site of Fracture
FIG. 8--Initiation sites q/i/?acture in specimens at a stress amplitude ~f 250 MPa shown in Fig. 3.
stick (non-slip) region. The outer area is the slip region. The stick region is wide. However, at
a contact pressure of 25 MPa, the stick region is very narrow. The main cracks are seen near
the boundaries between the stick region and the slip region. These boundaries were nearer to
the initiation sites of the main cracks when the cracks were initiated, since the boundaries
moved inward with the number of cycles. There was a tendency that the higher the contact
pressure, the wider the stick region. At the lower contact pressures, there was no stick region,
The amount of wear debris produced during testing was much more at low contact pressures
than at high contact pressures.
Discussion
In fretting fatigue, the frictional shear stress acts on the fretted surface and stress concentration occurs there. The decrease in fatigue life caused by the fretting damage is thought to be
brought about by the decrease in crack initiation life due to this stress concentration and also
by the acceleration of the early stage of crack growth by fretting [2,12,13]. Hence, the larger
the frictional stress, the shorter the life would be. As shown in Fig. 3, the fretting fatigue life at
a stress amplitude of 350 MPa decreased monotonously with increasing contact pressure. This
result was the same as those reported so far [1-7]. The frictional stress amplitude shown in
Fig. 7 also increased with increasing contact pressure. There was a good correlation between
the life and the frictional stress amplitude.
On the other hand, the fretting fatigue lives at stress amplitudes of 180 and 250 MPa exhibited a m i n i m u m at contact pressures of 15 to 35 MPa, and a m a x i m u m at contact pressures
of 55 to 80 MPa, then decreased again and became constant at high contact pressures. In the
titanium alloy, the fretting fatigue life also took a m i n i m u m at a contact pressure of 20 MPa.
A similar result has been found in fretting fatigue by torsion in an aluminum alloy [8], in
which the life took a m i n i m u m at a certain intermediate contact pressure. This phenomenon
was explained by the m a x i m u m fretting damage at a specific contact pressure, which resulted
probably from the decrease in relative slip amplitude accompanied by the increase in contact
122
FIG. 9--Scanning electron micrographs of/ketted surfaces near the initiation sites o/J?acture in steel
specimens at a stress amplitude ~f 250 MPa. Contact pressure. (a) 80 MPa, (b) 25 MPa.
pressure. In the present study, the relative slip amplitude between the fretting pad and the specimen decreased slightly with the increase in contact pressures at a given stress amplitude. It is
difficult to explain a m a x i m u m and a m i n u m u m in fatigue lives in Fig. 3 by the slight change
in relative slip amplitude. The contact pressure dependence of fretting fatigue lives at stress
amplitudes of 180 and 250 MPa cannot be explained simply or directly by the change in frictional stress amplitude, since the frictional stress amplitude increased monotonously with
increasing contact pressure as shown in Fig. 7.
The fretting damage is shown schematically in Fig. 10. Under a certain testing condition,
there exist a stick region at the middle portion of the fretted area and slip regions on either side
of it. The relative slip mode depended on the contact pressure, relative slip amplitude [14],
and number of cycles. Hereafter, the relative slip mode is restricted to the situation around 104
to 105 cycles when the crack was initiated. The stick region was narrow when the contact pressure was low as shown in Fig. 9. When the contact pressure was very low, the whole area was
occupied only by the slip region, which could be obviously judged from observing the fretted
area or the wave form of frictional force. In the steel specimens, the contact pressures for the
no stick region were limited below 10 MPa at a stress amplitude of 180 MPa, below 15 MPa
123
Normat Load
Fretting Pad
}"P
Stick
R ion sI,p
Weor
~,Debris
Cyclic Load
FIG.
lO--Schematic representation of~'etting damage.
at 250 MPa, and below 55 MPa at 350 MPa, respectively. The relative slip mode thus changed
with increasing contact pressure as follows: only slip region--narrow stick region plus wide
slip region--wide stick region plus narrow slip region.
The contact pressure and the frictional stress amplitude used in the present study were.the
average values and were calculated by dividing the normal load and the frictional force by the
apparent fretted area determined by the size of the pad shown in Fig. 2, respectively. Therefore,
it was assumed that they were distributed uniformly over the whole fretted area. In the slip
region, the contact surface was heavily damaged and wear debris was produced, and a part of
wear debris was removed out of the fretted area. The large amount of wear debris was removed
near contact pressures where the life exhibited a minimum. The net contact pressure acting in
the slip region was probably lower than the average contact pressure, since the normal load
was given through the medium of wear debris. On the other hand, in the stick region, the net
contact pressure was probably higher than the average contact pressure, since the normal load
was increased by the decrease in normal load in the slip region. Hence, the net contact pressure
and the net frictional stress amplitude acting in the stick region were higher, while those in the
slip region were lower than the average values. Therefore, stress concentration occurred near
the boundaries between the stick and slip regions, and the crack could be easily initiated. The
main crack was initiated near the outer edge of the pad when the contact pressure was high
and the stick region was wide. However, the main crack was initiated at the middle portion of
the fretted area, when the contact pressure was low and the stick region narrow. This correlation between the crack initiation site and stick region width also implies that the crack was
initiated near the boundaries between the stick and slip regions.
In the steel specimens at stress amplitudes of 180 and 250 MPa, the fretted areas were occupied only by the slip regions at very low contact pressures, while they were occupied by the
wide stick regions at high contact pressures. In both cases, the contact pressure was distributed
presumably uniformly over the whole fretted area. At a contact pressure where the life exhibited a minimum, the existing stick region was very narrow as shown in Fig. 9b. To this narrow
stick region, a contact pressure greater than average was applied, and the net contact pressure
and resulting frictional stress amplitude equal to or greater than those at high contact pressures
must have acted on this narrow stick region. The m i n i m u m life was probably caused by this
high concentration of frictional stress amplitude. The effect of contact pressure on fretting
fatigue life at a given stress amplitude is shown schematically in Fig. 1 1. There are two life
curves. The first has A B D E drawn assuming that the contact pressure is distributed uniformly
over the whole fretted area; the second with BCD assumes that the contact pressure is concentrated at the narrow stick region. Point C corresponds to a m i n i m u m in life observed at a low
124
Decrease due to
concentration of contact
pressure end frictional
stress
l
X~B
~',
\",
\ r
"~
~)
C
No
II
stick t= Narrow stick

region,
region
I
I
0
i
Wide stick
region
I
I
Contact Pressure
.~
FIG. 11--SchemaHc represenlalion q/the eff~'~7Q/'COnIClCt pressure on /Y~'llin,~ /~lJ~,lt~" ]J/~"Cli Ct g/VPn
stress amplitude.
contact pressure, and an intersection of the two curves D to a maximum in life at an intermediate contact pressure. In the steel specimens at a stress amplitude of 180 MPa (Fig. 3), the
applied stress amplitude was near the fretting fatigue strength at l 0 7 cycles. Therefore, the life
was greatly influenced by a slight change of frictional stress state, and the life at D probably
exceeded 107 cycles. Contact pressure dependence at stress amplitudes of 180 and 250 MPa
can be well explained by this model.
In the steel specimens at a stress amplitude of 350 MPa, the contact pressure range for the
relative slip mode with no stick region was extended to a higher contact pressure compared
with those at stress amplitudes of 180 and 250 MPa, since the relative slip amplitude was
higher [ 14] and the relative slip mode with the narrow stick region shifted to high contact pressures. Near contact pressures with a narrow stick region, however, the life decreased monotonously with increasing contact pressure without exhibiting a minimum. The amount of wear
debris removed from the slip region was small when the crack was presumably initiated,
because the crack initiation life was very short due to a high stress amplitude and a high frictional stress amplitude. Hence, there may not have been such a high local concentration of
contact pressure and frictional stress at the narrow stick region as that at stress amplitudes of
180 and 250 MPa. This explains the monotonous change in life with contact pressure at a stress
amplitude of 350 MPa.
At contact pressures greater than 100 MPa, the fretting fatigue lives remained constant for
stress amplitudes of 180, 250, and 350 MPa. As one of the reasons for this, the change of frictional stress amplitude should be pointed out, since it tended to be saturated at high contact
pressures as shown in Fig. 7. As other reasons, the influence of contact pressure on crack
growth has been pointed out [4, 7,13]. The contact pressure may have given a static compressive stress on stress field near the crack front, or crack closure effect. The increase in contact
pressure may have been compensated for the increase in frictional stress.
Summary
In the steel specimens, at a stress amplitude of 350 MPa, the fretting fatigue life decreased
monotonously with increasing contact pressure. At stress amplitudes of 250 and 180 MPa, it
125
exhibited a m i n i m u m at a low contact pressure and a m a x i m u m at an intermediate contact

pressure, then decreased again, and became constant at high contact pressures. The fretting
fatigue strength at 107 cycles was high at an intermediate contact pressure. In the titanium alloy
specimens, the fretting fatigue life showed a similar contact pressure dependence. The frictional force increased m o n o t o n o u s l y with increasing contact pressure at a given stress amplitude. The initiation site of main crack depended on the contact pressure. The main crack was
initiated near outer edge of the fretting pad when the contact pressure was high and the stick
region wide. The main crack initiation occurred at the middle portion of the fretted area when
the contact pressure was low and the stick region narrow. The contact pressure dependence
was discussed in terms of stress concentration at the fretted area.
References
[1] Nishioka, K. and Hirakawa. K., Bulletin (?~the Japan Society Of Mechanical Engineers, Vol. 15,
1972, pp. 135-144.
[2] Endo, K. and Goto, H., Wear, Vol. 38, 1976, pp. 311-324.
[?3] Gaul, D. J. and Duquette, D. J., Metallurgical TransacHonsA, Vol. 11A, 1980, pp. t555-t561.
[4] Edwards, P. R. in Fretting Fatigue, R. B. Waterhouse, Ed., Applied Science, London, 1981, pp. 6797.
[5] Sato, K. and Fujii, H., Journal of the Japanese Society jot Strength and Fracture ~?fMaterials, Vol.
18, 1984, pp. 98-113.
[6] Nagata, K., Matsuda, T. and Kashiwaya, H., Transactions cfthe Japan Society :?[Mechanical Engineers, Vol. 53, 1987, pp. 196-199.
[ 7] Mutoh, Y., Nishida, T. and Sakamoto, I., Journal :?fthe Society :?[Materials Science, Japan, Vol.
37, 1988, pp. 649-655.
[8] Harris, W. J., Metallic Fatigue, Pergamon Press, New York, 1961, pp. 166-204.
[ 9] Fenner, A. J. and Field, J. E., Revue de Metallurgie, Vol. 55, 1958, pp. 475-485.
[10] Wharton, M. H., Taylor, D. E. and Waterhouse, R. B., l~2~ar,Vol. 23, 1973, pp. 251-260.
[11] Nakazawa, K., Sumita, M., Maruyama, N. and Kawabe, Y., ISIJlnternational, Vol. 29, 1989, pp.
78t-787.
[12] Alic, J. A. and Hawley, A. L., l~2)ar,Vol. 56, 1979, pp. 377-389.
[ 13] Sato, K., Fujii, H. and Tamaki, H., Journal :?/'theSociety ~?/Materials Science. Japan, Vol. 33, 1984,
pp. 1065-1070.
[14] Vingsbo, O. and S6derberg, S., Wear, Vol. 126, 1988, pp. 131-147.
Fretting Fatigue Testing---Methods and

Equipment
D. B. R a y a p r o l u ~ a n d R. C o o k ~
A Critical Review of Fretting Fatigue

Investigations at the Royal Aerospace
Establishment
REFERENCE: Rayaprolu, D. B. and Cook, R., "A Critical Review of Fretting Fatigue Investigations at the Royal Aerospace Establishment," Standardization of Fretting Fatigue Test Methods" and Equipment, ASTM STP 1159. M. Helmi Atria and R. B. Waterhouse, Eds., American
Society for Testing and Materials, Philadelphia, 1992, pp. 129-152,
ABSTRACT: The Royal Aerospace Establishment (RAE) of the United Kingdom has been
involved in fretting fatigue research for about 15 years. This paper summarizes the work that has
been carried out and describes research that is currently being undertaken. It also presents the
authors' views of future requirements in fretting fatigue testing and modeling of contact stresses
in fatigue loaded structures. The main aims of the RAE research are to determine the important
factors which influence the initiation and propagation of fretting fatigue cracks and to develop
fracture mechanics models which could be applied to structural situations. To achieve these aims
considerable work has been undertaken to develop both fretting fatigue testing equipment and
theoretical models of the fretting fatigue process, This paper describes the developments made
in both areas.
A summary is given of the fretting fatigue variables that have been considered in the test programs to date. Early fracture mechanics models are briefly described and their ability to predict
fatigue endurances is assessed. These models which were based on a growing edge crack have
been modified to take account of the early growth of inclined cracks under fretting conditions.
A critical review is made of testing methods and the authors' views on essential testing requirements are presented. The Report concludes with a discussion of the direction of future work both
at the RAE and in the proposed ASTM activities.
KEY WORDS: fretting, fatigue, frictional force, fracture mechanics, stress intensity factors, aluminum alloys, standardization, test methods
It has been recognized for m a n y years that relative m o v e m e n t between contacting surfaces
can p r o m o t e the initiation and growth of cracks [1-5]. Subsequent loading of the cracked
c o m p o n e n t can cause these cracks to propagate and may lead to catastrophic failure of the
structure. The engineering solutions to this problem have variously been to avoid designs
where such m o v e m e n t s occur, lubricate the surfaces in contact, increase the contact pressure
between the surfaces, or reduce the bulk stresses in the contacting parts. While these various
solutions have often proved successful, the mechanisms which lead to the i m p r o v e m e n t s have
not always been well understood. This is supported by the n u m b e r of catastrophic failures
which have occurred and are still occurring. In m a n y situations it is not possible to entirely
r e m o v e the relative m o v e m e n t s which cause fretting, but designs can be i m p r o v e d to m i n i m i z e
the problem if the m e c h a n i s m s involved are understood. A further design problem exists, particularly in the aerospace industry, where, despite careful design, cracks are still likely to initiate. In this case s o m e m e t h o d must be available to predict the effects of contact stresses on
Royal Aerospace Establishment, Materials and Structures Department, Farnborough, Hants GU 14
6TD, England.
129
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
lntcrnational
www.astm.org
130
the probable rate of growth of these cracks. This process, known as fretting fatigue, has been
the subject of many investigations including an extensive program carried out by the Royal
Aerospace Establishment (RAE). This report describes the studies carried out at RAE over the
past 15 years and the work that is currently being undertaken.
The initial aim of the RAE research was to examine the effects on fatigue endurance of various contact parameters. A knowledge of the effect of these parameters would enable designers
to minimize fretting fatigue problems. An early qualitative investigation [6] identified the
need for a quantitative analysis of the local stresses and showed that cracks initiated extremely
readily under the action of fretting, and that most of the fatigue life was involved with the
growth of these fretting induced cracks. It was, therefore, decided that a fracture mechanics
approach should be adopted which would allow the prediction of the growth of fretting fatigue
cracks. It was reasoned that this would give a reasonable prediction of the total fatigue life since
most of it was consumed in the crack growth phase. Thus the aircraft industry requirement
for a model to predict the growth of fretting fatigue cracks would also be satisfied. Measurements were made of the local stresses [ 7,8] and fracture mechanics models were developed [911]. A number of other theoretical aspects are discussed relating to frictional waveform [12],
variable coefficients of friction [13], and fatigue crack orientations.
Fretting fatigue testing has evolved through the various requirements outlined above from
the simple S/N approach to the crack depth, length and orientation measurement requirements of current testing. Testing developments are discussed below and a test apparatus is
proposed as an appropriate standard for future testing.
Fatigue Test Methods
Test Objectives
The main aims of the RAE research have been to understand the effects of various geometric
and loading variables on the resultant fatigue behavior in structural situations and to develop
analytical models to describe this behavior. To achieve these aims a building block approach
was used in both the testing and modeling phases. The initial aim of the experimental work
was to determine the effect of contact variables on the fatigue life of a simple test specimen.
The test specimens used for the different RAE investigations varied in detail, but all conformed
to the basic type shown in Fig. 1. The early contact variable investigation [6] involved S/N
testing to establish the effects of different specimen geometries and loading variables on the
number of cycles to complete failure. To determine the effect of fretting on fatigue endurance,
tests were conducted with fretting pads present for various proportions of the fretting fatigue
life. The results from these tests indicated the period of crack growth during which the fretting
pads had an effect. Where the fretting pads were removed at an early stage of the test and no
subsequent failure occurred, specimens were broken and crack lengths measured. This gave
an indication of the size and time to initation of cracks which were induced by the fretting
action.
To apply fracture mechanics models (described below), some knowledge is required of the
local stresses induced by the fretting action and, in particular, the frictional stresses. A second
series of tests [7,8] was performed on suitably strain gaged specimens, to enable frictional,
normal, and bulk stresses to be monitored throughout the tests. It was not possible to assess
the accuracy o f the fracture mechanics models in terms of predicted crack propagation rates,
since no crack length measurements were made. It was possible, however, to test the models
by comparing measured fatigue life with life predicted by the models from an assumed initial
flaw size to failure. One of the aims of current investigations is to define suitable methods for
measuring crack propagation. The methods under consideration are specimen compliance,
ACPD, and visual surface measurements after pad removal.
RAYAPROLUAND COOKON FRETTINGFATIGUEINVESTIGATIONS
131
Fretting
pads
1.27 mm
'
Fretted /
specimen
15.2 mm dia
//-~-~,\
\
J
\X
Loading
ring
Crack
-\Loaabng} ( ~
JLoad adjusting
~ball J/ ~ s c r e w
FIG. l--Frettingj)lligue testing assembly.

All of the tests described so far involve constant normal load, frictional, and alternating
body loads. To separate these actions and study the action of the fretting forces alone a third
test series [14] was performed. The objectives of these tests were to examine the crack range
over which fretting stresses influenced crack growth, examine whether any body stresses were
required to initiate fretting cracks, and model the behavior in the absence of alternating body
stresses. Attempts to measure crack length were unsuccessful. Consequently, the models could
only be tested by comparing the predicted crack lengths at crack arrest with those measured.
In most structural situations the magnitude of the normal load will not remain constant
during the alternating body load cycle. For instance, in a pinned lug the normal load at a point
on the contact arc increases with applied body load. In some areas of the contact arc the normal
load may reach a steady value at which point sliding of the pin/lug surface takes place. The
normal force may remain at this value, increase at a slower rate or decrease with increasing
applied load. A fourth series of tests are currently being undertaken to examine the effect of
cyclic normal load variations on cyclic frictional load, crack initiation, crack propagation, and
fretting fatigue life of simple specimens.
Fatigue Test Configurations

The various fatigue test configurations used throughout these programs are based on that
used by Cornelius and Bollenrath [15], and by Fenner and Field [16]. The basic form shown
in Fig. 1 was used in the first test series investigating the effect of contact variables on fatigue
endurance. A number o f modifications to the test set-up were made for the other test series,
which will be described later in this section. There are, however, many features which are common to all of the test series and these are described below.
132
F R E T T I N G FATIGUE T E S T M E T H O D S A N D E Q U I P M E N T
The principle on which the test configuration is based is the difference in stiffness between
the fretting pads and the test piece which causes relative movement between them when the
test piece is loaded. The test piece material used throughout the four test series was an alum i n u m alloy (specification BS L65). The fretting pads were made of steel (specification BS
$98) and were clamped to the test piece by a calibrated proving ring.
In the second test series, where frictional force measurements were required, both fretting
pads were strain gaged between the pad feet and correspondingly on the opposite surface (see
Fig. 2). The four gages were joined in a full bridge and calibrated to measure the load passing
through the pads. This is equal to the shear load transferred to the pads which is referred to as
the frictional load. A loading pad or "rider" was placed over the fretting pad to allow the normal load to be applied (see Fig. 2). The rider consisted of a knife edge and a roller contact to
ensure that the loading was applied over the fretting pad feet and allowed for bending of the
rider when the normal load was applied. Frictional force variations were monitored throughout the fatigue tests. Output from the strain gage bridge was amplified and fed to the Y channel
of an oscilloscope with the alternating machine load signal connected to the X channel. A
time-lapse video recording was made of the entire test and frictional load measurements were
Straingauges~
Locatingbolt
,/
~~
I
tt
iI
1 tl
II iI
C~.
Padspan .I
II I
ql
+It
Ill
i1
/2
Loadingball
FIG.
II
, l,
II h,~
It iI
I]
I+
Roller
Dimensionsin mm
2--Strain ga~edJ?etlingpad and loading pad assembly.
RAYAPROLU AND COOK ON FRETTING FATIGUE INVESTIGATIONS
133
made from a playback at selected intervals. This enabled all single events as well as gradual
trends to be measured throughout the tests. A recent development has involved computer
sampling of the frictional load variations and analysis of the data.
In the third series of tests a completely different test set-up was used. A cage around the test
piece allowed a constant mean body load to be applied during the test, while alternating frictional displacements were applied by the fatigue test machine. A calibrated proving ring was
used to apply the normal loads to the fretting pads.
In the fourth series o f tests normal load is cyclically varied in phase with the body load. The
test set-up requires the use of a biaxial fatigue machine with phase linked actuators. The test
piece configuration is essentially the same as that used in the second test series where frictional
forces were measured.
Test Variables
Test Series" 1--The Effect of Contact Variables--In this test series the objectives were to
examine the effect on fatigue endurance of changing the contact and loading variables. The
fatigue life is governed by the stress fields produced by the different contact and loading parameters. It can be appreciated that the stress fields are complex and that the effect of changing
single parameters cannot be evaluated in isolation. Hence, a matrix of parameter changes had
to be evaluated. The parameters which govern the stress fields are pad span, pad load, and body
loads. The magnitude of the frictional stresses are governed by all three of these parameters
and can be varied up to a maximum value above which sliding or "macroslip" occurs. Below
this maximum value of frictional stress sliding may occur over small areas under the pad feet
(where the local normal loads are lowest), but gross sliding is not observed. Under these conditions "microslip" is said to occur. Combinations of pad load, pad span, and body loads were
chosen which gave various degrees of both microslip and macroslip. The test matrix consisted
of two pad loads and four pad spans. Constant and variable amplitude loading types were used
for the body loads at a range of alternating and mean stress values.
A number of partial damage tests were conducted in which pads were removed after a predetermined number of fatigue cycles. The tests were conducted under both constant and variable amplitude loading at a range of stress levels with one pad span and one pad pressure.
Test Series 2--Frictional Force Measurements--The tests were carried out with essentially
the same variables as for test series 1. In addition to these variables, tests were conducted with
various fluids present in the contact regions. A range of fluids were used, but of primary concern was a PX-24 water displacing penetrant. The penetrant was sprayed onto the test piece at
different intervals throughout the fatigue test. Spraying was done either before or after the pads
were assembled and then periodically throughout each test. The frictional force variations
were monitored closely during and after spraying.
Test Series" 3--Frictional Loading Only--ln these tests no alternating body stresses were
applied. There was, however, the possibility of introducing a mean stress into the specimen
via the frame described above. The first variable was, therefore, the magnitude of the mean
body stress. Additional variables were the magnitude of the normal load and the frictional
loads and displacements applied to the fretting pads.
Test Series 4--Cyclic Variations of the Normal Load--This test series has only recently
started; the precise range of variables to be investigated will depend on preliminary test results.
However, it is anticipated that they will include pad span, alternating body load magnitude
and sequence type, and magnitude and cyclic form of normal loads. Crack measurement techniques will be developed with the aim of monitoring crack growth and crack shape.
134
Observations from Test Results
Effects of Contact Variables

The most obvious observation was that significantly shorter fatigue lives were measured for
all fretted specimens when compared to plain specimens of the same material tested at the
same alternating body stress level (see Fig. 3). This is due to the additional local stress contribution from the frictional forces. The effect was most marked under constant amplitude loading where the frictional forces attain a higher average value than under variable amplitude
loading due to the undisturbed keying-in process. Pad span also affected fatigue endurance; as
pad span increased, fatigue endurance decreased (see Fig. 4). This again is attributed to the
higher frictional forces generated by the larger pad spans under microslip conditions which
predominated with the testing parameters chosen. The effect of pad pressure was extremely
small and under certain conditions negligible. This appears a surprising result, since the frictional forces developed under microslip conditions would be directly proportional to the pad
load, assuming a constant coefficient of friction. It was observed from the broken fracture surfaces that cracks did not grow initially perpendicular to the alternating loading axis of the specimen. They grew from the outer edges of the pads at angles between 10-40 ~ under the fretting
pads as shown in Fig. 5. As the pad normal load is increased the effect would be to close any
growing cracks and reduce the rate of propagation. This effect acts in opposition to the increase
in frictional load caused by an increase in pad normal load. The two effects tend to cancel each
other out in terms of crack growth. The overall effect of pad load is small as observed from the
fatigue test results. This will not always be the case. For example, as the pad load tends to zero
the frictional forces induced tend to zero and the fatigue life will tend to that of a plain
specimen.
fol ~
E 150
09
fol~
\
) ~ X
Zero mean stress
X XX
"~X~X
,,...,
~, 100
"0
0
r
~
E
<
50
3ymbol
o b'--.~
x Plain specimen
o Fretted specimen
o o ~-o--------o-~
(16.5 mm pad span
103.5MN/m 2 pad pressure)
I
105
o---.-
_.
106
107
Cycles to failure, N
FIG. 3--Fatigue lives ofplain and.lkettedspecimens.
108
RAYAPROLUANDCOOKONFRETTINGFATIGUEINVESTIGATIONS
200
0~,,~
7"
E 150
> ~X" x ~x
.5 and 34.35 mm
135
Pad pressure = 103.5 MN/m2

~xZ e r ~ mean stress
~_,
(~kk
x ~>ee.,~_
._x. . . .
ffl
~ 100
0
r
4+6"b-,q'b+o o"~--o o
Symbol
~E 50
<
z~
n
o
+
Pad span
Plain
2 mm
6.35 mm
16.5 mm
34.35 mm
105
~---------~=-
o ~ 5 " - ~" ~ .
- ~
..__
+-N-+ o "'""o------~ ~--~--~:~,fi--o
106
107
108
FIG. 4--Effect of pad span on fatigue life during constant amplitude loading.
From an examination of the fracture surfaces it was observed that the angled crack behavior
only persisted over a short distance (typically less than 1 mm) before turning to an angle of 90*
to the specimen surface. The angular crack growth was attributed to the biaxial nature of the
stress distribution near the fretting pads. As cracks grew away from the fretting pad area they
were less affected by the frictional stresses and more influenced by the body stresses, hence, the
crack angle became perpendicular to the direction of the applied body stresses. These obser-
FIG. 5--Section through fretting scar showing crack angle O.

136
vations were supported by the partial damage tests. It was observed that under constant amplitude loading, removing the fretting pads at 30% or greater of the fretting fatigue life had no
effect on the subsequent number of cycles to failure, i.e., fretting damage was only active in
the first 30% of the fretting fatigue life. Under variable amplitude loading the corresponding
observation was that fretting damage was only active over the first 5% of the fretting fatigue
life.
Frictional Force Measurements
Frictional forces were measured using strain gaged fretting pads, as described above. Under
constant amplitude loading an initial "bedding-in" phase was observed in the first few hundred cycles after which the frictional forces remained at a fairly constant value. The-beddingin period (typically less than 1% of the fatigue life) consisted of a gradual build up in the magnitude of the frictional forces and a reduction in the degree of macroslip. This was equated to
wearing away the larger contacting asperities, during which macroslip occurred, and to
increasing the area of asperity contact, which promoted microslip conditions. Under microslip
conditions during the bedding-in process a point is reached when the entire frictional load is
transferred by elastic deformation of contacting asperities [17]. From this point on a relatively
constant value of frictional force occurred throughout the remainder of the fatigue test. This
is illustrated in Fig. 6 at the lower alternating stress levels. At the alternating stress level of 99
M N / m 2 the limit of microslip is approached. The test at 14 l M N / m 2 is under macroslip conditions. In this case the frictional forces build up to an unsustainable level (an average coefficient of friction of unity), whereupon, an instantaneous drop in the magnitude of the frictional
force takes place and sliding occurs. The bedding-in process then starts again; this gives the
characteristic jumps in the frictional force trace.
The general drop-offin frictional force as tests progress (see Fig. 6) is attributed to the growth
of cracks. As cracks gradually propagate from the edge of the fretting pad feet frictional load
transfer decreases. This can be seen in Fig. 7 which shows frictional forces as a function of
applied body forces during individual cycles throughout the fatigue life. In the tensile part of
1.5
141 MN/m 2
- 10o
O
71 MN/m2
"v,.. 49 MN/m 2
Alternating body stress
09
Pad span
= 16.5mm
Pad pressure = 103.5MN/m 2
0
0.1
0.2
'
0'.4
0.3
0.5
0.6
0.7
Fraction of life consumed
0.8
0.9
1.0
FIG. 6-- Variation ~?~/Hctionalforce during fatigue te~ts at a range q#alternating body" str~,sses.
RAYAPROLUANDCOOKONFRETTINGFATIGUEINVESTIGATIONS 137
Percentage of life
Alternatingbodystress= 71MN/m2 0 ~
Pad span
= 16.5mm ~ 5 0
Pad pressure
= 103.5MN/m2
S
Frictional
.3
99
0.1
~~
force
Specimen
load
FIG. 7-- Variation of/rictional.fi)rce with ,specimen load at various stages during the fatigue I~[~~.
the load cycle the peak frictional force decreases as the crack grows. This is because the load
path through the specimen is around the crack which reduces the load in the area of pad contact and, hence, frictional load transfer decreases. During the compressive part of the cycle,
however, the load path is through the faces of the crack, the load in the area of pad contact is
undisturbed and the frictional force in the compressive part remains constant and independent of crack length. The cycle at 0.1% of the fatigue life illustrates a point in the bedding-in
process, described above, where macroslip is occurring and the alternating frictional forces are
relatively small. At 0.3% of the life, it can be seen that the magnitude of the frictional force has
reached a maximum, the bedding-in process is complete and microslip conditions have been
reached.
Stabilized frictional load cycles (when the bedding-in process is just complete) are shown in
Fig. 8 for the same alternating stress levels used in Fig. 6. The "cusp" in the frictional force
trace at 141 M N / m 2 (see Fig. 8) occurred due to the formation of a groove in the specimen
surface. It is caused by surface wear from the rubbing action of the fretting pads and is referred
to as "keying-in." As the pad feet come up against debris at the side of the groove, load is
transferred to the pad. This gives rise to the cusp shape and apparent average coefficients of
friction greater than unity (see Fig. 6). The groove can be seen in the photograph shown in Fig.
9a at the highest stress level tested. In contrast, Fig. 9b shows the surface of a specimen tested
at the lowest stress level of 49 MN/rn 2. It can be seen that wear of the specimen surface is quite
severe under macroslip conditions (high stress) and only slight under microslip (low stress)
conditions, where the machining marks on the specimen surface are still evident.
Testing performed under variable amplitude (Gaussian) loading exhibited a somewhat different behavior. The bedding-in process, observed under constant amplitude loading, was disturbed by groups of high amplitude load cycles, causing macroslip in the variable amplitude
loading tests. Following the high amplitude load cycles, the bedding-in process had to start
again. As a consequence, the frictional loads at the onset of macroslip did not usually reach
the peak bedded-in value measured in the constant amplitude tests. The measured frictional
forces, therefore, varied considerably throughout the fatigue tests with the fully bedded-in constant amplitude values forming an approximate upper bound. The measured values generally
138

Alternating
body stress
(MN/m 2)
Frictional force
41
Specimen load
Macro slip
FIG. 8--Stabilized fHctionalj~rce cycles at a range ~f alternating body stresses.

ranged from the upper bound to about half of this value. It was concluded, therefore, that the
action of fretting under variable amplitude loading would be less marked than under constant
amplitude loading, since the frictional forces induced were somewhat lower.
The greater degree of macroslip observed in the variable amplitude tests was accompanied
by a significant increase in the amount of keying-in. Close inspection of the fretting scars
showed a much higher degree of wear on the specimen surface and a smearing of material at
the scar edges caused by the periodic high loads.
Fatigue tests were also performed with a PX-24 water displacing penetrant sprayed onto the
specimen/fretting pad region at various intervals during testing. Figure 10 shows frictional
force measurements made throughout the fatigue tests. Frictional forces measured in tests
which were sprayed before loading commenced were about 25% of the values measured in
non-sprayed tests. No changes to this value were observed when spraying was repeated at regular intervals throughout the fatigue tests. In tests where spraying was applied some way
through the fatigue life, frictional forces decreased upon spraying, but to a lesser degree than
in those tests which were sprayed prior to fatigue loading. The decreased value of frictional
force was typically 50% of the values measured in non-sprayed tests. Fatigue lives of sprayed
specimens were consequently greater than those of non-sprayed specimens and were dependent on the cyclic delay before spray was applied.
Action o f Fretting Forces
To study the action of fretting forces alone a test configuration was designed in which a static
body load could be applied to a specimen and frictional loads could be applied via displacements of the fretting pads. Considerable difficulty was encountered in designing the test setup
[14] and the program was not entirely successful. While crack propagation rates were not successfully measured, crack angles and lengths were determined at crack arrest for a range of
body and frictional loads under microslip and macroslip conditions.
139
FIG. 9--Fretting scars for specimens with a pad pressure of 103.5 M N / m 2 and a pad span ~f16.5 mm.
1.5
,-,
,,2'k
~,
Fatigue life, N
(cycles)
No spray
5.65 x 104
(~
ov
---~o 1.0
Spray
.~. Spray
--
8.66 x 104
r-.__
~ NO.5
2.66 x 105
, 6x
1.91 x i
<
Sprayed before
testing
0.5
Fraction of life consumed
FIG. l O--EO"ect o f water displacing penetrants on.[Hctional,forces.
1.0
140
No cracking was detected when the mean body load was zero. In all subsequent tests a mean
body load was applied via a loading cage. Multiple cracks were evident in the majority of these
tests and grew at a range of angles to the specimen surface. Angular crack growth continued
until crack arrest was observed. Prior to crack arrest, the crack angle (measured from a direction normal to the specimen surface) was generally greater than the initial value, i.e., cracks
tended to turn towards the axis of the specimen. This behavior is in contrast to that observed
in the previously described test series in which cracks turned and grew perpendicular to the
specimen axis under the action of the alternating body loads. The depths of arrested cracks
were proportional to the load in the specimen body. This load is an aggregate of the static body
load and the frictional load induced in the specimen. Detailed analysis of the crack angles and
arrested crack lengths will be undertaken shortly.
Fracture Mechanics Modeling

Fracture mechanics modelling of fretting fatigue cracks would assist in predicting the life
estimates of structures in service from laboratory crack growth measurements carried out
under restricted loading conditions. For such predictions a knowledge of stress intensity factors at the tips of growing cracks under service conditions is required.
Basic Stress Intensity Factor Solutions

Work at RAE has given rise to several solutions for stress intensity factors for cracks under
fretting conditions. Fracture mechanics analysis of fretting fatigue cracks requires certain
assumptions to be made regarding the starting flaw size and the distribution of normal and
frictional forces acting over the contact area.
The stress intensity factor at the tip of a crack growing from the end of a fretting pad is
assumed to be made up from three individual contributions. The first two, from the alternating
body stress (a~) on the specimen and the frictional load (Q), are alternating and contribute to
crack growth. The third contribution, due to the normal force (P) on the pads, is static and
compressive, hence, in the bridge type configurations used it tends to be beneficial and only
affects the mean level of stress intensity factor. Both the normal and tangential forces contribute to Mode I and II stress intensity factors.
Under most of the testing conditions described in this paper, cracks tended to grow uniformly across the width of the specimen except for the very early stages of crack initiation.
Hence, the frictional and normal forces under the pads are assumed to be uniform across the
specimen width. Thus, the stress intensity factor solutions derived tbr inclined edge cracks in
a half-plane can be utitized.
In the analyses of Edwards, Ryman and Cook [9] cracks were assumed to grow normal to
the specimen surface. Assuming various distributions of the normal and tangential forces
along the pad feet, Fig. 11, the pad load contributions to the stress intensity factor were calculated by integrating over the fretted surface the product of the stress at any position and the
corresponding Green's function value derived by Rooke and Jones [18]. The contribution of
frictional force to the stress intensity factor is twofold. The first contribution, based on the
assumed distribution of force on the contact area, results in an increase in Mode I stress intensity factor, particularly in the early stages of crack growth. The second contribution is negative,
due to the fact that the specimen alternating body stress is lower under the pads because some
load is diverted through them.
Thus, for a crack normal to the surface the alternating Mode I stress intensity factor Ka is
given by
Ka Mon
= Mar
1.12
a a ~ EDT+2016
Q . GO - 1.12 7 9 w x / ~
28 00:05:25
(1)
RAYAPROLUANDCOOKONFRETTINGFATIGUEINVESTIGATIONS
141
Pad foot
Crack..... ~
Specimen .
i_/"
Line loads along

crack root
t.._
- _,J-
Positive lineardistribution
of line loads
L_2-__L-C-
Line loads along end of

fretting pad foot
.....
__)
Negativelineardistribution
of line loads
-%
Parobolic maximum
distribution of line loads
__
Parobolicminimum
distributionof line loads
Uniform distribution of line loads

FIG. 11--Assumed distributions q/tangentia/ and m)rma/ pad loads.
where a is the crack length, w is the area of the fretting scar, i.e., the true contact area, {ra is the
semi-range of alternating body stress, Q is the frictional force per unit pad length and unit scar
width, G ] is the Mode I Green's function due to a point tangential force on the surface, and
A is the cross sectional area of the specimen.
The static mean stress intensity factor K,~ is given by
K~ = 1.12~,,x/~ + P . Gf
(2)
where P is the normal force per unit length and unit fretting scar width, Gf is the Green's
function for Mode I crack propagation under the influence of a point normal force acting on
the surface of a half plane, and ~m iS the mean stress acting on the specimen body.
In the early stages of crack growth under fretting conditions the cracks tend to be small, less
than 1 mm. Peek [ 19] has suggested that at these short crack lengths the plastic zone at the tip
of the crack is more effective at opening the crack than at longer crack lengths. Hence, the
effective stress intensity factor at short lengths is greater. To account for the plastic zone effects
Edwards gt al. [9] applied an empirical length correction to the above computed stress intensity factors. Initially, this correction was added to the alternating/~, thus increasing its magnitude. However, this is open to criticism since the effective alternating stress intensity factor
at short by
crack lengths arises from a change in the mean stress intensity factor. In the subsequent
Downloaded/printed
142
predictions of crack growth rate and in the life prediction computations [I1] the length correction has been made to the mean stress intensity factor.
Both the alternating and the mean stress intensity factors computed as above were corrected
for the finite width [11] of the specimen. The correction was based on a specimen with
restrained ends [20].
The Influence of Slant Cracks

In the above analysis cracks were assumed to grow normal to the specimen surface. However, in the initial stages of crack growth the influence of Mode II propagation caused cracks
to be slightly slanted and to grow under the fretting pads. Therefore, the basic configuration
that needs to be analyzed is a slant crack growing under the influence of normal and tangential
forces acting on the surface of the specimen. The configuration is shown schematically in Fig.
12 with a crack of length a inclined at an angle 0 from the normal to the specimen surface.
To obtain the stress intensity factor values for the above configuration, crack line Green's
functions have to be determined for point forces acting on the crack. Hartranft and Sih [21]
determined them for the special case of a normal edge crack, and Hills, Sackfield and Uzel
[22], and Nowell and Hills [23] for a slant crack. Aliabadi, Rooke and Cartwright [24] used
boundary element analysis, coupled with a weight function technique, and obtained accurate
crack line Green's functions. Using these values and conventional stress analysis it is possible
to evaluate the Green's function for a point force acting on the specimen surface. Recently,
Rooke, Rayaprolu and Aliabadi [25] have obtained these functions for several crack angles.
For any arbitrary stress distribution on the surface, these Green's functions can be integrated
to give the stress intensity factor for inclined cracks under the influence of these stress distributions. Stress intensity factors for various configurations are being determined at present.
Assuming a unit magnitude of normal or tangential force acting on a pad of unit length and
unit fretting scar width, the Mode I and Mode II stress intensity factors have been evaluated
for cracks growing normal to the surface and at various angles 0 to the surface. The assumed
distribution of the normal and tangential forces are shown in Fig. i 1. For each of these distributions the variation of stress intensity factor with crack length was calculated. Figures 13, 14,
15 and 16 show typical results for K~ due to P at 0 ~ and 45 ~ and K~ due to Q at 0 ~ and 45*
respectively.
The results show that the crack inclination has a large influence on the magnitude of the
maximum value of Kt. In the case of K~, due to the normal forces (P), the maximum value of
K~ observed for a 45* crack (Fig. 14) is about five times that for a normal crack (Fig. 13). For
an assumed minimum parabolic distribution of normal force (P) or tangential force ( Q ) (see
Fig. 11), the stress intensity factor for inclined cracks (Figs. 13 to 16) shows a change of slope
b .!P
=Q
FIG. 12--Normal (P) and tangential (Q_)point forces on edge of half plane (x ~ O) containing a slant
crack.
-1.0
-0.8
\
\
~
'
"
'
-~
.............
........
- - - . . . .
-0.6
/ ~
,,
-0.4
//
'
~~ . . - - ~ c ~ . ~ _
143
Crack root
End of pad
Uniform
+ve linear
ve linear
Max parabola
Min parabola
........
-0.2
0.5
1.0
Crack length/Pad length
2.0
1.5
FIG. 13--Influence o f a s s u m e d normal stress distribution in the contact area on Kt (0 = 0~
-4
Crack root
End of pad
Uniform
........... +ve ~inear
.........
ve linear
Max parabola
Min parabola
-3
P
KI
-2
-1
9 ""
0.5
,,
1.0
--
1.5
FIG. 14--1nfluence o f a s s u m e d normal stress distribution in the contact area on KI (0 = 45 ~

2.0
144
"',.
-0.2
iI)~v.{
\ \ :'%_~'~
. - . ...............................
". ~. . - . ~ .:......
' . :~ _~
"
-0.4
"
K,I
....
-0.6
~
J
J
J
-0.8
-1'00
0.5
---- - ~
......
..........
.....
]
1.0
Crackroot
End of pad
Uniform
+ve linear
ve linear
Max parabola
Min parabola
1.5
2.0
FIG.15--Influenceo f assumed tangential stress distribution in the contact area on KH (0 = 0~
0.5
- - -
~....'
'
Crack root
End of pad
Uniform
............ +ve linear
ve linear
.......
Max parabola
----Min parabola
-
\
\', ~-.X
"'"
'"..
K,I
-0.5
-1.0
0
FIG.16--Influence
0.5
1.0
1.5

Mar 28 00:05:25
EDT 2016
o fMon
assumed
tangential
stress distribution in the contact
2.0
area on Kit (0 = 45~
145
at a crack length corresponding to the position where the force on the pad has a minimum
value. The assumed parabolic distribution is a rough approximation to the distribution of
forces during certain contact situations [26].
The main conclusion from the above analysis is that for any assumed force distribution,
stress intensity factors under fretting conditions are significantly affected by the angle of crack
inclination,
Application of Fracture Mechanics to the Prediction of Crack Growth on Fretting Specimens

To estimate the fatigue limit and crack growth rates certain assumptions have to be made
regarding the initial flaw size, crack inclination and the distribution of forces on the contact
area. Assuming a uniform distribution of normal and tangential stresses along the length of
the pad and a range of initial flaw sizes from 0.02 m m to 0.1 ram, initial stress intensity factors
were evaluated [9]. Utilizing Pearsons [27] crack growth rate data for BS L65 aluminum alloy,
the life of specimens with constant alternating body stresses have been evaluated for the chosen
range of initial flaw sizes (see Fig. 17). These curves show that the predicted fatigue limit is not
significantly dependent on the chosen initial flaw size. The predicted life at higher stress amplitudes is less satisfactory than the correlation at lower stress amplitudes.
F r o m the measured values of frictional forces, Edwards, Ryman and Cook [9] derived crack
growth rates and hence fatigue endurances for tests carried out with four different pad spans
and a chosen value of normal load (see Fig. 18). For these predictions the normal and tangential loads along the pad foot have been assumed to increase linearly from the crack root. The
crack was assumed to be normal to the surface and initially of length 0.02 ram. Based on similar assumptions, fatigue endurances were computed for two different pad pressures for a specific pad span of 16.5 ram. The computed results were compared with experimental results
(see Fig. 19).
250
&-" 200
v
,r 150
_8
....
0 01 Uniform distribution o f 'anuem'a'
\ \\\\~/~
.
and vertical pad loads.
\\\\\
>o<x Assumed initial flaw sizes shown (mm)
\ \\\'L~.j~0.02 Constant amplitude loading.
\ \\\\\
Zero mean stress
0.1 " ~~- ~ , " ~ 0 0 5 : ~
Pad span is 16.5mm
"
"~"~N
Pad pressure is 103'5MN/m2
0.2.~,~xx
loo
x x~
0"5 ~ ~ _ ~ x , ~ x
if: 50
X
X
03
104
105
106
107
108
FIG. 17--Eff2,ct of varying the initial flaw size on predicted./~z,ttingjatigue
I~/~~
146

&-- 200
E
z
16.50m - ~ m
150
nm
~\
~
t.,O
c")
6.35 mm
2.00 mm
~., 100
"O
O
~
E
Test points
+ 34 35 mm "]
~--,,,,~,~, x , x
~---,,~c~x:>-<>--x+-x+--x+
5o
6.35mm [[ spans
x 2.00 rnm 9
oio~__,u
mm
<
1
03
104
;Jq-.~b
105
10 6
107
108
FIG. 18--Predicted and achieved lives at a pad pressure of 103.5 M N / m 2 and four pad spans.
200
150
103.5MN/
~<
v
o9
o9
"10
O
..Q
O'}
C
100
50
<
o 103.5MN/m 2
x 41.4MN/m 2
0
103
104
[] [ ]
Pad pressure
J
105
106
0_ O ~
N--~-
.L.,.
107
10 8
FIG. 19--Predicted and achieved lives at two pad pressures and a pad span o f 16.5 rnm.
In the above two comparisons the predicted fatigue lives matched well at lower alternating
stresses. The predictions were better for smaller pad spans and lower pad pressure. At larger
alternating body stresses, pad pressures or pad spans, the predicted fatigue endurances are
much shorter than the measured values. Edwards [28] reported that under microslip situations
for a given alternating body stress the inclination of the crack increases as the pad span
increases. Thus, from Figs. 18 and 19 significant differences between measured and predicted
behavior were noticed when the cracks were growing at large angles of inclination. From Figs.
147
l 3 and 14 it can be seen that/(1 due to the normal force (which is compressive) increases rapidly with crack inclination. It follows that calculating the crack growth rates based on stress
intensity factors derived for inclined cracks would increase the compressive/s lower the predicted crack growth rates, and reduce the apparent disparity between the measured and calculated fatigue endurances. Hence, future measurements and modeling should involve crack
length and crack inclination so that appropriate values of stress intensity factors may be
determined.
R e c o m m e n d e d Testing Procedures
Fretting fatigue tests are carried out for a variety of purposes. Those at RAE have been in
support of research work to obtain a better understanding of the fretting fatigue process. The
test procedures developed at RAE are, however, applicable to most other test purposes. The
All d i m e n s i o n s in m m
Scale 1 9 1
45.10
45.00
Q_
(DO
dN
11.90
11.85
Oc
Oc
O_
O
O,I(
)
)
/
l,
1--10.021A-BI--
~45
FIG. 20a--Recornmended standard fretting specimen for fretting fatigue tests.

148
test rig used for frictional force measurements, described previously, may also be used to compare the fretting fatigue performance of different test piece and fretting pad material combinations, surface treatments, fretting palliatives, and many other parameter variations. It has
the advantage that not only comparisons of fatigue endurance can be performed, but also a
quantitative measure of frictional load transfer can be made. Both quantities are important in
assessing the effect of different surface parameter combinations. This is particularly true for
surface treatments which rely on the induction of surface residual stresses in the test piece. The
fatigue life indicates the effect of the sum of the residual and fretting stresses. The frictional
force measurements indicate the effect of the surface treatment on the fretting stresses alone
and, hence, the two effects may be separated. The test rig described can, therefore, fulfill a
number of essential requirements of fretting fatigue testing. It should be emphasized, however,
that the test rig was designed for fretting fatigue tests and not fretting wear tests. It is envisaged
that some form of spherical contact would be more appropriate for wear tests. This geometry,
would avoid deep keying-in of the fretting pad feet and subsequent rectangular groove formation, as described above. The fretting pads could, however, be simply modified to have
rounded feet for such applications.
The test rig described above is considered to be a suitable candidate on which to base a standard design. Details of the rig are given in Fig. 20. The specimen was designed to be tested in
All d i m e n s i o n s in m m
Scale 2 " 1
General tolerance -+0.1
C/bore to
suit screw
"'a.. _ _ q
..I---4
/Pad
Strain gauge
positions
span
"'l---4
-- ~-72__ t
o151
0.50
M3 c l e a r a n c e
FIG. 20b--Recommended standard fretting pad for frettingJi~tigue tests.

All dimensions in mm
Scale 2 " 1
General tolerance -+0.1
149
___~6.00
(DO
O(D
M3 clearance
Loading
bali
To suit
fretting
pad
_~
1!_.
!
\
Half pad
span
_F_-./
3.20
IFIG. 20c--Recommended standard loading pad for frettingjatigue tests.
a Schenck resonant fatigue test machine, which accounts for its characteristic dumbell shape.
Split collars are placed over the ends of the specimen and a nut is used to clamp the collars in
place. The specimen may be loaded in tension or compression and a simple threaded adaptor
can be made to fit any modern fatigue testing machine.
The strain gaged fretting pads may be calibrated in any test machine which can be controlled
in displacement mode. The following method is recommended. The specimen is cut in half
and assembled in a test machine with a small gap between the specimen halves. The fretting
pads and proving ring arc then assembled. Under displacement control, a sinusoidal displacement is applied to the specimen and the achieved load is monitored. This will gradually
increase as the pads bed-in. When the transmitted load reaches about half of the pad load, a
trace of the transmitted load as a function of the measured pad strain is recorded. The slope
of this trace will give the pad calibration value which enables the conversion of measured fretting pad strain to frictional load transfer. The fretting pads were designed as reusable items. In
the RAE investigations steel fretting pads (BS $98) were used in conjunction with aluminum
alloy specimens (BS L65). The fretting pads did not wear, but at the end of each test they
needed to be carefully cleaned to remove the aluminum/oxide buildup. The pad cleaning gradually wore away the fretting pad feet and the pads had a finite life of about 20 tests. This life
150

All dimensions in mm
Scale 1 9 1
General tolerance +0.5
M8
Load adjusting
screw
Strain gauge
positions
.
Loading
balls
68.00
75.00
FIG. 20d--Recommended standard proving ringfor fretting fatigue tests.
could be extended if the pad feet were deeper, but an increase in pad foot depth could lead to
fatigue failure of the feet themselves. It is recognized that one set of pads per specimen may be
required when testing similar specimen/pad materials or when the pad material is less hard
than the specimen material. The pad span may be varied to suit individual requirements. The
minimum span will be governed by the requirement for a strain gage to be bonded between
the pad feet and the maximum span will be governed by the length of the fiat portion on the
test specimen. In practice, however, increasing the pad span beyond 16.5 mm had little effect
on fatigue endurance. It could be argued that this may correspond to a maximum damage limit
and that increasing the pad span significantly beyond this value may result in an increase in
fatigue life. This is because extensive sliding would occur and the average coefficient of friction
would be lower than in the maximum damage limiting case.
The proving ring is also strain gaged and may be calibrated by loading through the axis of
the load adjusting screw. It is relatively flexible in order that wear of the specimen surface does
not result in a significant reduction in the normal applied load. The normal load may be varied
over a wide range within the elastic limit of the proving ring. It should be remembered, howCopyright
by ASTM
(all rights reserved);
Mon Mar 28pad
00:05:25
EDThas
2016 a dual effect; it causes an increase in the attainever,
thatInt'l
increasing
the normal
load
151
able value of frictional force before the onset of macroslip (detrimental), but tends to close
propagating slant cracks (beneficial). The net result is uncertain.
In summary, the test rig shown in Fig. 20 is recommended as a standard for adoption by
ASTM. It is versatile in its applications and a large database of experimental results already
exists. For wear tests it is suggested that the same test set-up be used but the fretting pads should
be redesigned with rounded feet.
Future Work
The authors believe that extensive further research on many aspects of fretting fatigue is
required despite the signficant efforts that have already been made. There is a need to improve
our understanding of the mechanisms involved, and our ability to predict failure sites, fatigue
endurances and crack growth behavior. The authors agree with the ASTM proposal that further work is also required to standardize fretting fatigue test methods and equipment. One
technique for investigating fretting fatigue is proposed in this paper. However, there are many
other areas where research effort should be directed. These areas include the effect of contact
parameters on crack initiation and growth (with particular reference to initiation sites, angular
crack growth, short crack growth and mean stress effects) and fracture mechanics and other
predictive methods (particularly comparisons, applications and shortcomings of existing life
prediction models, the development of stress distribution models and associated angular crack
growth models). Advances in these areas could eventually lead to the development of fretting
fatigue models suitable for airframe design applications.
References
[1] Milestone, W. D. and Janeczko, J. T., "Friction Between Steel Surfaces During Fretting," Wear,
Vol. 18, 1971, pp. 29-40.
[2] Wharton, M. H., Waterhouse, R. B., Hirakawa, K., and Nishoika, K., "The Effect of Different Contact Materials on the Fretting Fatigue Strength of an Aluminium Alloy," Wear, Vol. 26, 1973, pp.
253-260.
[3] Endo, K., Goto, H., and Nakarnura, T., "Fretting Fatigue Strength of Several Material Combinations,"Bull, ofJSME, Vol. 17, No. 92, 1973.
[4] Endo, K., Goto, H., and Fukunaga, T., "Behavior of Frictional Force in Fretting Fatigue," Bull. of
JSME, Vol. 17, No. 108, 1974.
[5] Endo, K. and Goto, H,, "Initiation and Propagation of Fretting Fatigue Cracks," Wear, Vol. 38,
1975, pp. 311-324.
[6] Edwards, P. R. and Ryman, R. J., "Studies in Fretting Fatigue under Variable Amplitude Loading
Conditions," RAE Technical Report 75132, 1975.
[7] Edwards, P. R. and Cook, R., "Frictional Force Measurements on Fretted Specimens under Constant Amplitude Loading," RAE Technical Report 78019, 1978.
[8] Edwards, P. R. and Cook, R., "Frictional Force Measurements on Fretted Specimens under Variable
Amplitude Loading," RAE Technical Report 78059, 1978.
[9] Edwards, P. R., Ryman, R. J., and Cook, R., "Fracture Mechanics Prediction of Fretting Fatigue,"
in Proceedings, Ninth ICAF Symposium Darmstadt, 1977.
[10] Edwards, P. R. and Cook, R., "Fracture Mechanics Prediction of Fretting Fatigue under Constant
Amplitude Loading," RAE Technical Report 77056, 1977.
[11] Edwards, P. R, and Cook, R., "Fracture Mechanics Prediction of Fretting Fatigue under Gaussian
Random Loading," RAE Technical Report 78086, 1978.
[12] Rooke, D. P. and Edwards, P. R., "Waveforms in Fretting Fatigue," RAE Technical Report 87032,
1987.
[13] Rooke, D. P. and Courtney, T. J., "Effect of Variable Friction Coefficient on Fretting Fatigue Waveforms," RAE Technical Report 88015, 1988.
[14] Walker, C., "An Experimental Investigation of Fretting Fatigue,'" MSc Thesis, University of Bristol,
1989.
[15] Cornelius, H. and Bollenrath, F., "Eintluss von Einspann unger auf die Wech Selfestigkeit yon unegieltem
stahl,"
Vol. 14,
335-340.
Copyright by ASTM
Int'l (all
rights reserved);
Mon1941,
Mar 28p.00:05:25
EDT 2016
152
[16] Fenner, A. J. and Field, J. E., "Fretting under Fatigue Conditions," NEL Report AB Div. 16/57,
1961.
[17] O'Conner, J. J. and Johnson, K. L, "The Role of Surface Asperities in Transmitting Tangential
Forces between Metals," Wear, Vol. 6, 1963, pp. 118-138.
[ 18] Rooke, D. P. and Jones, D. A., "Stress Intensity Factors in Fretting Fatigue," Journal of Strain Anal
ysis, Vol. 14, No. 1, 1979, pp. 1-6.
[19] Pook, L. P., "Various Aspects of the Fatigue Damage Threshold in Mild Steel," Proceedings, SEE
Conference on Fatigue Testing and Design, London, 1976.
[20] Rooke, D. P. and Cartwright, D. J., Compendium of Stress Intensity Factors, HMSO, 1974.
[21] Hartranft, R. J. and Sih, G. C., "Alternating Method Applied to Edge and Surface Crack Problems,"
in Methods of Analysis and Solutions of Crack Problems, G. C. Sih, Ed., Ch. 4, Leyden, Nordhoff
Intnl. Publ., 1973, pp. 179-238.
[22] His D. A. Sacked A and Uze A. ``The Green s Functin fr a Sant Edge Crack Engineering
Fracture Mechanics, Vol. 20, 1984, pp. 245-253.
[23] Nowell, D. and Hills, D. A., "Open Cracks at or Near Free Edges," Journal of Strain Analysis, Vol.
22, 1987, pp. 177-185.
[24] Aliabadi, M. H., Rooke, D. P., and Cartwright, D. J., "Mixed-mode Bueckner Weight Functions
Using Boundary Element Analysis," International Journal of Fracture, Vol. 34, 1987, pp. 131-147.
[25] Rooke, D. P., Rayaprolu, D. B., and Aliabadi, M. H., "Crack-Line and Edge Green's Functions for
Stress Intensity Factors of Inclined Edge Cracks," to be published in Fatigue and Fracture, Eng.
Marl. Struct.
[26] O'Connor, J. J., The Role of Elastic Stress Analysis in the Interpolation of Fretting Fatigue Failures
in Fretting Fatigue, R. B. Waterhouse, Ed., London Applied Science Publishers Ltd., 1981, pp. 2366.
[27] Pearson, S., "Initiation of Fatigue Cracks in Commercial Aluminum Alloys and the Subsequent
Propagation of Very Short Cracks," RAE Technical Report 72236, 1973.
[28] Edwards, P. R., "Fracture Mechanics Applications to Fretting in Joints," in Proceedings, 6th Int.
Conf. on Fracture, Vol. 6, New Delhi, 1984, pp. 3813-3836.
Trevor C. L i n d l e y 1 a n d Kevin J. N i x ~
Fretting Fatigue in the Power Generation

Industry: Experiments, Analysis, and Integrity
Assessment
REFERENCE: Lindley, T. C. and Nix, K. J., "Fretting Fatigue in the Power Generation Industry: Experiments, Analysis, and Integrity Assessment," Standardization of Fretting Fatigue Test
Methods and Equipment, ASTMSTP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 153-169.
ABSTRACT: Experimental procedures for carrying out fretting fatigue tests using either the
proving ring method or a specificallydeveloped servo-hydraulic biaxial rig are described. S-N
curves are generated with and without fretting in order to establish the important variables that
can affect fretting fatigue and to rank material fretting combinationsthat should be avoided. The
S-N tests and associated metallography indicate that small cracks, typically 0.25-0.5 mm deep,
initiate rapidly at only a small fraction of fatigue life. Fracture mechanics and near-threshold
fatigue concepts can be used to predict the conditions for continued growth or arrest of the small
fretting defects.
KEY WORDS: fretting fatigue, experimental fretting rigs, frictional force measurement, S-N
tests, fracture mechanics, small crack growth
Fretting is promoted by high frequency, low amplitude vibratory motion and commonly
occurs in clamped joints and "shrunk-on" components. It is particularly important in the
power generation industry. For example, in turbogenerators, fretting is possible at several locations including: turbine blade root to disc fixing: disc to shaft seating: key/keyway contact in
the disc to shaft assembly; end ring fixings in generators; button drives in turbines; and radial
stalks in generators. Tubes in boilers and heat exchangers, as well as the overhead cables which
transmit electricity, are also prone to fretting wear and/or fatigue.
This paper describes the experimental procedures developed to establish the fretting fatigue
properties of various materials. To establish the important variables which can affect fretting
fatigue life, many studies [1-5] have generated S-N curves, with and without fretting, which
allow fretting fatigue strength reduction factors to be established for the material combination
of interest. With the realization that small cracks are introduced at an early stage of fretting
fatigue life [6-8], more recent studies have often been concerned with measurement of the
rate of development of such cracks and the use of fracture mechanics concepts to model this
growth [ 7-1I ].
Two distinct methods have been used to achieve the controlled fretting between two contacting surfaces. Earlier experiments [8] involved clamping contact pads against a fatigue specPlant Engineer and Research Officer, respectively, National Power Technology and Environmental
Centre, Leatherhead, Surrey, United Kingdom, KT22 7SE.
153
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
lntcrnational
www.astm.org
154
imen (Fig. 1) using a load calibrated steel proving ring, the assembly being mounted in a standard uniaxial fatigue machine. The second more sophisticated method uses a biaxial test rig
with two pairs of actuators on horizontal and vertical axes specifically designed for fretting
fatigue experiments (Figs. 2a and 2b).
Proving Ring Method

The simulation of service conditions dictated the need for contact loads as high as 300 MPa
and therefore several steel rings of different sizes were manufactured to achieve optimum sen-
FIG. l --Proving ring used in fretting fatigue test assembly.
Copyright by
ASTM
Int'l (allBiaxial
rights reserved);
Mon Mar 28 00:05:25 EDTtest
2016rig and (b) detail of specimen~contact pad
FIG.
2--(a)
servo-hydraulicJretting
assembly.
LINDLEY AND NIX ON POWER GENERATION INDUSTRY
155
sitivity at either low or high loads. Strain gages were bonded to each proving ring and, using a
strain gage amplifier, load calibration was carried out in the usual manner.
The clamping load between the contact pads and specimen was transmitted via steel balls
to give the required contact pressure. The fretting ring assembly (Fig. 1) was installed and
fatigue tests carried out on an Amsler Vibrophore machine typically resonating at about 150
Hz with a sinusoidal waveform of constant amplitude. Although this machine had the advantage of high test frequency, a standard uniaxial servo-hydraulic fatigue machine would be
required where load waveform or test frequency were primary test variables. The servohydraulic machine, rather than a resonant machine, is also required for variable amplitude
loading.
In the present experiments, maintenance of the proving ring load required a brief settlingin period, thereafter remaining constant apart from material combinations showing pronounced fretting wear (see next section).
Biaxial Fretting Fatigue Rig

A biaxial test rig (Figs. 2a and 2b) has been specifically designed to give improved control
of the relative "slip" between specimen and pad. This rig has two pairs of actuators on the
horizontal (contact loading) and vertical (fatigue loading) axes and was usually cycled at 90
Hz. The load cell circuitry provided switchable ranges 0f25, 50, 100, and 250 kN.
The biaxial rig has the following advantages: (1) more precise control and measurement of
relative slip between pad and specimen; (2) more convenient application of a wide range of
contact loads and better suited to the application of large contact loads up to 250 kN, allowing
simulation of in-service loadings; (3) constant contact loads throughout the test duration when
substantial fretting wear would otherwise result in load relaxation; (4) easy interruption of tests
for inspection of surface damage prior to return to the original test parameters; (5) variable
amplitude loading can be readily achieved; and (6) possibility for out-of-phase fatigue and contact loadings.
Fatigue Specimen
Fatigue specimens of the type shown in Fig. 3 were machined from the materials of interest
(Table 1). Flat pieces were machined on opposite sides of the specimen gage length and then
polished down to a 600-grade silicon carbide (SIC) finish in the direction of fatigue stress and
99.5
48 mmx 1 mm PITCH
DIMENSIONS IN mm
2.5 x45 ~
,~? ,,.-,~/
<:
3--Frettingfati#ue specimen.
Copyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25

FIG. EDT 2016
156
FRETTING FATIGUETEST METHODS AND EQUIPMENT

TABLE l--Mechanical properties at room temperature (~/materials.
Material
3.5NiCrMoV
1CrMo
18Mn4Cr
18Mn 18Cr
2014A (Longitudinal
Orientation)
0.2% Proof
(MPa)
Tensile
Strength
(MPa)
Elongation
(%)
Reduction
in Area
(%)
Hardness
(VPN)
600
841
1000
1000
458
733
999
1180
1050
504
25
21
29
29
9.8
70
59
42
68
...
222
340
429
359
155
degreased. Flat plate specimens of similar gage section to that shown in Fig. 3 were machined
from 2014A aluminum alloy. For the turbo-generator materials, all specimens were machined
such that their axial orientations were coincident with the rotor axis.
Contact Pads
Several fretting pad geometries have been used in fretting fatigue tests, but the most commonly employed is either the cylindrical or the bridge type, the latter being shown in Fig. 4.
Cylindrical fretting pads have the advantage that the contact stress distribution is well defined
and the stress in the fatigue specimen can be found without the need for finite element analysis
[12]. On the other hand, bridge pads often allow the contact geometries found in practice to
be more closely simulated. In the present experiments (Tables 1 and 2), the fretting pad span
ranges from 12,5 to 50 mm, giving a wide variation in applied slip amplitudes. The pad feet
were ground to a 600-grit finish in the direction of fatigue stress and then degreased (the standard finish). Complications can arise however with bridge contact pads. In examining the
effect of contact pressure on fretting 1CrMo pads against 3.5NiCrMoV rotor steel specimens,
contact pressures of 30 and 300 MPa were chosen. The fretting fatigue curves were quite similar for the two contact pressures (Fig. 5), despite the markedly different fretting scars (Figs. 6a
DIMENSIONS IN m m
J
II
STRAIN GAUGE
II ....
COUNTERSUNK FOR LOADING BALL
I
PAD SPAN IS)
FIG. 4--Bridge-type contact pad.

157
TABLE 2 - - F a t i g u e strength reduction factors due to fretting.
Fatigue Specimen
Contact Pad
Contact
Pressure
(MPa)
3,5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
l 8Mn4Cr
18Mn4CT
18Mn 18Or
18Mnl8Cr
2014A AI
2014A AI
2014A AI Peened
1CrMo
1CrMo
1CrMo
ICrMo
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
3.5NiCrMoV
30
30
300
300
20.7
20.7
20.7
20.7
30.8
30.8
30.8
Fatigue
Mean
Stress
(MPa)
Fatigue
Strength
(Unfretted)
(MPa)
Fatigue
Strength
(Fretted)
(MPa)
Reduction
Factor
Wear
Rate
0
300
0
300
0
300
0
300
75
125
125
_+300
+_215
_+300
+ 215
_+250
+_ [25
+_250
+ 185
_+ 140
_+ 135
+_ 135
_+ 140
_+60
+_ ! 30
_+60
_+ 100
+_50
_+ 165
+_70
+_ 15
+ 12.5
+ 50
2.1
3.6
2.3
3.6
2.5
2.5
1.5
2.6
9.3
10.8
2.7
Medium
Low"
Medium
Low a
High
High
High
High
Low
Low
Low
"See text.
340
320
-300
UNFRETTED
30 MPa NOMINALCONTACTPRESSURE
260
24O
PAD SPAN24,1 mm
300 MPa NOMINALCONTACTPRESSURE
22O
200
180
.a
160
14o
12o
i
N
I00
8O
6O
40
20
I I
0
i0 s
It'll
~-'
106
1111111
i
107
L ,+l~ltl
,
10B
t IIIIL]
109
CYCLES TO FAH. URE
FIG. 5--S-N curves with and without/?eltingjbr 3 . 5 N i C r M o I 9rotor steel at zero m e a n stress.
and 6b). However, at the high n o m i n a l contact pressure of 300 MPa, the edges of the pads
apparently lift to give a " w i n d o w " type scar (Fig. 6b) in which the fretting cracks initiate. The
actual contact pressure where fretting initiates is judged to be m u c h lower than the nominal
value o f 300 MPa.
For plant application reasons, surface finishes other than to the " s t a n d a r d " 600-SIC final
polish are required. O n fretting 3 . 5 N i C r M o V steel pads against a 2014A a l u m i n u m specimen
in the as-machined condition, the fretting wear is uneven for tests close to the fatigue limit and
with a n o m i n a l contact pressure of 31 M P a (Fig. 7). A duplicate test again produces uneven
wear but a quite different scar pattern, making both estimates of actual contact pressure and
standardization of the fretting test extremely difficult for these particular test conditions.
158
FIG. 6--Wear scars produced by.fretting 1CrMo steel contact pads against 3.5NiCrMo V rotor steel
using nominal contact pressures ~f(a) 30 MPa and (b) 300 MPa.
FIG. 7--Wear scar from fretting 3.5NiCrMoV steel pads against 2014A aluminum at 125 +_ 25 MPa
applied stress.
Spink [13] fretted bridge type contact pads machined from various ferritic and austenitic
steels against a 2.5NiCrMoV rotor steel. He found that the fretting fatigue limit was largely
insensitive to changes in pad material despite major differences in surface damage. These
observations are in agreement with the present survey (Table 2). Spink also found that the wear
damage increased progressively with increasing slip amplitude and that fretting fatigue cracks
always appeared away from areas of maximum wear. Indeed, in many tests, cracks initiated
in areas where surface damage was barely detectable. The results in Table 2 for 3.5NiCrMoV
steel pads fretted against 2014A aluminum specimens indicate a very large reduction in fatigue
strength due to fretting and yet minimal wear was observed.
Frictional Force Measurement
Frictional forces between the fretting pad and the specimen can be measured by bonding
strain gages to the underside of the bridge type contact pads and measuring pad deflection, a
technique similar to that adopted by Edwards and Cook [ 14]. Such measurements fulfill the
following objectives: (1) detection of the initiation of a fretting fatigue crack; (2) monitoring
of subsequent crack development; and (3) the measurement of frictional forces for use in a
fracture mechanics treatment of crack growth. An outline of the experimental arrangement is
shown in Fig. 8. The approximately sinusoidal output from the pad strain gauge is recorded
by feeding the signal through a transient recorder and continuously plotting samples of the
159
sfroin
gauge
Sfrain OclugeAmphfier
OscULoscope
Trclnsienf ~ecorder
<
Charf Recorder
<
I
Pad sfrain range
FIG. 8--Measurement o]i/>'ictionalforces.
output on an x-time recorder. In this way, the range of surface strain (AcB) from the bottom
surface of the pad is determined continuously for each pad throughout the fatigue test. The
frictional force (AFt) at each fretting pad foot results in both bending and uniaxial extension
of the pad and pad strain (A~B)is given by
EpAD ' A6B =
B.D
" CB
(l)
where EpAD is the Young's modulus of the pad material, and B and D are the pad thickness
and height, respectively. The term CB is a correction factor which accounts for the bending
contribution to pad deformation; it is related to pad span S such that as S ~ ~ , CB ~ 1. The
factor CB was measured experimentally using a split fatigue specimen technique as described
in detail by Edwards and Cook [ 14], in which the static load applied to the specimen is diverted
160
entirely through the pads. A check was also made of the dynamic calibration by bonding strain
gages to the center of a standard fretting fatigue specimen and monitoring the reduction in
specimen strain due to load redistribution through the pads. Good agreement was found
between the static and dynamic calibrations.
Frictional force measurements were made for 2014A T6 aluminum in contact with bridge
type contact pads (span 25.4 ram) made from 3.5NiCrMoV rotor steel. A typical plot of frictional force range A F t v e r s u s load cycles in a fretting fatigue test is shown in Fig. 9. Frictional
forces initially rise as the fretting wear develops, followed by a "plateau" region when they are
essentially constant. The plateau value of frictional force is maintained throughout the test
unless the initiation of a macro fatigue crack occurs, in which case the frictional force measured on the side of crack initiation can decrease rapidly (Fig. 9).
The plateau value of frictional force for various contacting materials was determined for all
conditions of alternating stress, contact pressure and pad span. In all cases, frictional force
varies about a mean of zero and applied mean stress has no effect on measured values of,SF,.
Plots of the plateau frictional force versus applied stress amplitude 3.5NiCrMoV rotor steel
are given in Fig. 10 for various contact pressures, pad spans, and pad materials. The measurements [15] indicate that at low contact pressure, the peak frictional force LxFt/2 increases with
applied stress and achieves a limiting value approximately equal to the applied contact load
on each pad foot. This implies that a maximum coefficient of friction equal to unity is achieved
under fretting conditions at an applied stress amplitude greater than some critical value. This
critical value is itself dependent on pad span and pad material. These findings are in agreement
with those of Endo, Goto, and Fukunaga [16] and Edwards and Cook [14].
600
Ft
500
'~CRACK
(2)
DEPTH 4rnm
400
z
u.
300
CRACK iNITIATION
SIDE, Ft (I)
L==
200
100
POSITION OF
CRACK
3 4
~,
5 6
7 8
9 1 11 12 13 14 15 1 17 18
CYCLES x 10s
FIG.
9--Frictional /orce variation with endurance in 2014.4 a/umimlm.
161
; -~-CONTACT PAD PAD

! PRES
~SURE SPANMATERIAL
1200--
1000
]
BOO
60.
/;.,
'
400
/A
/
/
.~ A~&-- A
/x
A~o~
O"~r-
.,"
/~
.&"
O...~lI"
r~ t5/ _O'~'
d"
~.
..~ <>--C'~
COEFFICIENTOFFRICTION=,
FORCONTACTPRESSURE=30MPa
/ .~-~>
0 f AI ~/' t -IIr4"D""
200
0
0
I
i
I
i
i
1
I
75
100
125
150
175
200
225
APPLIED:STRESSAMPLITUDE,MPa
FIG. 1O--Relationship between plateau vahw of~'icHonaljbrce and applied stress.
25
50
The nominal relative slip range 3n at each pad foot is given by

S
6~ = , ~ E
(2)
where E is Young's modulus of the specimen and cris the applied stress amplitude. This nominal slip range can be corrected to give the true slip range/~ by employing the pad strain measurements [15] to account for pad deformation (Fig. 11). The data in Figs. 10 and 11 demonstrate that it is possible by varying pad geometry, material, or contact pressure to generate
a range of values for frictional force and relative slip range at any particular value of applied
stress amplitude. This method of controlling frictional force and slip range can be used to study
their influence on the initiation of fretting cracks [15].
It is relatively easy to follow changes in frictional force and crack development in 2014A
aluminum (Fig. 9), where large fatigue strength reductions due to fretting are associated with
small slip amplitudes and little wear. Behavior is much more complex with high rates of fretting wear found at large slip amplitudes when 3.5NiCrMoV pads are fretted against 18Mn18Cr austenitic stainless steel. A comparison of the test machine load cell and pad strain gage
outputs (Fig. 12) show that the latter is distorted due to the large amounts of fretting debris.
The sudden large drops in pad strain are probably associated with surface delamination effects.
Metallography and Fractography

In the present experiments, fretting between each specimen and two bridge type pads results
in four rectangular fretting scars. With small slip ranges (~ --< 10 #m), fretting scars were patchy
with little fretting wear. At higher slip ranges (6 >_ 20 urn), considerable wear is encountered
162
Ul:-
CONTACT
RE$SURE
1 ~
3C--
/
2.-. - -
2s
9/
10-
25
50
MATERIAL
30MP. ! 127mm
1CrMo
[]
30 MPa
254 mm
1CrMo
30 MPm 50.8 mm
ICrMo
300 MPa
254 mm
1CrMo
30 N~'=
264 mm
2014A
/o"
_/
.I n ~
'~
o/~ ~/'
J ~ x/
~'"~
"~
..0.--'3'
-O''"
75
100
125
150
APPLIED STRESS A M P L I T U D E , MPa
175
200
225
FIG. l l - - R e l a t i o n s h i p between slip range and applied stress.
600 -
550
Ft (N)
500
_.
450
CYCLESxI0-s
PARTOFM
AXJMU@~S
R
'IA
N
I OU_TPUTFOR0 MEANSTRESSTEST,
225MPaALT.STRESS
: COEFFICIENTOF FRICTION
I
LOADCELLOUTPUT
PADSTRAN
I OUTPUTAT
A
PADSTRAN
I OUTPUTAT
B
I
I
i
I
I
i
I
i
CORRESPONDINGCYCLICSTRAINOUTPUT
FIG. 12--Frictional force variation associated with high rates of wear of 3.5NiCrMoV fretting pads in
contact with 18Mn-18Cr austenitic stainless steel.

163
over the entire contact area. For any specimen/pad combination examined, surface damage
from wear is greater in the softer material. If metal transfer occurs from one component in the
fretting assembly to the other, then it is always from the softer to the harder material. Examples
of wear scars are shown in Figs. 6 and 7.
A longitudinal metallographic section through a scar reveals the shallow angle cracks up to
500 vm deep which are typical of fretting fatigue (Fig. 13). As the cracks develop, they
change direction to become transverse to the stress direction and this transition can be either
sharp or gradual. The "dominant" fatigue crack which leads to final failure tends to initiate at
the outer edge of the pad and grow beneath the pad foot.
Fractographic examination was made of "broken-open" small fretting defects produced
during interrupted fatigue tests carried out at stresses below the fretting fatigue limit [15].
Breaking open the fatigue specimen can be achieved by either cooling in liquid nitrogen, in
which case the tip of the fretting defect is marked by a change in fracture mode (Fig. 14a) or
by removing the fretting pads and continuing the test at an elevated stress. The crack tip then
coincides with an abrupt change in the crack angle (Fig. 14b).
FIG. 13--Longitudinal section through a .fretting scar showing shallow angle jatigue cracks in
3,5NiCrMo V rotor steel in contact with 1CrMo pads at 30 MPa contact pressure and zero applied mean
stress.
FIG. 14--Small fretting defects (tips indicated by arrows) in 3.5NiCrMo V rotor steel broken open by (a)
brittle fracture and (b) pad removal and continuation o f test at elevated stress.
164
Fretting Fatigue Analysis and the S - N Approach

The conventional method of establishing the important variables which can affect fretting
fatigue has been to generate S-N curves with and without fretting, allowing fretting fatigue
strength reduction factors to be established. Such information is given in Table 2 for materials
used in turbo-generators. This method is particularly important for ranking material fretting
fatigue performance and gives an indication of material fretting combinations which should
be avoided. The roles of contact pressure, frictional force, and range of slip between the contacting surfaces have received much attention [1-7,12-18]. Nishioka and Hirakawa [17]
developed the following relationship to predict fretting fatigue strength ~I-,~:
where c~w,is the fatigue strength without fretting, u is the coefficient of friction, Po is the contact
pressure, S is the slip amplitude, and K is a constant where a cylindrical pad is fretted against
a plain fatigue specimen. However, care must be exercised in applying fretting fatigue strength
obtained from a simple experimental fretting geometry to a service geometry which is much
more complex. The input parameters in Eq 3 are invariably difficult to assess in real machines
and structures. Furthermore, various researchers [6,9,12,13, ]5,17,19,20] have concluded that
it is the surface stresses resulting from frictional forces together with the bulk stress which
determine fretting fatigue behavior. The specific geometry of the contact and the associated
stress concentration factor will therefore be of vital importance.
The S-N approach can also be used to assess the use of palliatives to combat fretting fatigue.
Either solid lubricant surface coatings or glass bead peening was used [21] in attempts to
140 1 x" AS MACHINED

MACHINED +MoS2
A
"
+PTFE
<> GLASS BEAD

i 20+
PEENED
PEENED +PTFE
WITHOUT FRETTING
+PTFE
.LENED
6o
C2~
t---
PEENED--'-~.~
o._
<
40
AS MACHINEF'~
+MoS2 OR PTFE
"'~"-"~""~'~'~
x
........
"x~.....
X-I.--.
20"
AS M A C H I N E D
"~I'-'~'-~l
@
i,,
Ib @
........
107
0e
CYCLES TO FAILURE
FIG. 15--Effoct of pa[fiative treatments on the fretting fatigue S-N behavior o.[2014 A aluminum.
165
improve the fretting fatigue performance of 2014A aluminum in contact with 3.5NiCrMoV
steel (Fig. 15). These palliative treatments fall into two categories: (1) those which improve the
near surface fatigue properties by introducing compressive residual stresses, e.g., shot or glass
bead peening; and (2) those which reduce friction at the fretting interface such as the solid
lubricants MoS2 or PTFE (Fig. 16). In the presence of surface coatings, the relative slip range
6 increases under conditions of low friction (Fig. 17). Typical fretting scars produced on 2014A
aluminum surfaces with various surface treatments are shown in Figs. 18a to 18d, Little surface wear is apparent on machined or peened surfaces and dark areas of fretting damage are
confined to patches. The scars on the surface coated specimens show considerable damage to
the coating layer with underlying metal exposed in places. Fretting damage and fretting fatigue
crack initiation occurs in these exposed regions. The most effective single palliative treatment
to combat fretting fatigue in the 2014A aluminum is glass bead peening which introduces near
surface compressive residual stresses and inhibits the growth of small fretting cracks. It should
be noted that large relative slip ranges and high rates of wear could negate the benefits of peening by removing the near surface layer containing the compressive residual stresses.
Fretting Fatigue Analysis: The Fracture Mechanics Approach

The S - N tests and associated metallography indicate that small fretting cracks, typically up
to 0.5 mm deep, initiate rapidly at only a small fraction (usually less than about 10%) of fatigue
2000
9
x
MACHINED OR PEENED
MoSz COATED
PTFE COATED
Z
C:D
(_)
z1500
od
L_
CD
*--~=0 9 3 ~-
LO
Z
-~2
Od
Ld
l-Z
L~
c:~ 1000
LL.
L~
_d
*-g=0.2
LL
LLJ
Z
~--
Ld
LL 500"
i~--~=0. I L~
(29
LLI
0 ~
I0
20
30
40
50
50
STRESS AMPLITUDE, +/-MPa

FIG. 16--Ejs]~'ctof pallia6ves on the plaWau value qlyrictionaljbrce as a./imclion ~#applied s/ress.
166

2O
I MACHINED OR PEENED
x MoS~ COATED
PTFE COATED
15J
a7
Z
c3_.10
..-d
F-.-d
t~ 5"
cv"
I0
20
30
40
50
60
STRESS AMPLITUDE, +/-MPa

FIG. 17--Relationship between relative slip range and applied stress after palliative treatment,
life, In addition, non-propagating cracks can occur at stresses considerably below the fretting
fatigue limit [15]. Hence, fatigue life is likely to be largely determined by the conditions for
the continued propagation of small fretting cracks. Models based on fracture mechanics have
been developed to describe the conditions for growth of such defects [ 7-12].
The stress intensity factor at the tip of a crack growing beneath a fretting pad will arise from
both body stresses and from the tangential and vertical forces due to the fretting pads. The
composite applied stress intensity factor can be found by various methods: (a) finite element
analysis which is often necessary for the complex assemblies found in practice and (b) by using
opening (and shear mode in more sophisticated analyses) stress intensity factors arising from
tangential and normal forces at the fretting position as computed by Rooke and Jones [22].
This latter method requires the measurement of frictional forces (F, by strain gaging the underside of the fretting pads as already described) as a function of semi range of applied alternating
stress and strain for each pad span and contact load. Following the method of Edwards,
Ryman, and Cook [ 7], the alternating stress intensity factor AKAppis given by adding the components due to alternating body stresses ~a and frictional force F,:
AKApp = Ya:~/~a + FtKtp -- Y ~
(4)
where Y is a compliance function, a is the crack length, and A is the specimen cross-sectional
Copyright
by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
area.
167
FIG. 18--Fretting scars" on specimen surface o[2014 A aluminum. (a) As-machined. (b) MoS2 coated.
(c) PTFE coated. (d) Glass bead peened.
The mean stress intensity factor Km arising from mean body stress cry,and normal force F,
is given by
g m = ] / ' f f m ~ ~t- Fngnp
(5)
The stress intensity factor components due to frictional (F0 and normal (Fn) loads (per unit
load) are Ktp and Knp respectively. The terms K,p and Knp depend upon the distribution of
stresses beneath the fretting pad foot. The measured frictional forces Ft (as functions of specimen stress and strain) are required in order to scale Ktp and K,,p as indicated in Eqs 4 and 5
(see Nix and Lindley [9] for details). In their methodology, Edwards, Ryman, and Cook [7]
168

20
15
with fn~tin9
10
~/"
S~
~" ~ ~ c a ~ e f e c t
sizetor
growth without fretting
~Z~KA
frettin9
no f~etting
/
o
0
I
0.2
l
I
0.4
0.6
Crackdepth, mm
I
0.8
I
1
FIG. 19--Appliedstress intensityfilctor AKAand threshold AKo (with and withoutj?ening) as afimction
o/crack length.
integrated Eq 4 in order to obtain an S-Ncurve for fretting fatigue. By contrast, in our experiments, we compare applied AKAppwith the experimentally determined threshold AK0 at the
appropriate value of stress ratio R. When /XKApp< AK0, crack arrest is predicted. If AKApp>
AK0, then sustained growth of a fretting crack will occur. In particular, the applied stress intensity factor/XKApp for a small crack will be increased under fretting conditions, possibly promoting continued growth, whereas the crack would remain dormant in a non-fretting situation (Fig. 19). The fracture mechanics model needs development in that Mode II
displacements are presently neglected. Nevertheless, fracture mechanics models allow treatment of fretting fatigue on a more quantitative basis than was previously possible.
Conclusion
Laboratory practices for carrying out fretting fatigue tests have been described. The important variables which affect fretting fatigue can be established by generating S-N curves with
and without fretting. This approach gives fretting fatigue reduction factors, identifies material
fretting combinations which should be avoided, and can be used to assess the effectiveness of
palliatives to combat fretting fatigue. Since small fretting cracks are introduced at an early
stage in fatigue life (usually less than about 10%), fracture mechanics and threshold concepts
can be used to evaluate their subsequent growth behavior.
Acknowledgments
The work was carried out at National Power Technoiogy and Environmental Centre and is
published by permission of National Power.
References
[ 1] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, 1972.
[2] Nishioka, K. and Hirakawa, K., "Fundamental Investigation of Fretting Fatigue: Part 2--Fretting
Fatigue Testing Machines and Some Results," Bulletin of Japan Society of Mechanical Engineers,
Vol. 12, No. 50, 1969, pp. 180-187.
[3] Field, J. E. and Waters, D. M., "Fretting Fatigue Strength of En26 Steel: Effects of Mean Stress, Slip
and Clamping Conditions," Report 275, National Engineering Laboratory, 1967.
169
[4] Endo, K., Goto, H., and Nakamura, T., "Fretting Fatigue Strength of Several Material Combinations," Bullelin q]Japan Society of Mechanical Engineers, Vol. 17, No. 92, i973.
[5] Fenner, A. J. and Field, J. E.. "Fatigue Under Fretting Conditions," Revue Metallurgique, Vol. 55,
1958, pp. 475-478.
[6] Endo, K. and Goto, H., "Initiation and Propagation of Fretting Fatigue Cracks," Wear, Vol. 38,
1976, pp. 311-324.
[7] Edwards, P. R., Rymam R. J., and Cook, R., "Fracture Mechanics Prediction of Fretting Fatigue
Under Constant Amplitude Loading," Report TR 77056, Royal Aircraft Establishment, 1977.
[8] King, R. N. and Lindley, T. C., "Fretting Fatigue in a 3.SNiCrMoV Rotor Steel," Report RD/L/
N75/80, Central Electricity Research Laboratories, 1980.
[ 9] Nix, K. J. and Lindley, T. C., "The Application of Fracture Mechanics to Fretting Fatigue," Fatigue
of Engineering Materials and Structures, Vol. 8, No. 2, 1985, pp. 143-160.
[10] Tanaka, K., Mutoh, Y., and Sakoda, S., "Effect of Contact Materials on Fretting Fatigue in a Spring
Steel," Transactions of Japan Society c~fMechanical Engineers, Vol. A51-464, 1985, pp. 1200-1207.
[I1] Mutoh, Y., Tanaka, K., and Kondoh, M., "Fretting Fatigue in $45C Steel Under Two-Step Block
Loading," Transactions (~fJapan Society (~fMechanical Engineers, Vol. 52-478. 1986, pp. 14771483.
[12] Hills, D. A., Nowell, D., and O'Connor, J. J., "On the Mechanics of Fretting Wear," ~2~ar, Vol. 125,
1988, pp. 129-146.
[I3] Spink, G. M., "Fretting Fatigue ofa 2.5NiCrMoV Low Pressure Turbine Shaft Steel: The Effect of
Different Contact Pad Materials and of Variable Slip Amplitude," Wear, Vol. 136, 1990, pp. 281297.
[14] Edwards, P. R. and Cook, R., "Frictional Force Measurements on Fretted Specimens Under.Constant Amplitude Loading," Report TR 78019, Royal Aircraft Establishment, 1978.
[15] Nix, K, J. and Lindley T. C., "The Influence of Relative Slip and Contact Materials on the Fretting
Fatigue of 3.5NiCrMoV Rotor Steel," Wear, Vol. 125, 1988, pp. 147-162.
[16] Endo, K., Goto, H., and Fukunaga, T., "'Frictional Force in Fretting Fatigue," Bulletin of Japan
Society of Mechanical Engineers, Vol. 17, 1974, pp. 647-654.
[17] Nishioka, K. and Hirakawa, K., "Fundamentallnvestigation of Fretting Fatigue: Part 5--The Effect
of Relative Slip," Bulletin ofilapan Society o/Mechanica/Engineers', Vol. 12, No. 52, 1969, pp. 692697.
[18] Nishioka, K. and Hirawaka, K., "Fundamental Investigation of Fretting Fatigue: Part 6--Effect of
Contact Pressure and Hardness of Materials," Bulletin of Japan Society of Mechanical Engineers',
Vol. 15, No. 80, 1972, pp. 135-144.
[19] Wharton, M. M., Waterhouse, R. B., Hirakawa, H., and Nishioka, K., "The Effect of Different Contact Materials on the Fretting Fatigue Strength of an Aluminium Alloy," Wear, Vol. 26, 1973, pp.
253-260.
[20] Chivers, T. C. and Gordelier, S. C., "Fretting Fatigue and Contact Conditions: A Rational Explanation of Palliative Behaviour," Proceedings of the Institution of Mechanical Engineers', Vol. 199,
No. 4, 1985, pp. 325-337.
[21 ] Nix, K. J. and Lindley, T. C., "Palliatives to Combat Fretting Fatigue in 2014A Aluminum Alloy,"
Fatigue Prevention and Design. Engineering Materials Advisory Services Publishers, Warley, West
Midlands, U.K., 1986, pp. 343-352.
[22] Rooke, D. P. and Jones, D. A., "Stress Intensity Factors in Fretting Fatigue," Report 77181, Royal
Aircraft Establishment, t977.
C. R u i z , 1 Z. P. Wang, l a n d P. H. W e b b ~
Techniques for the Characterization of

Fretting Fatigue Damage
REFERENCE: Ruiz, C., Wang, Z. P., and Webb, P. H., "Techniques for the Characterization
of Fretting Fatigue Damage," Standardization of Fretting Fatigue Test Methods" and Equipment,
ASTM STP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing
ABSTRACT: The factors that determine fretting fatigue damage of a given material pair are relative slip between the two surfaces in contact, shear stress at the interface, and tensile stress that
forces the crack, once started, to grow. Three types of test are described: a biaxial loading test for
interlockingjoints, a uniaxial tension-compression test, and a three-point bending test. It is concluded that the latter is sufficient for most purposes besides having low cost and simplicity.
KEY WORDS: fretting fatigue, ceramics, biaxial loading, shear stress, tangential stress, interface
slip
It is universally acknowledged that the life of mechanical joints, embodying interlocking or
threaded components, depends on the contact conditions between the two parts of the joint
as much as on the actual fatigue strength of the materials involved. The basic fatigue strength
is assessed by means of plain specimen tests and may be presented in the form of S-N curves
or crack propagation rate against (AK), but it does not take into account the rapid initiation
process that results from the damage caused to the surfaces by rubbing them together during
service. The nature of the damage for a given material pair depends on the relative slip between
the two surfaces, which may range from several times the ruling dimension of the smaller of
the two surfaces in contact, as in the case of wear in the thrust pads of a bearing, to a few
micrometers, as in the fretting of a shaft with a shrunk-on wheel. The interlocking joint presents an interesting problem in that, as shown in Fig. 1, the amount of slip varies along the line
o f contact, indicating that in a small rigid body displacement of 20 um an even smaller elastic
slip is superimposed. The slip displacement data shown in the figure refers to a Moir~ interferometric study of a dovetail joint that models a typical turbine blade root [1 ]. Of course, for
damage to become severe, a resistance to sliding must also be present. Neither the slip nor the
interface shear, by themselves, cause the damage, but rather, the combination of the two. It
has been shown [2,3] that the severity of the damage characterized by the surface roughness
depends entirely on the product of the slip times the shear stress. This parameter is clearly
related to the energy dissipated while the two surfaces in contact rub against each other. Once
the surface is damaged, cracks start from the stress concentrations of the surface~ The crack
growth is governed by the tangential stress, leading to the combined fatigue-fretting damage
parameter (FFDP) of
F F D P = crrr6
(1)
Reader in Materials Engineering, Department of Engineering Science, Oxford University, Oxford OX

28BU, England.
170
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
RUIZ ET AL. ON FRETTING FATIGUE DAMAGE
171
65
L/" %.1
1"5
I!L
FIG. 1--Typical model of a dovetail joint and relative slip along the flanks under 4.5 kN load with a 6mm thickness.
where
~r = tangential stress (along the line of contact),
z = interface-shear stress, and
= relative slip.
The importance of getting these three components of the F F D P right in any test designed
to reproduce the conditions prevailing in a real joint has been emphasized in a series of exploratory tests [4], in which specimens under apparently the same stress levels behaved very differently depending on the relative slip. Equally, geometrically similar specimens differed
depending on the magnitude of the crack driving stress ~T. Three types of test used in our laboratory to obtain data for the design of interlocking joints subjected to cyclic loading are
described below.
Biaxial Tests
A 250 kN biaxial fatigue testing machine has been described in detail elsewhere [5] and is
shown schematically in Fig. 2. It consists of two actuators, one vertical, supported through a
spring and damper arrangement on a cruciform base, and one horizontal, suspended from a
rectangular frame through springs. The machine is balanced and tuned so that at a frequency
of 10 Hz loading takes place along two perpendicular axes only, with minimum bending, torsion, and shear applied to the specimens. Moving masses, i.e., piston, actuator rod, dynamometer, traverse beams and clamps, have to be kept as light and rigid as possible. In the Oxford
machine the beams are fabricated from mild steel plates to form box sections. Self-alignment
requires heavy and cumbersome clamping systems and is undesirable since it precludes
reversed load cycling. The whole machine was carefully assembled to ensure parallelism or
perpendicularity and centralization. The test procedure consists of gripping and loading the
172
250kN
r --,'~,
'j
FIG. 2--Diagram
of biaxial testing machine and test piece.
specimens very lightly, strain gauged or provided with mechanical extensometers and adjusting the alignment by tapping the specimen until the gauges or extensometers, register only pure
tension. The strains are monitored at room temperature or, for high temperature tests, only
for a few cycles prior to heating the specimen to the required test temperature in an electric
resistance furnace.
The value of biaxial testing when studying the behavior of dovetail joints has been demonstrated amply [1-6]. Not only is it possible to obtain data of direct application to design, but
also it provides fundamental information on the state of strain and on the relative displacements between the two parts of the joint that can be used for checking the accuracy of finite
element analysis. This can then be used with confidence to map out the stresses in the actual
blade-to-disk joint. However, the whole testing system which includes the machine itself, furnace, extensometry and Moir6 interferometer is too complicated and cumbersome for any but
the most highly specialized laboratories. For a general purpose test it is necessary to use a simpler alternative.
A detailed description of the Moir6 technique, as applied to the study of dovetail joints, is
found in [6]. In essence, an interference grating with 1200 lines per millimetre fixed to the
specimen under load is illuminated with two laser light beams. The interference pattern produced by the two first-order diffracted beams provides loci of quidisplacement points. One
increment in fringe order corresponds to a relative displacement of 1/2400 mm, i.e., the gross
sensitivity of the technique is approximately 40 ~m, although smaller displacements can be
measured by interpolation between fringes.
RUIZETAL.ONFRETTINGFATIGUEDAMAGE 173
Tension/Compression Tests
The familiar tension-compression test with fretting pads pressed against the specimen lends
itself to a simple modification that makes it possible to obtain information on three different
phenomena: (1) classical fretting, where the relative slip results only from the differential elastic deformation of the two surfaces in contact; (2) classical wear, where the damage is associated with a relative motion whose amplitude is larger than the characteristic dimensions of the
surfaces in contact, leaving the surfaces exposed over part of the cycle; and (3) combined fretting and wear, prevailing in interlocking joints.
Since the amplitude of the relative slip is the key variable, a testing technique has been
devised to cover all three possibilities. The specimen in Fig. 3 is fixed to the crosshead of the
testing machine through a standard grip. The loading stem linking the bottom grip to the table
of the testing machine is attached to a flange which is connected to the actual table through
Belleville conical springs. Varying the number of springs permits a change in the stiffness of
the connection within a wide range of values. Two friction pads are pressed against the specimen with a force, P, applied by two compressed air actuators acting on two arms hinged to
the top of a rigid frame. Stain gauges on the two beams measure the shear force between specimen and pads, and transducers measure the displacements at points A and B. Under slip con-
--
CrosS;head
'
Strain
gauges
_C 1
<o>
. . . .
~)J
Furnace
Loading
jacks
Ta~bte
A
(b)
1N*P
-----P
,z,z,~.z
I B
FIG. 3--Diagram of tension-compression test rig. (a) general arrangement, (b)jbrce diagram.
174
ditions, the vertical displacement of the center of the specimen is
(2)
where km is the combined stiffness of grip and loading stem, ks is the stiffness of the Belleville
spring joint and k is that of the specimen.
The displacement of the pad is
S
kc
FzP
kG
(3)
where k~ is the stiffness of the strain gauged beams. The amplitude of the relative slip, N/k~ uP/kc, can be adjusted for a given set of N, P by changing ks and maintaining kG unchanged.
This is done prior to the proper testing using a specimen that is discarded once the system has
been adjusted. Once set, the amplitude remained constant during the test. The technique permits a continuous monitoring of the pressure force P, the friction (shear) force Sand, consequently, of the friction coefficient ~. Typical results for a ceramic specimen pressed between
two Waspaloy pads are shown in Figs. 4 and 5 [ 7]. In these tests, the specimen was enclosed
in a furnace. The test temperature was 600~ The loading stems were water cooled and the
two beams were outside the furnace. The relative slip was _+0.5 m m at a frequency of 0.5 Hz
and a total number of cycles below 50 000. At such low frequencies the vibration response of
the system does not present any problems. At higher frequencies a dynamic analysis is needed,
but this does not offer any serious difficulties.
In this test, as in the biaxial fretting/wear fatigue test, the specimen is subjected to a cyclic
load at the same time as the surface damage is induced. The two processes of surface damage
and fatigue crack growth go together, while in many industrial applications they may be independent. An alternative to the tension/compression test consists of drawing the specimen back
and forth between the two pads after releasing the bottom loading stem from the table of the
testing machine. For brittle materials the residual tensile strength after a specified number of
cycles is a measure of the damage suffered by the specimen. The testing procedure is simple
and convenient, but it suffers from a serious drawback, namely that the nature and severity of
the damage depends on the magnitude of the stress ~ras shown, for the case of Si3N4 specimen,
in Table 1 [8]. The tabulated data was obtained from the bending test described below. Each
set consisted often specimens subjected to the same number of identical fretting/wear cycles.
The residual bending strength or modulus of rupture (MOR) drops consistently as the maintained stress a~ increases. The Weibull modulus, which is a measure of the scatter of results,
Friction force
.I0
(kgf)
-I0
l
I
I
2
I
3
Time(s)
FIG. 4-- Variation offriction force during a typical uniaxial fatigue test (Syalon against Waspaloy).
0.6
175
0.5
0.4
0.3
T T
TTTT,TT . . . . TT
0.2
O.l
0
10
102
103
104
105
Cycles
FIG. 5--Varmt~n of~wtion coefficwnt w#h/(~(Sya~n agamst ~hspaloy).
TABLE l--Eff~,ct of maintained stress on the residual strength
of Si3N4 specimens damaged by fretting~wear.
Average MOR
(MPa)
Weibull
Modulus
Maintained Stress
(MPa)
560
433
418
352
10
8.7
4.5
6.2
0
120
160
200
also drops at first before increasing slightly when the severity of the surface damage is the only
dominant cause of failure. The conclusion that must be drawn is that the maintained stress
drives cracks from the asperities introduced in the surface by fretting/wear. A meaningful test
must, therefore, be conducted at the correct stress level.
Bending Test
The three-point bending test offers the possibility of combining all the factors that contribute to the process of surface damage. It has been extensively used at Oxford University to study
structural ceramics. Beams are tested in a three-point bending arrangement as described in Fig.
6.
The main load, P, is applied though a loading pin midway between the two simple supports.
The pad is pressed against the top face with a force N and subjected to a reciprocating motion
with a given amplitude. The complete apparatus is illustrated in Fig. 7. It consists of an outer
stainless steel tube that supports the ends of the specimen and contains a piston connected to
the main loading pin. A smaller concentric cylinder and piston apply the load on the pad. The
whole assembly fits into a cylindrical furnace designed to operate at up to 900~ The reciprocating motion is imparted to the pad by an electric motor driving an adjustable eccentric at
40 Hz.
The test procedure consists of applying first the main bending load and the pad bearing load
once the test temperature of 6000C has been reached. The pad is then subjected to a pre-specified number of cycles. Once that number has been reached, it is released and the main load is
increased until fracture occurs. The load at fracture is then recorded.
176

%
zl-
1.2
__L2
so
FIG. 6--Diagram qf three-poml bending lest.
J
Z
Z
50
dia-
"~
compressed air
(rubbing pad load )
compressed air
(n',ain load )
cooling water
N
2o
dia.
adjus~Gble
o
r[L
q(I
"~--~S pecime~n
furnace
'"
FIG. 7--General arrangement diagram of bending test rig.
Conclusion
With the advances in computational techniques, it is now possible to know with great accuracy the true state of stress in a contact pair and the relative displacement between the two
surfaces in contact. The laboratories that provide design data can no longer be contented with
offering simple technological tests devised to grade qualitatively a number of materials or
material conditions. In response to the requirements of modern product engineering designs,
new tests have to be designed to evaluate the effect of those variables that have a fundamental
effect on the behavior of the product in service. In the case of fretting/wear fatigue, these are
combined in the form of the FFDP. In the tests described in this paper, these variables can be
controlled to reproduce the actual test conditions. The test ceases to be a material or quality
test. It becomes a designer's tool, while still capable of yielding information of scientific interest
for the study of the phenomenon of surface damage under a cyclic or static overall stress.
The full biaxial fatigue test is necessary when dealing with ductile materials, e.g., metals and
177
alloys. W h e n dealing with m o d e r n structural ceramics and other brittle materials, it is possible
to use one of the other two techniques described in this paper, since the residual static strength
test provides a better indication of the severity of damage than the residual fatigue strength or
fatigue life tests.
References
[1] Ruiz, C., Post, D. and Czarnek, R., "Moir6 Interferometric Study of Dovetail Joints," Journal of
Applied Mechanics, Vol. 52, 1985, p. 109-114.
[2] Ruiz, C., Boddington, P. H. B., and Chert, K. C., "An Investigation of Fatigue and Fretting in a Dovetail Joint," Exp. Mech., Vol. 24, 1984, p. 208-217.
[3] Ruiz, C. and Chen, K. C., "Life Assessment of Dovetail Joints Between Blades and Disks in AeroEngines," Proceedings, International Conference on Fatigue and Structures, I. Mech. E., London
1986.
[4] Boddington, P. H. B. and Ruiz, C., "A Biaxial Fatigue Test for Dovetail Joints," ASME International
Conference on Advances in Life Prediction Methods, D. A. Woodford and J. R. Whitehead, eds., New
York, 1983.
[5] He, M. J. and Ruiz, C., "Fatigue Life of Dovetail Joints: Verification of a Simple Biaxial Model,"
Exp. Mech. Vol. 29, 1989, p. 126-131.
[6] Ruiz, C., Webb, P., and Post, D., "Interferometric Measurements of Strains and Displacements in
Engine Components in Advanced Instrumentation for Aero Engine Components," Proceedings,
AGARD Conference No. 399, Philadelphia, May 1986.
[ 7] Wang, Z. P. and Ruiz, C., "Characterization of Contact Damage of Syalon in Contact with Waspaloy,
Wear, Vol. 140, 1990, pp. 107-118.
[8] Wang, Z. P. and Ruiz, C., "Characterization of Contact Damage of Si3N4in Contact with Waspaloy,"
British Ceramics Transactions Journal, Vol. 89, 1990, p. 12-16.
Gerhard Fischer, ~ Vatroslav Grubisic, ~ a n d Otto B u x b a u m
The Influence of Fretting Corrosion on Fatigue

Strength of Nodular Cast Iron and Steel under
Constant Amplitude and Load Spectrum
Tests
REFERENCE: Fischer, G., Grubisic, V., and Buxbaum, O., "The Influence of Fretting Corrosion on Fatigue Strength of Nodular Cast Iron and Steel under Constant Amplitude and Load
Spectrum Tests," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM
ABSTRACT: The fatigue life reduction caused by fretting corrosion was determined in com-
parative fatigue life tests for axially loaded flat specimens made of GGG 40 (cast nodular iron,
ASTM A 536-80: 60-40-12) and of St 34-2 (hot rolled structural steel, ASTM A 283-78: Grade
C). A special test rig was constructed for superimposing fretting corrosion onto constant amplitude and spectrum loading of the specimen. Various surface conditions and two levels of surface
pressure were investigated. Using the results of these tests and the parameters derived, comparative damage calculations were performed aiming at predicting fatigue life in the presence of
fretting corrosion. Means to avoid fretting corrosion were also explained.
KEY WORDS: fretting corrosion, constant amplitude tests, load spectrum tests, steel, nodular
cast iron, shot peening, fatigue life prediction
Introduction
Fretting corrosion (frictional oxidation, frictional rust, fretting fatigue) may occur under
high surface pressure between contacting metallic surfaces such as threaded fasteners, riveted
joints and press-fitted c o m p o n e n t s [1,2]. If the contacting surfaces in such connections are
subject to relative displacement as a result of time-varying loading, an increase of principal
stresses and, with time, local friction welding can occur. Initial cracks may then be indicated
at this point. In this case the resulting c o m p o n e n t ' s fatigue strength is decidedly lowered compared to that without the presence of fretting corrosion. Therefore, the influence of fretting
corrosion must be considered as an additional criterion for the design of components. This is
especially i m p o r t a n t if safe as well as economical structures are to be designed, i.e., products
having a low weight a n d yet fulfilling the required service life.
The following m a y serve as a n example of a c o m p o n e n t failure that was caused by fretting
corrosion. The n o d u l a r iron wheel hub of a truck (Fig. 1) failed because its outer bearing race
of the i n n e r bearing began to move circumferentially as a result of the repeated elastic deform a t i o n that occurred due to the wheel's rotation under service loading. The relative motion
between h u b and bearing race resulted in a catastrophic failure. The crack origin was found at
the contact zone between bearing race and hub [8].
Senior Scientist, Head of Department, and Director of Institute, respectively, Fraunhofer-lnstitut fiir
Betriebsfestigkeit (LBF), Darmstadt, Germany.
178
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
FISCHER ET AL. ON NODULAR CAST IRON AND STEEL
179
FIG. 1--Fracture surface ~f a truck rear wheel hub that resulted from,/retting corrosion in the area of
the outer bearing race seat. (a) Overall view. (b) Fracture surface at the bearing seat.
Test Program
It was decided to carry out this research program in form of a series of comparative tests
including constant amplitude and load spectrum tests under otherwise identical conditions.
This decision was based upon previous investigations in the field of fretting corrosion (with
the exceptions of [5,9,10]) which were limited to sinusoidal constant amplitude tests [3,4,6,
7]. However, the transfer of constant amplitude test results to spectrum loading has not been
part of past investigations. The comparative test results allowed us to derive the influence of
the load sequence and, in addition, to derive possibilities for theoretical life prediction
calculations.
The spectrum load tests were carried out using a random sequence of peak with a cumulative frequency distribution corresponding to that of a Gaussian process [11] (Fig. 2b). This
"amplitude" distribution was chosen because many machine components are exposed to this
or a similar type of loading. All tests were carried out in a servo-hydraulic closed-loop testing
machine with pulsating load, but constant mean loading.
The unnotched flat specimens (Fig. 2a) were axially loaded. The nodular cast iron specimens were 8 m m thick and were tested in surface ground and in shot-peened condition, while
the 10 m m thick steel specimens were tested in surface ground condition and with rolling scale.
A special test setup was designed for the purpose of creating and maintaining uniform fretting corrosion conditions throughout the tests. In this test setup two opposing steel friction
pads are pressed hydraulically against the test specimen (Fig. 2c). A control device ensures
constancy of the compressive force on the friction pads; thus, nominally constant surface presCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
180
So, z~.:_~--,,--,:-',:.
I..
,
_(So )2 inHo
,u~ I~.,....,..,..,~jH-Ho.;'-~o
' '
<
R=O
a) Shape of the specimen
b) Loading pattern for the load

spectrum test
b.
Connecting bracket
ut
"
Friction pads
Specimen
Hydraulic piston
c) Test rig for the creation of fretting corrosion

FIG. 2 - - Test details.
181
sures, regardless of the amount of abrasion that occurs at the contact areas, is obtained. The
test described above may last up to 650 h. The setup allows the simultaneous application of
tensile forces acting on the specimen and compressive forces at the friction pads. This results
in a defined distribution of the forces between specimen, friction pad, and friction-pad holder
so that the relative displacements (elastic slip), typical for components that are subject to fretting corrosion, will be achieved. The magnitude of these relative displacements will depend on
loading as well as on the magnitude of surface pressure and the rigidity of the friction-pad
holder. The levels of surface pressures selected for this test program were 100 MPa and 50
MPa; the values correspond to the nominal measured surface pressure between the wheel nuts
and the axle-to-wheel hub joint.
The following were the test conditions:
9
9
9
9
Material St 34-2 and G G G 40.

Constant amplitude and spectrum loading at various load levels.
Surface pressures of 100 MPa and 50 MPa.
Different surface conditions of the specimen, namely ground, shot peened and with rolling scale.
The material properties were:

9 G G G 4 0 : Su
=
(annealed) Sy~02) =
e
=
9 St 34-2:
Su
=
Sy~o.2) =
e
=
512MPa
332 MPa
22%
369 MPa
265 MPa
42%
Results of Fatigue Tests

The test results are shown in Fig. 3 in the form of S-N curves resulting from constant amplitude tests, and of fatigue-life curves resulting from load spectrum tests, for a probability of
survival P~ = 50% and a fracture of the specimens as failure criterion. The insights gained from
these results are described below.
Influence of Load Sequence

Under constant amplitude loading and with fretting corrosion at a surface pressure of 100
MPa the material G G G 40 shows decreasing fatigue strength with increasing number of load
cycles (Fig. 3a). Thus, the slope of the S-N curve increases and the point of inflection of the
S-N curve at the endurance limit shifts from 2 )< 106 to approximately 107 cycles; the fatigue
strength is reduced by approximately 35%.
Contrary to the above, the fatigue life in load spectrum tests with fretting corrosion decreases
almost constantly at least within the investigated range of load cycles (Fig. 3b). Here the fatigue
life reduction is less compared with the constant amplitude test results. If the ratio of the
fatigue strength with and without fretting corrosion is plotted versus the respective number of
load cycles. The loss of fatigue strength, in the case of constant amplitude loading versus the
case of spectrum loads at a surface pressure of 100 MPa, becomes even clearer (Fig. 4a). This
relative fatigue strength reduction amounts to approximately 30% under tension-compression-loading, which corresponds to a calculated seven-fold increase in fatigue life for the load
spectrum test. These results may be explained by the relative elastic slip between friction pads
182
FRETTING FATIGUE TEST M E T H O D S A N D E Q U I P M E N T
c5
~2
oo
~o
o o
n ~_
-~ ~
?: t'
ii
-or--'--,
I,/1
"0
~'u
,,
O.~
9 ."
1/
O- r" O.
~.) C
/ =i
d~
LL
(]~
>
it
"ID
i~]ly_o.
//4
7:
L.
CL
._,.2
,,if ~P z,p l -
o~
o 121
13. n
ii"
u_z
I' % '/
,,,/./
71'
{1'
CD
CD
- -
! L
oi
Itli
"~ E
.__ "~
Iz
o,, S.-~
~ ~
~ ~~
9-~! g ~
2; ~'s
ill
<
"
6u!pDo7
>i.o ed,'E I
z ~s
-o
o
~
0
11
rl
CI
a._
d
Z
q~
ul
o.
/
o
L~
r
0
&
o
c~
~b
(uo!soJJo3 Bu!l~aJl Inoq~!t~ / 6u!lleJ; ql! ,~)

o!IDJ q 1 6 u e J l s en6!~o-I
~,~ 0
C
o~ .0
II
0"~
C
II
.--~-
CE
c
o
,r
-5
o
o
cz
ii
&_
~c o ,0
9 o
(1)
0
,&
i
"'1
.~ o ' ~
O"
(uo!soJJo3 Bu!~.~a.~. ~.noq~.!,~ / 6u!~eJ~ q~JM)

O!),DJ U ~ B U e J ~ S anS!~D:l
~
-5
183
184
and specimen, which is always the same in the case of constant amplitude loading and causes
more pronounced embedding of the contracting surface, thus causing higher principal stresses,
higher abrasion and wear than under variable amplitude loading.
Influence of Surface Pressure

Surface pressure has been found to be an important parameter in the design of mechanical
joints, it depends on the size of the contacting surface involved in the joint. If the average surface pressure of the material G G G 40 is reduced from 100 MPa to 50 MPa, a marked increase
of fatigue strength occurs in covstant amplitude testing (Fig. 3a). In contrast, for the case of
spectrum loading, the fatigue life curves for the surface pressures 100 MPa and 50 MPa are
hardly changed (Figs. 3b and 4b). The reason is also tbund in the degree of embedding of the
contacting surfaces which, in the case of constant amplitude loading, is affected more by surface pressure, as mentioned earlier. In the case of variable amplitude loading, the influence of
the embedding process is reduced which, in combination with the graphite nodules at the surface, produces hardly any difference between the fatigue life curves. Accordingly, the decrease
in fatigue strength for both surface pressures is similar.
Influence of Shot Peening

Shot peening is frequently used to increase the fatigue life of areas with high local stresses.
Tests with shot peened G G G 40 specimen without fretting corrosion at constant amplitude
resulted in almost no increase in fatigue strength (Fig. 3c), because the strength-enhancing
influence is limited to a small fraction of the cross-sectional area of the homogeneous stress
condition. With shot peening material steel bolts used, the resulting Almen-value was A2 =
0.40 m m (shot peening intensity). In the case of constant amplitude loading with fretting corrosion, the shot peened surface section, having been strengthened and containing compressive
pre-stresses, prevents a significant decrease in fatigue strength compared with non-shot peened
specimens. Its effect is, therefore, similar to reducing the surface pressure in a fretting corrosion
test under constant amplitudes. In conclusion, shot peening will increase the fretting corrosion
fatigue life. But in load spectrum tests the opposite is true. The fatigue life curves under fretting
corrosion show an almost identical decrease in fatigue strength, as under constant amplitudes
for both shot peened and non-shot peened specimens (Fig. 3d). This results from the comparatively rapid reduction of the shot peening induced compressive pre-stresses under high load
amplitudes during the spectrum test and from the elevated principal stresses in the contact
surfaces [12,13]. Nevertheless, in the case of load histories with relatively low peak loads and
high numbers of cycles (N > 10s) and under bending stresses, shot peening may still be
advantageous.
Influence of the Material

The results presented so far may not apply to the material St 34-2 because of the complicated
process and limited understanding of fretting corrosion. However, for the material St 34-2, in
constant amplitude testing with a surface pressure of 100 MPa, the reduction in fatigue life is
similar to G G G 40 (Fig. 3e); but when the surface pressure is reduced to 50 MPa only a negligible increase of fatigue strength can be observed. Possibly the lubricating effect of the graphite nodules in G G G 40 is missing in this case.
Again, in contrast to the material G G G 40, the steel St 34-2, when tested with fretting corrosion under spectrum loading, shows a greater reduction of fatigue life than would be anticipated from the results of the constant amplitude tests (Fig. 3f).
185
The results of the spectrum loading tests allow the basic conclusion that the "friction pairing" of cast nodular iron with steel will be more favorable than the pairing of steel with steel.
Interpretation of Test Results

Influence of Fretting Action on Stresses in Contact Area
Since the processes during the formation of fretting corrosion are complicated, the importance of individual parameters investigated cannot always be anticipated. An interpretation of
the test results, however, is possible. For understanding the strength reducing process, the load
transfer at the contact points, i.e., between specimen and friction pads, has been analyzed
[3].
Very important are the tangential forces Fr~.~in the contact area between specimen and friction pad. These forces are determined by the coefficient of friction and the compressive force
on the friction pads FN; both affect significantly the principal stress in the contact zone. Any
increase in the tangential force increases shear stress, principal stress, and wear.
The tangential force acting on the friction pads was derived from the distribution of forces
throughout the friction pads, which in turn were determined by means of strain gauges. If the
tangential force is plotted versus specimen stress, the following effects become evident (Fig. 5):
1. For the material GGG 40 and constant amplitude loading (Fig. 5a), the tangential forces
increase continuously with increasing specimen stress. That is the result of embedding
of the friction pads. This influence depends also on the applied surface pressure, which
explains the difference in fatigue strength at constant amplitude loading between 100
MPa and 50 MPa (Fig. 3a).
2. When spectrum loading is applied to the material GGG 40, the tangential force peaks at
a maximum tensile stress in the specimen of approximately 150 MPa and decreases if
PA = 50NPe
static
o
ConstQnt ~mpl test (CA) 9
Load spectrum tesi {LST)
t00MPo
o
9
9
~: 100MPo
Constant ampl test {CA)

Load spectrum lest (LST)
~ ~ f r c
30:
~: 50 N P o
9
o
%
%:
/u = - ~rc
-
,100
//
Z 2O
::
L~
.,0
5O
LST
I-- 10-
./
I
200
Tensile s t r e s s
100
100
S O tn the specEmen
a) M a t e r i a l GGG LO s u r f a c e
ground
CA/LST
/,.o ~ 10o
L~
~- tO-
el
100
300
k,tPa
]
200
Tenslle stress S o do the speclmen
b) M a t e r i a l St 3 4 - 2 with ro[hng s c a l e
FIG. 5--Determination q/tangentlaljbrce.

.-4300
MPa
186
that stress is exceeded. This peak, however, is only one-half of the tangential force
reached at constant amplitude loading. This behavior is caused by a reduced embedding
process under variable amplitude loading produced by the continuously varying relative
slip between friction pads and specimens, where the lubrication provided by the graphite
nodules supports that process. This explains the greater loss of fatigue strength at constant amplitude loading over spectrum loading (Fig. 4a).
. For the material St 34-2, on the other hand, no difference can be observed between the
results of the two loading sequences (Fig. 5b). Constant amplitude loading as well as spectrum loading show a significant increase in the tangential force with increasing specimen
stress, their dependence on surface pressure is also different.
The different behavior of the two materials is finally based on their differences in microstructure. The graphite nodules which reduce the tangential forces in G G G 40 are not present in St
34-2 (Fig. 6). In addition, the more inclined course of the crack in St 34-2 also suggests higher
stress in this material. All this contributes to the more pronounced decrease in fatigue life for
St 34-2 under spectrum loading with fretting corrosion when compard to GGG 40.
This does not mean, a priori, that G G G 40 will resist fretting corrosion sufficiently as has
been shown in the earlier mentioned example of a truck wheel hub that fractured at the bearing
FIG. 6--Fretting corrosion cracks during load spectrum tests. (a) GGG 40. (b) St 34-2.
187
seat (Fig. I). In this case the designer has to compensate for the lower coefficient of friction of
nodular iron by providing a tighter press-fit in order to achieve adequate service life.
Theoretical Life Prediction

A reliable prediction of service life by calculation can only be achieved with "relative damage calculations" [14,15] because of the complex and partially opposing influences that are
active during fretting corrosion. The relative damage calculation must be based on the mean
damage sum ratio for individual numbers of cycles that has been determined with and without
fretting corrosion. Calculations carried out for conditions of light fretting corrosion (material
G G G 40) resulted in life values that were close to experimental results, while the calculated
life-values for heavy fretting corrosions were always shorter than experimental results. As an
example of 108 load cycles, this being a typical life for vehicles and aircraft components, the
fatigue life under fretting corrosion was predicted from normal fatigue life data. The results
for both materials showed approximately half the life as determined in load spectrum tests. It
follows that a relative fatigue life estimation can be made if this tolerable deviation to the conservative side is taken into account.
A Method for Avoiding Fretting Corrosion

An example of a component exposed to fretting corrosion is the flange mounting of vehicle
wheels and hubs. Since load spectrum tests have shown that neither mechanical surface treatments (like shot peening) nor design related measures (reduction of surface pressure) have a
sufficiently significant effect, a new approach is suggested. At the bolted connection between
wheel and flange, a thin sheet of high quality steel was glued to both interfaces so that a layer
of glue would separate the metallic friction partners. In this arrangement the sheet absorbs
partly the relative displacement (Fig. 7). The steel sheet also prevents coining of the flange by
the highly pretensioned wheel nuts. This method has been developed for steel wheels, railroad
wheel sets, and automotive aluminum wheels. Tests carried out with constant amplitude as
well as with spectrum loading showed that the cross section of the flange can be reduced by
15% to 30%, depending on design, on loading conditions, and on the basic material (Patent
No. DE 3147820, US 4842 338).
Conclusions
Because of the different behavior of fretting corrosion under constant amplitude and under
spectrum loading on fatigue life and because of the difficulty in assessing the effects of design
parameters, material and post-treatments in the presence of fretting corrosion only design data
that has been determined from service loading [16] should be used. In summary, the following
general conclusions can be drawn from the presented results regarding the eflbct of fretting
corrosion:
1. For the material G G G 40, the damaging effect of fretting corrosion under constant
amplitude loading is larger than under spectrum loading.
2. The parameters that affect fretting corrosion such as surface pressure, shot peening, and
material characteristics show different effects under constant amplitude and spectrum
loading.
3. Shot peening and reduction of surface pressure under fretting corrosion does not result
in increased fatigue life under spectrum loading at the chosen test conditions.
4. The material G G G 40 showed more fatigue resistance under spectrum loading at the
188
Wheelnut with col(or

( M22 x 1,5 }
Wheelnut
:ement disc
T2
tiff
Jmmum wheel
S~ee~ fic~nge
Reinforcement dFsc
! glued )
//
Layer of adhesive
FIG. 7--A mdhod of reducing fretting corrosion by a glued on reinforcement disc.
chosen test conditions than the material St 34-2; one can, therefore, conclude that a
material-pairing of nodular iron and steel will be more favorable than steel and steel if
functional requirements have to be fulfilled.
5. Using the data gained in these tests, it is possible to conduct a relatively accurate fatigue
life estimation under conditions of fretting corrosion; also a successful means to avoid
fretting corrosion has been proposed.
References
[1] Bartel, A. A., "Passungsrost (Reibrost) - Krebsgeschwiirder Metallkonstruktion," Metal/, Vol. 29,
1975, H. 8., pp. 828-832
[2] Kloos, K. H., "Werkstoffauswahl und Oberfl~ichenbehandlung unter tribotechnischen Gesichspunkten," Z. Werkstofftechnik, Vol. 10, 1979, H. 12, pp. 456-466.
[3] Kreitner, L., "Die Auswirkung yon Reibkorrosion und von Reibdauerbeanspruchung auf die
Dauerhaltbarkeit zusammengesetzter Maschinenelemente," Diss. TH Darmstadt, 1976.
[4] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, 1972.
[5] Waterhouse, R. B., "Fretting Fatigue," Appl. Sci. Publ., London, 198 I.
[6] Horger, O. 1., "Fatigue of Large Shafts by Fretting Corrosion," Proceedings, Int. Conf. on Fatigue
of Metals, Inst. of Mech. Engrs., London, 1956, pp. 352-360.
[ 7] Tanaka, S. and Hatsumo, K.; u.a., "Fretting Corrosion and Fatigue Strength of Wheel-Seat of CarAxle," Jap. Nat. Railways Quart., Rep. 17, Railway Tech. Res. Inst, 1976, No. 1, pp. 14-28.
189
[8] Grubisic, V. and Fischer, G., "Gewichtsoptimierunggeschmiedeter Radnaben ftir Nutzfahrzeuge,"

Automobil-lndustrie, Vol. 29, 1984, H. 4, pp. 457-464.
[9] Ga3ner, E., "()ber den Eintlu3 der Reiboxydation aufdie Lebensdauer gekerbter Proben aus einer
AlCuMg2-Legierung," Aluminium, Vol. 39, 1963, No. 9, pp. 582-584.
[I0] Edwards, P. R. and Ryman, R. J., "Studies in Fretting Fatigue Under Variable Amplitude Loading
Conditions," Roy. Aircraft Establ. Tech., Rep. 75132, Dec. 1975.
[11] Fischer, R., Hiick, M., et al., "Eine dem station~iren Gau3-Proze3 verwandte Beanspruchung-ZeitFunktion f'tir Betriebsfestigkeitsversuche," Fortschritts-Berichte VDI-Z, 1977, Reihe 5, Nr. 30, VDIVerlag Diisseldorf.
[12] Broszeit, E., Zwirlein, O., Adelmann, J., "Werkstoffanstrengung im Hertz'schen Kontakt--Einflu/~
von Reibung und Eigenspannungen," Z. Werkstofftechnik, Vol. 13, 1982, No. 12, pp. 423-429.
[13] Broszeit, E., Kloos, K. H., SchweighiSfer, B. O., "Schwingverschlei3priJfung im Reibkufenversuch__
Zweidimensionale Spannungsanalyse und Reibwegamplituden" Z. Werkstofftechnik, Vol. 16,
1985, H. 6, pp. 187-193.
[14] Buch, A.; Lowak, H.; Schiitz, D., "Vergleich der Ergebnisse yon Betriebsfestigkeitsnachweisversuchen mit Hilfe der Relativ-Miner-Regel," Z. Werkstofftechnik, Vol. 14, 1983, H. 6, pp. 207-219.
[15] Buxbaum, O., "Betriebsfestigkeit--Sichere und wirtschaftliche Bemessung schwingbruchgef~rdeter Bauteile," Verlag Stahleisen, Diisseldorf 1986.
[16] Fischer, G., "Zum Einflu3 der Reibkorrosion und das Festigkeitsverhalten von Stahl und Stahlgu3
unter sinusf6rmiger und zufallsartiger Belastung," Fraunhofer-lnstitutfiir Betriebsfestigkeit (LBF),
1987, FB-177.
Janusz Labedz ~
Adaptation of a Servohydraulic Testing

Machine to Investigate the Life of Machine
Components Operating under Fretting
Conditions
REFERENCE: Labedz, J., "Adaptation of a Servohydraulic Testing Machine to Investigate the
Life of Machine Components Operating under Fretting Conditions," Standardization of Fretting
Fatigue Test Methods and Equipment, ASTMSTP 1159, M. Helmi Attia and R. B. Waterhouse,
ABSTRACT: Fretting tests are of considerable practical importance to increase the reliability of
machinery and to reduce design costs. They are frequently conducted on expensive, special apparatus. A reduction of outlay on research work is possible through the use of a universal test rig
and commercially available equipment. In this paper a fretting test method is proposed using a
servohydraulic testing machine.
Tests on the effect of residual stresses on fretting wear of bearing steels were performed for the
purpose of verifying the proposed methodology. The test results are compared with data given
in other publications which validate the method.
KEY WORDS: fretting, bearing steel, servohydraulic testing machine, residual stresses
Fretting is a complex physical/chemical process and is a most agressive form of wear on

machinery parts. It arises from the contact of various materials with no lubricant or under
boundary lubrication conditions with average and high pressure per unit area and small relative displacements between contacting components. Typical fretting cases are press-fitted
items, riveted or bolted joints, rolling bearings, springs, and steel wire ropes [ 1,2]. Fretting tests
are of considerable practical importance to increase the reliability of machinery and reduce
design costs. They are frequently conducted on expensive special apparatus. A reduction of
outlay on research work is possible through the use of a universal test rig and commercially
available equipment.
Test Methodology
From among more than 50 parameters which affect fretting [3], the essential ones are:
9
9
9
9
9
Magnitude of the surface pressure distribution.

Material and surface layer condition of the machine elements.
Frequency and relative displacement of contacting parts.
Temperature in the contact zone.
Type of medium where the process occurs.
Adjunct Professor, Institute of Basic Problems of Mechanical Engineering, Academy of Mining and
Metallurgy, Cracow, Poland.
190
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
lntcrnational
www.astm.org
LABEDZ ON SERVOHYDRAULIC TESTING MACHINE
191
Tests to examine the processes occurring in the contact area require variations to be made
to the values of these quantities. This task is facilitated using a servohydraulic testing machine
and an additional loading rig. It enables most of the above cited parameters to be varied over
a wide range. To take full advantage of the test machine potential in the fretting test an additional rig is required which is clamped to the samples. The rig for testing plain samples is shown
in Fig. 1. The method of loading samples during testing is shown in Fig. 2. The force, P, is
exerted by a pneumatic clamp. The value of the force is measured by a strain gauge system
connected to a bridge amplifier and a digital voltmeter. This rig allows for simultaneous testing
of two samples with four loading points which considerably reduces the amount of testing. The
tests are continued until a predefined number of fretting pad displacement cycles is reached
(Fig. 2). Fretting-wear resistance of different materials can be evaluated by a comparison of
the depth of wear scars on the test samples.
f / ..--- j j
~mea'surinooo[nts
15
FIG. l--Fretting-wear testing rig used on servohydraulic testing machine: 1-body; 2-nut; 3-retaining
screw; 4-top mandrel; 5-strain gauge; 6-connection clip; 7-roller; 8-bottom mandrel; 9-pin ([ketting pad);
1O-clamping screw of sample; 11-clampingja w; 12-pneumatic clamp, 13-sample; 14-top clamp (stationary)
of testing machine; 15-bottom clamp (mobile) of testing machine.
192
samole
P -load force
P=const.
fretting pad
FIG. 2--Load scheme of test sample.
Tests on the effect of residual compressive stresses on the fretting wear of bearing steels were
performed to verify the proposed test methodology. Testing was carried out on six samples
(overall dimensions of samples: height H = 10 mm, depth D = 10 mm, length L = 60 mm)
which were made of the steel LH 15 (marking acc. Polish Standard PN-71/H-84041). The
chemical composition is as follows: C = 0.98%; Si -- 0.27%; Mn = 0.34%; Cr = 1.5%; P =
0.014%; S = 0.013%; Ni = 0.17%; Cu = 0.1%. These were heat treated to a hardness of 6061 Rockwell Hardness C-Scale (HRC). The engineering process for making the specimens
included the operations of milling, heat treatment, and grinding. The surface roughness met
class 8-9 (Ra = 0.5-0.3 urn). The samples were divided into three equal groups. Samples in
the first and second groups were subjected to an additional stress relieving process to produce
samples with different values of residual stresses. The stress relieving consisted of holding the
first group of samples in machine oil at a temperature of 423 K (20 K lower than the tempering
temperature of the heat treatment) during 6 h (for the first group of samples) or during 12 h
(for the second group of samples). Afterwards, measurements of residual stresses were made
using the X-ray method (sin 2 ~b). The results are presented in Table 1.
Tests of fretting wear were made under the following conditions:
9
9
9
9
9
9
9
amplitude of fretting--0.02 m m
normal load on mating p a r t s - - 5 0 N
four number measuring points (Fig. 1) on each sample
frequency of relative displacements-- 100 Hz
number of applied fretting cycles--1 l 0 6 cycles
elements assembled under dry friction conditions
fretting p a d - - b e a r i n g steel--radius 10 m m
The m a x i m u m depth of the scar was determined by measurements on a microscope for

surface roughness measurements. This was adopted as a measure of wear. The test results are
shown in Fig. 3.
TABLE 1--Results of residual stress measurements.
Tested Group of Samples
daN
Surface Resident Stress Value, mm 2
Group I
Group II
Group IIl
-25
0
-60
193
IO
7.8
~.3
~2
i
'
-6b "
-25
Residual compressive
+~gmax ]
~
~
daN/mm
z
stresses
/kgmax - limit of error of measuredavarage

value
Avarage value of maximumdepth cf the
wear SCeU
FIG. 3--Influence o/ residual compressive stresses on maximum depth of the wear scar of the test
sample.
Conclusions
The results obtained from the tests indicate less damage to the samples containing residual
compressive stresses compared with the samples containing no residual stresses. A probable
reason for this is the effect of residual stresses on the intensity of processes taking place in fretting and above all on abrasion, fatigue, and oxidation of material.
By testing the effect of residual compressive stresses on functional quality of machinery
components it was found that they may produce increased resistance to abrasive wear [4,5] as
well as boost surface fatigue strength [6]. The effect of residual stresses on oxidation processes
is similar. Residual compressive stresses can prevent stress corrosion [ 7,8].
Thus, in all the cases where residual compressive stress occurring in the material causes a
reduction in the intensity of basic phenomena accompanying the fretting, one can expect less
wear and, therefore, less wear scar depth. This correlation was found to exist in the tests
pertbrmed.
The test results obtained are consistent with data given in a number of publications [6,9,
10], which were obtained using different test methods. The results presented in Fig, 4 may be
an example. They were performed on samples in which two stress states were induced (Fig.
4a): 1) stresses arising from pure bending which were assumed to be residual stresses; and 2)
contact stresses, caused by pressure of a ball upon the tested objects with a periodically variable
force. Quantitative differences may be explained considering the complexity of the phenomCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
194

(b13 497
F s - load force - slnusoldal
a)
vaclable,
P - load m~oduclng bendlnq

which is equivalent to
5ample
the residual stresses,
~
tsIl
s
35
115
-q
RI,R 2- bearing reactllns
130
b)
gmax' pm
_}1.6
3O
23
16.2
7,71~]~1]
,
I I'
daN/mm2-?] -75 -~5 -[~5

Resldua. :ompresstve
daN/mm ~
Residual tensile
[~5
75 90
stresses
+Agmax/~//7~Avarage
-~gmax/~
/
~gmax
value of maximum depth of
the weac scar

- limit of error of measured avarage value
FIG. 4--1nfluence (#residual stresses on m a x i m u m depth ~?fthe wear scar o f a test sample. (a) Load
scheme o f the test sample. (b) Test results (nominal contact stresses m a x G = 225 daN~ram 2,.frequency q#
loading.# = 15.3 ttz, number c?fapplied load cycles N = 4.59 X 108) [ 10],
e n o n a n d the different test conditions used. It is concluded that the proposed methodology is
valid.
References
[1] Kocafida, S. and Szala, J., Podstawy oblicze6 zmr

PWN, Warszawa, 1985.
[2] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, 1972.
[3] Collins, J. A. and Marco, S. M., "The Effect of Stress Direction During Fretting on Subsequent
Fatigue Life," S T P 64, ASTM, Philadelphia, 1964, pp. 547-560.
[4] Kaczmarek, J. and Summer-Brason, P., "Wpt~w stanu naprr
na ~cieranie lu~nym ~cierniewem,"
Przegl~d Mech. No. 13, 1964, pp. 358-360.
195
[5] Kaczmarek, J., "Badania zale~no~ci zu~ycia ~ciernego niekt6rych gatunk6w stali od stanu napr~efl
wymuszonych si~ zewn~trzn~," Zeszyty Nauk. Pol. Krak. No. 9, 1965.
[6] Kosteckij, B. J., Soprotivlenie iznasivaniju detalej masin, Masgiz, Moskva, 1959.
[7] Bates, J. E., "Effect of stress on corrosion," Ind. Engng. Chem. Vol. 58, No. 2, 1965, pp. 18-29.
[8] Vijh, A. K., "The application of ice on the fretting corrosion of metals," Corrosion Science, Vol. 16,
No. 8, 1976.
[9] Lab~d~, J. and Skrzypifiski, A., "Experimental Research on the Influence of Uniaxial Residual
Stresses on Contact Fatigue Strength," Metal Treatments Against Wear, Corrosion, Fretting and
Fatigue, Pergamon Press, Oxford, 1988, pp. 59-66.
[10] Lab~d~, J., "Effect of Residual Stresses on Wear of the Surface Layer of Components Operating
Under Fretting Conditions," Metal Treatments Against Wear, Corrosion, Fretting and Fatigue, Pergamon Press, Oxford, 1988, pp. 87-98.
Environmental and Surface Conditions
Yoshiharu Mutoh, 1 Toyoichi Satoh, ~and Eiji Tsunoda 2
Improving Fretting Fatigue Strength at

Elevated Temperatures by Shot Peening in
Steam Turbine Steel
REFERENCE: Mutoh, Y., Satoh, T., and Tsunoda, E., "Improving Fretting Fatigue Strength
at Elevated Temperatures by Shot Peening in Steam Turbine Steel," Standardization of Fretting
Fatigue Test Methods and Equipment, ASTM STP 1159, M. Helmi Atria and R. B. Waterhouse,
ABSTRACT: Fretting fatigue tests at elevated temperature were carried out using shot-peened
specimens to improve the high temperature fretting fatigue strength of 12Cr-Mo-W-V steam turbine steel. Shot peening improved fretting fatigue strengths by a factor of 1.8 at an elevated temperature (773 K) as well as at room temperature. Residual compressive stresses first drop rapidly
by a factor of three and then level off. Residual compressive stresses of 200 MPa were measured
after a 30 000 h exposure at 773 K. Shot peening is thus shown to be effective over long periods
at 773 K.
KEY WORDS: fretting fatigue, fatigue at elevated temperature, shot peening, residual stress,
steam turbine steel
Introduction
Fretting fatigue has become a serious problem in steam turbines which operate at elevated
temperatures under increasingly severe conditions brought about by higher demands in power
and efficiency. The authors carried out high temperature fretting fatigue tests on two steam
turbine alloys (12Cr-Mo-W-V steel and Cr-Mo-V steel) and reported that the drop in fatigue
strength through fretting was significant at both elevated and room temperatures [ 1 ]. Improvement in fretting fatigue strength is required from a design point of view particularly at high
temperature.
Shot peening [2,3 ], coatings [4,5] and soft shims [6 ] have improved fretting fatigue strength
at room temperature but little work has been done at high temperature. The effect of shot
peening on high temperature fretting fatigue strength under representative working conditions
is discussed in this paper.
Two kinds of steam turbine steel (12Cr-Mo-W-V and I ICr-Mo-V-Nb) were used in this
study. 12Cr-Mo-W-V steel was used for the fretting fatigue specimen and 11Cr-Mo-V-Nb steel
for the contact pad. The chemical composition of these materials are shown in Table 1a. Heat
treatment conditions and mechanical properties at room and at elevated temperature (773 K)
are given in Table 1b. The strength levels of these materials were almost identical.
Professor and research associate, respectively, Department of Mechanical Engineering, Nagaoka University of Technology, Nagaoka-shi 940-21, Japan.
2 Toshiba Corporation, Tsurumi-ku, Yokohama 230, Japan.
199
Copyright
by ASTM Int'l
(all rightsASTM
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
www.astm.org
200
TABLE 1--Chemical compositions (a) heat treatments and mechanical properties (b) of the material
used.
(a)
Material
C Si Mn P
S Ni Cr Mo V W Co Sn A[ Ti Nb N
12Cr-Mo-W-V
0.23
),9
3.70
3.021
0.0010.67
,1.21 3.930.23:).953.020.01 0.010.01 . . . . . . . - '
Steel
11Cr-Mo-V-Nb
0.17 )~9. 3.66 0.019 0.007 0.35 11.4z~ 0.92 O.21 3.02 0.01 Tr Tr 0.01 0./,i D.O/,
Steel
(b)
1,4ateriat Temp Heat
Yield
Tensile
Elan- Reduction u
Vickers
( K ) treatment strength(~ys
MPo strength(jB
MPa gOtion$/,
% incarea% "nodulusE
GPa hordnes~Hv
12Cr-Mo-W-V R.T. 1325 OQ 852
923 AC
Steel
773 w AC 651
11Cr--Mo-V-Nb R.T, 1368 OQ
923 AC
Steel
773 898 AC
987
17.2
52.0
209
676
27.0
77.7
160
874
1000
18.9
54.8
206
"'"
700
22.5
69.9
......
3z,6
.....
342
R.T. :Room temperature

OQ :Oil quenching
AC :Air cooling
Specimen and pad dimensions are shown in Fig. 1. Two flats were machined at the specimen
and were peened to arc height for Almen A strip of 0.34 mm using 0.6 mm diameter steel shots.
The schematic illustration of the apparatus used for fretting fatigue tests at elevated temperatures is shown in Fig. 2a. A fretting fatigue specimen (1) was attached to a water-cooled rod
( 11 ) by using a divided ring (9) and collar (10). The water-cooled rod was settled in a hydraulic
grip of the fatigue test machine. A contact pad was attached to the grooved part of a jig (3) for
measuring the frictional force as shown in Fig. 2b. A pair of contact pads was pressed to the
specimen by using a proving ring (6). The other end of the jig was rigidly attached to the watercooled rod by using blocks (4 and 5). The relative slip between the specimen and the contact
pads was induced by the difference in deformation between the specimen and the jig. The contact pressure was applied using the output signal of strain gage glued to the proving ring. The
Fretting fatigue specimen

Contact
surface
Section AA
Pad
FIG. 1--Shapes and dimensions of the specimen and the contact pad.
MUTOH ET AL. ON SHOT PEENING IN STEAM TURBINE STEEL
Actuator
201
11 Cooling rod
12 Furnace
Contact pod
, Stro n g a u ~ e
(b)
F1G. 2--Schematic illustrations. (a)Freuingfatigue tests apparatus at elevated temperature. (b) FrictionalJbrce measurement jig.
frictional force between the specimen and the contact pad was measured using a strain gage
attached to the end part of the jig (3). The dimensions of the contact area were 2 m m in the
longitudinal direction (axial direction) and 4 m m in the transverse direction.
The test temperature of 773 K was selected to simulate the steam turbine service condition.
The distribution of temperature in the gage part of the specimen was in the range 774 to 768
K, and the variation of temperature during tests was controlled in the range of +_ 1 K. A servohydraulic fatigue test machine with a capacity of 49 kN was used. Fretting fatigue tests were
carried out under a load-control condition with a stress ratio ofR = - 1 and frequencies from
5 to 10 Hz. The contact pressure of contact pad to the specimen was controlled at a constant
value of 100 MPa during the tests.
In order to investigate the effectiveness of shot peening for long-term practical use, high temperature fretting fatigue tests using specimens shot-peened and subsequently exposed at elevated temperatures for long periods were also carried out. The exposed conditions are shown
in Table 2. Exposed temperatures of 773 K and 823 K were selected to simulate steam turbine
202
TABLE 2--Exposing conditions.

Expose Hold time Equivolenttime
temp.
(K)
at 773K
(h)
(h)
773
100
1000
100
1000
823
40
100
1000
2649
1000
3 0740
service conditions. In Table 2, the equivalent holding time at 773 K was calculated using the
Larson-Miller relation [ 7]:
P = T ( C + log t)
(1)
where
P
T
C
t
=
=
=
=
Larson-Miller parameter,
exposed temperature, K,
material constant (assumed C = 20 in this study [8]), and
holding time at the exposed temperature, h.
The Larson-Miller relation gives the equivalency between time and temperature under creep
condition. The correlation between deformations under creep and relaxation conditions is
well known. Then, the Larson-Miller parameter was applied to reduce the holding time at different exposed temperatures. According to the relation in Eq 1, the exposed condition at 823
K for 1000 h is almost equivalent to that at 773 K for 30 000 h.
The residual stresses induced by shot peening were measured by the X-ray diffraction
method. The distribution of residual stress in the depth direction of the specimen was also
measured by repeatedly removing the thin layer of the specimen surface by electropolishing.
Results
Fretting Fatigue Strengths

The S-N curves for plain fatigue and fretting fatigue of unpeened specimens are shown in
Fig. 3 [1]. The plain fatigue strength and the fretting fatigue strength were defined as the runout stress amplitude at 2 X 106 cycles. It was found from Fig. 3 that the effect of fretting on
fatigue life was significant at 773 K as well as at room temperature. Fretting reduced the fatigue
strengths to almost one third compared with the relevant plain fatigue strengths at both
temperatures.
The results of fretting fatigue tests of shot-peened specimens at both room and elevated temperatures are shown in Fig. 4. For reference, the approximate S-N curves for plain and fretting
fatigue of the unpeened specimens at both temperatures are also shown, The fretting fatigue
strengths of the shot-peened specimens at room temperature and 773 K were equal to or more
than 300 MPa and 200 MPa, respectively. The shot peening was found to improve the fretting
fatigue strength by a factor of around 1.8 at elevated temperature as well as at room
temperature.
The results of fretting fatigue tests at 773 K for the specimens shot-peened and subsequently
exposed at 773 K are also shown in Fig. 4. The fretting fatigue lives of these specimens were
1200
1000
I
I
1
I
I
i
12Cr-Mo-W-V Steel-11Cr-Mo-V-Nb Steel
--
Unpeened
800
=_. 6 0 0
400 - ~
"o
3
O ~
O--~-,~ -
- iI ~ ~ z ~ _ z ~ _
__-
(3.
E
m
ul
200
T;.Tm.P Plon Freeing
u3
773K
101
"%1
104
AA
9 I
--
~Ir-~--
105
106
Number of cycles to failure Nf
5x10 6
FIG. 3--S-N curvesfi)r plain and fretting fatigue of the unpeened specimen.
800
600
no 500
~E 400
300
.Q.....
(b
R=-I
"~..0
temp.time
~ 2001 'tempExposeHold
'Test
(K)
(K)
[(h)
AS s h o t . . o _
peened
1oo
773 1000 A
773
40 []
R.I
"~ I O O p
50t
823
104
Unpeened
Test teml~ Plain Fretting..
9
1000 9
R.T.
773K
100
105
Number of cycles to failure
106
Nf
FIG. 4--S-N curvesfor fretting fatigue ( f the peened specimen.

5x106
203
204
almost identical, regardless of the exposure conditions. Even after 1000 h at 823 K, which
corresponds to the exposing time of 30 000 h at 773 K, the fretting fatigue strength was as high
as 225 MPa. This value of 225 MPa means improvement of the fretting fatigue strength by a
factor of 1.8. Therefore, it is suggested that shot peening is effective to improve the fretting
fatigue strength, even for long-term practical use at elevated temperatures.
Friction CoeJficient
The relationships between the stress amplitude and the friction coefficient are shown in Fig.
5. In the figure, those for the unpeened specimens at room temperature and 773 K [1] are also
plotted. The friction coefficient u was defined as the ratio of the frictional force amplitude Fa
and the contact pressure P, u = FJP.
At room temperature, the friction coefficient of the shot-peened specimen increased with
an increase in stress amplitude, and attained a constant value of 0.8. At 773 K, the friction
coefficients were almost identical, regardless of the exposed conditions. These relationships for
shot-peened and unpeened specimens were almost identical at both room and elevated
temperatures.
The oxidation of the material used was not significant at the test temperatures, and the oxidized surface was removed by fretting action. The surface roughness for unpeened and peened
specimens were almost identical during and after the fretting fatigue test. In addition, the hardening of the peened surface was not significant. This particular surface behavior of the material
used seems to result in the unique relationship shown in Fig. 5, which is independent of the
test conditions.
Residual Stress Measurements

Examples of the compressive residual stress distribution in the depth direction are shown in
Fig. 6. For the as-shot-peened specimens, the compressive residual stress at the specimen sur-
1.0
Unpeened
-~
._u 0.6
X
@0
Xx ~
Test ExposelHo[d
tr~)p, temp.ltirne i
x@@
0.4
(K) l ( h ) '
R.T. As s h o t .
peenee
IOOI
773 1000
773
z,o
823 100
c
o
~U_
0.2
I
0
I
200
Stress
I
amp[itude
,
400
oA
[]
II
~ooo ~I,'
,
600
O'a (hlPa)
FIG. 5--Relalionship between stress amplitude and friction cocl~cient.

MUTOH ET AL. ON SHOTPEENINGIN STEAMTURBINESTEEL
205
_~I'r
-zoo
-400
o o
0
0
-600
00o
~ -800
cr
-I000
0
O As shot-peened
r 823K,1000h
I
01
I
02
[
03
[
04
I
Q5 0.6
Depth from surface(mm)

FIG. 6--Distributions of residual stress in the depth direction of the specimen.
face was almost 600 MPa, and it slightly increased with an increase in depth. At the depth of
about 150 urn, the compressive residual stress attained a maximum value. Beyond this point,
the compressive residual stress decreased with an increase in depth. On the other hand, for the
specimens exposed to elevated temperatures, the compressive residual stress increased with
depth. It attained a m a x i m u m at 200 to 300 urn. Then it decreased with an increase in depth.
Although the residual stress in the deeper region was not measured because of the difficulty in
the measurement, the residual stress in the central region of the specimen would be tensile
because the residual stress is a self-balanced force.
The relationship between the compressive residual stress and the equivalent holding time at
773 K is shown in Fig. 7. Both the residual stress on the specimen surface and the maximum
value of the residual stress in the depth direction are shown. The exposure time of 40 h at 823
K can be reduced to 1000h at 773 K according to Eq 1. As can be seen from Fig. 7, the compressive residual stress of the specimen exposed at 823 K for 40 h agreed well with that of the
specimen exposed to 773 K for 1000 h. Therefore, the reduction of exposing time, according
to Eq 1, seems to be reliable.
The compressive residual stress in the specimen surface induced by shot peening process,
which was initially about 600 MPa, was significantly decreased by exposure for 2 h at 773 K.
However, the subsequent relaxation of compressive residual stress with exposing time was not
significant as shown in Fig. 7. Even after exposing for 30 000 h at 773 K, a peak compressive
residual stress of 200 MPa remained.
Discussion
The fretting fatigue strength at 773 K increased by shot peening by a factor of around 1.8 in
the same manner as the case at room temperature. Therefore, it is suggested that shot peening
is effective to improve the fretting fatigue strength at elevated temperatures as well as at room
temperature. It is considered that the factors, which are induced by shot peening and influenced the fretting fatigue strength, are work hardening in the surface layer, roughening of the
surface and compressive residual stress.
206
'
A~~
'mm-------t
i~i
:E -200
v
(h
-400
ffl
9
A
9
~._]nitial residual
stress lever
_ -600
O
rl
.-g_
u~
D:
2 h (7?3K)
t 0 0 h (773K)
1000 h (773K)
1 0 0 0 h ( 8 2 3 K /-*Oh)
2 6 4 9 h ( 8 2 3 K 100h)
@ 30740 h ( 823K lO00h)
-800
slashed mark : peak value
-100C
I I
101
i [ a
102
Equivalent
hold
100
I i
103
I I
104
I
105
time at 773K (h)
FIG. 7--Relationship between the equivalent time at 773 K and the residual stress.
Work Hardening in Peened Surface Layer
The hardness in the depth direction of the specimen was measured to investigate the work
hardening in the peened surface layer. The measurement of hardness was conducted using a
micro Vickers hardness test machine under the load of 1.96 kN and the holding time of 15 s.
Since the resultant size of indentation was approximately 34 ~m, the nearest measuring point
to the surface was almost 25 to 30 ~m from the surface. The results are shown in Fig. 8. As can
be seen from the figure, significant work hardening in the surface layer was not found for the
material used. The microstructure of the peened specimen was also observed together with the
hardness measurements. Figure 9 shows microstructures in the subsurface region (a) and the
center region (b) of the specimen. Although the specimen surface was roughened by shot peen-
>4001oO
t]
-1-
i 3} ~
u~
ul
9~
o
~t- 350F**g @@
,-
9o
O As shot-peened
~U
8 2 3 K , 1000h
300
Q1
0.2
Distance
0.3
0.4
from
1.8
1.9
2.0
surface(ram)
FIG. 8--Distribution of hardness in the depth direction of the specimen.

207
FIG. 9--Microstructures of the peened specimen. (a) The subsurface region. (b) The center region.
ing, the deformation of the subsurface region was found to be inconsiderable. Therefore, the
plastic deformation region seems to be restricted in the very narrow thin surface layer for the
material used.
Consequently, the work hardening will not be the influencing factor for improving the fretting fatigue strength at elevated temperatures.
Roughening of Peened Surface

Examples of the fretted surface profile measured for peened and unpeened specimens are
shown in Fig. 10. SEM observations of the fretted surfaces are also shown. The maximum
roughness Rm,, of the initial surface for the unpeened specimen was less than 1 urn. On the
other hand, it was significantly high and equal to about 20 um for the peened specimen.
During the fretting fatigue tests, the surface roughness of the unpeened specimen was
increased, while that of the peened specimen was decreased by cyclic wear. As a result, the
fretted surface morphology of both unpeened and peened specimens became almost identical.
FIG. IO--Specimen surJi~ceprofile and SEM observations of[retted surfaces. (a) Peened. (b) Unpeened.
208
Therefore, the increase of surface roughness by shot peening will not be the main factor in
improving the fretting fatigue strength at elevated temperatures.
Residual Stress
It is generally known that fatigue crack growth rate is significantly affected by residual stress.
Although the compressive residual stress rapidly reduced by a factor of three by exposing at
773 K, the subsequent relaxation of residual stress with exposing time was not significant. The
compressive residual stress of 200 MPa remained, even after exposing for 30 000 h at 773 K.
Therefore, the main factor to improve the fretting fatigue strength for the material used will
be the compressive residual stress induced by shot peening.
The authors have successfully performed the fretting fatigue life prediction of the peened
specimen with compressive residual stress as well as the unpeened specimen at room and elevated temperatures on the basis of the fracture mechanics analysis. These predictions are given
in Ref 9 and are omitted from the present paper.
Conclusions
Fretting fatigue tests at elevated temperature were carried out using shot-peened specimens
to improve the high temperature fretting fatigue strength of 12Cr-Mo-W-V steam turbine
steel. The effect of shot peening on fretting fatigue strength under representative working conditions was also investigated. The main results obtained are summarized as follows:
1. The fretting fatigue strength of the material used was significantly improved by shot
peening by a factor of 1.8 at the elevated temperature of 773 K as well as at room temperature. The improvement of the fretting fatigue strength by shot peening was still held
even after an exposure time of 30 000 h at 773 K.
2. The relationship between the friction coefficient and the stress amplitude for the shotpeened specimen was in agreement with that for the unpeened specimen at both room
and elevated temperatures.
3. The compressive residual stress is the main factor to improve the fretting fatigue strength.
Although the compressive residual stress was rapidly reduced by a factor of three after
the specimen was exposed at a temperature of 773 K, the subsequent relaxation of residual stress with exposure time was not significant. Even after being exposed for 30 000 h
to 773 K, the compressive residual stress of 200 MPa remained. Therefore, the shot peening seems to be effective for long period practical use at 773 K.
References
[1] Mutoh, Y., Satoh, T., Tanaka, K. and Tsunoda, E., "Fretting Fatigue at Elevated Temperature in
Two Steam Turbine Steels," Fatigue & Fracture of Engineering Materials & Structures, Vol. 12, No.
5, pp. 409-421, 1990.
[2] Waterhouse, R. B., "AvoidanceofFrettingFatigueFailures,"FrettingFatigue, Applied Science Publishers, London, 1981, pp. 221-240.
[3] Leadbeater, G., Noble, B. and Waterhouse, R. B., "The Fatigue on an Aluminium Alloy Produced
by Fretting on a Shot Peened Surface," Proceedings, 6th Int. Conf. on Fract., India, Vol. 3, pp. 21252132, 1984.
[4] Gabel, M. K. and Bethk, J. J., "Coatings for Fretting Prevention," Wear, Vol. 46, pp. 81-96, 1979.
[5] Vardiman, R. G., Creighton, D., Salivar, G., Effatian, A. and Rath, B. B., "Effect on Ion Implantation
on Fretting Fatigue in Ti-6A1-4V Alloy," ASTMSTP 780, ASTM pp. 138-149, 1982.
[6] Tanaka, K., Mutoh, Y. and Sakoda, S., "Effect of Contact Materials on Fretting Fatigue in a Spring
Steel," Transactions, Japan Soc. Mech. Engrs(A), Vol. 51, No. 464, pp. 1200-1207, 1985.
209
[7] Tanaka, C. and Ohba, T., "Stress Relaxation Data of 1Cr-0.5Mo-0.25V, 12Cr-lMo-lW-0.25V and
19Cr-9Ni- 1.4W-Nb Bolting Steels," Transactions, NRIM, Vol. 20-2, pp. 138-153, 1978.
[8] Larson, F. R. and Miller, J., Transactions, ASME, Vol. 74, p. 765, 1952.
[9] Satoh, T., Machida, K., Mutoh, Y., Tanaka, K. and Tsunoda, E., "Improvement of High-Temperature Fretting Fatigue Strength by Shot Peening," Transactions, Japan Soc. Mech. Engrs(A), Vol. 56,
No. 528, pp. 1784-1791, 1990.
DISCUSSION
L. Vincent ~ (written discussion)--Your results clearly presented the beneficial effects of
residual stresses for fretting fatigue strength. However, it is admitted that specific fretting
regimes can favor particle detachment which can delay cracking. Do you think that residual
stresses can influence the surface behavior in the case of phase transformation and particle
detachment?
Y. Mutoh et al. (authors' closure)--Since a debris-like crack generally initiates at an early
stage of fatigue life under fretting condition, almost the whole life of fretting fatigue is spent in
propagating the crack [ 1 ]. Compressive residual stresses reduce the stress intensity factor at
the crack tip and the crack growth rate. Therefore, the fretting fatigue life is improved by the
compressive residual stress. Significant plastic deformation in the thin surface layer is induced
by fretting action on the contact surface. It is also known that the effect of residual stress is
diminutive for the crack initiation process. Therefore, not the residual stress but the heavy
plastic deformation can influence the surface behavior in the case of phase transformation and
particle detachment.
Ecole Ceptrale de Lyon, Ecully Cedex, France.

Deng Yunshu, ~Zhang Baoyu, ~and Luo Weili'
The Fretting Fatigue Properties of a Blade

Steel in Air and Vapor Environments
REFERENCE: Deng, Y., Zhang, B., and Luo, W., "The Fretting Fatigue Properties of a Blade
Steel in Air and Vapor Environments," Standardization of Fretting Fatigue Test Methods and
Equipment, ASTM STP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society
for Testing and Materials, Philadelphia, 1992, pp. 210-216.
ABSTRACF: Fatigue and fretting tests were carried out with 15 Cr 11 MoV steam turbine blade
steel in air and vapor environments.The steel was in the as-heat-treated and as-shot-peened conditions. Both testing ambience and shot peened residual stress could effect the fretting fatigue
strength of the steel. The AES and Mosshauer spectrometry experiments indicated that the oxide
layer in the fretted scar and the debris had different structures, depending on whether they were
formed in air or vapor. This resulted in a change in the tribological properties between the contact surfaces and in the fretting fatigue strength of the steel. However, the environmentaleffects
became less important in compressive stresses.
KEY WORDS: steam turbine blade steel, plain fatigue, fretting fatigue, environmental effects,
oxide debris, Mossbauer spectrometry analysis, shot peening
Introduction
Fretting fatigue in rotor blades of steam turbines is a complex problem which may result in
the increased risk of failure. Investigation to deal with this problem can be conducted with
specimens in the form of a dovetail. The information obtained is then directly concerned with
the stress field and cracking behavior under simulated practical conditions. Another test
method is to use typical fretting fatigue test specimen/pad assemblies to clarify the factors
affecting the fatigue strength of the material. The experiments in this paper are in the second
category.
The material investigated is a blade steel 15 Cr 11 MoV which is essentially developed from
the martensitic stainless steel 1 Cr 13 (AISI 403) by modifying the alloying elements to
improve its heat resisting strength. This material is subjected in-service to vapor adding environmental contribution to fretting fatigue. Corrosion may cause pits from which cracks could
be initiated. Fretting deprives the material of its inherent corrosion and heat resistance by continually disrupting the protective film. In addition, because vapor ambience, the debris produced during fretting could be modified in composition. The properties of the accumulating
debris would play an important role in governing the tribological behavior between the contacting surfaces and, therefore, change the additional alternative shear stress which is strongly
related to the fatigue strength. Furthermore, as a means of alleviating the damaging effect of
fretting, a layer of residual compressive stress could be introduced in the material surface by
shot peening.
Senior Research Engineer, Research Engineer, and Senior Research Engineer, respectively,Shanghai
Research Institute of Materials, 200 437, Shanghai, China.
210
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
lntcrnational
www.astm.org
DENG ET AL. ON BLADE STEEL
211
Experimental Details
The composition of blade steel 15 Cr 11 MoV is listed in Table 1. The material was: (1)
heated at 1050-1100 ~ in a salt bath and oil quenched; (2) and tempered at 720-740 ~ and
air cooled. The hardness of the steel is 228 HV.
The two groups of specimens were shot peened to Almen intensities of 0.027 A with steel
shot and 0.025 N with glass beads. The fatigue and fretting fatigue specimens were in the form
of cylinders, 12 mm in diameter and 226 mm in length, Two parallel flats were machined on
them in the middle, reducing the thickness to 6.4 mm. To produce fretting, a pair of bridges
of the same material as the specimen was clamped onto the flats by means of a proving ring.
The clamping pressure was 32 MPa.
Both fatigue and fretting fatigue testing were carried out in a rotating-bending machine at a
frequency of 25 Hz. In vapor testing, the specimen was sealed in a stainless steel container into
which the hot steam generated from the main water supply was introduced, The residual stress
distribution was measured with X-ray diffraction. The fretted scar and debris were examined
with SEM, AES, and Mossbauer spectrometer.
Results and Discussion
Figure 1 shows the experimental results of the S-N curves for the specimens tested in fatigue
and fretting fatigue. In air, the fatigue strength (curve 1) of 15 Cr 11 MoV is reduced by fretting
(curve 2). In comparison with air data, plain fatigue in vapor results in a certain reduction
(curve 3). Fretting fatigue strength to vapor (curve 4) is greater than fretting fatigue strength
in air (curve 2). Shot peening with steel shot alleviates the damage effect of the fretting both in
air (curve 5) and vapor (curve 6). Glass bead peening also provides benefits to fretting fatigue
strength in vapor (curve 7). However, the improvement is less than that achieved with steel
shot.
From the present results, the environmental effects are highlighted. Because of corrosive
effects, the fatigue strength of 15 Cr 11 MoV is lowered in vapor. Previous experiments [1]
indicated that the chloride ions could produce pits on the surface of the blade steel even at
room temperature. The vapor used in the present experiments contains chloride ions. This
was proved by AES analysis on the fretting scar. The temperature rise caused by hot vapor
could decrease corrosion potential, which leads to more pit formation. As a consequence, the
fatigue of the steel in vapor shows the same behavior as that of the corrosion fatigue.
A more noticeable effect of vapor is on fretting, in which the fatigue strength can be markedly improved from that in air, possibly due to the behavior of the debris. The properties of
debris generated during fretting has long been discussed. The structures of iron oxides, mainly
"y-Fe203 and Fe304, are too similar in crystal structure to be indentified even by X-ray or electron diffraction technique. This difficulty may be overcome by using Mossbauer spectrometry.
Fe304 is revealed in the Mossbauer spectrum with two groups of magnetic hyperfine structures.
Correspondingly, the internal magnetic field strengths are 489 KOe (A site) and 450 KOe (B
site) at room temperature. The spectrum ofs,-Fe203 shows only one group of magnetic hyperfine structures with the internal magnetic field strength of 493 KOe. The standard internal
magnetic field strength at room temperature [2] is shown in Table 2, from which the iron
oxides concerned could be distinguished,
TA BLE 1--Chemical composition ~)[15Crl IMo V (wt. %).
C
0.16
Mn
0.22
Si
0.31
Cr
10.80
Mo
0.59
V
0.33
Fe
Remainder
212
HPa
400
L,d
I--
300
LO
Z
Z
I.J.d
\-,%.\
"~, "',O,x~,.~
200
<
""-.~1) ,
\
\.
<>\
\\
5
-[7)
100
105
106
10?
CYCLES TO FAILURE
FIG. 1 - - S - N curves for steel 15 Cr 11 M o V : (1) o, PF, air; (2) ~, Ft~; air; (3) o, p I~. vapor; (4) O, FF,
vapor, (5) A, SP, FF, air," (6") ~, SP, FF, vapor; (7) X, GP, FF, vapor. (Notes: P F = plain fatigue; F F =
fretting fatigue; S P = peening with steel shot; GP = peening with glass bead; air = in air, and vapor =
in vapor).
For analyzing the oxide debris in more detail, comparisons are made between heat resistant
steel 15 Cr 11 MoV, mild steel 45 and nitrided steel 38 CrMoA1 in Table 3. The data indicate
that the debris have different compositions. There is a limited distinction in the internal magnetic field strength between the standards and that of the debris. This is due to the atoms of
alloying elements such as Cr, Mo, V and Mn by which the internal magnetic field strength of
an iron oxide is weakened [2]. The weakened strength data produce a problem to identify -yFe203 and a-Fe203. For this, a more accurate analysis of the spectrum with a computer is
TABLE 2--MOssbauer parameters for iron oxides at room
temperature.
Compound
c~-Fe
oL-Fe203
7-Fe203
Fe304
fl-FeOOH
Internal Magnetic Field

Strength, KOe
Electric Quadrupole
rams J
330
518
493
489 (Site A)
450 (Site B)
0.68 + 0.01
213
TABLE 3--Compositions of oxide debrisformed duringfretting fatigue.
Steel
Environment
In Air
Compound
Internal Magnetic Field

Strength, KOe
a-FezO3,~-Fe203
Fe304
a-Fe
~-FeOOH
499
475(Site A),419(Site B)
308
a-FezO3, ~-FezO3
Fe304
a-Fe
~-FeOOH
493
470 (Site A),435 (Site B)
315
a-Fe203, ~-Fe203
Fe304
a-Fe
~-FeOOH
508
487 (Site A), 452 (Site B)
329
~-Fe203, y-Fe203
Fe304
,-Fe
~-FeOOH
500
473 (Site A),43l (Site B)
329
Electric
Quadrupole
mms
Wt.%
0.6831
7
22
38
33
0.8101
,15
37
39
9
0.7354
37
42
15
6
0.8093
31
39
24
6
15Crl lMoV
In Vapor
45
38CrMoAI
In Air
In Air
needed. However, the existence of Fe304 is confirmed by the internal magnetic field strengths
corresponding to A site and B site. The presence of a-Fe is also demonstrated because the
datum is so distinct. The content of a-Fe could be related to the corrosion resistant properties
of the material. Perhaps this information is significant to the fretting mechanism [3, 4],
whether the debris formed during fretting is based on scraping off the oxide film, which is
immediately rebuilt, or the formation of metallic particles which are subsequently oxidized.
Considering again the steel 15 Cr 11 MoV, the Mossbauer spectra is presented in Fig. 2, from
which part of the data in Table 2 is derived. It is noted that the amount of Fe304 in the debris
produced in vapor is greater than in air. In comparison with Fe203, Fe304 is considered to have
better tribological properties, i.e., wear resistant and anti-friction. Thus, the debris combined
with the vapor could form a mud solid lubricant which acts more effectively than those formed
during fretting in air.
The examinations with AES (Fig. 3) indicate that at about 8 m m outside the scar there is
only a small difference between the thickness of the oxide film formed in air (30 nm) and vapor
(20 nm). This contrasts with oxide thicknesses found on the scar when the specimens are tested
in air (250 nm, followed by a constant tendency) and vapor (250 nm, followed by a rapidly
falling tendency). This is mainly due to the lower frictional temperature in the contact zone
of the specimen tested in vapor, as under this condition the more effective solid lubricant could
be formed. The deduction, therefore, is that the environment itself, air or vapor, provides only
a limited effect on the oxidation of the steel, whereas, the dominating factor is the interaction
between the environment and the fretting process, by which the oxidation and tribological
behavior of the surface could be determined. This could eventually relate to the fatigue
strength of the material.
If the specimens are not peened to a certain intensity, the induced residual stress should be
involved in the fretting fatigue, as discussed in Ref [5-6]. In the present work, shot peening is
found to restore significantly fretting fatigue strength, with no essential difference in air or
214
FRETTING FATIGUE TEST METHODS
~i.,J
ill -'ll''ll+'.i+~
"-~
AND EQUIPMENT
"~
4"
+:,- "r"c',+.l.~.~,i-t...r
,l-i+,q,i-.~
+
+ 4- ;i..
~mr
+'i~++,,+i+~.,.i+ "P +4+

~ %+
+!.++9.~ +.,. ~ii
t,
~,
,+ +
i++l.. +
8~t-
ml
++
~/
40
%
E
I--C
C:)
LJ
,t
-i-i
,,. ,'
~.
~+i +
.i.l
'
7
I
40
80
200
240
++, .,+. +'+% ,+'

+ +~ +-, ~+ V'% F "% +~
4 ~'~.+..~
" "~ i
"4.i
c
i
(a)
I
120
160
EHANNEL
80
%ui':
'
+l
'
i
m)
V
I
120
160
EHANNEL
,'
200
240
FIG. 2--Mossbauer ~pectra for oxide debris o.fstee115 Cr 11 Mo V fretted (a) in air and (b) in vapor.
vapor for constant peening intensity. This means that environmental effects are suppressed.
The most important role played by environment is in rubbing the surfaces concerned with
crack initiation and propagation in a range beyond which the additional frictional stress shows
no effect. In the presence of fretting, the proportion of life concerned with crack initiation is
shortened compared with plain fatigue. Thus, the fatigue strength is largely dependent on the
intensity of shot peening which determines the near surface residual compressive stresses (Fig.
4), promoting barriers which retard the propagation of cracks. This suggestion is clearly supported by a comparison made with the fretting fatigue strength and the residual stress distribution for the specimens peened with steel shot and glass beads.
Conclusion
Environment influences both the fatigue and fretting fatigue properties of the steam turbine
blade steel 15 Cr 11 MoV. The effect of corrosion on this heat resistant steel can be assessed
in terms of the reduction in fatigue strength for specimens tested in vapor. The most damaging
effect on fatigue properties is fretting in air, whereas, fretting the steel in vapor could restore
fatigue strength. This is presumably due to the favorable tribological properties provided by a
more effective solid lubricant. This lubricant is accumulatively formed as a result of the interaction between the environmental ambience and the fretting process, and contains more Fe30+
in its constituents as revealed by Mossbauer spectrometry analysis. When specimens are shot
peened, fretting fatigue strength is largely dependent on the intensity of the peening, no matter
215
FIG. 3--Element distribution versus depth o f the oxide layer as revealed by A E S (a) in the scar, fretted
in air, (b) about 8 m m out o f the scar, fretted in air, (c) in the scar, fretted in vapor and (d) about 8 m m out
q f the scar, .fretted in vapor.
MPahIi~,
-/+00
~-30C
t./)
- ?oc_
100l
~,~1
100
DEPTH
200
BELOW
300
600
SURFACE
~a
FIG. 4--Residual stress profiles for steel 15 Cr I I MoV: (l o, SP, prior to testing, (2 X, GP, prior to
testing; (3 ~, after FF, vapor, N = 7.88 X 10 ~ cycles to failure; S = 272 MPa, (4 o, SP, after FF, vapor, N
= 4.70 104 cycles to failure; and (5 S = 453 MPa.
216
which ambience, air or vapor, is involved. In this case, environmental effects are suppressed
by the influence of the residual stress induced in the surface of the material.
References
[1] Wu, J. et al., "The Atlas Metallography of Stainless Steels," (in Chinese), Research report, Shanghai
Research Institute of Materials, 1982.
[2] Xia, Y., et al., "Mossbauer Effect and Its Applications, ""(in Chinese), Atomic Energy Press, 1984.
[3] Waterhouse, R. B., Fretting Corrosion, Pergamon, Oxford, 1972.
[4] Pendlebury, R. E., "Unlubricated Fretting and Sliding Wear of Steels in Air," Proc. Int. Conf. Tribology--Friction,Lubrication and Wear--50 Years, on 1-3 July 1987, London, Instn. Mech. Engrs.,
pp. 267-275.
[5] Waterhouse, R. B., Fretting Fatigue, Applied Science, London, 1981.
[6] Deng, Y., Zhang, B. and Luo, W., "Fretting Fatigue of Some High Strength Aluminium Alloys and
Steels," Proc. C--MRS Int. '90, July 18-22, 1990, Beijing, China, Elsevier Applied Science, (in the
press).
Stephen Price ~and David E. Taylor 2
The Application of Electrochemical

Techniques to Evaluate the Role of Corrosion
in Fretting Fatigue of a High Strength Low
Alloy Steel
REFERENCE: Price, S. and Taylor, D. E., "The Application of Electrochemical Techniques to
Evaluate the Role of Corrosion in Fretting Fatigue of a High Strength Low Alloy Steel," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM STP 1159. M. Helmi Atria
and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992,
pp. 217-228.
ABSTRACr: An experimental assembly has been developed to enable aqueous fretting fatigue
tests to be performed with an orthogonally crossed cylinder arrangement of fatigue specimen and
fretting pad. The equipment has been employed in a study of the fretting fatigue behavior of the
high strength, low alloy steel, Corten A, in both air and synthetic seawater environments.
The use of impressed cathodic protection has identified electrochemical dissolution processes
as having a significant influence on fatigue life. Normal corrosion fatigue in seawater produces a
60% reduction in fatigue life compared to that in air, while the imposition of fretting causes a
further 24% reduction in fatigue life.
The initiation sites of fretting fatigue cracks occur at the boundary between slip and non-slip
regions in both air and aqueous environments. Cracks propagate at an oblique angle to the surface of the fatigue specimen. On leaving the zone of influence set up by the alternating shear
stress, the cracks propagate perpendicular to the applied cyclic stress.
KEY WORDS: fretting corrosion, fretting fatigue, seawater corrosion, high strength steel, cathodic protection
Introduction
Fretting is said to occur when two surfaces held in contact are subjected to relative cyclic
tangential displacement or slip. The term fretting describes the action and not the effect, gives
rise to the production of wear debris and surface deformation, and involves such processes as
surface film disruption, surface abrasion from resulting debris and the making or breaking of
cold welds. Two recognized problems associated with such localized damage are: ( 1) wear and
a c c o m p a n y i n g loss of fit between surfaces; and (2) nucleation of cracks which under fatigue
loading m a y propagate a n d lead to failure.
The e n v i r o n m e n t in which fretting takes place is a significant factor in determining both the
nature a n d degree of surface damage. In-air adhesion and tribo-oxidation are recognized as
the major wear processes active during fretting; however, in aqueous e n v i r o n m e n t s such as
aggressive oxygenated seawater adhesion between the surfaces, fretting is reduced as a result
of a lubricating effect of the electrolyte. More important, is the possible accelerated attack of
underlying metal when the surfaces in the contact zone are disrupted by mechanical action.
Design Metallurgist, NEI Parsons Ltd., Meaton Works, Newcastle Upon Tyne, England.
2 Senior Lecturer, School of Technology, Sunderland Polytechnic, Sunderland, England.
217
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
International
www.astm.org
218
In addition, the fretting contact often provides a situation in which crevice corrosion is possible due to differential aeration effects. Pearson et al. [1 ] have identified three components of
material removal during fretting in aqueous media:
(1) A mechanical component similar to that perceived when fretting occurs in air.
(2) A purely electrochemical contribution, where material may be removed as a result of
crevice corrosion.
(3) A synergistic effect of mechanical and electrochemical components, whereby continual
cleaning of the contacting surfaces occurs to enable the crevice corrosion to proceed.
In earlier studies [2,3], the authors have quantitatively assessed the relative importance of
these three major processes from fretting wear tests in aqueous environments. The work concluded that the conjoint action of corrosion and mechanical surface disruption was the significant factor in determining material removal rates with normal corrosion playing an insignificant role.
The present work examines fatigue behavior under fretting conditions in air and aqueous
oxygenated seawater and aims to establish the significance of corrosion processes during fretting corrosion fatigue.
Materials
Corten A
Corten A is a high strength, low alloy steel which is reported to exhibit improved mechanical
properties over conventional structural steels [4,5,6]. The composition and mechanical properties of the steel are given in Table 1. The microstructure ofCorten A consists of a fine grain
ferritic matrix with a small dispersion of pearlite grains.
Specimen Preparation
Fatigue Specimens--Fatigue specimens were produced from a rod of 150 mm length and
5 m m diameter. The rods were turned down at the centre to a diameter of 2.5 mm over a gauge
length of 25 mm. The specimens were polished in a pedestal drill chuck on SiC paper and then
finally on selvyt cloth impregnated with diamond paste to give a 1 um finish. The ends of the
rod were threaded over a length of 20 m m to allow gripping within the collets.
Fretting Pads--The cylindrical pads consisted of 30 m m lengths of a 5 m m diameter rod
polished as outlined above.
TABLE 1--Composition and mechanical properties o[Corten A.
Composition (wt.%)
C
Si
Mn
Ni
Cu
Sn
Cr
Mo
0.088
0.345
0.024
0.097
0.36
0.03
0.31
0.003
0.842
0.005
Ultimate tensile stress (MPa)

Yield strength (MPa)
Modulus of elasticity (GPa)
Elongation (%)
Hardness (Hv)
Surface finish (C.L.A.) (#m)
501
401
196
27
157
0.16
PRICE AND TAYLOR ON ELECTROCHEMICAL TECHNIQUES
219
Fretting Fatigue Equipment

Fatigue experiments were carried out on a push-pull mode Avery midget pulsator fatigue
machine fitted with threaded collets to retain the specimen. The fatigue machine operated at
a fixed frequency of 50 Hz and has a maximum load capability of 3 kN.
A cell (Fig. 1) was developed to carry out fretting fatigue tests in an aqueous environment
and to facilitate the employment of electrochemical control. The cell was fabricated from a 6
m m sheet ofPerspex T M in an open box construction with a lid clamped on top and sealed with
an O-ring. The fatigue specimen was situated across the cell, passing through seals incorporated in the side walls.
The fretting pads have a span of 20 m m and were retained in two Perspex blocks fitted to
solid cylindrical loading pistons (Fig. 2). The pistons were located through the base and lid of
the cell through seals. On the top face of each piston, a hexagonal socket bolt was fixed to hold
the bolts of the proving ring, which was used to apply pressure between the pads and the fatigue
specimen. The lid of the cell contained two ports which enabled consistent positioning of a
reference and counter electrode into the electrolyte when applying electrochemical control.
The procedure for setting up the equipment and testing required the six steps listed below:
(1) The lower piston incorporating two pads was inserted through the base of the cell. The
fatigue specimen was inserted through the side walls. The lid containing the upper piston and pads was then secured.
(2) The test load was applied between the pads and the fatigue specimen by rotating the
bolts of the calibrated proving ring evenly, while measuring the ring diameter until the
required loading was achieved.
(3) Ifa seawater test was to be performed, then the cell was filled with seawater through one
of the electrode ports in the lid. If, in addition, cathodic control was to be employed,
then the electrodes were incorporated into the cell.
(4) The cell was connected to the fatigue machine using the threaded collets.
(5) With the eccentric cam for adjusting the alternating load set at zero, the static load was
FIG. l--Schernatic of the./?etting&tigue cell.

220

F~'3"~IG pAD
"
SPEC
FIG. 2--Illustration (?fthefretting pad gripping arrangement.
applied by rotating the crack handle. The level of applied loading was determined using
the calibrated microscope reading of the strain gauge loop dynamometer.
(6) The alternating load was then superimposed on to the static load by rotating the eccentric cam and again using the microscope reading to evaluate the level of loading. The
apparatus used to carry out all testing is displayed in Fig. 3.
Fretting Fatigue Studies
Fretting fatigue studies were carried out in synthetic seawater over a range of mean and alternating stress values. The stress ratio, R, defined as the ratio of the minimum to maximum
FATIGUE
/
~a.ic
,.o,,,,'~--.---~---;L
S~ITC~
,~'-
k~
m',~Iri
~-trw::::kt
I.d,rlll
_tLL~.J/"
~.A~D~
.\
FIG. 3--Schematic d'the frettingJi~tigue testing apparatus showing the cell in position.
221
stress, was maintained at 0.3. The load applied by the proving ring to the assembly was 570 N
in all tests.
The slip amplitude was determined from the following equation for the alternating stress
range employed [ 7]
(1)
S = 6x/E
where S = slip amplitude, ~ = applied alternating stress, x = pad span, and E = elastic modulus of the fatigue specimen material.
Thus for alternating stresses between 100 and 275 MPa the estimated slip amplitudes are
between 10 and 28 , m .
Comparative tests were performed at an alternating stress of 208 MPa under the following
conditions: (1) fretting fatigue in air; (2) corrosion fatigue in synthetic seawater; and (3) fretting
fatigue in synthetic seawater with cathodic protection applied at - 9 5 0 mV with respect to the
saturated calomel electrode.
A comprehensive list of the test parameters and environment conditions is detailed in Table
2.
The tests involving cathodic protection were performed using the circuitry utilized in the
electrochemical studies of fretting wear [2,3]. The fatigue specimen formed the working electrode in the circuit and electrical connections were made to both ends of the specimen. This
ensured that, should the specimen fail, both fracture faces would remain protected from secondary electrochemical dissolution.
Specimen Examination
Fracture surfaces of the fatigue specimen were examined in the scanning electron microscope. The scars produced at the opposite end to that which fractured were mounted in clear
Perspex and polished and etched to reveal the scar profile, and enable examination for the
presence of cracks and their nature.
TABLE 2--Fatigue test conditions and parameters. ~
Type of Test
Alt.
Stress
(MPa)
Mean
Stress
(MPa)
Prov. Ring
Load (N)
Fret.fat.SW
Fret.fat.SW
Fret.fat.SW
Fret.fat.SW.cp
Fret.fat.air
Corr.fat.SW
Fret.fat.SW
Fret.fat.SW
Fret.fat.SW
Fret.fat.SW
275
250
208
208
208
208
190
175
150
100
508
463
386
386
386
386
352
325
278
185
570
570
570
570
570
570
570
570
570
570
aWhere Fret.fat.SW. = fretting fatigue in seawater.

Fret.fat.SW.cp. = fretting fatigue in seawater cathodic
protection applied (-950mV w.r.t.S.C.E).
Fret.fat.air = fretting fatigue in air.
and Corr.fat.SW. = corrosion fatigue in seawater.
222
Results
Examination of Test Data

Data compiled from fretting fatigue studies are shown graphically in Fig. 4. The S-Log N
graph shows the curve for Corten A steel under conditions of fretting fatigue in seawater. The
curve shows a limit at 160 MPa which is consistent with the findings of others [8].
Figure 4 also shows individual points determined at an alternating stress of 208 MPa for
comparative tests of corrosion fatigue in seawater, fretting fatigue in air and fretting fatigue in
seawater with cathodic protection employed. In the latter two tests, fracture could not be
achieved and run-outs in excess of 107 w e r e obtained. The comparative test results are summarized in the Table below.
Fretting
Fatigue
in SW
Fatigue
in Air
> 107
Fretting
Fatigue
in Air
Fatigue in
SW
Fretting
Fatigue in
SW
@
--950 mV
> 107
3.94 X 106
1.56 X 106
> 107
+ C.P.
The ratio of lifetimes compared to that of fretting fatigue in

seawater are therefore as follows:
>6.4
>6.4
2.5
>6.4
The data clearly indicate the important role of environment on the fatigue life of Corten.
The restoration in the fretting fatigue life in seawater to a level similar to that observed in air
through the use of impressed cathodic protection indicates that mechanical action plays a
minor role in the life reduction process. The results of corrosion fatigue and fretting fatigue
tests in seawater suggest that corrosion is the major effect when compared to fatigue in air.
Optical and Scanning Electron Microscopy

Fretting Fatigue in Seawater--Examination of the fatigue specimens fretted in seawater
under freely corroding conditions was carried out after each test. The four scars produced on
each specimen were elliptical in shape as a result of the 2:1 ratio in pad to fatigue specimen
diameter. The real scars were approximately equal in.area (3.04 X 10 -6 m2). Thus, an even
distribution in proving ring load was achieved. The loading at each pad of 142.5 N resulted in
a m a x i m u m stress of 70 MPa and a mean stress of 47 MPa.
A typical scar produced in stagnant seawater is shown in Fig. 5(a) and (b) for fretting fatigue
at an alternating stress of 208 MPa. The micrograph shows both halves of the fracture specimen and the crack is clearly seen to originate at the boundary between fretted and non-fretted
regions. During all tests in seawater, partial slip was observed at the extremities of the fretting
scar. The resulting fretting damage is well defined as shown in Fig. 6. Near the crack, wear
tracks are produced in the direction of applied stress and at the extremities of the scar smearing
results in a layered structure of plastically deformed material.
The fracture surface appearance is shown in Figs. 7(a) and (b). At the nucleation site (Fig.
PRICE AND T A Y L O R ON E L E C T R O C H E M I C A L T E C H N I Q U E S
223
Alternating stress vs cycles to failure

(Corten steel)
300
28O
250
240
220
%-
200
~Z
v
180
140
120
1O0
80
50
40
2O
0
5.4
5.8
9
FIG.
6.2
FF,olr
6.6
Log (cycle~ to foilure)
n
FF.sw
7.4
/~
CF.sw
7.8
O
FF.cp
4--FretHng,/btigue S-tog N curve for Corten stee/ tested in synthetic seawater in push-pull mode.
7(a)) ductile surface is apparent with possibly secondary brinnelling producing flattening of
asperity contacts on crack opening and closing. At a distance from the nucleation site, intergranular facets are observed along with transgranular striations (Fig. 7(b)). The intergranular
regions are indicative of an environmental influence, but may also appear when the size of the
crack tip plastic zone approaches the grain size of the steel.
Microsection examination of those specimens subject to cathodic protection revealed no
evidence of cracking. Microsections taken through the scars produced under freely corroding
conditions showed cracks at both ends of the scar propagating in towards the center of the scar
at an oblique angle. Typical cracks observed for alternating stress values of 208 and 275 MPa
are shown in Figs. 8 and 9, respectively. The cracks are approximately 100 and 140 um in
length. The oblique angle of the cracks is encouraged both by the alternating stress applied to
the surface by the fretting action and by the environment. In general, low alternating stresses
and a corrosive environment encourage this stage 1 crack growth [8]. The cracks occur at the
boundary between fretted and non-fretted regions as indicated on the micrographs. At 208
MPa the crack can be seen to follow the ferrite grain boundaries, but takes a transgranular path
through the pearlite phase. At 275 MPa the grain boundaries are not clear from the micrograph; however, a transgranular path through the pearlite phase is apparent.
Fretting Fatigue in Air--Fretting fatigue in air did not produce specimen failure within the
limits of the fatigue machine. However, sections taken through the resulting scars showed the
presence of single cracks at the periphery of the scar. The cracks were similar to those produced
in seawater, exhibiting the same intergranular path through the ferrite grains and transgranular
224
FIG. 5--Fretting fatigue specimen tested at an alternating stress of 208 MPa in seawater. The crack is
clearly observed to nucleate between the region of slip (a) and non-slip (b).
FIG. 6--Layered structure produced at the periphery of the scar (alternating stress of 208 MPa in
Copyright
by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
seawater).
225
FIG. 7a--Nucleation site of fretting fatigue crack produced in seawater at an alternating stress of 208
MPa. The surface has a ductile appearance with secondary brinnelling in evidence.
FIG. 7b--Surface appearance at a distance from the nucleation site. lntergranular facets can be seen
along with transgranular striations.
through the pearlite phase. A typical crack produced at an alternating stress of 208 MPa is
shown in Fig. 10. The crack is approximately 70 am in length and propagates obliquely to the
surface.
The similar intergranular nature of the cracks produced in air, obviously a much less aggressive environment than seawater, suggests that crack tip plastic zone size may be the important
influence. However, comparison of fretting in vacuo [8] to air has demonstrated the corrosive
nature of the air environment and its ability to enhance crack propagation.
Discussion
The combined effects of fretting and fatigue have been examined using Corten steel. In these
investigations, electrochemical processes are found to have the greatest influence. Of those
conditions investigated, fretting fatigue in seawater resulted in the shortest life with significant
improvements in life time being achieved under normal corrosion fatigue and a further
increased life time whilst fretting under fatigue conditions in air.
Cracks produced during fretting fatigue in seawater were clearly observed to initiate at the
interface between fretted and non-fretted regions, and propagate intergranularly through the
small grain ferrite phase and transgranular through the pearlite phase. Striations produced on
transgranular regions may be effected by carbide orientation, a feature which has been
observed by others in structural steels [9]. Fretting causes crack nucleation and also influences
the direction of crack growth because of the alternating shear stress imposed. However, the
mechanism of propagation is believed to be similar to corrosion fatigue. Crack growth rates
Copyright
by ASTM
Int'l (all rights
Mon Mar 28 00:05:25
EDT 2016 such as anodic dissolution at the crack tip and
are
accelerated
byreserved);
electrochemical
processes
226
FIG. 8--Microsection taken through a fretting fatigue scar produced at 208 MPa in seawater showing
oblique nucleation angle and intergranular path through ferrite and transgranular through r~ions ~?f
pearlite.
FIG. 9--Microsection taken through a fretting fatigue scar produced at 275 MPa in seawater showing
oblique nucleation angle.
227
FIG. 10--Microsection taken through a fretting fatigue scar produced at 208 MPa in air.
hydrogen embrittlement. Similar microstructural features observed in air compared to seawater suggest the former to be of greater validity. In corrosion fatigue, enhanced crack growth
rates in seawater are suggested to be a result of a mechanical cracking process which is
enhanced by anodic dissolution of plastically deforming material at the crack tip [9]. Once the
crack is outside the influence of the fretting action, propagation proceeds perpendicular to the
applied alternating fatigue stress. An important factor influencing propagation of the crack is
the ability of the electrolyte to penetrate the crack to the tip. The work of others [9] has shown
that replenishment of seawater to the crack tip is provided during fully reversed bending conditions. In addition, the pH within the crack is the same as the bulk solution. Thus, increased
reactivity by mechanical stimulation must be responsible for accelerated dissolution and crack
growth as opposed to an increase in acidity of solution within the crack.
Cathodic protection was found to restore the fretting fatigue life of Corten steel in seawater
similar to that observed in air. This infers that the mechanical action of fretting alone does not
produce cracks which propagate to failure. A conjoint action is responsible for the reduced
fatigue life with fretting in seawater compared to normal corrosion fatigue in seawater.
Conclusions
Preliminary investigations on the fretting fatigue behavior of Corten steel have been carried
out from which the following conclusions have been drawn:
1. Electrochemical dissolution processes are the significant factor determining material
integrity when subjected to fretting and fatigue in seawater. Corrosion fatigue in the absence
of fretting gives a life 2.5 times greater than that under fretting conditions, while in air a life of
at least 6.4 times that under fretting conditions in seawater is observed.
2. The role of fretting is to promote early crack nucleation at the interface between slip and
non-slip regions. Once formed the crack grows obliquely to the surface under the influence of
the alternating shear stress imposed by the fretting action. The crack having grown outside of
228
the influence o f fretting proceeds to propagate perpendicular to the alternating fatigue stress
under normal corrosion fatigue processes.
3. In both seawater and air, cracks were observed to propagate intergranularly through the
ferrite matrix and transgranularly through pearlite.
References
[1] Pearson, B. R., Waterhouse, R. B., Proceedings, International Congress on Metallic Corrosion, Vol.
2, 1984, Vol. 125, pp. 334.
[2] Price, S., Taylor, D. E., Wear, Vol. 125, 1988, pp. 107.
[3] Price, S., Taylor, D. E., "In Surface Engineering Practice," Datta, Gray, Horwood, Ltd., Stratford,
England, 1990, pp. 553.
[4] Rodgers, T. H., Marine Corrosion, Newnes International Monographs on Corrosion Science and
Technology, London, 1968.
[5] Larrabee, C. P., Corrosion, Vol. 9, 1953, pp. 259.
[6] "Properties and Selection: Irons and Steels," Metals Handbook, 9th Edition, Vol. 1, ASM, Ohio,
1978, pp. 419.
[7] Lindley, T. C., Nix, K. J., MultiaxialFatigue, STP853, ASTM, Philadelphia, 1982, pp. 340.
[8] Waterhouse, R. B., Fretting Fatigue, Applied Science, London, 1981.
[9] Hodgkiess, T., Proceedings, International Conference on Mechanisms of Environment Sensitive
Cracking of Metals, Univ. of Surrey, 1977, pp. 348.
Nonconventional Materials and Test

Methods
A l a i n Cardou, ~L o u i s Cloutier, ~M i c h e l St-Louis, 2 a n d AndrO L e b l o n d ~
ACSR Electrical Conductor Fretting Fatigue at

Spacer Clamps
REFERENCE: Cardou, A., Cloutier, L., St-Louis, M., and Leblond, A., "ACSR Electrical Conductor Fretting Fatigue at Spacer Clamps," Standardization (?/'Fretting Fatigue Test Methods
andEquiprnent, ASTMSTP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 231-242.
ABSTRACT: Overhead electrical conductors are subjected to aeolian vibrations which may
induce aluminum wire fatigue breaks at line suspension points and spacer clamps. Such wire
failures are essentially due to fretting fatigue occurring at interwire and clamp-wire contact
points. Tests have been performed on a typical ACSR (aluminum conductor steel reinforced)
conductor. The test benches are based on a new design in which a cyclic deflexion amplitude is
imposed at the center of the test specimen through an eccentric shaft. Tests have been conducted
with two types of spacer clamps. Individual wire breaks have been recorded, including those
occurring at inner layers. Analysis of each specimen has shown that almost all wire breaks initiated at fretting marks. A careful location analysis (both in the conductor cross section and axial
direction) has yielded some new results on the cyclic bending behavior of a conductor-clamp
system.
KEY WORDS: overhead electrical conductors, aluminum conductor steel reinforced, fretting
fatigue, spacer clamps
Overhead electrical conductors are generally manufactured using circular aluminum wires
stranded helicoidally in one or several layers around a central core. The core itself may be
made of a l u m i n u m or steel. In each category there are many models differing as regards both
matei'ial properties and various geometrical parameters such as the wire diameter, number of
layers, and helix angle. Under natural conditions these conductors are subjected to wind
induced small amplitude oscillations called aeolian vibrations, which can lead to wire breakage
at critical points of the transmission line, such as at the suspension clamps and bundle spacer
clamps. Such fatigue breaks are essentially due to fretting fatigue occurring at inter-wire contact points.
The fretting fatigue phenomenon in overhead electrical conductors has been described by
Rawlins [1 ]. He has also made a review of the various ways of relating fatigue at suspension
clamps to vibration data, illustrated by some of the available experimental data and field
results. Another review of similar fatigue data has also been published by a CIGRE committee
[2].
Endurance Tests on Electrical Conductors
As underlined above, a great deal of data is available regarding the fatigue strength of overhead electrical conductors coming either from laboratory tests or from field observations.
Professors and Graduate Student, respectively, Department of Mechanical Engineering, Laval University, Ste-Foy, Quebec, Canada G 1K 7P4.
2 Engineer, Hydro-Quebec, Montreal, Quebec, Canada H2L 4P5.
231
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 1992by
lntcrnational
www.astm.org
232
However, this data is often difficult to correlate for several reasons: (1) it characterizes a given
conductor-clamp combination; (2) some tests are performed under alternating in-plane bending and others under alternating tension; (3) in bending, various controlling parameters may
be used such as amplitude at a standard distance from last contact point, strain on a given wire,
m a x i m u m amplitude for standing wave, vibration, etc.; and (4) the failure criterion is not
always clearly defined (number of broken wires at which a conductor is considered to have
failed). For example, many fatigue tests have been performed on long specimens with one end
passing into a typical suspension clamp, the other being held by the tensioning device. The
transverse amplitudes are obtained using an electromagnetic vibrator driving the conductor
at or near to resonance.
Endurance test data are generally analyzed and presented in the same fashion as those
obtained from tests on solid specimens. A classical analysis is the one carried out by Poffenberger and Swart [3] which yields a bending stress value from a measured amplitude at a given
distance from a clamping device, as illustrated in Fig. 1. The Poffenberger-Swart formula can
be written as
d
~.=~X
E~,Xk a
e ~x_ 1 + kx x Yb
(1)
with
k~ -
T
EI
{2)
where
E1 =
Ea~ =
T =
d -x -Yb =
assumed bending stiffness of the conductor, N 9

2,
Young's modulus of aluminum, MPa,
static tensile load on the conductor, N,
aluminum wire diameter, mm,
distance from clamp at which vibration amplitude is measured, mm, and
peak-to-peak vibration amplitude at distance x from clamp, mm.
Usually, the assumed conductor bending stiffness is a lower bound, corresponding to the
case in which all individual wires act independently. The position x at which the bending
amplitude Yb is measured has been standardized, at least in North America, at a distance of
89 m m (3.5 in.) from the last point of contact between the conductor and the clamp.
The alternating stress value % obtained from Eq 1, is a nominal value which may be quite
/////////
~lalt
point of contact
:'7-.
FIG. 1--Conduclor bending in clamp region.
CARDOU ET AL. ON ELECTRICAL CONDUCTOR FRETTING FATIGUE
233
different from the actual values obtained by direct strain-gage measurements on the conductor. In spite of this, the cra value presents the advantage of normalizing results from tests performed under different axial loads and with conductors of different geometry. However, this
normalization does not completely eliminate such effects. For example, Rawlins [1] has
shown that single layer conductors have a quite different endurance behavior compared with
multilayer conductors.
Another limitation on the use of the Poffenberger-Swart formula is that wire fracture is not
induced by the alternating stress, per se, as is the case in a standard fatigue test on a solid specimen but, rather, by the fretting occurring at inter-wire contact points. It is well known that
fretting fatigue is mainly controlled by interface pressure and the relative motion in contact
areas, not to mention material and surface characteristics. The tensile stress in the wires plays
a role in crack propagation once the latter has been initiated. However, the crack propagation
phase is a very small fraction of the total life of a broken wire, most of it having been spent in
the crack initiation stage. Thus, the alternating stress c~ given by Eq 1 should be interpreted as
an index of fretting parameters, rather than a stress level to be used in the wire material S/N
curve.
This has been shown clearly by Lanteigne et al. [4] in tests performed on individual wires
of EC-H19 aluminum (electrical conductivity aluminum wires cold drawn to 90%). First,
plain bending fatigue tests were performed on wire specimens under a given level of axial tension. Subsequently, tests were performed on specimens pressed across a similar aluminum
wire located at the critical cross section. An adjustable clamping load was applied on the
crossed wires. The test wire was subjected to the same nominal average and alternating stresses
as in the plain bending tests. This resulted in a drastic decrease of the number of cycles leading
up to fracture, except in low-cycle fatigue, where plasticity plays a major role. The decrease
was strongly influenced by the clamping force. However, as contact conditions in multilayer
overhead electrical conductors are quite difficult to determine in general and in particular near
a clamp, the aim of the present work is to show how a conductor-clamp combination can be
tested in bending and what conclusions can be drawn.
Fatigue Test System

The fatigue test system used in this work has already been described [5]. However, for completeness the main features, including some further improvements to the test rig should be
reviewed. The bending amplitude is imposed on the conductor specimen by an adjustable
eccentric (Fig. 2) equivalent to a slider-crank mechanism. The clamp is attached to the slider.
Both ends of the specimen are fixed to a tensioning system which allows slight displacements
in order to maintain a nearly constant tensile load. The up and down clamp motion induces
a slight variation of the conductor angle at the clamp mouth. This corresponds, for a given
point of the conductor axis, to an alternating amplitude with respect to the clamp, equivalent
to the vibration taking place in the field, where the clamp can be considered as fixed.
In order to obtain a required relative amplitude, one may vary the eccentricity of the slidercrank system. However, for practical reasons, the variable eccentricity system used in previous
tests has been replaced by a set of fixed eccentricity shafts yielding clamp motion peak-to-peak
amplitudes of 50, 60, and 70 ram. In order to span a continuous range of relative amplitudes,
adjustable blocks have been added on both sides of the clamp. These can be moved horizontally, vertically, and angularly. Of course, they induce some bending in the conductor as the
clamp moves up and down. To avoid undesirable fatigue breaks at these locations, they are
padded with rubber cylinders. This procedure has proved to be adequate since no wire break
has been noted up to now in the region of the blocks. Total specimen length has been set at
about 3.6 m (12 ft), or half this value between the clamp and each end fitting. Cycling freCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
234
conductor
..Quspenslon
clamp
amplitude
control
end fitti~
load control
FIG. 2--Principh, qf conductor fatigue test bench.
quency has been set at 10 cps in order to minimize dynamic effects both on the test bench and
the specimen. Two identical benches have been built and used in the present test program.
Following an Institute of Electrical and Electronics Engineers (IEEE) standard [ 1 ], the relative
amplitude is measured at 89 m m (3.5 in.) from the last contact point between conductor and
clamp. Conductor specimens are cut out from commercial 300 m (1000 ft) reels which impose
a strong initial curvature. Specimen end fittings consist of a steel tube swaged onto the core,
and an aluminum cylinder swaged on the aluminum layers and the steel tube.
In a conductor fatigue test, wire breaks often occur in the inner layers. It is important to be
able to record these wire breaks, firstly, to know when to stop a test, and secondly, to be able
to analyze the time sequence of the breaks. The principle of the system used to detect each wire
break has been described in [5]. It is based on the fact that a very minute rotation of the specimen occurs whenever a wire breaks in any one layer. This rotation is transformed into a linear
movement by two arms clipped midway between the clamp and each adjustable block. These
arms are connected to the cores of two linear variable differential transformer (LVDT)
transducers.
Test Parameters
For a given fatigue test, the primary parameters to be defined are the following: (a) the conductor type, (b) the kind of clamp, and (c) the tensile load applied on the specimen. Secondary
parameters may also have an influence on fatigue strength, such as conductor slope at the
mouth of a suspension clamp or pressure between keeper and conductor. Fatigue tests
reported in this work have been performed on a typical (aluminum conductor steel reinforced)
ACSR conductor--namely, the Bersfort 48/7 ACSR (48 aluminum wires around a 7 steel wire
core). This is a standard Canadian conductor consisting of three aluminum layers wrapped
around a seven wire steel core. A listing of its relevant physical characteristics is shown in Table
1. This ACSR conductor is very similar to certain American conductors such as the Dipper or
Martin types.
There are many types of clamps commercially available. In the present investigation, two
types of spacer clamps have been used. The first type is a so-called pre-twisted rod spacer clamp
developed by the Hydro-Qu6bec power utility (Fig. 3). It is an " X " used in four-conductor
235
TABLE 1--Main physical characteristics of Bersfort ACSR

conductor.
Overall diameter, mm
Number of wires per layer:
Steel core
Aluminum layer
Aluminum wire diameters, mm
Aluminum Young's modulus, GPa
Bending stiffness lower bound, N. m2
Rated Tensile Strength RTS, kN
35.56
1-6
10-16-22
4.27
69
65.04
180
bundles. Each arm is connected to a conductor via a set of four pre-twisted aluminum rods.
The holding end of the arm is padded with a rubber cushion. With such a system, one can
hardly speak of a clamp since little external pressure is applied on the conductor. The conductor specimen is mounted horizontally and the up and down amplitude is symmetric with
respect to the rectilinear initial position.
The second spacer clamp (Fig. 4) is actually a home-made clamp similar to the clamping
system found on several commercially available spacers. These spacers are actually bolted
onto the bundle conductors. This clamp is made of a steel bracket which can be bolted directly
onto the slider. Two 606 l-T6 aluminum pads are fitted in the part gripping the specimen. The
edge of these pads has a 3.2 m m (~ in.) radius. The keeper is bolted onto the clamp body
through four 12.7 m m (V~in.) diameter bolts. In all tests torque on these bolts has been set at
8.8 N.m. This value was calculated to give an average pressure of about 4 MPa on the con-
FIG. 3--Pre-twisted rod spacer clamp.

236
~.~
89~2
114.30
12.70dia.
1
dimensions in milimetre
FIG. 4--Bolted spacerclamp.
ductor, a value falling in the range of those found in the field when calculations are made with
the usual semi-empirical relation between bolt torque and tensile load. It is realized that this
method of clamping pressure control is not precise. Further tests are planned in which bolt
load will be directly monitored.
The average tensile load applied in each test is defined as a percentage of the RTS (rated
tensile strength) of the conductor. This RTS is a nominal value which is calculated according
to ASTM Standard on A l u m i n u m Conductors Concentric Lay-Stranded Coated Steel Reinforced (B 232) or its Canadian counterpart CSA Standard on Aluminum Conductors Steel
Reinforced (C-49). For the Bersfort ACSR conductor, the RTS is 180 kN (40 500 lbf) and all
tests have been performed at 25% of this value.
Tests have been stopped at the first occurrence of either 4 recorded breaks or 100 million
cycles. Test data have been analyzed in two ways: (1) the number of cycles to individual wire
breaks versus the imposed amplitude (the equivalent of the usual fatigue, or S/N, diagram);
and (2) wire break location within the conductor.
237
Test Data Analysis
Fatigue Diagram
Recording of the first four wire breaks in each test is plotted in Fig. 5. Bending amplitude is
shown in terms of both relative amplitude Yb and the stress amplitude, obtained from Yb
through the Poffenberger-Swart equation. For the Bersfort ACSR conductor, the relation is (o,
in MPa, Yb in ram):
era = 35.14 Yb at
25% RTS
Results from tests with both types of spacer clamps are shown in Fig. 5. Since the pre-tw]sted
rod clamp exerts very little pressure on the specimen, one may consider the corresponding
loading as almost pure bending. Indeed, the contact points between the twisted rods and the
conductor outer layer had almost no fretting marks and no fracture originated from these
points. As can be seen from Fig. 5, the pressure imposed by the bolted clamp on the conductor
produces a drastic decrease in fatigue strength for an equivalent bending amplitude.
Fracture Location Analysis

Once a test was completed, the central part of the specimen was cut out symmetrically, some
300 m m away from the last contact points. The wires were then removed one by one, layer by
layer, noting the location of each fracture, i.e., distance from axis of support and position in
the section with respect to the neutral bending axis. First, it should be mentioned that no partially cracked wire was observed in these tests. Figures 6 and 7 show the axial distribution for
each type of clamp. For the twisted rod spacer clamp (Fig. 6), broken wires seem to be ran-
,,
,I
ll Bersfort ACSR
SDacer clamos
Tension, 25% R T S
i,
35.
9 Bolted =pacer ckirn~

O Pre twisted rods sDacer c l a r r m
E
E
OO9
>-
-25.
o.6-
c6
r
.,.I
a.
UJ
>
0.4-
-~5. I-~)
W
nr 0 9
-5.
0-~
gjO7
106
CYCLES TO FRACTURE, N
FIG.
5--Fatiguediagram. Berslort-spacerclamps(25%RTS).
9108
W
~
<
238

2 5 "-t
CLAMP
BERSFORT ACSR
SPACER C L A M P
2 0 't
P R E - T W I S T E D RODS
I,- I
OI
Oi
i...i
n" 1 5 .
n,.,
LLI
re
zl
O I
o.I
pI
11.1
11.1
'r
11.1
11.1
'
(.9
< 10I-Z
LU
O
re
LLI
0.
0
40
20
20
40
60
80
AXIAL FRACTURE LOCATION, (mm)
FIG. 6--Axia/J~acture location. Bers/~rt-twisted rod spacer clamp.
domly distributed over a 120 m m region, inside and outside the clamp, with a slight percentage
increase at 40 m m inside the clamp from the last point of contact, which is the middle section
of the clamp. Here, fractures occurring on both sides of the clamp have been aggregated. However, due to the small number (9) of broken wires obtained with this clamp, even for tests at a
comparatively high amplitude, the significance of these percentages should probably only be
viewed qualitatively. As for the bolted spacer clamp, the fracture axial distribution is shown
in Fig. 7. It shows that most wire breaks occur within the clamp, generally in the last point of
contact region.
In order to locate fracture positions within a conductor cross section, wires have been numbered in each layer. Wire numbering is shown in Fig. 8. Fractures are due to the alternating
bending of the conductor and it is interesting to study if wire distance from the cross section
neutral axis is a significant parameter in break occurrence. Since reversed bending is involved,
no distinction is made between the upper and lower halves of the cross section. Thus, there
are a number of positions which can be considered as equivalent with respect to bending. For
example, positions 3 and 8 of the inner layer are equivalent. Accordingly, each break in these
positions will be multiplied by a factor of~. For equivalent positions 2, 4, 7, 9, the weighting
factor will be 88 etc. Thus, in this case the weighting factor takes one of two values, ~ or 88
Hence, multiplying each fracture by the corresponding weighting factor w, and adding, a total
equivalent number of fractures, N~q is obtained. For the bolted spacer clamp, N~q = 20.5. For
all equivalent cross sectional positions, the weighted number of breaks can be expressed as a
percentage of the total equivalent number of fractures (100N, w,/Neq). Each equivalent position
in layer n of radius Rn is at a distance d, from the conductor neutral axis. This distance is nondimensionalized as di/R,. A sample calculation is shown in Table 2. Results for both spacer
239
CLAMP
20
]
fl
1I-.-~
[ BERSFORTACSR
[ SPACERCLAMP
/
BOLTED
19,2
I-1 t~
Io
io
16.4
Wm~1 5 "
IJ
U.I
m
Z
s.2
ee
Q.
55
5
2.8
0-
40
20
20
40
60
80
AXIAL FRACTURE LOCATION, (mm)
FIG. 7--Axial fracture location. Ber4"[ort-boltedspacer clamp.
UPPER
HALF-SECTION
---
8~iTe B
2.
NEUTRAL
AXIS
1
ALTERNATING
AMPLITUDE
LOWER,
HALF-SECTION
FIG. 8 - - Wire numbering. Bersfort cross-section.
clamps are shown in Figs. 9 and 10. These figures show the percentage of equivalent number
of fractures in each layer and for each equivalent position in that layer. This percentage can
be viewed as a fatigue sensitivity index for each one of these positions. The bar charts show
that, in general, the inner and intermediate layers conform more closely to beam bending theory, where fibers furthest from the neutral axis are most subject to fracture. On the contrary,
fractures in the external layer seem to depend less on the distance parameter d,/R,. Thus, the
external layer tends to behave more like a collection of independent wires.
240
TABLE 2--1ndividual wire fatigue sensitivity within conductor cross section. Bersfort-bolted spacer
clamp.
Failures
Number
Failures
Weighted
Number
Failures
Weighted %
Normalized
Distance
(IOON~w,/Neq)
(d~/R,)
Layer
Wire No.
Weight
Factor
(n)
(i)
(wi)
(N,)
(N, wi)
88
88
~
6
20
11
1.5
5
5.5
7.3
24.4
26.8
0.309
0.809
1.0
16
15
14
13
88
88
88
88
5
3
3
10
1.25
0.75
0.75
2.5
6.1
3.7
3.7
12.2
0.195
0.556
0.832
0.981
11 12 22
10 13 21
9 14 20
8 15 19
7 16 18
617
88
88
88
88
88
89
2
2
1
2
6
0
0.50
0.50
0.25
0.50
1.50
0.
2.4
2.4
1.2
2.4
7.3
0.
0.142
0.415
0.655
0.841
0.960
1.
1 5 6 10
2479
38
18
27
36
45
1
2
3
4
5
9
10
11
12
INNER
INTERMEDIATE
LAYER
LAYER
OUTER
LAYER
60
>_
co
z
ill
~9 4 0
iii
UJ
~_ 2 o
1S.4
_.1
W
7Y
.5
7.7
.5
.5
NORMALIZED DISTANCE, (di/Rn)
FIG. 9--Fatigue sensitivity within conductor cross section. Bers[ort-twisted rod spacer clamp.
T o characterize the relative fatigue sensitivity o f each layer, wires are separated into three
classes with corresponding weighting factors: ~0, ~,~, and ~ . T h e results presented in Table 2
clearly show that fatigue sensitivity is m u c h greater in the inner layers than in the outer one.
Since specimens have been tested in alternate bending, one would expect an approximately
s y m m e t r i c fracture distribution with respect to the c o n d u c t o r neutral axis. This is not what
has been found experimentally. The upper-lower half section fracture distribution is also
shown in Table 3. While in tests with suspension clamps almost all fractures occurred above
241
FIG. l O--Fatigue sensitivity within conductor cross section. Bersfort-bolted spacer clamp.
TABLE 3--Global fatigue sensitivity analysis.
Layer
Inner
Intermediate
Outer
Half-Section
Upper
Lower
Contact
Inner
Outer
Lateral
Bersfort
Twisted Rod
Spacer Clamp
Bersfort
Bolted Spacer
Clamp
86.4%
7.4
6.2
66.1%
23.2
10.7
78.0%
22.0
68.0%
32.0
11.1%
77.8
11.1
8.1%
82.4
9.5
the conductor neutral axis [5], a majority of them are found to occur below the neutral axis
in the Bersfort-spacer clamp specimens. The spacer clamp results are tentatively explained in
the following fashion. Test specimens come with an initial curvature. In suspension clamp
tests this curvature makes the specimen conform more closely to the average or static loading
configuration than in the case of spacer clamp tests where the specimen has to be straightened
out. Indeed, this straightening superimposes a static tensile stress in the lower half section and
a static compressive stress in the upper half (specimens have always been mounted with their
natural concave side down). Obviously, each clamp design results in a distinct pressure distribution pattern between conductor and clamp as follows: (1) one-sided at the edge of a suspension clamp; (2) two-sided and spread-out in the pre-twisted rod clamp, and (3) two-sided and
concentrated in the bolted clamp. This distinct pressure distribution probably explains much
of the observed differences in fracture distribution.
242
Finally, for each broken wire, the probable point of crack initiation on the wire surface has
been noted. At a given section, there are four possible contact points for internal wires, and
three for external wires, except in clamp contact regions: inner, outer and lateral contact. In
most cases (95%) a fretting mark, at times very small, could be associated with a cracked section. This shows that electrical conductor fatigue is clearly a fretting fatigue problem. Table 3
also shows the distribution of crack initiating fretting marks. It can be seen that most of them
correspond to an outer contact (contact with an outer layer). Lateral contact initiation was
almost always in the outer layer at wire sections near the conductor neutral axis. Outer contact
initiation was also noted in the outer layer. It was always at wire sections most remote from
the neutral axis and inside the clamp.
Conclusions
Overhead electrical conductor fatigue is basically a fretting fatigue phenomenon in which
the interaction of the various contacting solids needs to be better understood in order to
improve both transmission line design and inspection. In this work a fatigue testing procedure
has been described in which two types of data are collected. Firstly, the wire fracture time
sequence allows an evaluation of a given conductor-clamp combination bending fatigue
strength. Secondly, a fracture location analysis provides information on the critical wire sections. This procedure has been applied to two ACSR spacer-clamp combinations. These
results show the crucial influence of the spacer clamp design on conductor fatigue strength.
Although conductor damage in bending fatigue is widely recognized as one of fretting
fatigue, it does not yet seem possible to relate fatigue data obtained on full-size conductors with
single wire fretting fatigue behavior. Indeed, in order to establish such a relationship, one needs
a theoretical conductor bending model allowing for a reliable prediction of stick-and-slip conditions, in particular, in clamp regions. Although some simple models have already been proposed, much progress remains to be accomplished in this respect. Thus, it seems that extrapolation of fatigue results from one type of conductor to a quite different one and from one
conductor-clamp combination to another, although often made in practice, should be viewed
with caution.
Acknowledgments
This paper is the result of work supported by the Hydro-Qurbec power utility and by Canada's Natural Sciences and Engineering Research Council, Grant No. A8905.
References
[1] Rawlins, C. B., "Fatigue of Overhead Conductors," Transmission Line Reference Book, Electric
Power Research Institute, Palo Alto, Calif., 1979, pp. 51-80.
[2] Bfickner, W., "Summary of results of endurance tests on transmission line conductors," internal
report CIGRE SC 22 WG 11, International Conference on Large High Tension Electric Systems,
Paris, France, 25 Sept. 1989.
[3] Poffenberger, J. C. and Swart, R. L., "Differential Displacement and Dynamic Conductor Strain,"
IEEE Transactions, Vol. PAS-84, 1965, pp. 281-289.
[4] Lanteigne, J., Cloutier, L., and Cardou, A., "Fatigue Life of Aluminum Wires in All-Aluminum and
ACSR Conductors," CEA Report No. 131T241, Canadian Electrical Association, Montrral, Canada,
July 1986.
[5] Cardou, A., Cloutier, L., Lanteigne, J. and M'Boup, P., "Fatigue Strength Characterization of ACSR
Electrical Conductors at Suspension Clamps," ElectricPowerSystems Research, Vol. 19, No. 1, 1990,
pp. 61-71.
Olaf Jacobs, 1 Karl Schulte, 2 and Klaus Friedrich 3
Fretting Fatigue of Carbon Fiber-Reinforced

Epoxy Laminates
REFERENCE: Jacobs, O., Schulte, K., and Friedrich, K., "Fretting Fatigue of Carbon FiberReinforced Epoxy Laminates," Standardization of Fretting Fatigue Test Methods and Equipment, ASTMSTP 1159, M. Helmi Attia and R. B. Waterhouse, Eds., American Society for Testing and Materials, Philadelphia, 1992, pp. 243-260.
ABSTRACT: Continuous carbon fiber reinforced epoxy resin laminates were exposed to a tension-tension fatigue loading (R-ratio R = 0.1). Simultaneously, cylindrical metallic pins with
flat, polished front surfaces were pressed with an apparent contact pressure between 10 and 23
MPa onto two opposite sides of the specimen. This paper studies the mechanisms of damage
development of carbon fiber reinforced laminates under such a fretting fatigue situation and proposes a quantitative measure for the degree of fretting damage. Fatigue life of the composite
could drastically be reduced by an additional fretting component, if load bearing 0~layers were
damaged by fretting. Fretting fatigue damage was found to penetrate proportionally with time
into the bulk material; any action of notch effects was not found. However, cracking and delamination of fiber layers pre-damaged by fretting can cause synergistic interaction between fretting
and fatigue loading. The influence of the loading conditions is investigated.
KEY WORDS: polymer composites, carbon fibers, fretting wear, fatigue, fretting fatigue, delamination, stress redistribution
Continuous fiber reinforced polymer composites are used in an increasing number of engineering applications, mainly for reasons of weight reduction, high resistance against fatigue,
and good vibration damping properties. In many of these applications, an additional fretting
load damages the material's surface, e.g., at joints with other parts, causing a reduction in
fatigue performance. However, when compared with metals [1,2], there exist only few studies
on fretting wear of polymer composites [3] and in particular of carbon fiber reinforced epoxy
resin (CF/EP) laminates [4,5]. Fretting fatigue behavior of this group of materials is still an
open field.
A pilot study in the framework of this project [6, 7] indicated that additional fretting surface
damage may reduce fatigue life of CF/EP laminates if load bearing 0~
are exposed to
fretting. Off-axis plies contribute only little to the load bearing capacity of a laminate, so that
their damage due to fretting affects the material's fatigue properties insignificantly(Fig. 1).
The current project systematically investigates the influence of several system components
on the fretting wear and fretting fatigue performance of CF/EP laminates. For this purpose, a
fretting fatigue testing device has been specially designed.
Junior Researcher, Polymer & Composites Group, Technical University Hamburg-Harburg, Harburger Schloflstra/3e20, 2100 Hamburg 90, FRG. New affiliation since December 1990: Materials and
Processes Development (Dpt. EV 31), Deutsche Airbus GmbH, Htinefeldstrafie 1-5, 2800 Bremen 1.
2 Senior Researcher, Institute for Materials Research, DLR, Linder Hrhe, 5000 Krln 90, FRG.
Currently: Professor, Polymer and Composites Group, Technical University Hamburg-Harburg,
Denickestrasse 15, 2100 Hamburg 90, FRG.
3 Professor and Head of the Materials Science Group, Institute for Composite Materials, University of
Kaiserslautern, Erwin-Schrtklinger-Stra~3e,6750 Kaiserslautern, FRG.
243
Copyright
by ASTM
(all rights
reserved);
Mon Mar 28 00:05:25 EDT
2016
Copyright
9 Int'l
1992by
ASTM
lntcrnational
www.astm.org
244

~'
I1.
3E
250
u)
9
r/)
2oo
u.
Q.
9
o
150
Fretting Fatigue
Plain Fatigue
I
03
104
10 s
10 6
L o a d C y c l e s to Failure, N
FIG. 1--Fatigue lffb versus m a x i m u m stress in a .fatigue cycle for plain fatigue and fretting.&tigue.
Specimen. CF/EP, stacking sequence. [ +_45 ~ 0 ~ (+_ 45~ 90 ~ +_45~
Counterpart. aluminum pin
(diameter d = 5 mm) under a normal load FN = 450 N [6].
Materials
The samples consisted of an epoxy matrix (EP) reinforced with aligned continuous carbon
fibers (CF). The fiber volume content amounted to approximately 60%. Table 1 lists the two
laminates (L 1 and L2) tested. The covering layers possessed 0~
This was required
because fretting damage of off-axis plies only insignificantly affects the fatigue performance of
CF laminates [6].
Most of the laminates were manufactured using a peel ply. This is a layer of an open-weave
material separating the laminate from the bleeder cloth, which picks up excessive matrix resin
during curing of the laminate. After curing, the peel ply is removed, but it leaves a resin rich
layer with the impression of the fabric on the surface of the sample.
For fatigue and fretting fatigue tests, the laminates were cut with a diamond saw to rectangular beam shaped test pieces (free length g = 280 mm, thickness of L1 te~ = 2 mm, tL2 = 6.3
mm, width for L1 specimens WLI = 8.3 mm, WL2 = 6.3 mm) that were clamped to the servohydraulic test machine by grips. Aluminum end tabs enabled easy load transfer and protected
the specimen surface against the action of the grips.
Cylindrical metal pins (diameter = 5 mm) with flat front surfaces were pressed against the
two opposite width sides of the specimen, thus serving as fretting pins. The front surfaces were
ground and polished to a final surface roughness of approximately 0.12 ~m. The edges of the
front surfaces were rounded off in order to avoid cutting of the laminate by sharp edges. Before
starting the tests, the pins were cleaned with acetone. A low carbon NiCr steel (Vickers hard-
TABLE l--Materials tested under fretting fatigue load.

Laminate
Matrix
Fiber
LI
BASF
R 5212
Ciba Geigy
914C
T 300
[0~ 900212s
850 MPa
T 300
([0~ -+45 ~ 002, -+45 ~ 90~
799 MPa
L2
Stacking Sequence
Strength
JACOBS ET AL. ON CARBON FIBER-REINFORCEDEPOXY LAMINATES
245
ness HV = 296), an aluminum alloy (HV = 135), and a brass (HV = 160) were chosen as
pin materials.
Fretting Fatigue Test Device

Fretting fatigue studies are usually performed in order to explore how an additional fretting
load may affect the fatigue performance of a material under certain fatigue loading conditions.
Fretting fatigue testing should be carried out in the same servohydraulic test machine as that
for plain fatigue experiments.
Derived from a fretting fatigue test device used by Gaul and Duquette [8], for this inyestigation a system was developed which can be attached to the grips of the servohydraulic test
machine [ 7]. The system consisted of a positioning device (Fig. 2) for cylindrical fretting pins
and a fretting load frame (Fig. 3). The positioning device was fixed to the upper grips. This
device consisted mainly of a steel block into which two mutually perpendicular grooves had
been machined, to which two plates opposite to each other were fixed. The plates could be
positioned with their central holes either parallel or perpendicular to the length axis of the
gripping system. Each plate contained five central holes through which two fretting pins could
symmetrically act on the two opposite sides of the specimen. Thus, by changing the gage
length, five different slip amplitudes could be selected. Besides the holes for the fretting pins,
the plates contained two parallel rows of holes for positioning the fretting load frame.
The fretting load frame allowed symmetrical loading, exact adjustment, and control of the
fretting load. The heart of this system was a small load cell on one side and a screw-adjustable
coil spring on the other side. Depending on the stiffness of the coil spring, the fretting load
could self-reduce more or less quickly with material removal at the fretting contact points. As
the load was permanently controlled by the load cell, readjustment could be made from time
to time in order to maintain a constant contact pressure during the test.
Figure 4 is a photograph of the whole system. The upper grips were connected to the load
cell of the servo-hydraulic test machine, while the lower grips were attached to the actuator
applying the fatigue load. Fretting pins were pressed against the specimen. The cyclic straining
of the laminate caused a relative motion between fretting pins and specimen.
Test Conditions
For this investigation, tension-tension fatigue load was applied with a frequency of 10 s-~ at
an R-ratio o f R - ~/~, - 0.1 (au = upper and ~ = minimum fatigue stress in each consec-
Plates
Steel Block
/
~
~
/
Load Frame in Position
FrettingPin
Piontdf~H~
trh~tlFrgtllng
FIG. 2--Positioning device/brfrening pins.

246
18
12 13
16
/4
19
18
11
.lo
2
/ /
I/,
17
15
I0
F I G . 3--Assembly of the fretting load frame: ( 1) = cross section through the positioning device, ( 2 - 6 )
= tension plates, (7) = guide rod, (8) = fretting pins, (9) = positioning pins (D in Fig. 2), (10) = load
cell, (16) = coil spring, (17) = stabilization bolt, ( 18 a n d 19) = ball bearings.
F I G . 4--Assembly of the complete fretting fatigue device: (B)--positioning device, (C)--~lkettingJatigue

load frame with coil spring ( D ) and load cell (E), (F)--test sample, (G)--zfretting pins.
J A C O B S ET AL. ON C A R B O N F I B E R - R E I N F O R C E D E P O X Y L A M I N A T E S
247
utive load cycle). The relative slip, A, between the pins and the laminate can be estimated
according to
z21o-
A = --"
~T" s
(1)
(7 T
where Act = stress amplitude, crT = ultimate tensile strength ( = 850 MPa), ev = strain to
failure (1.2%), and g~ = distance of the fretting pins from the upper, fixed clamping grip.
For a stress amplitude of 700 MPa (upper fatigue stress au = 780 MPa), Eq 1 gives a slip of
650 um. A measurement with a mechanical displacement transducer resulted in a value of
about 700 urn, which is slightly higher than the calculated value because the laminate undergoes a small strain motion within the clamping grips.
Results and Discussion
Experimental Results
Mechanical Tests--Figures 5 and 6 present plots of the upper fatigue load applied versus
the resulting lifetime for fatigue and fretting fatigue conditions. In a semi-logarithmic scale,
the plain fatigue curves can be approached by a linear function [ 9]:
au = ~T" (1-m
log N)
(2)
where m = experimental parameter.

In the present case, m ~ 0.02 for Laminate L1 and 0.06 for L2, respectively. Obviously, L2
reacts much more sensitively to a change of the fatigue load than L1.
Application of an additional fretting load leads to a deviation of the ~r-log N-curve from the
simple logarithmic rule of Eq 1. This deviation may strongly depend on the particular loading
conditions. The hard steel pins as counterparts did not produce any significant fretting fatigue
effect up to a contact pressure of 23 MPa (FN = 450 N). When the fretting pins consisted of
aluminum, the fretting fatigue life of the laminate strongly depended on the contact pressure.
Thiswas also found for brass pins.
ft.
800
~ ~qM" ~ ,
1 ,~,~ ~,,&_
700
._~
600
14.
I~.
D.
9
u
o
' 9
500
10 =
Plain Fatigue
AI, F N = 2 0 0 N
AI, FN = 300 N
AI, FN = 4 0 0 N
Steel FN = 450 N
103
104
-'1~
"1
10 s
10 s
107
Load Cycles to Failure, N

FIG. 5--
Upperfatigue load versus l~,time Jbr plain.[i~tigue andfrettingjatigue (against aluminum pins)
of Laminate L1.
248
FRETTING FATIGUE TEST M E T H O D S AND EQUIPMENT
700
:S
v)
600
4)
500
1,1.
Q.
Q=
Plain Fatigue I
--T
400
1 03
10 4
1 0s
1 0s

FIG. 6--Upper fatigue load versus l~fetime.[brplain fatigue and fretting fatigue (against aluminum pins)
~f Laminate L2.
Examination of Damage Mechanisms--The development of fretting fatigue damage in
homogeneous, isotropic materials can be described by means of fracture mechanics concepts
[ 10,11 ]. In the area of the fretting contact, local stress peaks due to friction and normal forces
initiate cracks. These cracks act as sharp notches and lead to a high concentration of the fatigue
stress at their tips. As a result, the cracks propagate into the bulk and, thus, cause premature
failure.
In continuous fiber reinforced composites, the situation is considerably different. The simple application of fracture mechanics is not possible because cracks do not always propagate
perpendicularly to the main loading direction but advance preferentially parallel to the fibers
[ 12]. Neither the initiation and accelerated growth of single cracks, as in homogeneous, isotropic materials, nor morphological changes characterize failure mechanisms in laminated
composites under fatigue loading [13]. Instead, multiple matrix cracking along the fibers
causes a reduced load carrying capacity of the off-axis plies and accordingly enhanced stresses
in the 0~
Starting from intersections of matrix cracks and from edges, delamination
between the differently oriented plies develop [14,15]. Subsequently, the stress in the off-axis
plies is reduced while the 0~
have to carry an increasing part of the applied load. Final
failure occurs when the stress in the 0~
locally exceeds their strength [15], which may
additionally be reduced by randomly cracking of 0~
[16]. Figure 7 schematically illustrates the stress situation in a 0~
of a cross-ply laminate containing transverse cracks in
the 90~
FIG. 7--Schematic of the stress distribution in a O~

~?[neighboring 90~
c~[/ercracking (C) and local delamination (D)
JACOBS ET AL. ON CARBON FIBER-REINFORCED EPOXY LAMINATES
249
Laminate L2 contains a large number of 45~

which have a higher load carrying capacity than the 90~
in LI. Therefore, it is reasonable that the residual strength after a given
number of load cycles is more reduced in Laminate L2 than in L 1.
The effect of an additional fretting component on the fatigue damage mechanisms of
CF/EP laminates is described in detail in a previous publication [17]. Pressing the pins against
the specimen hinders the formation ofdelamination in the interior of the laminate. However,
plies pre-damaged by fretting tend to crack and peel offwhen the specimen is exposed to static
or dynamic tensile loading (Fig. 8a).
The cracking of 0~
could be caused by high equivalent stresses in the region of the
fretting contact. But this explanation seems not to be very probable, because the applied apparent contact pressures ( < 2 0 MPa) are very small in comparison to the fatigue stress level
( > 7 0 0 MPa). In fact, when a contact pressure of 22 MPa is applied via steel pins instead of
aluminum pins, no cracking of the 0~
was found (Fig. 8b). Obviously, the effect of the
normal load (in the range investigated), with which the pins are pressed against the specimens,
on the equivalent stress is not the crucial parameter which controls the development of fretting
fatigue damage.
Another assumption considers that the initiation and advance offrettingJatigue damage is
controlled by the fretting wear performance of the laminate. This explanation is supported by
the fact that fretting wear of CF/EP versus aluminum is more severe than against steel counterparts [5]. Figure 9 presents a scanning electron micrograph of a CF/EP surface subjected to
plain fretting wear versus aluminum. Besides polishing of fibers and matrix, cracking and
removal of broken fibers are visible. From the area denoted by the letter D, a chip of about 100
um width is delaminated. This very severe type of wear mechanism occurs only if the loading
parameters (pressure, amplitude, frequency) exceed characteristic critical values, which mutually depend on each other. This delamination wear was not observed for steel counterparts up
to an apparent contact pressure of about 40 MPa [18].
FIG. 8--Optical m icrographs of specimens (Laminate L 1) a~er frettingfatigue. (a) Versus an aluminum
(b) Versus a steel pin (Fu = 450 N, 270 000 load cycles).
pin (FN = 400 N, 30 000 load cycles).
250
FIG. 9--SEM micrograph (?[a CF/EP surface subjected tofretting wear against aluminum. FN = 400
N (p = 20 MPa), peak-to-peak oscillation width A = 700 urn, 110 000 load r
Once the fibers or fiber bundles are broken, they can no longer support the load carrying
capacity of the laminate, although they are actually not yet worn. Therefore it must be
expected that the fretting fatigue damage proceeds considerably faster than the plain fretting
weargravimetrically measured. When the specimen shown in Fig. 9 is simultaneously exposed
to a tensile fatigue load, shear stresses arise along the interface between the cracked and the
undamaged fiber bundles which are enhanced by the friction force (Fig. 10). This leads to the
observed delamination and peeling offofthe cracked 0~
(Fig. 8a).
Discussion
Mathematical M o d e l - - T h e observations of the fretting fatigue failure mechanisms suggest:
1. The absence of notch effects causes a relatively uniform distribution of tensile stresses
across any cross section of the undamaged 0~
Therefore, the change of the load carrying
capacity, dF, should be proportional to the reduction of the cross section dQ of the load bearing 0~
in the fretting contact region between N and N + dN load cycles:
d F = ~ru" dQ
(3)
where au = upper fatigue load at which the specimen fails after N load cycles.
2. Damage development can be considered to proceed proportionally to the time:
dQ
= const
(4)
The reduction of the cross section of the O~

in the fretted region leads to an enhanced
fatigue stress in the remaining O~
The specimen finally fails when this stress exceeds the
FIG. lO--Schematic illustration of de/amination of cracked surface layers. The advance ~f delaminaCopyright
by ASTM
Int'l (all
reserved);
Mon Mar 28 00:05:25
EDT 2016
lion
isforced
byrights
crack
openingmode
I and H.
J A C O B S ET.AL. ON C A R B O N FIBER-REINFORCED EPOXY L A M I N A T E S
251
strength of the 0~
The missing of a fretting fatigue effect when the contact pressure was
applied via steel pins rejected a significant influence of the stress concentration at the crack tip
in accordance with the non-applicability of fracture mechanics to continuous fiber composites
[ 12]. However, the actual cross section is not continuously monitored and, thus, the true stress
in the laminate between the fretting pins cannot be determined. But the apparent upper fatigue
stress ~,, at which the specimen fails after N load cycles (failure load related to the initial cross
section), diminishes proportionally with the cross section of the load bearing 0~
Accordingly, a, can be calculated by modifying Eq 2 as follows:
( 1 + Q 0dQ
)
~ 7 ~ . N 9 (1 -
c~o = ~r"
m. logN)
(5)
where Q0 = initial total cross section of the 0~

Inserting Eq 5 into Eq 3 and integrating the resulting differential equation [19] leads to a
new parameter Ao'rel~
av
~ Q~ - 0.5 9
(6)
where
c~v = upper fatigue load at which the specimen fails after N load cycles, and
avv = upper fretting fatigue load at which the specimen fails after N cycles.
Presuming the validity of Eq 4, Eq 6 suggests that, in a first-order approximation, the relative

fatigue strength reduction,/Xar~, increases proportionally with time. To check this assumption,
Fig. 11 presents a plot of A ~ versus fretting fatigue life, N, for two different laminates. Ini30
i
a) Laminate L1
~9
20
- ~'~
o
9
<3
AI, F N = 300 N
AI, FN 400 N
=
,,
10
530000
~
~ooooo
200000
o~ I~
IO
<]
/
0
b) Laminate L2
Aluminium Pins
FN = 450 N
100000
200000
300000
400000
FIG. 1l--Relativejiztiguestrength reduction ( zXcrrej)as a /imction ~?/fretling latigue If/e Jot Laminate L 1
Copyright
ASTM
(alla!
rights
reserved);
Mon Mar
28 00:05:25 EDT 2016
(a)byand
L2Int'l(b)
several
conlacl
/oadf.
252
tially, all curves increase proportionally with time but reach a constant value after a certain
n u m b e r of load cycles. M a x i m u m fretting fatigue damage is expected when the covering 0%
layers are totally penetrated by the fretting pins. Further fretting only damages off-axis plies
and influences the fatigue performance of the laminate insignificantly [6]. Figure 12 presents
an optical micrograph of a specimen of Laminate L2 which failed after 345 900 load cycles
(i.e., within the plateau region of the dxr162- N curve). The covering 0~
were totally penetrated a n d fretting already took place on the 45~
Based on these considerations, the
m a x i m u m value for/~r
can be calculated from the m a x i m u m reduction of the cross section
of the 0~
(Fig. 13):
d iF,m~x
AQmax = -- - - W
/tot
(7)
where
d =
w =
iF.... =
/tot =
diameter of the pin,

width of the specimen ( = 6 . 3 m m for L2 and 8.4 m m for E l )
m a x i m u m n u m b e r of fretted 0~
( = 4), and
total n u m b e r of = 0~
FIG. 12--Optical micrograph of a specimen of Laminate L2, sut~jected to j?etting.&tigue against an

aluminum pin. FN = 450 N,.failure after 345 900 load cycles.
FIG. 13--Cross section of the specimen in the region ~fretting contacl.

253
Accordingly, m a x i m u m fretting fatigue damage can be calculated by Eq 8. It is a simple

geometric quantity depending only on the stacking sequence of the laminate and on the contact geometry:
A 0"tel . . . .
AOmax
Q~-
0.5
' (AQmaxt2
(8)
Q0 J
= [0.25 for L1
[ 0.125 for L2
These values are in good agreement with the experimental results when consideringaluminum pins at a contact pressure o f p = 23 MPa (FN = 400 N, Fig. 11). However, at lower
contact pressures, the m a x i m u m fretting fatigue damage as found experimentally is considerably smaller than calculated. This effect will be presented and discussed later.
From the linear part of the Acrre~-- N curve, the rate of damage development dQ/dNcan be
derived. For reasons of comparability with the fretting wear tests, this quantity will be transformed to a "specific pseudo-wear rate" vi,s*in analogy to the concept of the specific wear rate
Ws:
AV
w s * - 2 . FN" L -
~r
d " Q0
16" F N ' A 9 N
(9)
where
A V = volume of removed or cracked 0~
L = total sliding distance ( = 2 9 A 9 N), and
FN = normal load.
Ws*
~r
16
--
d . Q0
FN.A.N
"
{1
--
'~1
--
"
AO'rel}
(10)
The quantity vOs*measures how deep fretting fatigue damage penetrates into the laminate
and includes material removal due to pure fretting wear as well as cracking of pre-damaged
fibers and fiber bundles. Therefore, v~'s*should be greater than vi,~.
Experimental Proof of the Model In order to avoid confusion, it should be emphasized
that the measured points along the A~rre~-- N curve represent different specimens which were
subjected to different fatigue stress amplitudes. The fact that these points follow a straight line,
primarily, means that the fretting fatigue damage accumulated until final failure is proportional to the fretting fatigue life, regardless of the fatigue stress level. On the other hand, the
presented model states that during the test of one particular specimen the effective cross section of the load bearing 0~
when exposed to fretting, decreases proportionally with time.
Therefore, an additional test was necessary to check this assumption of the model.
Several specimens of Laminate L1 were exposed to a fretting fatigue loading equivalent to
one half of the expected fretting fatigue life. The contact load was set to FN = 300 N. The upper
fatigue stress was ~, = 700 MPa. From Fig. 4, a lifetime of 79 000 load cycles can be expected
under these conditions. After 45 000 load cycles, the test was interrupted according to a theCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
254
oretical fatigue strength reduction of 4.5% (Fig. 11). Subsequently, the specimens continued
to be tested under plain fatigue conditions (~ru = 740 MPa) until final failure. Table 2 lists the
resulting lives and the according relative deviation from the a versus the log N curve for plain
fatigue of the undamaged laminate (Aare0. The measured mean value of Aar,~amounts to 4.5%,
which is in good agreement with the theoretical value calculated under the assumption that
the fretting fatigue damage proceeds proportionally with time during one single test.
Effect o f Contact Pressure and Counterpart Material--Figure 14a shows the specific pseudowear rate ws* calculated from the linear part of the Aa~e~versus N curve of Laminate L 1 as a
function of the contact pressure p. Obviously, there exists a critical value where the propagation of damage development j u m p s from an insignificant to a high level (boundary level). The
specific fretting wear rate w~ follows similar behavior (Fig. 14b) [20].
For aluminum pins, w~* is smaller than 8 - 1 0 - 6 mm3/Nm at a contact load of 10 MPa (FN
= 200 N) and, thus, has a magnitude similar to the specific fretting wear rate (6 9 10 -6 mm3/
Nm). At 15 MPa, still below the boundary value of the contact pressure, w~* has already
increased considerably to a value of about 3 9 10 -s mm3/Nm, while the specific fretting wear
rate, w~, still remained constant below its boundary value.
Conclusively, one can say that fretting fatigue damage proceeds as fast as material removal
due to plain fretting wear below a critical contact pressure (Table 3). Above this critical contact
pressure, both fretting wear and fretting fatigue damage become accelerated. But under fretting
fatigue conditions, this boundary pressure lies at considerably lower values (p~n, ~ 17 MPa)
than under plain fretting (P~nt ~ 30 MPa for aluminum pins). Above the boundary value of
the contact pressure, fretting fatigue damage proceeds much faster than the pure material
removal due to plain fretting wear:
ws*(P ~ 20 MPa) = 5 9 10 4 mm3/Nm
ws(P ~ 45 MPa) = 2.8 9 10 -5 mm3/Nm
(11)
A similar behavior of the specific pseudo-wear rate as a function of contact pressure can be
found for the brass pins. However, the boundary value of the pressure is shifted to lower values,
if compared to aluminum. This correlates with the fretting wear of CF/EP, which was found
to be higher for brass than for aluminum counterparts [20]. When steel pins were used as counterparts, no stepwise increase in the specific pseudo-wear rate was observed up to a contact
pressure of 23 MPa (FN = 450 N) according to the low fretting wear rate of CF/EP against
steel pins (Table 4).
TABLE 2--Residual fatigue Ire and relative fatigue strength

reduction qf CF/EP laminates (o, = 740 .MPa) pre-damaged by
45 000 load cycles under fretting fatigue (FN = 300 IV, ou = 700
MPa) conditions.
No.
Number of Fatigue
Cycles to Failure
AO~l(%)
Ao~z,me~.
(%)
1
2
3
7320
29 970
15 630
5.1
3.8
4.5
4.5
6O a) FrettingFatigue
9 Aluminium
o Brass
40" 9 Steel
=,,,"r
O
I-
20
:~
,"
30
255
lO
i?
i
J
2o
lo
C o n t a c t Pressure [MPa]
4
~'
'1 b) FrettingWear I
3" AluminiumPin i
~'o
~t
_I
lJ
i
10
20
30
40
50
C o n t a c t Pressure [ M P a ]
FIG. 14--Specific pseudo-wear rate, w,*, (a) and specific fretting wear rate, fVs (b) of Laminate LI as a
.[imction of contact pressure for different counterpart materials.
TABLE 3--Fretting fatigue life ~f Laminate L1 (number of

cycles to failure).[or difJ+,rent pin positions gl and fatigue stress
levels a,. Counterpart. aluminum, FN = 400 N.
Ou
s = 40 mm
s = 65 mm
s = 90 mm
650
650
700
700
16 600
14 500
6800
5400
15 000
12000
6800
6400
18 000
- 99
6500
5800
A probable reason for the small influence of the hard steel pins on the fatigue life of Laminate L1 is their high resistance against abrasion by fiber debris. The surface of the steel pins
remains rather smooth during fretting and, in turn, acts only slightly abrasively to the sample
material. A l u m i n u m pins, however, can easily be roughened by the fiber debris but the arising
asperities are rather soft and, thus, less abrasive than in the case of brass. A significant influence
of electrical contact corrosion seems to be unambiguous, because a similar trend of the effect
of counterpart material was observed for glass fiber (GF/EP) composites. Table 4 compares
the specific fretting wear rates of CF/EP and GF/EP worn against steel and aluminum, respectively. Fretted against steel, CF/EP and GF/EP exhibit similar wear rates. Changing to aluCopyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016
256
TABLE 4--Relative specific fretting wear rates for several

sample~counterpart combinations normalized to that of CF/EP
versus steel [20]. Matrix: R 5212, CF = T 300, GF = E-glass
fibers. The fibers were aligned parallel to the sliding direction. Fu
= 300 N, peak-to-peak oscillation width = 500 i~m, number of
load cycles = 72 000.
CF/EP
GF/EP
Steel
Aluminum
1
1.3
1.8
26.7
m i n u m pins, the wear of G F / E P is much more accelerated than that of CF/EP. This cannot
be explained by contact corrosion which is not effective in the case of glass fibers. But the glass
fiber particules act more abrasively and, thus, more effectively in roughening the counterpart.
Another effect of the contact load can be seen in Fig. 1 la. At a contact pressure o f p = 15
MPa (FN = 300 N), the relative fatigue strength reduction, 2xa~, becomes constant after about
150 000 load cycles at a value of 2Xaro~= 14% and does not reach the theoretical value of 25%
as it was found f o r p = 23 MPa (FN = 450 N).
Several authors [21-23] reported that also in metallic materials the influence of an additional fretting component on material fatigue is effective only during crack initiation and early
stage of crack growth. After a certain number of load cycles, the crack propagation rate under
fretting fatigue approaches that under plain fatigue [24]. This can be explained as follows. The
initial value for the stress concentration at the crack tip is enhanced by normal and shear
stresses applied by the fretting pin. While the crack propagates into the interior, this stress concentration diminishes to that value which is also active under plain fatigue conditions [25].
This crack propagation behavior would be in agreement with the fact that A~rrelremains constant after a certain number of load cycles.
However, the absence of notch effects in continuous fiber reinforced layers with 0 ~ orientation opposes the application of this model to the CF/EP laminates investigated here. In fact,
it was observed that specimens exposed to fatigue stresses above 640 MPa (contact load FN =
300 N) exhibited a more pronounced surface damage after a given number of cycles than specimens which were fatigued at a lower stress level. Especially, no peeling off of fiber bundles
occurred at fatigue stress levels below 640 MPa, even after some hundred thousand load cycles
(Fig. 15a). These observations suggest that exceeding a special number of load cycles is not
responsible for the lowering of the propagation rate of fretting fatigue damage but rather the
reduction of the fatigue stress level below a critical value (Cru.cn,= 640 MPa). In this case, probably, tensile stresses in the fretting region and shear stresses between cracked and undamaged
layers are not high enough to cause amplification of the surface damage due to fretting. In this
context, it should be remembered that the several points along the Aa,el curve (Fig. 11 ) represent differently loaded specimens. The damage progress in the specimens which failed in the
plateau region may not have proceeded along the drawn line but any other course with a
smaller slope which ends at the measured points.
According to the above considerations on damage development, it can be assumed that the
dXare~versus N curve for each individual specimen follows a straight line until final failure.
Table 5 represents an experimental check of this concept. Some specimens were exposed to
fretting fatigue at an upper fatigue stress of 640 MPa and a contact pressure o f p = 15 MPa
(FN = 300 N) for 200 000 load cycles. According to Fig. 4, the expected fretting fatigue life of
these specimens amounts to about 500 000 load cycles. Figure 15a depicts that these specimens did not exhibit any peeling off after 200 000 load cycles in contrast to specimens subjected to higher fatigue loads. Presuming a linear increase of 2Xare~until final failure after
257
FIG. 15--Optical micrographs of Laminate L1 after exposure to fretting fatigue. Specimen (a) was
exposed to fretting fatigue against aluminum ( F N = 3 0 0 N) at an upper fatigue load of au = 640 MPa for
200 000 load cycles. Afterwards it was run to &ilure (b) without fretting at an upper fatigue load of 750
MPa.
TABLE 5--Relative fatigue strength reduction due to j?etting

fatigue pre-damage. Specimens (Laminate L 1) were first
sztbjected to 200 O00ji'ettingjatigue load cycles (1~. = 300 N, au
= 640 MPa) and subsequentO' exposed to plain .&tigue (~v =
750 MPaj until./allure.
No.
Number of Cycles
to Failure
~r~l (%)
1
2
3
390
1380
780
6.6
5.4
6.O
A~rrel.mean
(%)
6.0
500 000 load cycles at A~el = 14%, a relative fatigue strength reduction after 200 000 load
cycles of about 5.5% would be expected (Fig. 11). Subsequent to fretting fatigue loading, the
specimens were run under plain fatigue at ao = 750 MPa until final failure occurred. During
this time, the layers pre-damaged by fretting rapidly cracked and delaminated (Fig. 15b). Table
5 lists the number of plain fatigue cycles to failure and the resulting deviation from the au versus log N curves for the undamaged laminate. The mean value for this difference is Acrrol =
6%. The coincidence of this result with the expected value confirms the assumptions that:
9 Fretting fatigue damage proceeds proportional with time in continuous fiber reinforced
laminates.
9 At low contact loads (FN --< 300 N), the rate of fretting fatigue damage development is
controlled by the upper fatigue stress, ~u.
However, for high contact loads (FN ~> 400 N) the development rate is independent of the
fatigue stress level. This can be explained by assuming the equivalent stress amplitude in the
258
subsurface region instead of the fatigue stress level to be responsible for the amplification of
fretting induced surface damage.
Conclusions
1. An additional fretting component may drastically reduce the fatigue life of an CF/EP
laminate if the fibers subjected to fretting are under 0~
2. The occurrence and magnitude of a fretting influence on fatigue performance sensitively
depends on the particular loading conditions (counterpart material, contact pressure). The use
of hard steel counterparts prevents a fretting fatigue effect, at least up to a contact pressure of
23 MPa under conditions given in the previously described experiments. For softer counterparts (aluminum, brass), a fretting fatigue effect occurred above a critical contact pressure.
3. As a quantitative measure for the degree of fretting fatigue damage, the relative fatigue
strength reduction
ffF -/k~rel
ffFF
-~F
was proposed. No notch effects were observed.

4. The mechanisms of interaction between surface damage due to fretting and fatigue are
different for different loading conditions. Figure 16 schematically distinguishes three regimes
of fretting fatigue. At low contact pressures (Regime I), there exists no synergism between fretting and fatigue. The rate of damage development is small and is determined by the resistance
against plain fretting wear. The specimens live nearly as long as under plain fatigue. At higher
pressures (Regime lI), the fretting fatigue damage proceeds considerably faster than the pure
material removal due to plain fretting wear. Fiber bundles exposed to fretting tend to crack
and delaminate. However, if the fatigue stress descends below a certain value, the mutual
amplification of fretting and fatigue damage becomes less effective because cracking and peeling off of fiber bundles decelerates. Above a critical contact pressure (Regime III), fretting
fatigue damage proceeds about 15 times faster than the pure material removal due to fretting.
Fiber bundles predamaged by fretting rapidly crack and peel off. The fretting fatigue damage
proceeds until the fretting pins reach off-axis plies, which carry only a small part of the applied
fatigue stress. The particular value of the critical contact pressure (Pen,) in Fig. 16 depends on
the hardness of the counterpart material.
G)
"O
O
O
G)
Q.
o~
Regim[Reg~
er
P
ContactPressure
crlt
FIG. 16--Sehernatic presentation Of'the course Of specific pseudo-wear rate as a Jimction q f contact
pressure.
259
Acknowledgments
This investigation was financed by the Ministry for Research and Technology of the FRG
(BMFT 03 M 1022). Thanks are due BASF and MBB for supplying the sample materials. Professor K. Friedrich is grateful for the personal research fund received from Fond der Chemischen Industrie E.V., Frankfurt.
References
[1] Waterhouse, R. B., (Ed.), Fretting Fatigue, Applied Science, London, 1981.
[2] Wear, Vol. 125, 1988, Special issue on fretting wear and fretting fatigue.
[3] Heinz, R., and Heinke, G., "Die Vorg~ngebeim Schwingungsverschleif3in Abh~ngigkeitvon Beanspruchung und Werkstoff," TriboIogie-Reibung, Verschleifl, Sehmierung (Documentation of the
German Ministry for Science and Technology), Springer, Berlin, Heidelberg, New York, 1981.
[4] Ohmae, N., Kobayashi, K., and Tsukizoe, T., "Characteristics of Fretting of Carbon Fibre Reinforced Plastics," Wear, Vol. 29, 1974, p. 345.
[5] Jacobs, O., Friedrich, K., Marom, G., Schulte, K., and Wagner, H. D., "Fretting Wear Performance
of Glass-, Carbon-, and Aramid-Fibre/Epoxy and Peek Composites," Wear, Vol. 135, 1990, p. 207.
[6] Schulte, K., Friedrich, K., Kutter, S., "Fretting Fatigue Studies on Carbon Fibre/Epoxy Resin Laminates, Part It: Effects of a Fretting Component on Fatigue Life," Composite Science and Technology, Vol. 30, 1987, p. 203.
[ 7] Friedrich, K., Kutter, S., and Schulte, K., "Fretting Fatigue Studies on Carbon Fibre/Epoxy Resin
Laminates, Part I: Design of a Fretting Fatigue Test Apparatus," Composite Science at:d Technology.
Vol. 30, 1987, p. 19.
[8] Gaul, D. J., and Duquette, D. J., "The Effect of Fretting and Environment on Fatigue Crack Initiation and Early Growth Properties in Quenched and Tempered 4130 Steel," Metallurgical Transactions, Vol. llA, 1980, p. 1555.
[9] Reifsnider, K. L., Schulte, K., and Duke, J. C., "Long Term Fatigue Behavior of Composite Materials," Long Term Behavior of Composites, STP 813, T. K. O'Brien, Ed., American Society for Testing and Materials, Philadelphia, 1983, p. 136.
[10] Edwards, P. R., "The Application of Fracture Mechanics to Predicting Fretting Fatigue," fretting
Fatigue, R. B. Waterhouse, Ed., Appl. Sci. Publ., Barking, 1981.
[11] Hills, D. A., Nowell, D. and O'Connor, J. J., "On Mechanics of Fretting Fatigue," Wear, Vol. 125,
1988, p. 129.
[12] Krey, J., Friedrich, K., and Schwalbe, K.-H. "Fracture Toughness and Fatigue Crack Propagation
of Single Fibre-Bundle Reinforced Model Composites," Journal of Materials Science Letters, Vol.
9 6, 1987, p. 851.
[13] Schute K. ``Damage Devepment under Cycic Lading Damage Develpment and Faiure Prcesses in Composite Materials, I. Verpoest and M. Wevers, Eds., Leuven (Belgium), 1987, p. 39.
[14] Jamison, R. D., Schulte, K., Reifsnider, K. L., Stinchcomb, W. W., "Characterization and Analysis
of Damage Mechanisms in Tension-TensionFatigue of Graphite/Epoxy Laminates," ASTM STP
836, 1984, p. 21.
[15] O'Brien, T. K., Rigamonti, M., and Zanotti, C., "Tension Fatigue Analysis and Life Prediction for
Composite Laminates," International Journal of Fatigue, Vol. 11, 1989, p. 379.
[16] Bader, M., "Modelling Fiber and Composite Failure," Damage Development andFailure Processes
in Composite Materials, I. Verpoest and M. Wevers, Eds., Leuven (Belgium), 1987, p. 8.
[17] Schulte, K., Friedrich, K., and Kutter, S., "Fretting Fatigue Studies on Carbon Fibre/Epoxy Resin
Laminates, Part III: Microscopy of Fretting Fatigue Failure Mechanisms," Composite Science and
Technology, Vol. 33, 1988, p. 155.
[18] Jacobs, O., Friedrich, K., and Schulte, K., "Schwingverschlei~von kohlenstoff-, aramid- und glasfaserverst~irkten Epoxidbarz- und PEEK-Verbundwerkstoffen," Reibung und Verschlei/3 bei metal
lischen undnichtmetallischen Werkstoffen. K.-H. Zum Gahr, Ed., Bad Nauheim, 1990, p. 199.
[19] Jacobs, O., Friedrich, K., and Schulte, K., "Fretting Fatigue of ContinuousCarbon Fibre Reinforced
Polymer Composites," Wear, Vol. 145, 1991, p. 167.
[20] Jacobs, O., Friedrich, K., and Schulte, K., "Fretting Wear of Continuous Fibre Reinforced Polymer
Composites," Symposium on Wear Testing and Advanced Materials, San Antonio, 14 Nov. 1990,
American Society for Testing and Materials, Philadelphia, to be published as ASTM STP 116 7.
[21] Endo, K., and Goto, H., "Initiation and Propagation of Fretting Fatigue Cracks," Wear, Vol. 38,
1976, p. 311.
260
[22] Alic, J. A. and Hawley, A. L., "On the Early Growth of Fretting Fatigue Cracks," Wear, Vol. 56,
1979, p. 377.
[23] Sato, K., "Damage Formation During Fretting Fatigue," Wear, Vol. 125, t988, p. 163.
[24] Sato, K., Fujii, H., and Kodama, S., "Crack Propagation Behavior in Fretting Fatigue," Wear, Vol.
107, 1986, p. 245.
[25] Hoeppner, D. W., "Comments on "Initiation and Propagation of Fretting Fatigue Cracks," Wear,
Vol. 43, 1977, p. 267.
DISCUSSION
L. Vincent ~ (written discussion)--(1) Can you be precise on the ranges of displacement
between pads and specimen during fretting fatigue tests? (2) Have you considered a possible
creep effect under the pad periphery which can induce rapid compressive failure of the
specimen?
O. Jacobs et al. (authors'closure)--(1) The ranges of displacement between pads and specimen depend on the compliance of the specimen, on the distance of the pads from the upper
fixed grips (in the text denoted as g~, and on the stress amplitude. Equation 1 explicitly
describes this relation. Discussion Figure 1 depicts the displacement ranges calculated according to Eq 1.
1000
80O
PPin Position l ...............'
~469~mmmmmm
...............
6OO
400~
200
600
700
800
Upper Fatigue Load [MPa]
Discussion FIG. l--Ranges qf dLsplacement between J?etting pins and specimen jor dfffl,rent [)in positions (distance of pins from upper fixed grips). The tests reported in this paper were all conducted with pins
at pos#ion g~ = 65 mm.
(2) We have not considered possible creep effects as a cause for compressive failure of the
specimens. The occurrence of a fretting fatigue effect depended on the counterpart material.
When the contact pressure was applied via steel pins, no significant fatigue life reduction was
found up to a contact pressure of 22 MPa. In contrast, fretting against aluminum pins led to
a clear relative fatigue strength reduction already at a contact pressure of 15 MPa. Obviously,
the wear process in the contact region rather than the contact load or according creep effects
is the crucial parameter. Furthermore, final failure is initiated by fracture on the 0~
which themselves do not creep at all. Creep could occur only in transverse directions.
Ecole Centrale De Lyon, Ecully Cedex, France.

Closing Paper
M. H e l m i Attia I
Fretting Fatigue Testing Current Practice and

Future Prospects for Standardization
REFERENCE: Atria, M. H., "Fretting Fatigue Testing: Current Practice and Future Prospects
for Standardization," Standardization of Fretting Fatigue Test Methods and Equipment, ASTM
ABSTRACT: A literature review has been carried out to identify the interactive role of the vari-
ables governing the fretting fatigue process and to examine the wide spectrum of the present testing practice. It has been shown that the present practice does not ensure an acceptable degree of
uniformity and results repeatability. The issues and areas that should be addressed for future
development of standardized test methods and equipment are also identified. A special emphasis
is placed on the conceptual development needed for modeling the thermal mechanical aspects
of the contact problem under fretting.
KEY WORDS: fretting fatigue, standardization, test methods, equipment, thermal constriction
resistance, contact pressure

When one of two contacting solids is subjected to an alternating stress, while the other is
rigidly held to overcome the frictional force, an oscillatory relative motion between the two
solids takes place. The accompanying fretting action will lead to fretting wear (removal of
material) and fretting fatigue (reduction in fatigue life). The fretting fatigue process, which is
primarily attributed to alternating frictional and bulk body stresses, involves two stages:
namely crack initiation and crack propagation.
The high probability of having two surfaces in contact under varying stresses or vibrations
explains the reason for the occurrence of fretting fatigue problems in a wide range of practical
situations, such as contacting strands in wire ropes, spring washers, leaf and coil springs,
splines, pivoted and bolted joints, press-fit joints, pinned connections, bearings, etc. [1-4].
Fretting fatigue is still, however, not well understood or appreciated compared with other
fatigue phenomena. With the present state of knowledge, fretting fatigue data is commonly
derived empirically by testing the components under simulated conditions. Whether fretting
fatigue testing is carried out to rank materials for their fretting fatigue resistance, to obtain
generic material-dependent data or to verify analytical (e.g., fracture mechanics) models, the
need for standardization of test methods and equipment is urgently needed to: (a) improve
the exchange of test results among research and testing laboratories, (b) be able to isolate interacting parameters, and thus evaluate their individual and combined effects.
A typical fretting fatigue test setup is shown schematically in Fig. 1. In the present article,
current test practice is reviewed and future prospects for standardization of test methods and
equipment are discussed. To appreciate the complexity of the fretting fatigue process, the
mechanics of the process is reviewed. A critical literature review is presented to highlight the
Research engineer, Mechanical Research Department, Ontario Hydro Research Division, Toronto,
Ontario, Canada M8Z 5S4.
263
Copyright
by ASTM Int'l
(all rightsASTM
reserved);
Mon Mar 28 00:05:25 EDTwww.astm.org
2016
Copyright
9 1992by
International
264
PROVINGLOADINGRING
FI'I'
,OAOAOJOSTINGSO.EW
;4,
PADSPANL
[~11[!~
"~
~ FATIGUESPECIMEN
~ FRETTINGPAD
FIG. 1-- Typicaljretting jhtigue test setup.
sources of uncertainty and poor repeatability, and to identify the issues and areas that should
be addressed for future development of standardized test methods and equipment.
Mechanics of Fretting Fatigue

The application of fracture mechanics to the fretting fatigue problem, as in [5-7], provides
the following insights into the interactive role of various governing variables and their effect
on the sequence of events:
(i) Under the influence of the applied and frictional alternating stresses, fatigue cracks will
initiate, due to dislocation movement, at points of high stress concentration or at the
boundary of the micro-slip region.
(ii) If the applied stress intensity range AK, exceeds the threshold value AK,h, the crack
propagates at a rate dg/dN, where g and N are the crack length and the number of
cycles, respectively. This process is nonlinear due to the mutual interaction between
AKa and g. The threshold stress intensity range AK,h is determined experimentally in
terms of the stress ratio R and the critical crack length go:
5 K , . = AK, h{R.
e. eo}
and
R = Sm,./S~a, = (Sin -- S ) / ( S m + S )
where S and Sm are the amplitude of the alternating body stress and its mean value,
respectively. A c o m m e n t on the dependence of AK~ on the process variables is in order
here. For two contacting elements of a given geometry, the applied clamping pressure
Pa gives rise to certain distributions of normal and frictional tangential contact stresses
pc{x,y} and q,.{x,y}, respectively, over the contact area Ac. Both, the opening and shear
265
ATTIA ON STANDARDIZATION
mode stress intensity factors KI and Kn, due to Pc, qc and alternating body stress S, at
the tip of the crack can be obtained from the following relation:
K~ = [Kp + Kq + K,],,
i = I and II
where
I,:p= f f kp{x,y}dxdy,
kp
kp{pc,x,y,g}
kq
kq{q,,,x,y,g}
Ac
Ac
K, = k,{S,e}
where kp and kq are the stress intensity for an extended edge crack in a half-space subjected to point normal and tangential forces P and Q, respectively [6,7]. The stress
intensity factor k, due to the alternating body stress is proportional to S ~ . The significant effect of the distribution of the contact pressure on the Mode I stress intensity
factor kp is demonstrated in Fig. 2 [5]. The mean stress intensity factor km and the
applied stress intensity range ~K, can, therefore, be expressed in the following functional form:
K~ = K,.{ e, S,., po, K~}
AK~ = /XK~{g,S, AF,, Kq}

C R A C K LENGTH ! (mm)
0,000
LLB
- 0.001'
Kp
(MNm3/2)
-0.002Palabohr
(PI)
-0.003
Uniform
{U}
Tr162
{T2)
Parabohr
(P2)
FIG. 2--Stress intensity factor for different contact pressure distribution (after Edwards [5]).
266
The relation between the frictional tangential force range 2xF, and the amplitude of the
applied body stress S is shown schematically in Fig. 3a for different pad spans L (see
Fig. 1). It has been demonstrated in [7] that AF, increases with the increase in the
applied pressure p~ and the pad span L due to the action of elastic slip. It is interesting
to observe the effect of the latter on the fretting fatigue life, as shown in Fig. 3b [5].
(iii) With the progress of the loading cycles N, the crack grows at a rate dg/dN (which
is controlled by K~ and AK~) until the maximum intensity factor (which corresponds
to Sm,~ = S~ + S) reaches the value of the fracture toughness K,. At this point, the crack
propagates rapidly to complete failure, ending the fatigue life Nfofthe component.
Two important issues concerning fretting fatigue testing emerge from the above discussion:
I. The first issue is the effect of the dimensions of the fretting pad and the fatigue specimen
on fatigue strength. The investigation carried out by Atria and Kops [8] concluded that
contact pressure distributions pc{x} of distinctively different shapes can practically be
generated by changing the relative flexural rigidity of the contacting solids. The photoelastic results shown in Fig. 4 indicate that the variation of the height-to-contact length
ratio (h/b) of one of the two elements causes a significant change in its elastic response
behavior from a flexible to a rigid body. These results were confirmed by finite element
analysis. It can, therefore, be concluded that the contact length b and height h of the
fretting pad (see Fig. 1) play a significant role in controlling the shape of the contact pressure distribution and consequently the results of the fretting fatigue test, as can be seen
from Fig. 2. The effect of the pad span L on the frictional tangential stresses and on the
S-N test results has already been presented in Fig. 3.
2. The second issue is the self-induced tribological changes during the test. A major source
of uncertainty in fretting fatigue testing is attributed to self-induced changes in the fretting conditions and test parameters. The contact geometry (constant conforming vs concentrated nonconforming geometry) plays the most critical role in controlling the generation and retention of the fretting debris. It was reported in [3] that, for a given average
contact pressure, a greater reduction in fretting fatigue strength was seen with cylindrical
fretting pads than with conforming ones. For a mechanically "constrained" system, such
as the one shown in Fig. 1, the accumulation of the oxidized debris leads to an increase
in the clamping pressure p~ when the debris volume is greater than that of the original
metal from which it originates [1]. Usually such an increase in p~ is associated with a
LI<L2<L3<L4
l<L2<L3<L4
m
L1
L~
L4
STRESS AMPLITUDE, S'

(~)
CYCLES TO FAILURE, N f
(b)
FIG. 3--Effect of applied alternating body stress and pad span on (a)frictional foree range and (b) frettingj~ttigue Ire.
FIG. 4--s
267
of the relative stiffness of contacting bodies on the distribution of contact pressure [8].
reduction in the slip amplitude a. The process of debris retention in the contact zone
contributes also to the change in the frictional force. The effect of the latter can easily be
assessed from Fig. 3.
For the purpose of clarification, a summary of the dependence of fretting fatigue results on
Pa and the slip amplitude a is shown schematically in Fig. 5. The increase in the clamping
pressure p~, up to a certain critical level Pa, results in a reduction in the fretting fatigue strength
SI~ Above fia, the fatigue strength remains unchanged. It is particularly interesting to note that
the increase in the width of contact between a cylindrical pad and a flat specimen leads to the
reduction of the fretting fatigue life NI under the same maximum contact pressure [9]. The
effect of the slip amplitude " a " on the fretting fatigue strength Sris coupled with the magnitude
of the mean body stress Sin. As Fig. 5b shows, for Sm --- 0, the fretting strength, Ssdecreases as
the slip amplitude " a " increases up to a certain value ~. For all values of a >-- ~, the value of
St remains nearly constant or increases slightly. For Sm >-- O, Fig. 5b shows that as the mean
body stress Sm is increased the fretting fatigue strength decreases first but then remains
unchanged. Some reported results indicate, however, that Sm does not affect Sr at relatively
small slip amplitudes, a < ac,. For 2.5 Ni-0.6 Cr-0.5 Mo steel, the value of at, is approximately
7.5 m (as reported in Ref 1). The drop in the fretting fatigue life in a two stage test with the
height h of the fretting pad feet (Fig. 5b) can be attributed to a reduction in the slip amplitude
due to the elastic deformation of the feet [1].
268
=,
o
(3
I
CLAMPING
PRESSURE Pa
MAX. CONTACT PRESSURE Pcma x

(a)
$-
(Sin= O)
pad
a<acr
h1
h1 9 h 2
-a
SLIP AMPLITUDE a
MEAN APPLIED STRESS Sm
CYCLES TO FAILURE Nf
(b)
FIG. 5--Effect of(a) the clamping pressure and (b) the slip amplitude and mean body stress on the
fretting fatigue strength and the fatigue life.
Current Test Practice

A literature review of the test methods and equipment used to evaluate the fretting fatigue
strength of metals is presented in this section [e.g., 1,9-40]. A summary and examples [cited
in brackets] for the variability in the test parameters and experimental setup is presented under
the following categories:
( i) Test Strategy and Mode of Control."

9 One-stage test, two-stage test, or two-stage test at a constant stress level.
9 Fatigue loading cycle:
9 Constant force versus contant strain (of specimen or actuator).
9 Constant stress amplitude versus block load program (staircase, Markov, or
random).
9 Stress ratio, state of stress (magnitude, direction, and time variation).
( ii) Test Equipment and Fixtures."
9 Type (pull-push, four point rotating bending, cantilever type rotating bending, twisting or in-house design).
9 Dynamic Load Verification: load frame size, type, compliance of loading train
9 Alignment.
9 Design of grips to prevent fretting and slippage.
9 Test equipment rigidity.
( iii) Control of Test Conditions:
9 Contact Pressure (no control, "Keep it as constant as possible" by continuous manual adjustment or by using a "sufficiently" flexible loading frame, or by the use of a
servo-hydraulic system).
269
9 Slip Amplitude (controlled in two-stage test procedure, otherwise "it cannot be

maintained constant" in a one-stage test due to the coupling between stress and
strain. An exceptional case was due to a special design arrangement by Nishioka and
Hirakawa [35].
9 Contact Temperature: well recognized as a critical factor, but no attempt has ever
been made to control it.
9 Frequency, for accelerated tests.
(iv) Specimens.
9 Geometry (sheets with variable or constant stress-cross-section, rectangular rod or
round rod).
9 Size: the negative scale effect is recognized, however, overall dimensions vary considerably (2 to 7 m m thick sheets, 65 by 20 by 25 m m to 254 by 38 by 1.6 mm rectangular rods, 4 m m diameter by 160 mm long to 44 m m diameter to 203 mrn long
round rods).
9 Contact Area: area versus line contact depending on the shape of the specimen and
the fretting pad (flat or cylindrical-ended). Wide range of nominal contact area.
9 Fretting Pad: a wide variation in the pad size, geometry, span, height as well as location along the specimen (which affects the slip amplitude).
9 Surface Preparation and Finish: Critical factor but yet arbitrarily chosen (emery
paper [ # 1500 or 600, etc.], polishing paste, metallographic polishing, grinding, etc.).
Only in a few cases were the degree of surface roughness and directior., of scratches
cited. Cleaning procedures differ from one laboratory to another.
9 Residual Stresses: an extremely important factor which is usually (if not always) not
considered.
(v) Environmental Conditions:

9 Surrounding medium, temperature, pressure and relative humidity.
( vi) Measurement:
9 Parameters: contact load, slip amplitude, friction force, electrical contact resistance,
environment (temperature, pressure etc.). In a very few cases, the elastic deIbrmation of the fretting frame is measured and accounted for. The problem of contact
temperature measurement is still unsolved.
9 Sensors: displacement; contacting versus non-contacting (e.g., capacitance, eddy
current, optical, laser beam, etc.)/axial load; type of load cell, its mounting method
and compliance/friction force, etc.
9 Automation of test procedure and data acquisition system.
It is quite striking to see the lack of the uniformity in current test practice, e.g., in specimen/
pad geometry and size and in the rigidity of the fretting apparatus and the loading train. The
latter was deliberately used, in some reported cases, to change the initial level of slip amplitude.
From the measurement point of view, the actual slip and contact temperature present a serious
challenge.
Future ProspectsJbr Standardization

As seen from the above, the difficulty in fretting fatigue testing manifests itself not only
through the numerous affecting variables (Collins [39] cited as many as 55 variables), but also
through their mutual interactions and the self-induced changes in the tribological system. The
latter poses a serious challenge when one faces the question of maintaining constant conditions
at the contact interface during the test. The discrepancy among published data is, therefore,
not surprising. The possibility and the potential for improving the repeatability of test results
270
do, however, exist. The issues to be addressed in developing standard test methods and equipment can be grouped under the following two categories:
1. Conceptual Development and Process Modeling."

Through a comprehensive system approach, a conceptual engineering framework should
be established to relate the stress, thermal and chemical aspects of the fretting fatigue
process, as well as fracture mechanics principles, to fretting fatigue failure.
The outcome of this task will lead to:
(i) Identification of the principal variables that are to be considered, controlled and monitored during the test, by setting-up a hierarchy of the significance of various governing
variables.
(ii) Development of similarity and scaling laws that are required for conducting testing
under simulated conditions and for extrapolation of results. Due to the fact that the
contact conditions (stresses and temperature) cannot directly be measured, there is an
urgent need for the development of realistic models for predicting these parameters.
A concentrated effort to develop mathematical models for the contact temperature and
the thermal constriction resistance in fretting is being pursued at the Fretting Laboratory of Ontario Hydro Research Division [41-46]. The first model, developed by Atria
et al. [4] to predict the contact temperature in fretting, was based on the assumption
that a single asperity is in contact with a semi-infinite body. Such a model neglects the
interaction between adjacent contact spots, as well as the finite thermal capacity of the
body. Using the concept ofsuperposition of image sources, these limitations were overcome [44] by solving the heat transfer process in an elemental heat flow channel
(HFC), which encompasses a single micro-contact and extends some distance in the
solid. In a recent study [45], this model was used to produce numerical data on the
temperature field in the vicinity of an asperity. Since these micro-contacts are thermally connected in parallel, the analysis of the heat transfer process at a single contact
spot constitutes the basic cell for predicting the contact temperature and thermal constriction resistance in a real tribological system. This study revealed the steep temperature gradient in the subsurface layer (Fig. 6) and near the edge of the contact spot (Fig.
7). For a given frequencyfand amplitude of motion a, the temperature rise 0 is presented in terms of the Heat Accumulation factor (HA):
HA =oCpP J L
4 #pa
where Cp and p are the specific heat and density of the material. The symbols # and p
stand for the coefficient of friction and contact pressure respectively, while J is the
mechanical equivalent of heat. The symbol L is the characteristic length of the microcontact spot. The numerical data obtained in [45] provide the framework for conceptually developing a theoretical model for the quasi-steady state thermal constriction
resistance. By integrating this model into a conventional finite element or finite difference analyses the temperature in the fretting zone can readily be calculated without the
need for preknowledge for the division of frictional heat between the contacting bodies.
(iii) Development of fracture mechanics models for failure analysis and prediction.
(iv) Development of a model for the interaction between dynamic normal and frictional
forces, and for prediction of friction-induced vibrations.
o0o664 i
271
HA
0.008001
0.005301
I
0.00395
Z 4.0
+5
x
0.00
FIG. 6--Temperature field in the X - Z plane at the start of the quasi-steady state cycle (contact area:
-I<X<
+I,-1
< Y < +1).
z=0
HA
0.00800
0.00665
0.00530
0.00395
+5
5
+I .0
y-1.0
+I .0
-I.0 X
-5 -5
FIG. 7--Temperature field in the contact plane at the start of the quasi-steady state cycle (contact area.
--1 < X <
+ 1 , - - 1 < Y < +1).
. Test Strategy, Equipment Design, and Measurement System:

In light of the review presented in the previous section, a concentrated effort should be
directed towards standardization of the geometry and the dimensions of the fretting pad
and of the fatigue specimen. The need to control the slip amplitude independently of the
fatigue stress level and to maintain the clamping pressure constant during the test should
not be underestimated. Due to their effect on test results, standardization of surface preparation and finish is in order. Other critically important issues have been identified in
[3]; namely, the control of the test environment, characterization of lubricants, and the
method of applying lubricants to the fretting zone. One of the issues which requires special attention is the temperature measurement in the subsurface layer and near the contact interface. The closeness of the temperature sensing elements to the interface or to
each other causes a distortion in the temperature field and may lead to significant measurement errors. Guidelines developed by Attia et al. for thermometric design considerations are given in [42,47,48].
In 1988, the ASTM Committee E-9 on Fatigue formed a Task Group, chaired by the author,
to establish fretting fatigue testing standards. The main objectives of the Task Group are:
1. Identification of the areas of uncertainties in conducting testing under simulated conditions, including measurement and control aspects.
272
9.~
~
~'~,
,.nr~
. . . .
._
~ ~.~
,,~
e.
o
'~
~0~
~.~ e
oo
e-
~ ~
e~
.~.
~3
=o
+ + +
.~.~
+
0
~2
b
~.-~ .~ ,.
0
+~
c~
I I I I I I I I
I
<
o~
~'=
.~.~_
oe'~~ ~
'-~
273
2. Identification of the measures that should be taken to improve the repeatability of test
results and to minimize their dependence on the design of test equipment.
In addition to the assembly of relevant data available in the literature, the above mentioned
objectives will be achieved through an international cooperative effort. This effort will include:
1. Revising the available ASTM standards for conventional fatigue tests, e.g., E 466, and E
467, with the purpose of extending the recommended practice to fretting fatigue tests.
2. Organizing the present symposium on the subject matter to exchange views and firsthand experience of participants from industry and academia.
3. Organizing a series of round-robin tests by reputable research/testing laboratories.
The first activity of the Task Group was to conduct a survey to solicit the input from 65
internationally renowned and active researchers in this field from 12 nations. The responses
received from 50% of these researchers are summarized in Table 1. The consensus was overwhelmingly in favor of standardizing fretting fatigue test methods and equipment; in particular, specimen geometry and preparation as well as monitoring and controlling test conditions. The need for conceptual development and modeling was expressed by 30% of the
respondents.
Concluding Remarks
The mechanism of the fretting fatigue process was reviewed to identify the interactive role
of various governing variables. It has been shown that present testing practice does not ensure
an acceptable degree of uniformity among different laboratories and attributes, therefore, the
repeatability of test results. A critical literature review has been carried out to identify the issues
and the areas that should be addressed for future development of standardized test methods
and equipment. The initiative taken by ASTM Subcommittee E09.08 to develop test standards is indeed a timely one.
References
[1] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, England, 1972.
[2] Fretting Fatigue. R. B. Waterhouse, Ed., Applied Science, London, England, 1981.
[3] "Control of Fretting Fatigue," NMAB Committee on Control of Fretting Initiated Fatigue, National
Research Council, Report No. NMAB-333, 1977.
[4] Forsyth, P. J. E., "Occurrence of Fretting Fatigue in Practice," Fretting Fatigue, R. B. Waterhouse,
Ed., Applied Science, London, England, 1981.
[5J Edwards, P. R., "The Application of Fracture Mechanics to Predicting Fretting Fatigue," Fretting
Fatigue, R. B. Waterhouse, Ed., Applied Science, London, England, 1981.
[6] Hattori, T., Nakamura, M., and Watanabe, T., "Fretting Fatigue Analysis by Using Fracture
Mechanics," Winter Annual Meeting of the ASME, 1984, Paper No. 84-WA/DE-10.
[ 7] Nix, K. J. and Lindley, T. C., "The Application of Fracture Mechanics to Fretting Fatigue," BNL,
Central Electricity Research Laboratories, U.K., Report No. TPRD/L/2648/N84, 1984.
[8] Attia, M. H. and Kops, L., "A Method for Generating Desired Contact Pressure Distributions in
Experimental Interfacial Studies," Trans. ASME, J. Engrg, Industry, Vol. 107, Aug. t985, pp. 241246.
[9] Waterhouse, R. B., "The Effect of Clamping Stress Distribution on the Fretting Fatigue of Alpha
Brass and AI-Mg-Zn Alloy," ASLE Transactions, Vol. II, 1968, pp. 1-5.
[10] Nix, K. J. and Lindley, T. C. "Palliatives to Combat Fretting Fatigue in 2014A Aluminum Alloy,"
BNL, Control Electricity Research Laboratories, U.K.. Report No. TPRD/L/3191/R87, 1987.
[11] Nix, K. J. and Lindley, T. C., "The Influence of Relative Slip Range and Contact Material on the
Fretting Fatigue Properties of 3~ANi Cr MoV Rotor Still," BNL, Central Electricity Research Laboratories, U.K., Report No. TPRD L/2958/N85, 1986.
274
[12] Tanaka, K., Mutoh, Y., Sakada, S., and Leadbeater, G., "Fretting Fatigue in 0.55C Spring Steel and
0.45C Carbon Steel," FatigueFract. Engrg. Mater. Stract., Vol. 8, 1985, pp. 129-142.
[13] Poon, C. J. and Hoeppner, D. W., "A Statistically Based Investigation of the Environmental and
Cyclic Stress Effects on Fretting Fatigue," J. Engrg, Materials and Technology, Vol. 103, 1981, pp.
218-222.
[14] Wharton, M. H., and Waterhouse, R. B., "Environmental Effects in the Fretting Fatigue of Ti-6AL4V," Wear, Vol. 62, 1980, pp. 287-297.
[15] Gaul, D. J. and Duquette, D. J., "The Effect of Fretting and Environment on Fatigue Crack Initiation and Early Propagation in a Quenched and Tempered 4130 Steel," Metallurgical Trans. A,
A1ME, Vol. 11A, 1980, pp. 1555-1561.
[16] Duquette, D. J., "The Role of Cyclic Wear (Fretting) in Fatigue Crack Nucleation in Steels,"
Strength of Metals and Alloys: Proc. the 5th Int. Conf., Aachen, W. Germany, 1979, pp. 213-218.
[17] Poon, C. and Hoeppner, D. W., "The Effect of Environment on the Mechanism of Fretting Fatigue,"
Wear, Vol. 52, 1979, pp. 175-191.
[18] Hamdy, M. M. and Waterhouse, R. B., "The Fretting Fatigue Behaviour of a Nickel Based Alloy
(Inconcl. 718) at Elevated Temperature," Proc. Conf. Wear of Materials, ASME, 1979, pp. 351-355.
[19] Waterhouse, R. B. and Wharton, M. H., "The Behaviour of Three High Strength Titanium Alloys
in Fretting Fatigue in a Corrosive Environment," J. ASLE, Lubrication Engrg., 1976, pp. 294-298.
[20] Dutta, M. K. and Waterhouse, R. B., "Initiation of Fatigue Cracks due to Fretting in C.P. Titanium
and Titanium Alloys," Trans. Indian Inst. of Metals, Vol. 27, 1974, pp. 102-104.
[21] Gross, G. L. and Hoeppner, D. W., "Normal Load Effects in Fretting Fatigue of Titanium and Aluminum Alloy," Wear, Vol. 27, 1974, pp. 153-159.
[22] Petukhov, A. N., "Procedure Features of Studying Fretting Corrosion in Connection with the
Fatigue of Metals," Translated from Zavodskaya Laboratoriya, Vol. 40, 1974, pp. 1246-1250.
[23] Endo, K., Goto, H., and Nakamura, T., "Fretting Fatigue Strength of Several Materials Combinations,"Bull. JSME, Vol. 16, 1973, pp. 143-150.
[24] Wharton, M. H., Taylor, D. E., and Waterhouse, R. B., "Metallurgical Factors in the FrettingFatigue Behaviour of 70130 Brass and 0.7% Carbon Steel," Wear, Vol. 23, 1973, pp. 251-260.
[25] Tedford, J. D., Carse, A. M., and Crossland, B., "Comparison of Component and Small Specimen
Block Load Fatigue Test Data," Engrg. Fracture Mechanics, Vol. 5, 1973, pp. 241-258.
[26] Malkin, S., Majors, D. P., and Courtney, T. H., "Surface Effects During Fretting Fatigue ofTi-6A I4V," Wear, Vol. 22, 1972, pp. 235-244.
[27] Harris, W. J., "The Influence of Fretting on Fatigue, Part Ill," AGARD, NATO, Report No.
AGARD-AR-45, 1972.
[28] Waterhouse, R. B., Dutta, M. K., and Swallow, P. J., "Fretting Fatigue in Corrosive Environments,"
Proc. Int. Conf. Mech. Behaviour ofMetals, Vol. 3, ASTM, 1972, pp. 292-298.
[29] Waterhouse, R. B. and Dutta, M. K., "The Fretting Fatigue of Titanium and Some Titanium Alloys
in a Corrosive Environment," Wear, Vol. 25, 1973, pp. 171-175.
[30] Starkey, W. L., "A New Fretting Fatigue Testing Machine," Corrosion Fatigue: Chemistry, Mechanical and Microstructure, NACE-2, 197 I, pp. 642-645.
[31] Hoeppner, D. W. and Goss, G. L., "A New Apparatus for Studying Fretting Failure," Review of
Scientific Instruments, Vol. 42, 1971, pp. 817-821.
[32] Waterhouse, R. E., "The Effect of Fretting Corrosion in Fatigue Crack Initiation," Corrosion
Fatigue: Chemistry, Mechanics and Microstructure, NACE-2, 1971, pp. 608-616.
[33] Waterhouse, R. B. and Taylor, E. D., "The Initiation of Fatigue Cracks in 0.7% Carbon Steel by
Fretting," Wear, Vol. 17, 1971, pp. 139-147.
[34] Nishioka, K. and Hirakawa, K., "Fundamental Investigation of Fretting Fatigue, Part lII: Some Phenomena and Mechanics of Surface Cracks," Bull. ofJSME, Vol. 12, 1969, pp. 397-407.
[35] Nishioka, K. and Hirakawa, K., "Fundamental Investigation of Fretting Fatigue-Part 2: Fretting
Fatigue Testing Machine and Some Test Results," Bull. ofJSME, Vol. 12, 1969, pp. 180-187.
[36] Endo, K., Goto, H., and Nakamura, T., "Effects of Cycle Frequency on Fretting Fatigue Life of
Carbon Steel," Bull. ofJSME, Vol. 12, No. 54, 1969, pp. 1300-1308.
[37] Collins, J. A. and Marco, S. M., "The Effect of Stress Direction During Fretting on Subsequent
Fatigue Life," Proc. ASTM, Vol. 64, 1964, pp. 547-560.
[38] Gassner, E., "On the Influence of Fretting Corrosion on the Fatigue Life of Notched Specimens of
an AL-Cu-Mg2 Alloy," Proc. Symposium on Fatigue of Aircraft Structures, W. Barrois and E. L
Riplay, Eds., Paris, France, 1961, pp. 87-95.
[39] Collins, A. J., "Fretting Fatigue Phenomena with Emphasis on Stress Field Effects," Ph.D. thesis,
Ohio State University, 1963.
[40] Fenner, A. J. and Field, J. E., "A Study of the Onset of Fatigue Damage Due to Fretting," Northeast
Coast Inst. of Engrs. and Shipbuilders Trans., Vol. 76, 1960, pp. 183-228.
275
[41] Attia, M. H. and D'Silva, N. S., "Effect of Mode of Motion and Process Parameters on the Prediction
of Temperature Rise in Fretting Wear," Wear, Vol. 106, 1985, pp. 203-224.
[42] Attia, M. H. and Ko, P., "On the Thermal Aspect of Fretting Wear--Temperature Measurement in
the Subsurface Layer," Wear, Vol. III, 1986, pp. 363-376.
[43] Attia, M. H., "A Thermally Controlled Fretting Wear Tribometer--A Step Towards Standardization of Test Equipment and Methods," Proc. Int. Conference on Wear of Materials, Denver, Colo.,
8-14 April 1989, American Society of Mechanical Engineers, New York.
[44] Attia, M. H., "Friction-lnduced Temperature Rise in Fretting--Elemental Heat Flow Channel
Model," Proc. 5th Int. Congress on Tribology, Eurotrib '89, Helsinki, Finland, June 1989.
[45] Atria, M. H. and Camocho, F., "On the Thermal Constriction Phenomenon in Fretting," accepted
for publication in the Proceedings of the Egytrib Second Tribology Conference, to be held in Cairo,
Egypt, Jan. 1991.
[46] Attia, M. H. and Yovamovich, M. M., "Development of Thermal Constriction Resistance in Fretting," to be submitted for publication in the Journal of Physics D: Applied Physics and presentation
at the International Conference Frontiers of Tribology, April 1991, organized by the Institute of
Physics, U.K.
[47] Attia, M. H. and Kops, L., "Distortion in the Thermal Field Around Inserted Thermocouples in
Experimental Interfacial Studies," Trans. ASME, Journal of Engineering for Industry, Vol. 108,
Nov. 1986, pp. 241-246,
[48] Attia, M. H. and Kops, L., "Distortion in Thermal Field Around Inserted Thermocouples in Experimental Interracial Studies--Part II: Effect of Heat Flow Through the Thermocouple," Trans.
ASME, Journal of Engineeringji)r Industry, Vol. 110, Feb. 1988, pp. 7-14.
Author Index
Arab
Atria, M. Helmi, 1,263

Baoyu, Zhang, 210
Berthier, Yves, 33
Buxbaum, Otto, 178
CmD
Cardou, Alain, 231

Cloutier, Louis, 231
Cook, R., 129
Dobromirski, Jan M., 60
N
Nakamura, Masayuki, 101
Nakazawa, Kozo, 115
Nix, Kevin J., 153
Nowell, D., 69
P--11
Price, Stephen, 217
Rayoprolu, D. B., 129
Ruiz, C., 170
Frog
Fischer, Gerhard, 178

Friedrich, Klaus, 243
Godet, Maurice, 33
Grubisic, Vatroslav, 178
H--3
Hattori, Toshio, 101

Hills, D. A., 69
Hoeppner, David W., 23
Ishizuka, Tatsuro, 101
J~cobs, Olaf, 243
L--M
Labedz, Janusz, 190
Lelond, Andrr, 231
Lindley, Trevor C., 153
Maruyama, Norio, 115
Mutoh, Yoshiharu, 199
Satoh, Toyoichi, 199

Sato, Kenkichi, 85
Schulte, Karl, 243
St-Louis, M., 231
Sumita, Masae, 115
T--V
Taylor, David E., 217
Tsunoda, Eiji, 199
Vincent, Leo, 33
Vingsbo, Olof B., 49
W--u
Wang, Z. P., 170
Waterhouse, Robert B., 1, 8, 13
Webb, P. H., 170
Weili, Luo, 210
Yunshu, Deng, 210
Copyright
1992by
ASTM
International
www.astm.org
Copyright
by ASTM 9
Int'l
(all rights
reserved);
Mon Mar 28 00:05:25 EDT
2016
Subject Index
A--B
Aeolian vibrations, 231

AES spectrometry, 210
Aluminum alloys, testing at Royal
Aerospace Establishment, 129
Aluminum conductor steel reinforced
electrical conductor, 231
Bearing steel, fretting wear, 190
Biaxial loading test, 170
Boundary element method, 85, 101
C
Carbon fiber-reinforced epoxy laminates,
243
Cast iron, nodular, fatigue strength, 178
Cathodic protection, high strength steel,
217
Ceramics, 170
Clamping pressure, 13
Coefficient of friction, 60
Constant amplitude tests, fatigue strength of
nodular cast iron and steel, 178
Contact fatigue, fretting maps and, 49
Contact pressure, 263
distribution, 85
fatigue of high strength steel and titanium
alloy, 115
Corten A, 217
Crack, nonpropagating, 101
Crack formation, 33
Crack initiation site, 115
high strength steel, 217
Crack propagation, 8
D--E
Debris beds, shear, 33
Delamination, carbon fiber-reinforced
epoxy laminates, 243
Elastic displacements, 33
Electrical conductor, aluminum conductor
steel reinforced, 231
Environmental effects, 1
fretting fatigue, 210
Epoxy laminates, carbon fiber-reinforced,
fretting fatigue, 243
F
Fatigue life prediction, 178
Fatigue limit, 101
Fatigue properties, 8
steam turbine steel, 210
Fracture mechanics, 60, 101, 129, 153
Frequency, 13
Fretting, machine components under,
190
Fretting bridges, contact pressure
distribution, 85
Fretting corrosion, 23
fatigue strength of nodular cast iron and
steel, 178
high strength low alloy steel, 217
Fretting device, 13
Fretting fatigue, 33
aluminum conductor steel reinforced
electrical conductor, 231
carbon fiber-reinforced epoxy laminates,
243
contact pressure distribution, 85
corrosion role, 217
experiment with well-defined
characteristics, 69
fretting maps and, 49
history, 8
mechanisms, 23
power generation industry, 153
strength improvement model analysis,
101
variables, 60
Fretting fatigue damage
characterization techniques, 170
nucleation, 23
Fretting fatigue testing
appraisal of methods, 33
conceptual framework, 1
current practice, 263
280
Fretting fatigue testing--continued

at elevated temperature, 199
equipment, 1
Hertzian contact, 69
methods, 1
review, 263
Royal Aerospace Establishment, 129
nonconventional materials and methods,
1

problems, 13
Royal Aerospace Establishment, 129
strength evaluation method, 101
variables, 263
Fretting maps, 33, 49
Fretting figs, experimental, power
generation industry, 153
Fretting wear, 23, 33, 49
bearing steel, 190
carbon fiber-reinforced epoxy laminates,
243
Frictional force, 129
high strength steel and titanium alloy,
115
Friction log, 33
GBH
Grooving, on contact surface, 101

Gross slip regime, 49
Hertzian contact, 69
High strength steel, 115
contact pressure and fatigue, 115
fretting corrosion, 217
I--L
Interface slip, 170
Knurling, on contact surface, 101
Load spectrum tests, fatigue strength of
nodular cast iron and steel, 178
M
Machine components, life testing, 190
Mechanics of contact, 1
Method of caustics, 85
Microstructural studies, 49
Mixed stick-slip regime, 49
Mossbauer spectrometry, 210
NmO
Nonpropagating crack, 101

Overhead electrical conductors, 231
Oxide debris, 210
pmR
Partial-slip regime, 69
Particle detachment, 33
Polymer composites, 243
Power generation industry, fretting fatigue,
153
Residual stress, 190, 199
Royal Aerospace Establishment, fretting
fatigue testing, 129
Seawater corrosion, 217

Servohydraulic testing machine, adaptation,
190
Shear stress, 170
Shot peening, 178, 199
steam turbine steel, 210
Slip amplitude, 13
Small crack growth, 153
S - N plots, 85
Spacer clamps, 231
Standardization, 129, 263
Steam turbine steel
fretting fatigue properties, 210
shot peening, 199
Steel, see also Steam turbine steel
high strength low alloy, fretting
corrosion, 217
structural, fatigue strength, 178
Stick region, 49, 115
Strength improvement models, analysis,
101
Stress concentration, high strength steel and
titanium alloy, 115
Stress intensity factors, 69, 101, 129
Stress redistribution, carbon fiber-reinforced
epoxy laminates, 243
Surface residual stresses, 1
Tmu
Tangential stress, 170

Temperature, elevated, fretting fatigue
testing, 199
INDEX
Tensile stress, 170

Tension-tension fatigue loading, carbonfiber reinforced epoxy laminates, 243
Thermal constriction resistance, 263
Thermal mechanics, modeling, 263
281
Three-point bending test, 170

Titanium alloy, contact pressure and
fatigue, 115
Uniaxial tension-compression test, 170
Velocity accommodation mechanism, 33

Standardization of Fretting Fatigue

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Standardization of Fretting Fatigue

Enviado por

Direitos autorais:

Formatos disponíveis

Copyright by ASTM Int'l (all rights reserved); Mon Mar 28 00:05:25 EDT 2016

M. Helmi Attia and R. B. Waterhouse, editors

ASTM Publication Code Number (PCN)

Library of Congress Cataloging-in-Publication Data

Peer Review Policy

FUNDAMENTAL ASPECTS OF FRETTING FATIGUE TESTING--CONCEPTUAL FRAMEWORK

Testing Methods in Fretting Fatigue: A Critical Appraisal--L. VINCENT,

Variables of Fretting Process: Are There 50 of T h e m ? - - J . DOBROMIRSKI

FUNDAMENTAL ASPECTS OF FRETTING FATIGUE TESTING--MECHANICS OF CONTACT

Determination and Control of Contact Pressure Distribution in Fretting F at i g u e- K. SATO

Fretting Fatigue Analysis of Strength Improvement Models with Grooving or

Effect of Contact Pressure on Fretting Fatigue of High Strength Steel and

FRETTING FATIGUE T E S T I N G - - M E T H O D S AND EQUIPMENT

A Critical Review of Fretting Fatigue Investigations at the Royal Aerospace

Fretting Fatigue in the Power Generation Industry: Experiments, Analysis, and

Techniques for the Characterization of Fretting Fatigue Damage--c. RUIZ,

Z. P. WANG, AND P. H. WEBB

Adaptation of a Servohydraulic Testing Machine to Investigate the Life of

ENVIRONMENTAL AND SURFACE CONDITIONS

Improving Fretting Fatigue Strength at Elevated Temperatures by Shot Peening in

The Application of Electrochemical Techniques to Evaluate the Role of Corrosion

NONCONVENTIONAL MATERIALS AND TEST METHODS

ACSR Electrical Conductor Fretting Fatigue at Spacer Clamps--A. CARDON,

Fretting Fatigue of Carbon Fiber-Reinforced Epoxy Laminates--o. JACOBS,

Fretting Fatigue Testing: Current Practice and Future Prospects for

Overview of the Papers of the Symposium

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

FundamentaI Aspects of Fretting Fatigue Testing--Conceptual Framework

Fundamental Aspects of Fretting Fatigue Testing--Mechanics of Contact

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

Fretting Fatigue Testing--Methods and Equipment

Environmental and Surface Conditions

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

A Historical Introduction to Fretting Fatigue

WATERHOUSE ON A HISTORICAL INTRODUCTION

The Problems of Fretting Fatigue Testing

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

The Object of Fretting Fatigue Testing

FIG. l--Design of specirnen with machinedflats.

WATERHOUSE ON PROBLEMS OF FRETTING FATIGUE TESTING

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

CLAMPING PRESSURE, MPa

FIG. 3--Fretting Jatigue stren~h versus clamping load.

WATERHOUSE ON PROBLEMS OF FRETTING FATIGUE TESTING

FIG. 4--Co~,~cienl (~/i/riction versus amplitude of slip.

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

WATERHOUSE ON PROBLEMS OF FRETTING FATIGUE TESTING

Fundamental Aspects of Fretting Fatigue

Mechanisms of Fretting-Fatigue and Their

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

or both o f the m e m b e r s m a y experience fretting fatigue. Reference 4 d o c u m e n t s m a n y specific

HOEPPNER ON MECHANISMS OF FRETTING FATIGUE

Verification Life ~Tlesting

FIG. 1--Reasons.[or conductingfrening-]atigue experimental studies.

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

constant load amplitude

FIG. 2--A conceptual view of#retting-fatigue emphasizing the numerous parameters.

" t Cyclic Loading ] -

FIG. 3--A systemsframework previously proposedjbr planning experimental studies q[[?etting-[atigue.

HOEPPNER ON MECHANISMS OF FRETTING FATIGUE

TABLE 1--"Damage"produced underJkettingz/btigue conditions.

FRETTING FATIGUE TEST METHODS AND EQUIPMENT

FIG. 4--Fatigue life estimationJbrJkeuing-fatigue.

FRETTING FATIGUE TEST METHODS AND EQUIPMENT