You are on page 1of 8

Rainflow Cycle Counting : A Historical Perspective

Darrell Socie
Mechanical Engineering Department
University of Illinois at Urbana-Champaign

Abstract: This paper reviews the historical


development of the rainflow cycle counting method
invented by Tatsuo Endo in 1967. Key
contributions that have extended the original work
of Endo are briefly reviewed.
Keywords:

Rainflow, Cycle Counting, Fatigue

1. Introduction
The local strain approach for fatigue analysis is
now routinely used throughout the world for assessing the
durability of structures and components. This approach
was made possible by four key developments:
* Coffin's and Manson's work that established
plastic strain as the controlling parameter for
fatigue damage,
* Morrow's and Topper's work on simulating the
stresses and strains of notched members,
* Commercial development of the closed loop
servohydraulic materials testing machine by
Johnson,and
* Endo's invention of the rainflow cycle counting
method.
All of this work has stood the test of twenty-five years
of careful scrutiny by researchers and engineers. In
this conference, Fatigue Damage Measurement and
Evaluation Under Complex Loadings, we honor the memory of
Tatsuo Endo for his lasting contribution to science and
industry. This paper briefly reviews the original work
followed by a series of major events that have led to the
widespread acceptance of the method.
2. Endo's Original Work
The original presentation of the rainflow method was
contained in a series of two papers presented to the
Kyushu District Meeting of the Japanese Society of
Mechanical Engineers in November, 1967:

T. Endo, K. Mitsunaga and H. Nakagawa, Fatigue of


Metals Subjected to Varying Stress - Prediction of
Fatigue Lives
T. Endo, K. Mitsunaga, H. Nakagawa and K. Ikeda,
Fatigue of Metals Subjected to Varying Stress - Low
Cycle. Middle Cycle Fatigue.
This was followed by the third paper in March of 1968:
M. Matsuishi and T. Endo, Fatigue of Metals
Subjected to Varying Stress - Fatigue Lives under
Random Loading.
Three additional papers were published including the
first English language publication of the method by the
original authors:
T. Endo, M. Matsuishi, K. Mitsunaga, K. Kobayashi
and K. Takahashi, Rain Flow Method - the Proposal
and the Applications in Memoir Kyushu Institute
Technical Engineering, 1974
T. Endo, K. Mitsunaga, K. Takahashi, K. Kobayashi,
and M. Matsuishi, Damage Evaluation of Metals for
Random or Varying Load - Three Aspects of Rain Flow
Method in Proceedings of 1974 Symposium on
Mechanical Behavior of Materials
T. Endo, K. Kobayashi, K. Mitunaga and N. Sugimura,
Numerical Comparison of the Cycle Count Methods for
Fatigue Damage Evaluation, and Plastic-Strain
Damping Energy of Metals under Random Loading
presented at the 1975 Joint JSME-ASME applied
Mechanics Western Conference.
These six papers form the basis of the rainflow cycle
counting method. Details of the method will not be
repeated here since they will be described in detail in
the next paper in the symposium. The early papers all
have diagrams of the stress strain response for simple
variable amplitude loading histories. An example from
the first paper is given below.
7- (kg/W)

El 5. fcfiM*^U**^*tf<^*
4 XiM *>*&*? **&&&J $-&*&
' (5. SOtt **JX

Figure 4 and 5 from Endo's first paper.

The cycle counting method was called " the effective


range count method. H Endo writes " . . . the method
corresponds to the work hardening properties of metals
under complex change of strain. It is compatible with
the abrupt change of tangent modulus which is
conditionally observed under the process of reloading
. . . " A cycle is identified when the material
remembers its prior deformation history and changes its
tangent stiffness to follow the original loading path.
Only closed hysteresis loops that are formed during a
short repeating history are shown in the first two papers
as shown in this example.

12fe.**-:& KM * %ru&.

(3 7fc) fctt.-3MUB.Mfc fS 7(b) *3kfij5*>*!l

Figure 6 and 7 from Endo's first paper.


Half cycles are introduced in the third paper. All of
the stress strain curves and hysteresis loops were
plotted in terms of shear stresses and strains because
all of the original experiments were conducted in torsion
on a solid bar.
The first English language description is the
Masters Thesis of Masanori Matsuishi in March of 1969.
In his thesis, he described the effective range count
method in great detail with many examples in about six
pages of text. At the end, he describes an alternate
procedure for obtaining the same ranges and called it
rainflow. This description took two paragraphs and
contains these familiar rules:
The rain-flow started from a maximum ( minimum )
peak stops when a following maximum ( minimum ) peak
larger ( smaller ) than the peak appears.
If a rain-flow meets the rain-drop failed from a
upper roof the rain-flow stops at that place.

He implemented these rules in a FORTRAN program that


required 350 lines of computer code for an IBM 360
computer.
Endo in the concluding remarks of the 1974 paper notes "
In order to have better prediction of the effects of mean
stress, the cyclic hardening or softening of metals,
nonlinear damage law, the extension of the method to the
crack propagation period etc. must be investigated." All
of the have indeed been studied and with the exception of
nonlinear damage laws are widely used in engineering
practice.
3. Dowling's Confirming Experiments
Dowling [1] published the first description of
rainflow
in 1972 in the Journal of Materials. He conducted an
extensive series of axial strain controlled fatigue tests
with variable amplitude loading where he concluded " . .
. the counting of all closed hysteresis loops as cycles
by means of the rain flow counting method allows accurate
life predictions. The use of any method of cycle counting
other than range pair or rain flow methods can result in
inconsistencies and gross differences between predicted
and actual fatigue lives." He conducted tests to
investigate the mean stress effect with a number of
loading histories designed to differentiate between the
average mean stress of the loading history and the mean
stress of each cycle. An appendix of this paper
describes the rainflow method in detail and gives
examples of the half cycles identified by rainflow
counting and the corresponding stress-strain hysteresis
loops.
4. SAE Round Robin Test Program
The Society of Automotive Engineers (SAE) Fatigue
Design and Evaluation Committee began a testing and
analysis program to evaluate cumulative damage models.
[2] A number of experiments were performed on a notched
specimen with three variable amplitude loading histories
that were obtained from ground vehicles on a test track.
In addition to providing valuable data, the program
served a common forum to discuss techniques for life
predictions. Most of this work was being accomplished by
industrial participants with real problems to solve
rather than by researchers at the universities. Life
prediction software developers were using the extensive
data generated in this program to validate their codes.
Soon a common feature of all successful models began to
emerge. Rainflow counting or some equivalent counting
method was used to determine the equivalent constant
amplitude cycles from the variable amplitude loading
histories.

By 1975 minicomputers became commonplace in many


industrial laboratories and life prediction software
developed for them. Computers were taken out of the
laboratory and into the field. These computers required
more efficient algorithms to process the large amounts of
data that could be collected in the field during a one or
two hour test. Downing [3] was instrumental in this
development. The first microprocessors became available
and commercial products for rainflow counting became
available in the late 1970's. Ten years after its first
publication in the Japanese literature the rainflow
counting method was in widespread use throughout the
world for fatigue analysis.
5. ASTM Cycle Counting Standard
Many variations of the rainflow counting method have
been proposed and used. Standard terminology and
methodology was needed. The American Society for Testing
and Materials (ASTM) Committee E9 on Fatigue adopted the
first standard for cycle counting methods for fatigue
analysis.
The standard E1049 Standard Practice for
Cvcle Counting in Fatiaue Analysis was developed and
approved by ASTM in 1985 using the simple rules for
rainflow cycle counting that were suggested by Downing.
6. Application for Multiaxial Loading
A strategy for fatigue life estimates for multiaxial
nonproportional loading histories was proposed by
Socie[4].
Three dimensional stress strain simulation is
used to obtain the stresses from the measured strains
after which a tensor rotation is used to determine the
stress and strain history on any potential failure plane
in the material. This results in a stress and strain
history for each potential failure plane in the material
as shown in this example from reference 4.

Multiaxial Stress and Strain Time History


7

Fatigue damage is computed by rainflow counting the


strain history to identify cycles. Subcycles are easily
identified as the strain is increased from point A to
point B. In multiaxial loading the corresponding stressstrain response have hysteresis loops which are on the
outside of the hysteresis loop formed by the major cycle.
1000

CL

I
-1000
-0.005

0.005

Multiaxial stress-strain response


Compare this figure with Endo's figures. The normal
strain history is counted on each plane and fatigue
damage computed from a tensile damage model. The shear
strain history is also counted and a shear damage model
is used to compute shear damage. The plane experiencing
the greatest damage is identified as the critical plane
for crack nucleation.
7. Application to Crack Growth
Crack growth measurements were made during the SAE
testing program. In 1976, Nelson and Fuchs[5] used the
overall range method which is a counting method that is
equivalent to rainflow to make good estimates of the
crack propagation lives. Later, Socie and Kurath[6]
conducted a series of tests with loading histories that
were designed to differentiate between range and rainflow
cycle counting methods. They concluded that the rainflow
method gave the best life estimates and that range
counting could lead to nonconservative results.
8. Application to Load History Reconstruction
Instrumentation for obtaining rainflow counted data
is available to collect months of operating data because
it is stored in the reduced histogram format. There is
considerable interest in reconstructing a time history
from this data for cycle-by-cycle damage analysis or to
develop a loading sequence for fatigue testing. The new

sequence is not the same as the original but does have


the same fatigue damage. Perrett [7] concludes in his
work, " In general the process of rainflow counting and
reversed range-mean-pairs reconstruction produces
sequences that create fatigue damage at the same rate
that it accumulates in the original. In the majority of
the tests described in this paper reconstituted lives
were the same as the originating FALSTAFF lives in
conditions dominated by crack initiation and crack growth
and also in conditions where fretting forces are
significant.
Bishop and Sherratt[8] provide a method for the
estimations of rainflow range density functions using
statistics computed directly from power spectral density
data. The rainflow range mechanism is broken down into a
set of events which can be analyzed using Markov process
theory.
Summary
The Rainflow counting method has been shown to give
the best fatigue life estimates. It has been applied for
both fatigue crack initiation and fatigue crack
propagation problems with great success. The method has
been extended for multiaxial loading and to
reconstructing fatigue test sequences.
References
1. Dowling, N.E. "Fatigue Failure Predictions for
Complicated Stress-Strain Histories" J. of Materials, Vol
7, No 1 , 1972, 71-87
2. Wetzel, R.M. editor Fatigue Under Complex Loading;
Analysis and Experiments SAE, AE-6, 1977
3. Downing, S.D. and Socie, D.F. "Simplified Rainflow
Counting Algorithms" Int. Journal of Fatigue, Vol 4, No
1, 1982, 31-40
4. Socie, D.F. "Multiaxial Fatigue Damage Assessment"
Low Cvcle Fatigue and Elasto-Plastic Behavior of
Materials, 1987
5. Nelson, D.V. and Fuchs, H.O. "Prediction of Fatigue
Crack Growth Under Irregular Loading" ASTM STP 595,
1976, 267-291
6. Socie, D.F. and Kurath, P. "Cycle Counting for
Variable Amplitude Crack Growth" ASTM STP 791, 1983,
II19-II32

7. Perrett, B. "An Evaluation of a Method for


Reconstructing Fatigue Test Loading Sequences From Load
Data Acquired via Rainflow Counting" Proc. 14th ICAF
Symposium, 1987
8. Bishop, N.W.M. and Sherratt, F. "A Theoretical
Solution for the Estimation of Rainflow Ranges From Power
Spectral Density Data" Fatigue and Fracture of
Engineering Materials and Structures, 1990, 311-326

10

You might also like