You are on page 1of 4

Stress Analysis and Design

for Cyclic Loading


P. Carter
Staff Consultant, A finite element implementation of rapid cycle analysis is described and demonstrated. It
Stress Engineering Services, Inc., forms part of a comprehensive framework for static structural analysis which consists of:
Mason, OH 45039 linear elastic analysis, limit load or nonlinear elastic analysis, and rapid cycle analysis.
e-mail: pcarter@ses-oh.com This approach allows for complex material and loading behavior, but is computationally
Mem. ASME more efficient and easier to perform than full inelastic analysis. It indicates more complex
behavior than can be inferred from linear elastic analysis. The objective of this paper is
to calculate shakedown, reverse plasticity, ratcheting, and the increase in strain rate as a
result of cyclic mechanical and thermal loading. Results are presented in the form of
interaction diagrams, similar to the ODonnell-Porowski plot in the ASME BPV Code,
which are effective design tools. S0094-99300001604-8

1 Introduction The objective of this paper is to show how rapid cycle analysis
generates this information for a structure. The methodology and
Effective techniques for structural design and assessment rely
solution algorithm are given in Sections 2, 3, and 4. Examples of
on a trade-off between rigor and economy. In terms of finite ele-
rapid cycle analysis and the generation of interaction diagrams to
ment models, the most economic option usually consists of linear
characterize a structure under cyclic loading are given in Section
elastic material models, minimal mesh refinement, small strains,
5.
and no nonlinearities because of large displacements, contact, etc.
The effectiveness of the model typically relies on:
2 Methodology
The realism or conservatism of the assumptions. There are three building blocks that have to fit together to make
The existence of design codes or rules which allow results of a useful design tool for cyclic loading. These are:
linear elastic analysis to be interpreted for more complex real
situations. 1 Finite element rapid cycle analysis.
The ability of the analyst to interpret linear elastic analyses 2 Use of n-power creep law and the limit as 1/n tends to zero.
for realistic materials and structural details, for example, 3 Generation of interaction diagrams.
notches. These will be discussed briefly in the next sections.
This paper presents a framework for structural analysis which is
more rigorous than elastic analysis, but does not sacrifice practi- 3 Rapid Cycle Solution Definition
cality or efficiency. The focus is on inelastic creep and plasticity The definition of the rapid cycle concept was due to Ponter 1.
ductile materials, subject to quasi-static cyclic mechanical and In its simplest form the theory is based on an elastic, n-power
thermal loading. creep material. Strain is the sum of elastic, thermal, and creep
It is proposed that three kinds of analysis are generally neces- inelastic components
sary to characterize such structural problems:
e t c
1 Linear elastic analysisthis gives a preliminary indication of
structural behavior. Stress is related to elastic strain via the stiffness E
2 Limit load analysis, or non-linear elastic stress analysis E e
these analyses give similar information in the case of highly non-
linear materials. Which technique is used may depend on their Inelastic strain rate is proportional to Mises or effective stress to
relative computational complexity. Elastic-plastic and nonlinear the power n
elastic analysis is now hardly more complex than linear elastic c / tk n1 /
analysis in most commercial finite element codes. The benefit of
these analyses is that effects of stress redistribution caused by Thermal strain is proportional to temperature minus a reference
plasticity and creep are clarified. Without this information, stress temperature.
classification schemes are necessary. t TT 0
3 Rapid cycle analysis for load cycles. Stress classification
schemes and design rules are generally less adequate for cyclic Consider a structure with volume V and surface S, subject to
and thermal loading than for steady loading. For such problems, cyclic mechanical loads P(t) over S, and cyclic thermal strains
reliance on design codes rules may not be a satisfactory option for t (t) over V. P(t) and t (t) have the same period T.
designers, particularly when new or different classes of problems Let the elastic solution for loading P(t) and t (t) be a stress
are faced, and when accurate information is required on factors of history e (t). The rapid cycle solution is defined by the constant
safety, that is the margin between design and failure. For these residual stress field rc , such that


problems, structural assessment requires determination of: shake- T
down, reverse plasticity or ratcheting behavior in the absence of c k n1 / : e t rc dt
shakedown, and creep strain accumulation. 0

can be derived from a displacement field u. This means that the


Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the PVP
creep strain c calculated from the elastic solution plus a con-
Division, July 26, 1999; revised manuscript received July 6, 2000. Associate Tech- stant residual stress field, must be kinematically admissible. There
nical Editor: C. Jaske. are a number of theorems 13 which give more detail on the

Journal of Pressure Vessel Technology Copyright 2000 by ASME NOVEMBER 2000, Vol. 122 427

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


rapid cycle solution, particularly its bounding properties. Of im-
portance is that the rapid cycle solution satisfies a minimum prin-
ciple. Define

0 V
k n1 e t dV dt

Then, the rapid cycle solution satisfies


rc
The calculation of the rapid cycle solution uses an algorithm
which calculates a correction to and then calculates the magni-
tude of the correction so that the energy term is minimized.

4 Rapid Cycle Solution Algorithm


The steps in the algorithm are as follows:
i Define the load cycle in terms of a finite number of steps.
ii Calculate and store elastic stress distributions for these
load steps.
iii Initially, assume zero residual stress. Subsequently, use the
updated .
iv Calculate the creep strain c associated with the stress
history e (t) .
v Calculate residual stress associated with c .
vi Calculate the value of such that is mini-
mized.
vii Update .
viii Go to step iii and repeat until the solution converges.
The solution may be characterized by the compatible strain field
c , or by the residual stress field rc .
ix Calculate the cyclic effective stress CES, which is the
constant stress which would produce strain c in time T.
x Calculate the shakedown stress, which is the maximum
Mises or effective stress during the cycle e (t) rc .
This algorithm was developed and applied for third order axisym- Fig. 1 ASME III ratcheting and creep strain accumulation
metric elements. diagram

4.1 Effect of Creep Exponent n. The output of a rapid


cycle solution is the maximum elastic plus residual stress during
the cycle, and the cyclic effective stress, at each point in the body. and mechanical loading. There is clearly a need to generate such
The maximum values of these quantities over the volume V char- an interaction diagram for any structure and cyclic loading.
acterize the structure and loading, for the creep law used in the The procedure followed to generate an interaction diagram for a
analysis. The maximum stress over the volume and over the cycle particular creep exponent is as follows:
is an upper bound estimate of the shakedown stress. When differ-
ing values of the creep exponent n are used, it is found that the i Perform mechanical and thermal elastic analyses to gener-
maximum stress and creep effective stresses vary nearly linearly ate say 12 combinations of mechanical and thermal loading.
with 1/n. As n gets larger, the maximum stress decreases and ii Normalize maximum elastic stresses for each combination
provides a better estimate of the shakedown stress, that is, the by these shakedown and cyclic effective stresses. This gives nor-
lowest value of the yield stress for which the structure would malized coordinates of shakedown and effective stresses.
shake down. As n gets larger, the creep effective stress typically iii Join coordinates to give contours.
increases. For n1, CES is unaffected by the thermal stress, as iv Generate contours of cyclic effective creep stress.
indicated in Fig. 2. Increasing material nonlinearity n accentu- The manipulations in steps iii and iv are readily performed
ates the effect of thermal cycles. This can be understood by con- using a spreadsheet.
sidering the effect on strain accumulation of a uniaxial tensile
stress history consisting of followed by for equal times.
For n1, CES. In the limit of 1/n0, CES. The limit-
ing case therefore generates the shakedown stress and the upper 5 Applications
bound for the cyclic effective stress. In this paper, a finite value of
n10 will be used as a limiting case. 5.1 Thin Shell Subject to Pressure and Thermal
Cycling. Figures 2 and 3 are the results for the BODP problem
4.2 Interaction Diagrams. Figure 1 is a well-known inter- on a thin spherical shell, namely constant pressure and cyclic
action diagram, due to Bree 4 in its early form, developed by through-wall thermal loading. At one extreme of the cycle,
ODonnell and Porowski 5, and now part of the ASME III NH stresses are due to pressure loading alone. At the other extreme,
high temperature code 6. The regions of elastic, shakedown, stresses are due to pressure and a through thickness thermal gra-
reverse plasticity and ratcheting behaviors are shown. The BODP dient. The procedure described in the foregoing produced a con-
plot refers to the case of constant membrane stress and cyclic verged solution which had stresses which varied by less than 1
through-wall bending in a uniform cylindrical shell. In the ASME percent from one step to the next. In these plots, the heaviest line
III NH code it is used as a general way of assessing cyclic thermal is the shakedown boundary, and the lighter lines are lines of con-

428 Vol. 122, NOVEMBER 2000 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 2 Thin shell: pressure and cyclic thermal loading n 1
Fig. 4 Tubeplate: pressure and thermal cycling n 10

stant creep effective stress CES within the shakedown boundary.


Continuation of the CES lines outside the shakedown boundary is The shape of the shakedown surface is in reasonable agreement
with the lightest lines. with Fig. 1. Unlike the BODP plot, the thermal stress axis has a
Figure 2 is the solution for the creep exponent n1, and Fig. 3 finite intercept for the creep effective stress for the rapid cycle
is for n10. The solution for n1 has creep strain accumulation solution; but in the limiting case of zero mechanical load, there
independent on thermal cycles, and the maximum stress in the can be difficulties with round-off errors, and so the results here
cycle is the primary stress plus the thermal stress amplitude. need to be treated with caution. However, the numerical difficul-
The results in Figs. 2 and 3 are only strictly valid in the shake- ties probably reflect real physical instabilities, in the sense that a
down region, since plastic strains are not calculated in the proce- net zero creep strain accumulation over the cycle depends on an
dure. In this region, residual stresses can be set up so that no exactly balanced cycle, which is unlikely given the temperature
plasticity occurs in the rapid cycle, and so the creep analysis used differences in the different parts of the cycle. The CES contours
in this approach is valid. This is indicated by the heavier CES generally predict higher creep strain accumulation rates than the
contours within the shakedown boundary. The CES contour be- BODP plot.
ginning at the value 1 on the primary stress axis is therefore not In the elastic region of the BODP plot, thermal stresses have no
strictly meaningful if it goes outside the shakedown boundary. It effect, whereas the rapid cycle solution shows an effect for all
really defines the shape of the other CES contours within the thermal stresses. There are some possible explanations to be
shakedown boundary. It does also, however, provide an indication noted. The BODP problem has creep strain occurring in one part
of the structural behavior if shakedown does not occur. If the CES of the cycle only. The rapid cycle analysis has creep strain occur-
contour and the shakedown contour are similar, then it is likely ring throughout the cycle. This combined with multiaxial effects
that ratcheting will occur if the structure fails to shakedown. If in the spherical shell probably accounts for the differences in the
they diverge, then ratcheting is less probable. This would have to shape of creep contours in the region of the knee, where the
be confirmed using a full inelastic analysis. horizontal and angled shakedown lines meet. The rapid cycle
analysis indicates that the ratcheting region might extend into the
region above the horizontal shakedown line.
5.2 Thick Plate Subject to Pressure and Thermal
Loading. Consider a circular plate subject to constant pressure
loading, and through-thickness cyclic thermal loading. The plate
has a thickness-diameter ratio of 1 to 10. Figure 4 is the interac-
tion diagram for this problem.
The primary stress intercept is 1.74 yield stress, showing the
strength associated with the built-in edge. The shape of the creep
contours suggests that if shakedown is not achieved, the structure
might ratchet, except possibly in the region where primary
stresses are less than 20 percent of yield. As before, the plots
distinguish between contours of creep effective stress inside and
outside the shakedown boundary.
5.3 Nozzles Subject to Cyclic Pressure Loading. Figure 5
shows the effect of adequate reinforcement on a nozzle. Three
geometries are represented. The plain spherical shell has D
200, T2. The unreinforced nozzle has a cylinder with d
100, t2. The reinforced nozzle has the amount of reinforce-
ment based on the area replacement rule.
The loading consists of cyclic pressure loading only, varying
Fig. 3 Thin shell: pressure and cyclic thermal loading n from a zero mean pressure to constant loading. Figure 5 shows the
10 shakedown boundaries for the three cases. The advantage of rein-

Journal of Pressure Vessel Technology NOVEMBER 2000, Vol. 122 429

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


6 Conclusions
The technique of rapid cycle analysis has been demonstrated as
a tool to understand the cyclic behavior of structures. It can be
used to generate interaction diagrams giving the shakedown
boundary, and the contours of cyclic or effective creep stress
within the shakedown region. The results for n1, and for a high
value of n, say 10, provide the limits of reasonable material be-
havior. This provides a powerful design tool so that designers can
understand structural behavior and factors of safety over a range
of loading conditions.
Detailed comparisons with other techniques are beyond the
scope of this paper. The only available alternatives which give
similar information are, to the authors knowledge, full cyclic in-
elastic analysis using an elastic visco-plastic material, and the
ASME III NH rules. It is difficult to visualize the generation of
the interaction diagrams using full inelastic analysis, although
analysis of one combination of thermal and mechanical loads
would be feasible if time-consuming. The ASME rules do not
constitute a directly comparable technique, since they cannot be
truly general as the different interaction diagrams in this paper
show.

References
1 Ponter, A. R. S., 1972, Deformation, Displacement and Work Bounds for
Structures in a State of Creep and Subject to Variable Loading, J. Appl.
Fig. 5 Comparison of nozzle behaviors Mech., 39, pp. 953963.
2 Ainsworth, R. A., 1977, Bounding Solutions for Creeping Structures Sub-
jected to Load Variations above the Shakedown Limit, Int. J. Solids Struct.,
13, pp. 971980.
forcing the nozzle is obvious, in that its strength and behavior 3 Carter, P., 1985, Bounding Theorems for Creep-Plasticity, Int. J. Solids
approach that of the plain shell. The unreinforced nozzle has only Struct., 21, pp. 527543.
60 percent of the strength of the plain shell. 4 Bree, J., 1967, Elastic-Plastic Behavior of Thin Tubes Subjected to High
The unreinforced nozzle shows a region close to the case of Internal Pressure and Intermittent High Heat Fluxes with Applications to Fast-
Nuclear-Reactor Fuel Elements, J. Strain Anal., 2, No. 3, pp. 226238.
zero minimum load where the behavior of the nozzle could be 5 ODonnell, W. J., and Porowoski, J. S., 1981, Biaxial Model for Bounding
limited by shakedown. The reinforced nozzle is more like the Creep Ratcheting, ORNL Report ORNL/Sub7322/2, ORNL, Oak Ridge, TN.
plain shell. 6 ASME III Division I Subsection NH. Appendix T, 1998.

430 Vol. 122, NOVEMBER 2000 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like