THERMAL RATCHETING OF A BEAM 

THTHERMAL RATCHETING OF A BEAMABSTRACT

Thermal ratcheting has been analytically investigated for a beam
subjected to a linear temperature variation across its solid rectangular
cross section. A linear strain-hardening material response exhibiting an
idealized Bauschinger effect was assumed. Formulas are given for the
associated strain accumulation which are valid over a large range of

strain hardening, temperature variation, temperature cycles, and axial

i
load. Specific results are tabulated for the materials and operating
conditions typically associated with liquid metal breeder reactors.
 INTRODUCTION
Much of the current motivation for understanding and estimating
therMl ratcheting is due to the design requirements of nuclear reactors.
Presently, the codes require designs which exclude ratcheting based
on the conservative criteria developed for an elastic-perfectly plastic
one-dimensional continuum [1,2]. However, the extreme operating conditions
envisioned for liquid metal breeder reactors may require designs which
i

account for some ratcheting. Since the materials proposed for these
reactors exhibit definite Bauschinger effects [3J» then information on
kinematic hardening ratcheting is required to optimize designs.
Comprehensive studies of the thermal ratcheting of an elastic-
perfectly plastic one-dimensional continuum have been made and reviewed
by Burgreen [4]. In the discussions which follow the term ratcheting
shall always imply thermal ratcheting as defined by Burgreen. For
strain-hardening materials some idealised problems have been considered.
Idealizations are required because of the variety and the complexity of
strain-hardening metal response which exist even for one-dimensional
stress states.
Thermal ratcheting of a one-dimensional continuum exhibiting an
isotropic hardening material response has received some attention [2)»
wherein a very small constant hardening modulus was assumed. The strain
accumulation per cycle was found to decrease with cycling resulting in
a finite strain accumulation for an infinite number of cycles. Thus,
the constant strain accumulation per cycle of a perfectly plastic material
. greatly overestimates isotropic strain-hardening ratcheting. The range
of mechanical and thermal loads for which ratcheting occurred was found
to be very nearly the same for both types of hardening.
 A study £51 which compared the basic ratcheting mechanism assoclatad
with kinematic and isotropie hardening has base recently completed.
Tba two-element assembly used by Burgratn (6] to Illustrate the basic
per£eetly-plastic ratcheting mechanism was employed. No assumption was
atade with regard to the sixe of the constant hardening aodulua. Both
the aanunt of strain aecuaulation and range of aechanlcal and thenul
loads for which ratcheting occurred was found to be greater for isotropic
*
hardening materials than for klneatatlc hardening aatarlals, especially
at higher tesperatures and smaller hardening Moduli. Although such
results are useful for assessing the need for kinematic hardening
ratcheting calculations* its usefulness in the design of components ±9
limited to those which can be idealized as a two-element assembly.
Many nuclear reactor components are better represented tm beams, plates,
or shells having solid cross sections„
 PROBLEM DESCRIPTION
The load response relations for a solid rectangular beas having a
cross-sectional area of 2A are presented here. Plane cross sections are
assumed to reaain in plane and perpendicular to the axis of the bean,
in keeping with the analogy to a cylindrical or spherical shell proposed
by Bree [2]. The one-diaensional piecewise linear eiastic-klneaatic
hardening material response shown in Fig. 1 is assuaed. Mo restriction
is aade regarding the size of the constant hardening aodulus. Initially,
a constant centroidal axial load of P is applied resulting in a unifora
•tress which is less than the elastic liait a.
A linear teapctrature distribution across the depth of the beaa is
assuaed. The difference between the teaperature at the end of the heating
half cycle and the ambient teaperature is

where H Is the half depth of the beaa and T is the teaperature at the top
of the beaa y - H. Distance aeasured perpendicular to the centroidal
axis is represented by y. At the end of the cooling half cycle the
aabient teaperature is again attained. Without loss of generality*
the aabient teaperature is assuaed to be zero and the aaxiaua teaperature
variation in (1) is between -T and 0. Strains associated with other
teaperature Halts* say (-T + extra teaperature) to (extra teaperature),
can be obtained by superposition of the unifora teaperature strain
associated with (extra teaperature) upon the results developed herein*
According to Fig. 1 the stress strain relation during the initial
heating half cycle

I <»y) + aAT
where Y is 1 or 0 if the elastic limit is or is not exceeded, respectively.
For Increasing (decreasing) c, the minus (plus) sign is used in (2).
The coefficient of thermal expansion is a, Young's elastic modulus is E,
and fc is related to the linear hardening modulus gE by

If the fiber of the beam is plastically deformed during the n-1 half
cycle* then for the nth subsequent, half cycle of heating or cooling
the material response is

where Ac , Ao , A(AT) are the changes which occur in the strain, stress,
and temperature, respectively, during the half cycle, y. Is +1, -1,
or sero for Aon positive and Acn~x. negative, Aan negative and AoB"*x, positive,
and A«rn and AoB.—L, the same sign, respectively. If the fiber of the beam

is elastically deformed in the (n-l)th half cycle, plastically deformed

in the nth half cycle, elastically deformed in the (n+l)th half cycle,
etc., then during the nth, (n+2)th, etc., half cycle the material response
is given by

Ae - -=* + k(Ao +Aff_.) + aA(AT) (4b)

B B n n°*x
Elastic cycling response is included (4a). Note that the subscripts on
parameters,associated with deformation,indicate the number of the current
half cycle.
 ELASTIC CYCLING. SHAKEDOWN, AND PLASTIC CYCLING
Tor a given beam, several modes of deformation can occur depending
upon the magnitude of the axial load P and the temperature range T.
The loading conditions which produce elastic cycling in every half
cycle are the same for any type of initially isotropic hardening response

Relation (5) defines Region E in Fig. 2.

Shakedown is defined as cycling wherein plastic deformation occurs
during the first heating half cycle, but only elastic deformation occurs
in subsequent cycles. Two different first half modes of plastic
deformation can precede shakedown. Plastic deformation can occur either
in the top beam fibers alone, Fig* 3a, with £, £ H, or in both top and
bottom fibers, Fig. 3a, when 1. < H.
For the stress distribution of Fig. 3a with 1. i H, the constant
•slues of e. and h. at the end of the first half cycle are determined
by equilibrium condition

LdA (6)

end the yield condition o. » a at y • h,. Substituting the appropriate

forms of the stress-strain relations (2) into these conditions gives
two expressions relating e, to h., respectively
aTE „ . x, (7a)

(7b)
Combining (7a) and (7b) gives
The mode of. deformation represented by Fig. 3a with 1. > H remains valid
as long as
-0 at y - -H (8a)

(8b)

With (7), both of (8) can be expressed in terms of P and T,

respectively. One can show that if F and T satisfy (9) and violate (5),
Region S1 in Fig. 2, then only elastic response occurs in subsequent
half cycles. Curves C. and C, in Fig. 2 are graphical representations
of (9a) and (9b), respectively, when the equality sign is employed.
If (8a) is violated and (8b) is satisfied, then the other shakedown
•ode represented by Fig. 3a with t, i H may occur. The values of e.,
h., and I. arc determined by the equilibrium condition (6), 0 • 0
it y - h., and 0. « -0 at y - -A-. Combining these conditions with the
appropriate form of the stress-strain relation (2) gives

H-STE-H <»•>>

In the second half cycle, a probable stress distribution including plastic

deformation is given by Fig. 3b. Assuming I, > hi» n ? >
*i» the

conditions of equilibrium for any half cycle

 IIA
Aa dA ,
n
.
(11)

Aa, « -2a at y * *,, and Aa. • 2a at y * -h, can be combined with the
z y & £ y *
appropriate form of the stress-strain relations (4) to determine As.,
h_, and 1« to *>e
""" ("a)

(12b)
• "F"T"aTE
Comparing (12) with (10), one sees that &, > & aad h^ > ^ for positive
, if

y
and
u
oTE > 4a (14)

For oTE which violate (14) but not (13), Region S^ of Fig. 2, shakedown
occurs.
When both conditions (13) and (14) are satisfied, Region P in
Fig. 2, one can show that repeated plastic strain cycling occurs with
a total change of strain

(15)

in each half cycle. The values of h and % are given by (12b). The
curve C. in Fig. 2 is represented by (13) with the equality sign. Note
that curve C_ is asymptotic to the line

Further, as k -*- » (0 -> 0), the intersection of curves C-, C., and CL
approach the coordinate lx a , 4o 1 and the line given by (16) coincides
with the line P - 0.
 8

RATCHETING

Obviously, ratcheting cannot occur for the deformation modes described

in Figs. 3a and 3b since the central beam fibers are never permanently

deformed. The deformation mode illustrated in Fig. 3c can be expected

to produce ratcheting.

Case 1

As might be anticipated, the first heating half cycle analysis and

i

results, equations (lOa-c), are the same for the deformation modes of

Fig. 3a, with I. < H, and Fig. 3c wheu the h are defined as shown.

For the nth subsequent heating half cycle, the analysis is based on the

simultaneous solution of the equilibrium condition (11), the condition

ACT » 0 at y • -h , and Ac • -2a at y • -I . For the (n+1) cooling

half cycle the conditions are the same, but the latter two are evaluated
at y * +h ., and y = +1 .., respectively. That is, I and h always
nrj. MM. n n
are defined so that they are positive. Combining the appropriate form

of the stress-strain relations (4) with the three conditions listed

above gives the explicit results

4o

H + (-W J (17C)

where tv. is given by (10a).

Fiom (17a) and (17b) one can deduce that h < % < %., Therefore

the condition %. > 0 is the condition which determines the values of

T and P for which the deformation mode of Fig. 3c is valid. Since the

results for the first cycle are the same as determined for the deformation
•ode of Fig. 3a with Jt. < H, the bounding curve CL is applicable.
Region (FR-) in Fig. 2 is the region for which (17) is valid.

The strain at the end of any half cycle where n 2 2 can be determined
by summing (17c) and (10s)

iB+ ^ [ T + ••
enE - e l E + ^ j ^ + •• ++-ir]
....
+ ^ [1 + (-1)"]
» | + *F ti (18)

where e^E is given by (10c). Combining (18) with (17a) results in a

geometric series which has the sum'

n V • « [I + (-I)", • toy £ )
Note that the sum is finite for n -*• ».
Case 2

The results of Case 1 for the P and ctTE defining curve C. indicate
that another mode of deformation occurs for which strain accumulates.

Figure 3d illustrates the deformation mode, where again the h values

n
alvays are defined so as to be positive. During the first half cycle,
the two conditions to be satisfied are the equilibrium condition (6)
and o . « a at y » -h.. Combining these conditions with the appropriate
form of the stress-strain relations, one finds

which is the negative of (7c). The value of e^, is found by combining

(10c) and (20).

For subsequent half cycles, the equilibrium conditions (11) and the
condition La = 0 at y =» (-1) h must be satisfied. Combining these

conditions yields, for this case, a recursion relation for h

n
(21)
 10

Knowing -^ from (20), Ae can be calculated via (17c). As can be seen

by inspection of (21), the deformation in the nth half cycle depends upon

the deformation in the (n-l)th half cycle. Both h and e decrease as

n n

cycling continues. Ultimately, as n + «, h •*• 0 and Ae •+• 0. The

dependence of the current deformation on past deformation is unique to

this mode of deformation.

To determine h for a finite number of cycles, the successive

i
numerical evaluation of (21) is required. Little chance exists that an
explicit expression for h can be determined for such a complicated

recursion relation. The change in h with n is easily calculated on a

digital computer for the range of values of k of interest. The results

are plotted in Fig. 4 and can be used for computations. The starting

( hj\ p

1 - -g-j, for N = 1, corresponds to the value of (20) for T T * O ;

the largest value of — for a given k. To determine fl — — 1 one simply

enters the graph

h
on the appropriate k = constant curve at the N corre-
l
sponding to — , say N » N,, and then reads the value of the curve at
p

H » N, + n - 1. For instance, suppose k = 999 (g = 0.001), -^ • a ,

and fl - - ^ p ) is desired. According to (20), fl - - — ] . - 0.061. For

the k = 999 curve in Fig. 4, 0.061 corresponds to H • B. • 1. Therefore
example, suppose
p •0.95 1*400
-57- = [1 \
a —,I oTE « 2afor
, k the
» 999,
/ fyooA
and [1 =—I is
at N = 400 one reads y = 0.80 y answer. \As another
a f
p / hi \ / fyooA
desired. According to (20), [1 - -g-j » 0.450. On the graph of Fig. 4,
h
f 400^
0.450 corresponds to N = N- = 72. The value of II =—I is read at

N - 471 to be 0.831.

To determine Ae for a finite n, — is obtained as described above

n H
and substituted into (17c). The value of c
is obtained from (18) which
a
n hn
in its present form requires calculation of ) — . A more convenient
m»2 H
 u
font can be obtained by use of (21). Noting that (21) can be rewritten

continued resubstitution of (21*) leads to the relation

(22)

Combining (22) with (18) gives ,

E. h 2 h 2-j
1 - r=-1 - ll - -g-j + —r~ [1 + (-1) ]
where e- is determined by (10c) and (20). With (23), evaluation of e
(23)

only requires that — be determined as outlined previously. For n •*• <»

H
^aTE ii - U ±1 j I wx" ri + (_1\"1 (1L\
1 8 |_

The sum is explicit and finite.

P
As an example, suppose -rr = 0.95 a , aTE = 2a , k = 999, and
*** J . y
the strain at the end of the 400th temperature half cycle is desired.
The value of [l - ~ | = 0.450 and fl —] = 0.831 is determined
E H
\ J V J
from (20) and the graph of Fig. 4, respectively, as previously described.
The value of e.E is determined from (10c) to be 0.55 a . Entering these
values into (23) yields e 4 oo E ~ 1 2 3 * 4 3 ° • F o r n * "» e««,E * 200.73 o
as determined from (24).
The deformation mode of Case 2 is valid for all cycles where h £ 0
and Jt 2: H. The first inequality is seen from (21) to be satisfied for
all n. However, the second inequality, which is equivalent to the condition
|Ao | < 2a , requires that
U j

Thus, Case 2 cycling will occur for all cycles (n -> °°) only if aTE < 4a ,
Region FR in Fig. 2.
 12

Case 3
For the remaining values of T and P, Region (FR-FR.) of Fig. 2,
the deformation will be that of Case 2 for (R-l) half cycles and that
of Case 1 for subsequent half cycles, n 2 R + 1. The Rth cycle, during
which the transition from Case 1 to Case 2 type cycling occurs, is the
one for which the inequality of (25) is first violated.
For n < R, Fig. 3d represents a typical stress distribution and h ,
Ae , e , etc., are calculated as discussed in Case 2 above. During tha
Rth half cycle the stress distribution changes from that of Fig. 3d to
that of Fig. 3c. Following procedures similar to those outlined previously,
the following expression for h_ can be determined:
1

T"[_ ka
y
1 kS
[ --5£)' r 1
l "-^! H ^+ toJ <26>
Expressions (17b) and (17c) remain valid for £_ and Ae R t respectively.
For n > R, the solution is the same as for Case 1 except that Case I
deformation mode is beginning in the (R-tt)th half cycle. In direct analogy
to (17a)

and (17b) and (17c) remain valid. In direct analogy.to (19)

(28)
where — is given by (26) and e^ is determined as discussed in Case 2.
For n -*• », a closed form upper bound to e E can be obtained via (28)
• Note that the part of the last term involving R goes to zero for n •»• •,
hR
thus only values of e_E and -=- are required. An upper bound on c«E can
Kb. K
hn h-
be obtained from (23), by choosing -~- to be the smallest value of ~
E
hR
allowed by (25). Ah upper bound on — can be obtained from (21) and
n
 13

(25). By choosing ~ in (21) to b« eh* saallest value of -~ allowed by

(25), the ealculabl* value of ~ ~ U always larger than ~ . Th«

closeness of the upper bound to eJE depends upon the rate of change of
K
~ per cycle, which depends upon P* T, and k. Tht saaller the rate of
H
change, .the closer the approximation will be. Fortunately* the bounds
are poorest for the larger rates of change where the cycle by cycle
numerical
As anevaluation
exsaple, eis
^ Sthe easiesc.
will *>* determined for j^ * 0.95 « y * oTE • 60 ,
and k - 999. From (20), fl - -£) • 0.264 and Case 2 is valid as long
as [1 - -n-J < 0.667, according to (25). Frow Fig. 4, 0*264 corresponds
to M, - 21.5 and 0.66? corresponds to H m 213. Therefore, Case 2 revains
valid for ^ 191 half cycles, and R * 192 and fl - — " J • 0.660. With
V
^
Che above values (23), (26), and (17c) give c E - 274.37 a. •»192
— - 0.339,
m

and ^* 1 92 E * **02 ° y respectively. Using these results, C^QQE " 435.5 o

is detemined frosi (28).
 DISCUSSION
The values of (P,T) for which ratcheting occurs, call them (P,T) K>
arc dependent upon the hardening coefficient k. The (PtT) R and associated
•train accumulation will be determined for widely different k to show
extremes in behavior. Figure 1 shows the stress-strain curve for a material
with large strain hardening k » 3, and one with none k - •: a perfectly
plastic response. Also shown is the idealized response of a typical
cyclically hardened 304 stainless steel at 1200°F [7], for which k « 44.5.
The (P»T) R corresponding to k • 3 and k - 46.5 are indicated in the
legions FR, FR-FR1, and FR1 shown in Figures 2 and 5. The ratcheting
region bounding curve for k • • also is provided for reference in both
figures. The range of oTE/o included in the figures should be sufficient
for application to liquid metal breeder reactor design; since
aTE/c < 17 for typical cyclically hardened and annealed 304 stainless
steels [7] undergoing a 1200*F temperature change.

Comparison of Figures 2 and 5 show that the (P»?) R for widely

different k have most values in common: for k * 46.2 the (P,T)_ are
almost identical to those for k • •; and even for the extremely different
k « 3 and k • •», the ratcheting regions only differ in size by about
25Z. However, the amount of strain which accumulates during ratcheting
Is a sensitive function of k. To assess strain accumulation the previously
developed relations were assembled in a computer program which calculates
•train accumulation for any number of half cycles N, temperature otTE/o* ,
or axial load P/2Ao . Consider the strain which accumulates after the
first half cycle for an infinite number of cycles

«eP . e- - « x _ 221 U + <_!)*] (29)

Figures 2 and 5 show 6e£ for k • 3.0 and 46.2, respectively. Obviously,
 • • 15

for a given P and T, considerably more strain accumulates for larger k» P» and
T. However, the accumulation is most sensitive to the value of k. This fact is
perhaps better illustrated by Fig. 6, which shows for P - 0.70 the variation in
6e£ with aTE/o_ and several values of k.
Figures 2 and 5 also illustrate that 5e£ is relatively insensitive to changes
In dTE/o at large aTE/o_, as compared.to the sensitivity of 6e£ at swill values
of aTE/o . A similar but more extreme trend was found for a two-element sandwich
beam [5]: no change occurred in 6e£ for variations in oTE/a at large aTE/o ,
Another extreme trend of the two-element beam was that repeated plastic cycling
would occur for aTE/o sufficiently large. For the solid beam, a much lest
pronounced trend toward smaller 6e£ at larger aTE/o can be seen in Fig. 2 for
P/2Ao_ > 0.9. In general, more strain will accumulate for a sandwich beam than
for a solid beam of the same cross sectional area, except at high temperature
variations. The fact that repeated plastic cycling occurs for a sandwich beam
at high temperature variations, whereas ratcheting occurs for a solid beam,
means that for some conditions an optimal cross section may be determined which
minimizes ratcheting and cost while maintaining static strength.
The strain which accumulates for a finite number of half cycles
after the first half cycle

«cg - e N - ^ - *f [1 + <-l)n] (30)

shows variations similar to those of 6e£: for a given N, more strain
accumulates for larger P, T, and k. For different N and a given P,
T, and k; the rate of strain accumulation per half cycle is a decreasing
function of the number of half cycles. Additional trends are shown in
Figures 7 and 8, which are indicative of the general variation of fiejj
with N. In Figure 7, the range of the number of half cycles necessary
to accumulate 90% of «e£, N 9 Q % , is plotted versus k. For a given k, the
maximum values of N Q 0 » correspond to the smallest values of aTE/o , for
which ratcheting occurs* and the largest values of P/2Ao. The minimum
values of N g 0 . corresponds to the largest value of oTE/a - 20 and are
nearly the same for any P/2A» . In general, the greatest cyclic strain
accumulation rate always occurs for the largest valua of T, and it is a
relatively insensitive function of k and P. The minlatum value of the
cyclic strain accumulation rate is a sensitive function of k, and it
always occurs for the smallest values of T and the larger values of P.
t
The larger amounts of strain accuaulation and the lafgef cyclic rates of
•train accumulation associated with larger values of Tare vividly
Illustrated in Figure 8* The desirability of avoiding large temperature
variations is dear.
The calculation of strain accumulation is most tedious when 6ef. is
H

required for Cases 2 and 3» because consecutive cycle by cycle calculations

must be made instead of the evaluation of a closed form solution. A
method of obtaining a closed form upper bound solution to 6e p for Case 3
has already been outlined. Of course, the closed form solutions and
bounds for 6c p are upper bounds on def.. The closeness of these bounds
can be estimated from results like those presented in Fig. 7. For instance,
for typical reactor engineering materials and thermal fluctuations, if
•ore than 23 full thermal cycles are expected in a design application;
then 6c p will be within approximately 10% of tiefs* * ° rotne * design
applications, more pertinent diagrams similar to Fig* 7 can be constructed
in order to simplify calculations.
Up to now the strain accumulation discussed has been that which
occurs after the first half cycle. The strain which accumulates during
the first half cycle can be determined from (19) and (23) to be
(31)
 17

This accumulation will be the sane for any ambient temperature base.
.For Case 1 (Region FR), the specific form of (31) Is

(32)

for Cases 2 aud 3 (Region FRl and FR1-FR), the specific form of (31) is

Inspection of (32) and (33) shows that e? depends directly upon T and
?, but is only a sensitive to variations in k for small k. Figure 9
demonstrates the variation with k and T for P « 0.7. Comparison of
Figures 8 and 9 show that e? can constitute a significant part of the total
strain accumulation, especially for' small n.
 18

CONCLUSIONS

Thermal ratcheting has been investigated for a solid rectangular beam

made of a material exhibiting an idealized Bauschinger effect, as modeled

by kinematic hardening. Constant strain-hardening, or equivalently, a

constant tangent modulus, and a linear temperature variation across the

beam was assumed. Results were obtained for a wide range of parameters:

a tangent modulus of zero to a tangent modulus equal to the elastic modulus,

*
an axial load of zero to a yield point axial load, a temperature variation
from zero to infinity, and temperature cycles from one to infinity.

Closed form exact expressions or upper bounds are given for determination

of the thermal ratcheting strain accumulation. For cases where bounds are

available but a more accurate evaluation of the strain accumulation is

desired, cycle by cycle numerical evaluation procedures are outlined and

examples presented. Also, a simplified graphical method to save calculation

time is included. For material response and loads typical of nuclear

reactor operating conditions, the strain accumulation for an infinite

number of temperature cycles has been calculated and presented in graphical

format. The strain accumulation reported for an infinite number of

temperature cycles is within 10%, at least, of that which occurs after

twenty-five temperature cycles.

Although no simple functional relation exists between the strain

accumulation and the parameters, over their entire range, certain general

trends were evident. The axial loads and temperature variations for which

strain accumulation occurred were found not to be a sensitive function

of the tangent modulus; except when the tangent modulus was nearly equal

to the elastic modulus. For material response typical of reactor operating

conditions, the ratcheting loads and temperature variations were nearly

the same as those for a perfectly plastic material. However, the amount
 19

of strain accumulation was found to be a most sensitive function, nearly

inversely proportional, to the tangent modulus. The functional relation
between the strain accumulation and the temperature variation was less
sensitive. It ranged from being nearly proportional, for small temperature
variations, to being very insensitive, or a constant, for large temperature
variations. The strain accumulation was nearly proportional to the axial
load. .
The strain accumulation response for a solid beam with a rectangular
cross section was found to be smaller than that of a sandwich beam of the
same cross sectional area, except for large temperature variations, where
the sandwich beam does not ratchet* For the latter case, sandwich beam
construction may be more desirable.
 20

References

1. D. Burgreen, "Structural Growth Induced by Thermal Cycling," Trans. ASMS,

Journal of Basic Engineering, Vol. 90, 1968, pp.

2. J. Bree, "Elastic-Plastic Behavior of Thin Tubes Subjected to Internal

Pressure and intermittent High-Heat Fluxes with Application to Fast-Nuclear

Reactor Fuel Elements," Journal of Strain Analysis, Vol. 2, 1967, pp. 226-

238.
«

3. "Thermal Stress Ratcheting," LMFBR Piping Design Guide,

Chap. 6, p. 9 (1971)

4. D. Burgreen, "Review of Thermal Ratcheting," Symposium on Fatigue at

Elevated Temperatures, June 18-22, 1972, Storrs, Connecticut.

5. T. M. Mulcahy, "An Assessment of Kinematic Hardening Thermal Ratcheting,"

to appear in the Journal of Engineering Materials and Technology.

6. D. Burgreen, "The Thermal Ratchet Mechanism," Trans. ASME, Journal of

Basic Engineering, Vol. 90, 1968, pp. 319-324.

7. W. L. Greenstreet, J. M. Corum, and C. E. Pugh, "High-Temperature

Structural Design Methods for LMFBR Components," Quart. Prog. Report,

ORNL-TM-3845, 64 (March 31, 1972)

 er(lb/in2)
60,000

40,000

€ (in/in)

Fig. 1 Uniaxial stress-strain curve for a linear, kinematic

strain-hardening material (drawn for a = 33,500 lb/in.2,
• E - 19 x 106lb/in.2,. and k = 44.5) y
 0.25 0.50 0.75

0.25 0.50 0.75 1.00

P/2Acr«
Fig. 2 Typical regions of elastic (E), shakedown (S,, s 2 )
repeated plastic (P), and ratcheting cycling (FR/ '
> PRi-FR, FRi); and curves ( ) of constant strain
a
accumulation. <Se".(drawn for k 3.0).
 (a) (b)

(d)

Fig. 3 Typical stress distributions for shakedown (a), repeated plastic cycling and
shakedown (b), Case 1 thermal ratcheting strain accumulation (c)', and Case 2
thermal ratcheting strain accumulation (d).

•i
 I I I I !

I I I 111!
10
Fig. 4 Graphical solution of Case 2 recursion relation (21). Curve
for p k 103 - 1, 1.5, 2, 3, 4, 5, 6, 8, 10, etc. '
 * I

1 f I 1 I I
100
shown
N
 I I TTT

, # • * ' • • . " ; "

I I I 1 1 1
1000

\
 I ;

0 5 10 15 20 25 30

0.25 0.50 0.75 1.00

P/2A<7y
Fig. 5 Regions of elastic (E), shakedown ( S ^ ^ ) , repeated plastic (P), and
• ratcheting cycling- (FR, FRi-FR, FRi); and curves ( ) of constant
strain accumulation, 6e p , for k • 46.2. ' • '
 k= co 999249

8
aTE/o:,
Fig. 6 Variation of 6e£ with aTE/a for P * 0.70. Each curve represents a different value of fc.
 120

100

80 —

S 60

40 —

20

20

Fig. 7 The range of the number of half cycles, N o n », necessary to accumulate

90% of 6eP for a given k. 90%'
••» '• ' . • . . - .

25 1 1 1
i
20
15
20 -
HlO

- 7

15

hi - 5
V
CO
10
— " '" - 4

V
• •

*
- 3

I
2.5
50 100 150 200
N

Fig. 8 Variation in Se£ with N and selected oTE/a , for P/2Ac = 0 . 7 and
y y
k - 46.2. .
 • \

6 —

5 —

t
UJ
3 —

2 —

I —

10 15 20 25
aTE/cru
Fig. 9 Variation of e£ with T for selected values of k and P/2Aa =0.7.