Vous êtes sur la page 1sur 215

SIM PLIFIED METHODS FOR THE SOLUTION

OF THERMAL RATCHETTING PROBLEMS

Thesis submitted for the degree of

Doctor of Philosophy

at the University of Leicester

by

Raymond John Mark Veness

Department of Engineering

University of Leicester

May 1992
UMI Number: U641550

All rights reserved

INFO RM ATIO N TO ALL USERS


The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

Disscrrlation Publishing

UMI U641550
Published by ProQuest LLC 2015. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
X ia_c\
Simplified Methods for the S o l ut io n of Th er ma l R a t c h e t t i n g
Problems.

Ph.D. Thesis.

By R.J.M. Ven ess

Abstract.

This thesis p r e se n ts a simple and c o n s e r v a t i v e m e t h o d for


the an a l y s i s of in el a s t i c cyclic thermal stress problems. It
is d e v e l o p e d for the a n a l y si s of severe cyclic thermal shocks
c o m b i n e d w i t h m e c h a n i c a l loads in thin w a l l e d tubes. The
m e t h o d s are int en de d to p r e d ic t both the onset of ratchetting
and the str ain s a c c u m u l a t e d by small e x c u r s i o n s into the
r at ch et ti n g regime.
The text be gi n s by s um ma ri s i ng the d e v e l o p m e n t of the
u pp er b o u n d s h a k e d o w n the or y- the main tool in this approach.
Pre v i o u s m e t h o d s for s i m pl i f i e d a na ly si s us i n g the upper bound
m e t h o d are re v i e w e d and some ideas adopted.
B u i l d i n g on this p re v i o u s work, a n e w m et ho d of predi cti ng
the s h a k e d o w n bo u n d is developed. This is ba sed on the upper
bo un d a n a l y s i s of p a r t i c u l a r ratchet m ec ha ni sm s, re su lting in
a li m i t i n g e q u a t i o n for each mechanism. These e q u a t i o n s are
u s e d to p r o d u c e ra tchet bo unds for several th ermal stress
example s. Th ese results are c om pa re d w i t h the b ou nd s pro duced
by d es ig n code rules c u r r e n t l y u se d for cyclic thermal stress.
One o b s e r v a t i o n re ached in the p r e d i c t i o n of these bounds
was the lack of s u p p o r t i n g evi de nc e available. Therefore, a
set of e x p e r i m e n t s w e r e p e r f o r m e d to show the bounds
c o r r e s p o n d i n g to two of these me ch anisms. These u se d a new
resist anc e h e a t i n g m e t h o d for app ly in g the thermal cycles.
The results are c o n s e r v a t i v e l y p r e d i c t e d by u pp er bound
analysi s. However, in some regions the eff ect s of cyclic
hardening redu ces the strain accumulated, making the
p r e d i c t i o n s over con ser vat ive .
To in ve s t i g a t e this eff ec t some simple h a r d e n i n g rules are
ap pl ie d to the sh a k e d o w n bounds. These are d e v e l o p e d to give a
p r e d i c t i o n of a c c u m u l a t e d strains w i t h i n the ra tchet region.
C o m p a r i s o n is made bo th w i t h the e x p e r im en ts and a finite
el em ent com pu te r model.
Finally, these p r e d i c t i o n s of sh ak e d o w n and cyclic
h a r d e n i n g c o n t r o l l e d str ain are co mb ine d to give a complete
pict ure of c y c l i c a l l y he a t e d tubes b e l o w the cre ep range.
Acknowledgements.

Firstly, I wish to thank Pr of essor Alan R.S. Po nter for

his in val uab le gui da nce and direction.

In addition, the f ol lo win g are some of the ma ny wh o have

h e l p e d make this work possible:

Dr. Keith Carter, Dr. Alan Cocks and Dr. John B e y n o n for

t ec hni cal advice and discussion.

Co li n Morrison, Sa ndy L igh tfo ot and Geoff O ' Co nn or for

help w i t h the e x pe ri me nt al pro gr am me and in m a n y other ways.

D ou g Pratt and his staff for some of the better figures.

Paul Smith for the Photographs. Pam G i b s on and R i c h a r d Mobbs

for h el p wi th co mp ut in g and w or d processing.

The e n g i ne er in g w o r k s h o p staff, e s p e c i a l l y Paul In gh am and

(the late) Dave Linnett.

My friends in the E n g i ne er i ng Department, e s p e c i a l l y Mark

Walters, A n d r e w Rowbottom, Ju lian De vo y and A n d r e w Searle.

Mrs. B. Couzens, my mo th er for he lp in g wi th the typing.

Last, but not least, Jane Barrow, w i t h o u t wh ose b a d g e r i n g

I m a y not have fin is he d w r i t i n g up.


Contents.

Chapter 1 A Ba ck g r o u n d to the Problem of Th ermal


Ratchettin g.

1.1 Introduction: The proble ms of R e p e a t e d T h e rm al 3


Shock.
1.2 Approaches to the Sol ut ion of Thermal 6
R a t c h e t t i n g Problems.

Chapter 2 Sh a ke do w n Theory: A L it era tu re survey, w i t h


Pa r t ic u la r Em phasis on Rec en t D e v e l o p m e n t s in
the Field.

2.1 Introduction. 10
2.2 B e h a v i o u r of Stru ctu re s Un der Co mb i n e d T h er ma l 10
and Me cha n ic al Loads: The i n t e r a c t i on Diagram.
2.3 A Br ief Ba c k g r o u n d to the D e v e l o p m e n t of 14
B o u n d i n g Theory.
2.4 St a t em en t of the Upper Bo und S h a k e d o w n Theorum. 17
2.5 Re c e n t D ev e l o p m e n t s of the Up per Bound 19
S h a k ed o wn Theory.
2.6 The Ex ten ded Upper Bound S h ak ed ow n Theorem. 21
2.7 A Linear P r o g r a mm i ng M e t h o d of S o l u t i on for the 23
Upp er Bound Equation.
2.8 Si mp l i f i e d M e t ho ds of Sh ak e d ow n Anal ysi s. 26
2.9 Summary. 29

Chapter 3 A Si mp li fi ed Me th od for the So l ut io n of the


Up pe r B oun d Sh ake do w n Equation.

3.1 Introduction. 35
3.2 The Upper Bou nd Method, as A p p l i e d to The rma l 38
R a t c h et Mechanisms.
3.3 The Upper Bo und Sha ke d o w n E q ua ti on in terms of 39
a Tresca Yi eld Condition.
3.4 D e s c r i p t i o n of R a t ch et Me ch anisms. 42
3.5 O p t i m i s a t i o n of Bounds for M e c h a n i s m III and 45
IV.
3.6 E xa mpl es of results a c h i e v ed us i n g m e c h a n i s m 48
bounds.
3.7 C o m p a r i s o n w i t h Other A n a l y s i s Methods. 50
3.8 Conclusions. 57

Chapter 4 E xp er i me nt s on Thin W a l l e d Tubes Un der Severe


Cycli c Thermal S tr es s T

Introduction. 75
A R e v i e w of Thermal R a t c h e t t i n g Experiments. 76
Ex pe ri me nt al Methodology. 80
De s i g n of Test Specimen. 82
316 Stai nle ss Steel as a St ruc tur al Material. 84
D e s c r i p t i o n of Test Equipment. 85
Exp er im en t al Procedure. 87
Re su lts of Thermal R at c h e t t i n g Experiments. 89
Co m p a r i s o n of E x p er im en tal Re su lts w i th Upp er 91
Bo u n d Solutions.
4.10 Conclusions. 93
Ch ap t er 5 E s ti ma ti on of P la sti c Strain Accumulation due
to R a t c h e t t i n g ?

5.1 Introduction. 110


5.2 A D e s c r i p t i o n of Simple H a r d e n i n g Models. 112
5.3 S i mp l if ie d Mo de ls of B e h a v i o u r O u ts id e the 113
Sh a ke do w n Bound.
5.4 Finite El ement A n a l y s i s 117
5.5 Re s u lt s of Finite E l e m e n t and S i mp l i f i e d 120
Ha r d e n i n g Model An al y s e s .
5.6 Modifications to the Simplified I s ot ro pi c 123
H a r d e n i n g Model.
5.7 In t er a ct io n between mechanism st rain 128
components. 132
5.8 Conclusions.

Ch a pt er 6 S i m p l i f i e d M e t h o d s for the Sol ut io n of Thermal


R a t c h e t t i n g Problems; A Summary.

6.1 Introduction. 148


6.2 A S i m p l i fi ed Method. 148
6.3 Co m p a r i s o n s W i t h D e s i g n Codes. 149
6.4 E x p e r i m e n t s on T hi n W a l l e d Tubes. 151
6.5 The P re di c t i o n of R a t c h e t Strains. 152
6.6 Pr o po s al s for Future Work. 154

Appendix A D e r i v a t i o n of M e c h a n i s m Equations. 157

Appendix B L i s t i ng of 'Mecalc' Co m pu te r Program. 168

Appendix C S p e c i m e n and M a t e r i a l Data. 179

Appendix D E x p e r i m e n t Co ntrol and Data C o l l e c t i o n Program. 183

Appendix E D e v e l o p m e n t of E x p e r i m e n t a l Ap p a r a t u s . 188

Appendix F E x p e r im en t al Results. 193

Appendix G D e r i v a t i o n of Wo r k H a r d e n i n g Equations. 196

References. 200
Ch apt er 1.

A Background to the P r o b l e m of Thermal Ratch et tin g.

1.1 Introduction: The p r o b l e ms of R e pe a t e d Thermal Shock.

This thesis deals wi th the problems of d e f o r m a t i o n caused

by r e pe at ed thermal stresses in tubes made from du ct il e

ma te ri al s. The dan ge rs of th ermal stresses can si mply be

il l u s t r a t e d for bri tt le ma terial s. For example, plunging a

ch eap glass into hot wa ter will cause it to shatter. Gl ass is

a br i t t l e solid with internal flawes. S t re ss es in duced by

uneven thermal e xp an si o n cause crack propagation and rapid

br it tl e failure. A less ob vi ous dan ger is re pr es en te d by

thermal shocks to ob jects made from duc ti le mat erials. There

are no d ra ma ti c exa mpl es of steels fa iling due to rapid

heating. Their useful ab il it y to r e d is tr ib ut e str esses ensures

that the strains ca used un der an y t h i n g but the m os t abnormal

c ir c u m s t a n c e s are small. However, in many structural

applications these tem per at ur es are not im po sed just once, but

vary r e pe at ed ly be tw ee n limits. If the small strains caused by

these cycles of thermal stress accumulate, then d a ng er ou s and

e x p e ns iv e di st o r t i o n s can occur.

Over recent years, the impetus to improve the e f fi ci e n c y

of he at tra nsfer wi t h i n power generation systems has led to

the ne ed for c o n t in ua ll y higher w o r k i n g temperatures. This has

fueled res earch into several rel ated fields. Creep at high

tempera tur es, fatigue, and ne w m a t er ia ls are all subjects

seeing development for a dv anc ed power plants. In ad d it io n the

a er os pa ce ind us tr y has become increasingly interested in high

t em pe ra tu re d es ig n problems, as sp ac ec r af t and fast jets push


the limits of mat er ia l technology. The nu cl ear power ind us try

has a re p u t a t i o n for be in g at the fore fro nt of this research.

S o l vi ng the un usu al p r ob l em s e n co u n t e r e d in nu clear

engineering requires innovation and ne w technology. However,

it is crucial that these i nn ov at io ns are cl ea rl y and

completely understood to a vo id some potentially d i sa st ro us

consequences.

In the mid 1970's a p r o j e c t was e s t a b l i s h e d by the U n i t e d

Kingdom Atomic En e r gy Authority (UKAEA) and the E u ro pe an

Atomic Energy Community to d e v e l o p a c o mm er ci al Li qu i d Metal

co o l e d Fast Bre ed er R e a c t o r (LMFBR). The conc ept relied on the

excellent heat transf er p r o p e rt ie s of liquid sod ium as the

primary coolant. While running, this coo la nt w o u l d be he at e d

to a maximum te mp e r a t u re of 600®C. However, at in tervals

du r i n g a design life, the reactor would be e xp ec te d to be

sw i t c h e d ra pidly on or off. The effect of this would be to

cause a pulse of li quid sodium to flow th r o u g h pipes pe rh ap s

250®C cooler. Figure 1.1 shows a schem ati c of a typical LMFBR.

During a rapid 'off' transient, or 'trip', cold so di u m wo u l d

flow from the primary pump into the reactor core ca using a

sharp drop in te mp e r a t u r e or 'downshock'. In addition,

fl uc t u a t i o n s between the hot level marked on figure 1.1 and

lower levels corresponding to cooler sodium result in thermal

shocks moving al ong the le ng th of the above core structure.

De si gn codes were in e x i s t an c e for assessing thermal stress

problems in p r e s s u r i s e d wa t e r reactors (PWRs). Unfortunately

these were not suitable for use on the ne w fast reactors, as

PWRs operated under a different combination of thermal and

me c h a n i c a l loads. The realisation of this problem led to the

decision to develop a new design code. This code would be


ai m e d at solving these, pr o b l e m s at the designers level,

without recourse to e x p e n s iv e experiments.

The des ig n code would n ee d to co nsider s t ain le ss steel

und er at least two i nd e p e n d a n t loads. One load would be

constant with time, representing either the weight of the

str uc t u r e and contents, or a p r e s s u re imp osed by the contents

of the vessel. The other load w o u l d be v a r y i n g cyclically with

time and p e r ha ps also w i t h p o s i t i o n alo ng the tube. This would

r ep re se nt an imposed thermal shock on the structure. Some 300

such thermal cycles co uld be expected over the lif et im e of a

reactor. The pur po se of the code would be to ensure the

prevention of si gn i f i ca n t str ai n gro wth over the reactor

lifetime.

Considering the r el at i v e l y lo w nu mb er of these thermal

cycles, high cycle fatigue sh o u l d not cause a signi fi ca nt

problem in this aspect of the design. However, at the highest

temperatures (above 480*C for the m a t e r i a l s c o m m o n l y used in

reactor pipework) the stru ctu re may suffer some d ef or ma ti on

through creep, or time dependent plast ic it y. The other cause

of deformation is thr ou gh the accumulation of incremental

pl a s t i c strains, an effec t called ratchetting. It is at the

understanding and p r e d i c t i o n of the in fl u e n c e of this thermal

ratchetting ef fect that this thesis is aimed. N e i th er creep

nor fatigue are directly considered. However, it has been

shown by Ponter [1] that the ef fe ct s of creep can be est imated

from a know le dg e of the ratchet behaviour.


1.2 Approaches to the S o lut io n of Th er mal Ratchetting

Problems.

The combination of loads o ut l i n e d in the p r e v i o u s section

can lead, in the si mplest terms, to two results. Up to a

certain combination of thermal and me c h a n i c a l st resses the

s t r a i n a c c u m u l a t e d by a large nu mb er of th ermal cycles will be

of the order of el astic st rains (~ 0.1%). Above this level,

c a l l e d the ratchet boundary, significant plastic st rain cycles

can occur, res ul ti ng in the possibility of large deformation

over a numbe r of cycles. The actual behaviour of the mat er ia l

is s o m ew ha t more co mplex than this. The effects of 'reverse

pl a s ti ci ty ' and 'cyclic harde ni ng ', co n ce pt s discussed in the

following chapters, will significantly reduce ra tchet strains

in many cases. However, as far as nuclear structures are

concerned, it is this b o u n d a r y that limits safe designs.

Th ere are three approaches for s o l vi ng these thermal

stress problems. Finite E l e me nt (F.E.) analysis, us i n g

constitutive mo de ls of the pla st ic response of the ma ter ial

can be us e d to ca lculate the p l a st ic st ra in s on each load

cycle. Although the computing power available to the engineer

has increased g r e at ly over the last decade, analysis of

sufficient cycles to be sure of the respons e is still a major

exercise. However, this method is becoming increasingly

po p u l a r for the validation of completed designs. Cur re nt

de s i g n codes generally allow a F.E. solution as pr oo f of a

safe design.

A n a l t e r n a t i v e method, currently adopted in the R C C _ M R [2]

design code is ob ta in e d by c ol le c t i n g t o g e t h er a large group

of results from thermal r a t c h et t i ng ex pe ri me nt s. The point


corresponding to the onset of r a tc het tin g for eac h exp eriment

is then p l o t t e d on d i m en s i o n l e s s axes corresponding to the two

applied loads and a line d ra wn inside all the points. This

method of p r o d u ci n g de sig n codes has pro ve d suc ces sf ul in

other engineering ap plications. However, the nu m be r of and

range of thermal rat ch et ti n g e xp er i m e n t s performed to date is

re la ti v e l y small. It is argued in chapter 3 that there has

b ee n an ins uff i ci e nt variety of experiments to use such a

method. There are c o n si d e r a b l e ad v a n t a g e s i n h e re nt in such

gr ap h i c a l methods. The bo u n d is p l o t t e d on a set of general

axes such that any p r o b l e m can be q u i c k l y v a l i d a t e d with only

a fr a c t i o n of the w or k inv ol ve d in any F.E. analysis.

The third approach to these p r o b le ms is to use sim plified

solutions. In particular, the upper and lower b o u n d shakedown

the or em s can be applied to cyclic thermal r at che tt in g

problems. The method used is to a s c e r t a i n if a st e a d y elastic

response can be re ached after an i n d e t e r m i n a t e l y large number

of cycles, us i n g an e l a s t i c - p e r f e c t l y plastic m a te r i a l model.

If this can be shown to occur, then the s t ru ct ur e can be said

to have 'shaken down'.

The lower bound shakedown th eo r e m has b ee n used in some

form for d e s i g n codes for a nu mb er of ye ars [31]. It has the

ad v a n t a g e of the empirical methods in that a nu m b e r of simple

general equations will give a so l ut io n for any thermal

ratchetting problem. It also has some of the ad v a n t a g e s of

F.E. c a l cu la ti on s, in that it is ba s e d on a p ie ce of general

theory, and sh ou l d be equally applicable to any problem.

However, the solutions produced however are of t e n very

co ns er va ti ve . Attempts to improve the so lu t i o n s are h a mp er ed

by the detailed stres s fields req ui re d for the gove rn in g


equations.

This thesis is ai med at fi n d i n g a simple al te rn at iv e

s o lu ti on based on the up per bound sh ak e d o w n theorem for

calculating both the on set of r a tc he tt in g and the plastic

strains a c c u m u l a t e d w i t h i n the ra tchet region.

To this end, chapter 2 d e s c r i b e s the u pp er b o un d sha kedown

th eo ry and summarises rel ev an t m e t h o d s for p r o d u c i n g sh ak edown

bo u n d predictions. Cha pt er 3 proceeds to pr op o s e a ne w met ho d

for investigating s h a k e d ow n bo u n d s b as ed on predicting the

onset of me c h a n i sm s by which r a t c h e t t in g can occur. This

method is comp ar ed with the oth er design cr i t e r i a cu rr e n tl y

available.

In the fou rth ch ap te r results of a series of ex pe rim en ts

on t h e r m a l l y and m e c h a n i c a l l y s t r e s s e d tubes are p r e s e n t e d and

com p a r e d w i t h p r e d i c t i o n s made u s i n g the up pe r b o u n d sha ke do wn

theorem. Chapter 5 looks at the d i f f e r e n c e s be t w e e n pr ed i c t i o n

and experiment. An explanation of these differences is given

in terms of work h a r d e n i n g b e h a v i o u r and a m o d i f i e d p r e d i c t i o n

produced.

In the last chapter, the w or k of c h ap te rs 3 and 5 is tied

to gether to give a method for predicting both ela stic and

pl a st ic ma t e r i a l behaviour over a range of loads and

conditions. Finally the thesis is s u m m a r i s e d w i t h emp ha si s on

the o b j e c t i v e s a c h i e v e d and issues remaining.


Complex thermal
transients are induced
within the I H X during
reactor trips

Fluctuations in the
sodium level
induces moving
Intermediate
temperature fronts
\ Heat
Exchanger
PRIMARY
PUMP

o-i

w I
Na hot leve

ABOVE CORE
Na cold
STRUCTURE

CORE

'Cold' sodium \\
impinges on hot
above core '
structure during
transients

^ 'Cold'
socium
PRI MARY V E S S E L ente-s core

GUARD V E S S E L

F ig u re 1.1 Main Vessel of a Liquid Metal Cooled Fast Breeder


Reactor.
chapter 2.

S h a k ed ow n Theory; A L it er at ur e survey, with P ar ti cu la r

Emphasis on Recent D ev el op me nts in the Field.

2.1 Introduction.

M uc h of the work presented in this th esis is based u po n

the up per bound shake dow n theorem. It is important, ther ef or e

to have, as a basis for this work, a cl ear d e s c r i p t i o n of the

principles and ap pl ic a t i o n s of this theorem. To that end, the

f o l l o w i n g chapter gives a su mm ar y of the relev ant work in this

field.

It is not the inte nt io n to p r o v i d e ei th er a full acc ou nt

of the hi st or i c a l d e v e l o p m e nt of the theory, or rigorous

proofs. The former has been cl e a rl y p r e s e n t e d in the thesis by

K a r e de n iz [3]. The latter can be fo und in several sources,

notably Koiter [4] and Ma rt i n [5]. Rather, this chapter

fo cuse s on recent de ve lo p m e n t s in the applications of the

th e o r y w h i c h have b ro ug ht it close to the stage of a useful

design tool for st ruc tur es su bject to severe cyclic thermal

loading.

2.2 Be h a v i o u r of S tr uc tu res Un d e r Combined Thermal and

Mechanical Loads; The I nt er ac ti on Diagram.

This wor k is c o n ce rn ed wi th st r u c t u r e s un der a c o m b i n a t i on

of two a p p li ed loads. One is a P r i m a r y load, so called as it

is not self-li mit ing , i.e. it does not undergo stress

r e la xa ti on as a result of small plastic st r ai n adaptions. The

primary loads co ns id e r e d in this i n s t a n ce are axial loads

10
(both tensile and compressive) and internal or external

pr es s u r e in closed ves sel s (i.e. inclu din g the effe cts of

p r e s s u r e on the ends).

The other app lie d load arises from thermal stress es caused

b y a cycle of n o n - u n i f o r m tempe ra tu re d i s t r i b u t i o n s w i t h i n the

structure. The f l u ct ua tio n of this stress, n o r m a l l y m ea su re d

by the maximum variation of effe ct iv e stress is generally

r ef err ed to as a sec on da ry stress in de si g n code rules. This

is beca use it arises from residual stress fields w h i c h are

se l f - l i m i t i n g and unable i n d iv id ua ll y to cause a pl as tic

str ain growth m e c h a n i s m w i t h o u t the addition of p r i m a r y load.

Th ere are, however, c ir cu ms ta nc es wh ere cycles of thermal

l oa di ng are capable of p r o d uc in g strain gr ow th in the absence

of ap pl ied pr im ar y load. In the des ig n codes there is some

c o n f u s i o n over these definitions. For example, in RC C _ M R there

is an ass um pt io n that sec on da ry stresses are always self

l i mi ti ng and unable alone to cause cyclic strain growth. A

clear e x p l an at io n of the di ff ere nce be tw ee n primary and

s e c o n d a r y loads is given by Cle ment et.al. [6].

The loading hi sto rie s co ns ide red consist of a num ber of

cycle s of se con dar y load in c on ju nc ti on with a c o ns ta nt ly

ap pl ie d p r i ma r y mec han ic al load. In sh ake dow n the or y the

mat er ia l is con sid er ed to be elas tic - perfectly plastic,

a l t h o u g h this ass um pt io n will be subject to mo di fic ati on.

As the be hav iou r is due to a in te rac tio n be tw een the two

loads, it is co nve nie nt to show both a p p li ed loads on one

diagram. An exa mple of such an 'interaction diagram' is given

in figure 2.1, along w it h a de fi n i t i o n of the axes. The

diagram is di vi de d into a number of regions corresponding to

the be ha vi ou r of the str ucture under p a rt ic ul ar co mb in at io ns

11
of load. The types of be ha vi ou r e n co u n t e r e d are as follows:

El as ti c (E); The stress h i s t o r y stays w i t h i n the bo un ds of the

yi el d surface t hro ugh ou t the lo ading cycle. Only

elas tic d e f o r m a t i o n s occur.

Sh ak ed ow n (S); The str es se s ex cee d yi e l d at some po i n t in the

lo ad ing cycle. However, after one or more cycles have

elapsed, a residual stress field bu il ds up which, when

superimposed on the ela st ic stresses, results in a

stress d i s t r i b u t i o n w h i c h is e n ti r e l y w i t h i n the yi e l d

surface. The result is that some initial pla st ic

s t r a in in g occurs. However, over a number of cycles the

st rain rate a s y mp t o t e s to zero. In this asymptotic

state further p la sti c yielding is prevented by the

presence of the residual stress field. The a c c u m u l a t e d

pl a s t i c st rains which pr oduce this residual stress

field are generally of the order of the el astic

strains, i.e. ra rely gr eater than 0.1- 0.2%. Sh ak e d o w n

the o r y predicts the boundary to this region in terms

of an el a s t i c p e r f e c t l y pla st ic material.

Reverse Plasticity (F); If the maximum ef fe ct iv e

thermo-elastic stress ex ceeds twice the yield stress

then it is no longer p o ss ib le to find a residual

stress field w h i c h fulfills the sh ak e d ow n condition.

However, if the stress distribution is such that

plastic st rain occurs at more than one po i n t in the

cycle and the i nc re me nt s are equal and op p o s i t e then

the net accumulated strain over a cycle is zero.

Figure 2.2 is a Tr esca yi e l d diagram for a biaxial

state of stress w i t h p r in ci pa l stresses and 0 2 * It

shows a typical thermal stress cycle gi vi ng rise to a

12
reverse plasticity condition. Where thermal be nding

str ess es occur, such as thr ough the wall of a non-

u n i f o r m l y h ea te d plate, then part of the mate ri al can

see a stress cycle such as figure 2.2, wh i l s t the rest

remains elastic. Un le s s this region can con ta in a

ra tchet mechan ism , then reverse plasticity results.

Fai lur e can still occur by fatigue, but ratch et tin g

will not take place. For an el astic p l a st ic mat erial

the upp er boundary of this region is d e f i n e d as the

load c o m b i n a t i o n at w h i c h r a t c he tt in g begins.

Ratchetting (R); The c om bi ne d residual and the rm o- elastic

str ess es cannot be co nt a i n ed wi t h i n the yi el d surface,

th er ef or e pl a s ti c strain growt h occurs on e v e r y cycle.

This g ro wt h rate can be high, so de s i g n is often

limited to the ratchet boundary. However, for strain

hardening materials the ratchet rate in the 'R' region

directly adjacent to an 'S' region can reduce

asymptotically to zero. In the process, some plastic

str a i n a c c um ul ate s, giving contou rs of co nstant

accumulated plastic strain wh i c h tend to be parallel

to the ratchet boundary. The effec t of strain

hardening in ratchetting mat er ia l ad j a c e n t to an 'F'

region is not fu lly understood. Ch apter 5 d e s c r i b es a

method for calculating these p l a st ic strains for a

range of d i f f e r e n t problems.

The presence of a reverse plasticity region gives rise to

an important classification between s t ru ct ur es un de r cyclic

thermal loads. This has b ee n d i s c u s s e d by Ponter [7]. Thermal

loads that give rise to a large comp on en t of b e n d i n g stress

t h r ou gh the thickness, such as a t h r o u g h - t h i c k n e s s tempe ra tu re

13
grad ien t, a l lo w a reverse p l a s t i c i t y region. In this case, the

maximum stress es occur at the surfaces, whilst in the centre

th ere will always be a region of zero thermo-elastic stress.

Even if the m a x i m u m stresses exc ee d yield, there will still be

a v ol um e of ela sti c m at er ial in the centre of the section.

Providing this ma te ri a l is su ff i ci en t to carry the primary

load, then reverse plasticity will occur rather than

ratche tti ng. This results in an i n t e r a c t i on diagram of the

f or m shown in figure 2.3a, with no p o s s i b i l i t y of rat ch et ti ng

without some me ch an i c a l load. This type of lo ad ing was

c l a s s i f i e d as class 'A' by Ponter.

Other thermal loads, such as a tem pe ra tu re front mo v i n g

a x i a l l y al on g a tube, give rise to thermal st resses w i t h a net

membrane stress th ro ugh the thickness. Un der these so ca lled

'class "B"' loads there is no d e f i n i t i v e l y zero stress point,

so it is pos si bl e to exc ee d yi e l d at ev er y po int in the

st ruc tu re with just the thermal loads. This results in the

possibility of r at ch et ti ng un de r no p r i m a r y load, such as in

the i n t e r a c t i o n d i a g r a m of figure 2.3b.

This is of pa rt ic u l a r im po rta nce in co mp a r i s o n w i t h d e s i g n

codes wh e r e it ap pe ar s that in some cases class 'B' st ruc tur es

have been ove rlooked. This problem is exa mi ne d in some detail

in ch ap ter 3.

2.3 A Br ie f Background to the Development of B o u n d i ng

Theory.

The co nc ep t of bounding th eo r y is particularly powe rfu l

when applied to cyc li c lo ad ing problems. The fact that the

c al c u l a t i o n s are history ind ep en de nt simpli fie s sol ution

14
vastly co m pa re d with classical e l a s t ic - p l a s t i c ca lc ulations.

The m ai n d i s a d v a n t a g e w i t h this c o m p a r is on is that no d e t a i l e d

i n f o rm at io n about p la sti c response is obtained. However, it is

possible to e xt en d the results from b ou nd in g ca lc ul a t i o n s to

inc lud e es ti ma te s of p la sti c strain for v a ri ou s mate ria l

models, as d i s c u s s e d in ch apter 5.

The conc ept of a bo und on shakedown, b e h a v i o ur was first

discussed by Bleich [8] in 1932 for trusses and frame

structures. This was expanded by Me l a n [9] as a general

theorem for elastic-plastic structures. This theorem can be

expressed in the f o l l o w in g terms: If any ti m e - i n d e p e n d e n t

distribution of residual stresses can be found for a

particular combination of cyclic loads such that on

superposition with ela st ic stresses, a safe stress state

results at every po int in the structure and thr ou gh ou t the

cycle, then the struc tur e will shake down for this state of

re pea ted va r i a b l e loading.

The proof gi ve n by Me l a n for this theorem is complex.

However, several a ut ho rs have restated, and p r o v i d e d simpler

p ro of for the theory, notably Koiter in 1964 [4]. This m et ho d

of formulation results in a lower b ou nd on the region of

shakedown. One other impo rt an t result ob t ai na bl e from the

lower bo un d t h e o r e m is that if shak ed ow n can be shown to occur

for one initial residual stress distribu tio n, then it will

occur for all other a d mi s si bl e initial residual stresses. This

me an s that the initial residual stress state of the m a ter ia l

need not be considered in the shak ed ow n calculation and the

sh ak ed ow n limit is in de pe n d en t of any initial f a br ic at io n

stresses.

The theor y was first e xt en de d to cover the eff ec ts of

15
temperature gra di en t s by Prager [10]. At the same time, his

re-formulation of the th eo re m in terms of a ' k i ne ma ti ca lly

ad m i s s i b l e distribution of residual strain rate' showed that

the phy si c a l source of the stresses was not relevant. An early

consideration of the adv ant ag e s o f fe r e d by the lower bo und

th eo ry for the d e v e l o p m e n t of de si gn codes was by Leckie [11],

which was later e x te n de d and incl ude d in p r e s su re vessel

d e s i g n codes [12].

In 1956, Koiter [13] no te d the s i m i l a r i t y in form be tw ee n

the lower bo u n d the or y for plastic coll aps e and Me l a n s lower

bo u n d theorem.

'The in v e s t i g a t i o n of plas tic co llapse of a structure,


or limit a n al ys is is gr ea tl y f a c i l i t a t ed by two
complementary theorems the two theore ms m a y be
used to get her to ob tai n both upp er and lower bounds
for the co lla ps e load.
On the other hand, only one theorem, due to Me la n
[] is av a i l a b l e for the much more d i f f i c u l t sh ak e- d o w n
(sic) p r o b l e m . '

Koiter w en t on to derive the e q u i v a l e n t upper bo und sh ak edown

th eo re m in terms of a bo und on the p l a s t i c work requir ed to

create a ratchet mechanism.

This theorem is best stated as follows: Shakedown cannot

occur if it is p o s s i b le to find an a d m i s s ib le cycle of pla stic

strain and a pr o g r a m m e of load v a ri a t i o n s between prescribed

limits for w h i c h the exte rna l work done over a cycle by this

pr og ra mm e is gr e at er than or equal to the e n e r g y d i s s i p a t e d in

plas tic wo rk associated with this admissible strain rate

cycle.

This was later e x t en d ed for the case of a te mp e r a t u r e

cycle by Ma i e r [14] and others. However, it has not been until

recently that this upper bo un d theory has been a p p l i e d to the

pr ob lem s a s s o c i a t e d w i t h thermal ratchetting. The d i f f i c u l t i e s

16
inher ent in this m e t h o d are explained, al o n g w i t h a statement

of the theorem, for a c om bi na t i o n of th ermal and me ch an i ca l

loads, in the fo ll ow in g sections.

2.4 St at em en t of the Up per Bound S h a k e d o w n Theorum.

The up pe r bo un d sha ke do wn t h eo re m has been stated in,

p r o b a b l y the sim pl es t po ss ib l e form by M a r t i n [15].

"...If s h a k ed ow n occ urs in a given s t r u c t u re un der cyclic

loading,

"ext < Wint 2.1


for all a d m i s s i b l e p l a sti c strain rate cycles

(in the curr ent notation)

is a cycle of work done by e x ter na l age nc ie s to the

s tr uct ure and a cycle of internal work. These two terms

m ust result from a cycle of strain which is kinematically

admissible. For a pl as ti c strain rate cycle to be

k i n e m a t i c a l l y ad mis sible, the in cr ement over a cycle.

G ? . dt - 6e?^ 2.2

must be co mp a t i b l e with the displacement field 6U^. The

su pe rs cr ip t c denotes co mp at ibility. N ot e that the rate of

strain G ? . ne ed not be k i n e m a t i c a l l y admis si ble .

This can be expressed for the load s y s t e m c o n s i s t i n g of a

c on sta nt mechanical load and a c y c l i c a l l y v a r y i n g te mp er at ur e


0
field, ca us in g thermo-
­ e l a s t i c stresses o^j, by ex p a n d i n g the

terms of e q u a t i o n 2.1.

17
The external wo rk done to the sy st em comes from two

sources, the m e ch an ic al load (signified by P) and thermal

s t re ss es (signified by 0):

W Wp + W g 2.3
ext
The wo rk done by the me cha nic al load \|/P in ca us ing a

displacement is

W p - ip/gP.ûu'^ ds 2.4

w he r e s is the surface area of the b o d y and * is a scalar load

multiplier.

The wor k done by thermal stresses is equal to the integral


rT
w 0 ^ . ( x, t) G? .( x, t) dt dV 2.5
0 1 1] —
V' 0
over a cycle 0<t<T and th rough the vol ume V.

The internal wor k term, cor re sp on ds to the ene rg y

d i s s i p a t e d by the po s t u l a t e d pla stic d e f o r m a t i o n mechanism.

*ij(X'G't)G9j(x,t)dtdv 2.6
" in t
0
wh ere is rel ated to e^jthrough the a s s o c i a t e d flow law.

C om b i n i n g these work terms gives the up per bo und


A
* P. AU dS + ( x , t ) G? .( x, t) dt dV <
13 — 1] —

/T
*ij(X't)G?j(x,t)dtdV 2.7
0
this can be r e -w ri t t en as
A
. I
P.AU dS < :cr?.(x,e,t)-a?.(x,t) ]c ?. (x ,t) dtd v 2.8
s
Note that e q u a t i o n 2.8 is un c h a n g e d if both the strain and

displacement fields are scaled by a co ns tan t factor. Hence

by d e f i n i n g P.A U ds = 1 , i.e. pl ac in g a r es tr ic ti on on the


s

18
m ag n i t u d e of 6U^, the ine qu al it y can be reduced to
A
[o9.(x,e,t)-sf.(x,t))e?.(x,t)dtdv 2.9
13 — 1] — 1J —
V' 0
Thus, by d i v i d i n g the mechanical load into co mp on e n t s of

magnitude (if/) and configuration (P) the inequality has been

c o n s i d e r a b l y simplified.

Note that G?j and AU^ are linear terms in the inequality.

He nce the b o u n d is u n c h a n g e d if both q u a n t i t i e s are ch anged by

the same scalar multiple.

If the postulated mechanism is the ratchet mechanism

r e qu ir in g the lowes t value of this w or k d i f f e r e n c e for such a


0
combination of if/P and &ij(x,t) then the bo un d wil l equal the

exact ratchet limit.

The form of the u pp er bo u n d th eo ry e x p r e s s e d in eq ua tions

2.8 and 2.9 wil l form the basis of m a n y of the ca lc ul a t i o n s in

this thesis.

2.5 Re ce n t Developments of the U pp er Bo u n d S h ak ed ow n

Theory.

The u pp er bo u n d s h a ke do w n theor y has been an acc ep ted

des ig n tool in the field of civil e n g i n e e r i n g st r u c t u r es for a

number of years. In particular, Horne [16] gi ves several

worked examples of the s o lu ti on of repe ate d cy cl ic loading of

portal frames using up per bo u n d techniques. In these cases,

the po ss ibl e fail ur e mechanisms were l i m it ed to a small known

set, giv ing rise to an in te r ac ti o n d i a g r a m c o n s i s t i n g of this

set of m e c h a n i s m s b o u n d i n g a region of s h a k e d o w n behaviour.

However, this route has not been taken in mec han ic al

de sign codes due to the di ff i c u l t y in choosing from the

19
p o t e n t i a l l y v as t range of m e c h a n i s m s poss ib le in, say, a tube

under a com pl ex thermal stress field, such as occurs in the

LMFBR. The codes cu rr e n t l y use a m uc h s i mp lif ied ve rs io n of

the lower bo u n d theorem.

The lower bo un d so lution has been used in the full

c a l c u l a t i o n of thermal ratchet bounds by Leckie [11]. However,

this requires a value for the residual stress field, wh i c h

pr oves co mplex for all but the si mplest l o ad in g cases. In

1980, Ponter [7] showed that if the o p t i m u m ratchet m e c h a n i s m

could be found, for example, by numer ica l methods, then the

solut i on to the upper bo und e q u a ti on became a m uc h more

prac tic al solu tio n to complex loa din g problems. He went on to

pr opo se ap pl ic a t i o n s for the up per bo und th e o r y to several

st andard sh ak e d o w n problems. He also gave e x am pl es of the

bounds that would result from a small gr oup of postulated

ratchet mec hanisms.

Cocks [17] ca l c ul a te d the solu tio n to the problem of a

simply sup po rt e d plate s ub je cte d to a cyclic thermal stress.

Both lower bound and upper bound (using a postulated

mechanism) ca l c u l a t io ns were performed. The co rr ela tio n

betw ee n results was v er y good.

The same p r o b l e m was also solved by W e b s t e r et.al. [18]

u si ng a finite ele me nt method. This gave a v e r y similar result

to the bounding methods. However, a different ratchet

m e c h a n i s m was predicted than the one used in the upper bound

solution.

It is clear that due to l i m i ta ti on s and a s su mpt ion s

ado pt ed to allow so lution of the upper b ou nd equation, the

exact form of the op t i m u m m e c h a n i s m will not always be found.

This will result in some overestimation of the size of the

20
s ha ke do wn region. However, co mp ar iso ns b e tw ee n upper bound

s h a ke do wn solutions and theor et ic al solutions performed by

Carter and Ponter [19] have shown that the results are still

normally very close to each other. This is su pp o r te d by

Websters results.

2.6 The ext en de d Upper Bo un d Sh ak e d ow n Theorem.

It was obs er ve d in the sec tion d e s c r i b i n g the i n t e ra ct io n

diagram that sh akedown can only occur w he re the maximum

v a r i a t i o n of the th e r mo -e l as ti c stress can be c o nt ai ne d wit hi n

the yield surface by a cons ta n t residual stress field.

N e v e r t he le ss , it was shown that ex ce ed in g the sha ke do wn bound

does not necessarily result in ratchetting. As long as

suf fi ci en t mat er ia l remains el astic at ev e r y point in the

cycle to support mec han ic al loads, then reverse p l a s t i c i t y can

take place.

The fact that classical upper bo un d sha ke do wn the or y does

not ac co unt for this reverse plasticity region was no t e d by

Ponter and Kar adeniz [20,21]. They co mm e n t e d that the

p ot en t i al impact on the safe de si gn region, es p e c i a l l y for the

types of loadin gs oc cur ri n g in LMFBR's was significant. As a

result, a th eo ry was d e v e l o p e d to es timate the extent of this

reverse plasticity region. The basis of this 'extended upper

bo u n d theorem' is d e s c r i b e d below.

The vo lu me of the section, V, is di vi de d into two sections

Vg and V_. W i t h i n V- a residual stress field p . . can be found


O £ D 1J
Q
such that still lies e n t i r e ly w i t h i n the yi el d

surface. Within V^, no such residual stress field can be

con structed. However, it is still po ssi ble to define a

21
residual stress field such that a minimum isotropic

expansion of the yi el d surface is required to con ta in the


0 *
combination, + pUj.
Q *
This results in a stress field p^^ + p^^ wh o s e ex treme values

are s y mm et ri c w it h the yi eld surface. The e x t e n d ed upper bound

t h e o r e m def in es an a s s y m pt oti c state w h i c h is co ns is t an t with

the as s u m p t i o n that the ma te ria l in Vp has c y c l i c a l l y h a r de ne d

to an ela stic state (i.e. i s o t r o p i c a l l y ) w he ra s no strain

hardening has oc cu rre d in V g . These ex treme assumptions,

o v e r e s t i m a t i o n of the effect of cyclic ha r d e n i n g in the stress

history in and underestimation in Vg is consistant,

a cc o r d i n g to Ponter and Kara den i z [20], w it h a c o ns erv ati ve

es ti ma te of the ratchet boundary.

It is as s um ed that rat che tti ng occurs w h e n Vg reaches its

s h a ke do wn limit, with Vp c ar ryi ng no m e ch an ic al load,

t he ref ore the upper bo und e q u a t io n can be re- wri tt en as


A
* P.AU ds < [Pij(x,t)-p9j(x,t)-p^j]e. jdtdv 2.10
s /v_/Q
This requires the ca lc ul at io n of residual stress field p^j

within V g . Applications of the lower bo und t h e o r e m have shown

that this can be difficult. However, by virtual work:

Pi jACijdV P^AU^dS 0
'V
be ca u s e Pj^ 0. i.e, the residual stress is in equilibrium

with zero appl ied load.


/T
PijAC ijd V = PijêijdtdV + p ^ j e ^ j dt dv 2.11
'V 0 0
Vs Vp
thus the work done in the two regions on p^^ are equal and

opposite. Us ing this result, in e q u al it y 2.10 can be re-written

as

22
A
V/ PAU ds < [ ( X , t ) - p f .( x , t )1 h. .dtdv
1] — 1] — 1]
s
°

p.jûeijdV 2.12
'Vp
By ag ain d e f i n i ng ïgPAU ds = 1, i.e. p l a c i n g a r es tr ic ti on

on the m a g n i t u d e of AU^, the i ne qu ali ty can be reduced to


/T
[ p?j (x, t ) - p ? j ( X / 1 ) ] jd tdv + P i jAS ij dV 2.13

V O Vp

As the criteria for p^j w i th i n Vp have been defined, it is

straightforward to add and cal culate this new term in the

upper b ou nd equation.

This ex t e n d e d upper bound eq u a t i o n (2.12) was used

directly to calculate the so lution to some simple e x am pl es

[22]. However, for more complex p r ob le ms a nu me ri ca l tec hnique

u s i n g a linear p r o g r a m m i n g m e t h o d was developed.

2.7 A Linear P r o gr a m m i n g M e t h o d of So l u t i o n for the Upper

Bou nd Equation.

The linear programming ap p r o a c h is explained, with

examples, by Ponter and Kara den iz [23], however, the basic

method is as follows: The structure is first su b d i vi de d into

sections with similar ge om et ri ca l chara ct eri sti cs , i.e.

cylindrical, conical toroidal etc. Each se ction is s ub div id ed

elem en ts u s i n g auto ma ti c m e s h re fining such that the v a r i a t i o n

of thermo-elastic stress is r e a s on ab ly the same in each

element. As the thermal stress distribution tends to con ta in

sharp peaks as s o c i a t e d with te mp er at ur e gradients, m e sh

ref inement is an imp or ta nt me t h o d for attaining a c c ur at e

23
solutions. The nodal variables consist of the val ues of the

displacement c om po ne nt s and the six pl as tic m u l t i p l i e r s for

a Tresca yi el d condition. In the fo rm u l a t i o n developed by

Ponter and K a ra de ni z [23], Ponter and Carter [26] and Franco

[28] the fu nda men tal a ss um p t i o n has been that the X^'s vary

linearly within an element. For cy lin dri cal el eme nts it is

ea s y to show that this assumption is c o ns ist ent w i t h a s sum in g

linear variation of the radial displacement and qu ad r a t i c

variation of the axial displace men t. For non- cy lin dri ca l

e le men ts there is p r o b l e m over c o n s i s t e n cy as the v a r i a t i o n of

displacements and the X^'s wi t h i n an e l em en t need to be

separately de fi ne d such that the strain distribution

produced by each of them are reconcilable. This problem has

b ee n a d d r e s s e d by b ot h Ponter and Carter [26] and Franco [28].

The c a l c u l a t i o n can n o w be reduced to a linear p r o g r a m m i n g

problem with the nodal displacements and Xj^ va lu es as

vari ab les. The constitutive r e l at io ns hi ps w i t h i n ele me nt s and

between ge om e t r i c sec tions pro vi des the restraint equations,

a u g m e n t e d by the r e l a t i o ns hi p

J-PAU^ds - 0
O— —
The cost fun ct io n is p r o v i d e d by from e q u a t i o n 2.9 for the

up per bo und equation, and from e q ua ti on 2.13 for the ext en ded

u pp er bound.

The m i n i m u m va l u e s of the sum of these cost fu nc tions that

satisfy the c o n s tr ai nt co nd it i on s are then c a lc ul a t e d by

iteratively varying va l u e s of X^'s us ing the simplex linear

programming method. This yie ld s both the load at which a

ratchet mechanism forms for a specific thermal cycle, and the

shape of the mechanism. An in ter act ion d i a g r a m is built up by

sca lin g the thermal st resses up and then repe at ing the

24
o p t i m i s a t i o n process.

There are two prob lem s associated with this method of

solving the upper b ou nd equation. Firstly, the solutio n

re quires a set of ca lc ul a t io n s for every po in t on the

interaction diagram, de spite the fact that, frequently, the

same m e c h a n i s m is responsible for r a tc h e t t i n g over a range of

the boundary. This means that the s o l u t i o n takes some time to

ca l c u l a t e (although this time is still very small w he n

co m p a r e d with a full elastic-plastic an al ysis). The second

p r o b l e m is the number of v a r ia b le s for w h i c h a sol ut io n can be

found, i.e., the m a x i m u m nu mber of n o d e s and hi n g e s w i t h i n a

mecha ni sm , is li mited by the number of co ns t r a i n t s provided.

In some cases, this can m ea n that the optimum ratchet

mechanism cannot be identified. However, the lowest en er gy

mechanism within the limits of the method will be found. In

most cases this will give a result for the ratchet load w hi ch

is close to the optimal value.

De sp it e these drawbacks, the li near p r o g r a m m i n g m e t h od of

so l vi ng the upper bo und equation has proved a very useful

tool. S ol ut io ns to a large nu mber of i d e a l i s e d thermal loading

p r o b l e m s have been produced, both for tube ge o m e t r i e s [24] and

general axisymmetric g eo me tr ie s [25]. In addition, solutions

p r o d u c e d by this me t h o d have c om pa re d well w i t h the small fund

of experimental results [26]. Several e xa mp le s of results

produced by the linear programming method are gi ven in the

fo l l o w i n g three chapters. They are compared with des ig n code

rules, experimental results, full elastic-plastic an alysis

results and a mechanism ba se d method of so l v i n g the upper

b ou nd e q u a t i o n p r op os ed in chapter 3.

25
2.8 Sim pli f i e d Met hod s of Sh ake dow n Analysis.

The last four years has seen c o ns ide rab le effort go into

the development of s im pl ifi ed m e th od s of sol ving the upper

bo un d sha ke do wn equation. There have been primarily two

driving forces beh ind this work. Firstly, there has been a

strong desire wi t h i n the nuc le ar and power generation

i n d u st ri es to develop simple and co ns er v at iv e d es ig n code

rules b as ed on bounding theory. In addition, the linear

programming te ch nique ou tl in e d in the p r e vi ou s se ction has

p r o v i d e d a w e al th of in fo rm at i on about the sh ak e d ow n behav iou r

of ma t e r i a l s wi th vari ous loads and geometries. This

in fo rm at io n wo u l d be of more value if the trends and

di f f e r e n t classes of be hav iou r could be c l e ar ly id e nt if ie d and

related. Si mp l i f i c a t i o n and abstraction are usefu l tools to

this end.

Two me th ods of si mpl ifi ed analysi s are discussed here.

Although both are ba sed on the ex t en de d upper bo und sha ke dow n

theorem, they are v e r y d i f f e r e n t in approach.

a ) Sl op e- Ba s ed Method.

The temper a tu re his to r ie s of thermal ra tc h e tt in g prob lem s

t y p i ca ll y occ ur ri n g in fast reactor des ig n can be ide alised

into three types. These are the axial spike, the axial step,

and the th rough thickness distribution (see figure 2.4). For

each type, a number of in te ra ct i o n d ia gr am s can be p ro du ce d

for d i f fe re nt temper atu re gradi ent s by, for example, the

E EC _S h a k e d o w n program. These int er ac ti on dia gr am s can be used

d i r e c t l y to give a simple de si g n rule. Each d i a g r a m is reduced

26
to a pair of straight lines giving a c o n s e r v a ti ve ratchet

boundary. These lines can be ch a r a c t e r i s e d by three

parameters, 0^, 02 and x.

Figure 2.5 shows a typical i n t e ra ct io n in te r a c t i o n di a g r am

characterised by these two lines and their respective

parameters. The values of 0^, 02 and x can be ca l c u l a t e d for a

range of these three ideal ise d t e mp era tur e pr o f i l e s and

pl ot t e d aga ins t d i m e ns io n le ss v a r ia bl es re pr es e n t i n g the

se ve ri ty of the temper atu re profile.

The resulting par ame tr ic graphs allow construction of a

range of ide alised in te rac tio n diagrams. Ca rter et.al. [27]

have gi ven two exam ple s of this m et ho d for thermal ra tc hetting

experiments. The sim pli fi ed me t h o d gives a reasonable,

co ns er va ti ve e s ti ma ti on of the linear programming sol ution in

both cases. Practical t emp era tu re distributions differ

si g n i f i c a n t l y from these ideal ise d cases. Ho we ve r good

c om pa ri so ns can be made by m at c h i n g either the m a x i m u m t h e r m o ­

elas tic stress or maximum t emp er at ur e gr ad i e n t of the

pr act ica l di st ri b u t i o n w it h the a p p r op ri at e ideal ise d

parameter.

The main ob je c t io n to this s i mp lif ied method is that the

range of pro bl em s that can be an a ly se d are li mi te d by the data

av a i l a b l e on the pa r a me tr i c graphs. Providing suff ic ie nt

i n f o rm a t i o n for a co mpr eh e n si ve de s ig n code w o u l d require an

open ended 'book' of examples.

27
b ) Simple M e c h a n i s m Approach.

Franco [28] took a wholly different a p p r oa ch to solving

the up per bo und equation. The linear programming method has

produced both s ha ke do wn bo un d a r y load and ratchet m e c h a n i s m

as the solution. Kara de ni z and Ponter [23] have shown that for

the case of an axial te mpe rat ur e step m o vi ng along a length of

tube, o nl y a small number of ratchet m e ch an i s m s bo un d the

sh ake dow n region. Franco used the kn owl edg e of these

mec hanisms, comb ine d wi th an a na lyt ic al t h er mo -e l a s t i c stress

distribution to derive an equ at io n for each mechanism. These

e qu ati ons each gave a line on the int er ac ti on diagram. The

ad van tag e of this m e t h o d is that a reasonable sol ut io n to the

upper bound sol ution over a range of st resses could be

c a lc ul at ed wi th little effort.

There are two pro bl em s w ith the a p p l i c a t i o n of the me th od

p r op os ed by Franco. Firstly, the equ at io ns were de ri ved

s pe ci f i c a l l y for the case of a te mp era tur e step m ov in g along a

tube. This im me di at el y limited the ap pl ic a t i o n s of the method.

Secondly, a sim pl if yi ng as su mp t io n was made to ignore the

b en di ng compo nen t of the th er mo -e la st ic stress. This pr ecl ude d

the de t e c t i o n of a group of me ch a n i s m s known to appear in

these problems.

However, this m et ho d is still potentially very powerful.

In the next chapter, a more general m e c h a n i s m ba sed solution

is adopted. It is shown that this a p pr oa ch can give accurate

results to a ve ry wi de range of pr obl ems with c o ns id e r a b l y

less work than the linear p r o g r a m m i n g method.

28
2.9 Summary.

It has been shown that sh ake do wn th eo ry offers a useful

way of calculating bounds to sh ak e d o wn problems. The main

advantage to the th eory is that so lutions to a wh ol e class of

problems can be solved by a small nu mber of equations.

S o l ut io n s to the lower bound th eo r y are lim ite d by the

ability to ca lc ulate residual stress fields. The up per bound

theory is likewise limited by kn owl edg e of the ratchet

mechanisms. However, it has been shown that the exact ratchet

m e c h a n i s m is not required to obtai n a good solution.

Finally, met ho ds of ap p l y i n g the up per bo u n d th eo ry have

be en discussed. These show that trends of be ha v i ou r for a

class of p r ob le ms can be established which allow s im pl if ie d

up p e r b o u n d met ho ds to be applied.

29
0.5

- Maximum effective thermoelastic stress during a cycle,


normalised by the yield stress at the initial temperature.
P - Mechanical load, normalised by the limit load of the structure
at the initial temperature.

F ig u re 2 .1 A Tvoical Interaction Diagram Showing Regions of


Material Behaviour.

30
-de

F ig u re 2 .2 A Tresca Yield Surface. Showing a Typical Reverse


Plasticity Cvcle.

31
Qt

F ig u re 2 .3 a ) A Typical Interaction Diagram for a Class 'A' Structure.

Ot

F ig u re 2 .3 b ) A Typical Interaction Diagram for a Class 'B' Structure.

32
e -
Schematic of a Stationary Spike
Temperature Front.
\ A0
AXs
AXs =
/ R h

Schematic of a Moving Step


Temperature Front.
AXm
00 AXm =
/ R h

Schematic of a Through-
AG Thickness Temperature Profile.

00

Where R = Tube Radius


h = Wall Thickness
X = Distance along Tube
0 = Temperature

F ig u re 2 .4 Schematics of Soike. Step and Through- Thickness Temperature


Profiles.

33
Prediction Line 1

EEC_Shakedown Interaction Diagram

Prediction Line 2

F ig u re 2 .5 Interaction Diagram Showing Parameters for the ’Slope- Based'


Method of Predicting the Onset of Ratchet Behaviour.

34
Chapter 3.

A S i mp li f i ed Me th o d for the Sol ut io n of the Upper Bound

Sh ak e do wn Equation.

3.1 I n t r o d u c t i o n .

The upper bo und and ex te nd e d up per bo un d the ories have

been applied, u si n g a linear programming technique, to give

so lut ion s to thermal r atc he tti ng p r o b l e ms in the form of

i n t e ra ct io n diagrams. The results o b ta i n e d have been shown by

Ca rter and Ponter [26] to compare well w i t h the l i mi te d range

of e x pe ri me nt al evi dence available. In addition, the

i nt er ac ti on diagram format of the results allows for a whole

range of poss ibl e des ig n opt ions to be determined from a

single calculation.

However, there are prob lem s associated with the linear

programming technique, both from a de s i g n viewpoint and as

regards in te rp re ta ti on of the results. The comput er code

req uir ed to c alculate these bou nds is both ex te ns iv e and

specialised. A l t h o u g h still under develo pm en t, it is d i ff ic ul t

for an u n t r ai ne d op erator to use, as several internal

variables eff ec ti ng the sta bi li ty of the solu tio n mu st be

defined. Use of the upper bo und m e t h o d itself gives rise to a

problem. So lut ion of the upper bo un d e qu at io n requires

kno wl ed ge of the specific ratchet mechanism for the region

under consideration. This has been one of the main reasons why

the upp er bound met ho d has not been fully exploited until

recently. The linear p r o gr a m m i n g method tackles this problem

by i n s t i t u t i n g a search thr ough a range of p o s si bl e m e ch a n i s m s

until the d e f o rm at io n field re qu iri ng the lowest activation

35
energy is found. Although this is re as onable for small

problems, the time required for c o n v er ge nce increases in

proportion to the cube of the number of var iables. This makes

so l u t i o n of larger pro bl em s by this m et ho d impractical.

There is another p r o b l e m related to this m e t h o d of solving

the u pp er b ou nd equation. The 'black box' ap pr o a c h resulting

from a comple x num eri cal method ma kes it difficult to

de t e r m i n e the com bi na ti on s of stresses r e sul ti ng in the

fo rm at io n of a ratchet mechanism. A clearer understanding of

the causes of pa rt ic ul ar me ch an i s m s may well lead to simpler

and more reliable m e t ho ds of prediction.

There is an al te r n a ti v e ap pr oa ch to the so lu tio n of the

up per bound equation. If one pa r t i c u l ar ratchet m e c h a n i s m is

chosen, the res ulting di s p l a c e m e n t field can be s u b s ti tu te d

into the upper bound e q ua ti on and a bo un d ca lc ul at ed

corresponding to that mec hanism. This will lead to a solution

very rapidly. Unfortunately it will only be the correct

sol ut io n if the correct mechanism is used. In this context

'mechanism' refers to the change in shape of the structure

w h i c h allows pla sti c strain to occur.

A large bo dy of results has been p r o d u c e d for a range of

load cases, giving bo th int er ac ti on d i a g r a m and o p t im um

ratchet mechanism, us i n g the EEC_Shakedown program. Some has

b een p u b l is he d by Carter and Ponter [24], [25] and Karadeniz

[3] whilst a large number of cases are as yet unpublished.

After ex am i n a t i o n of all of these results for tube geome tr ie s

it became clear that only a small c h a r a c te ri st ic group of

mechanisms was id ent ifi ed by the EEC_Shakedown m et ho d as the

route by wh i c h ra tch ett ing occurred. In each case, either all

of the plas tic strain o cc ur re d wi t h i n a small number of

36
adj ac en t elements, giving a so ca ll ed 'local' mechanism, or

three local pl astic hi nges formed along the tube and it

deformed into a cone. This was re ferred to as a hi nge -co ne

mechanism.

Although the linear programming method is l i m it ed in the

range of m e c h a n i s m s it can search by the nu mb er of const ra int s

[23], this range is still v er y large. It is the refore

s ig ni fi ca n t that no other ratchet mechanisms have occurred.

This suggests that it ma y be p o s s ib le to cal cu la te the ratchet

b o u n d a r y by choo si ng the m e c h a n i s m from this small gr oup w hi ch

yi e l d s the lowest me c h a n i c a l load for a p r e s c r i b e d te mp erature

history.

Deriving pa rt ic u l a r eq ua t i o n s for these mec hanisms, in

terms of features of the t h e rm o - e l a s t i c stress h i s t o r y w ou ld

give a simple me ans of c a l c u l a t i n g the ratchet b ou nd in a wide

range of cases. A n a p p oa ch of this kind w o u l d be more suitable

for the solu ti on of p r e l i m i n a r y d e s i g n pr o b l e m s and could also

be a d a p t e d for de s i g n code rules.

This 'm ec hanism based' method is ex p o u n d e d in the

fo ll o w i n g chapter. The results o b t a in ed are com pa re d both with

predictions from E E C _S h a k e d o w n and cu rrent d e s i g n codes, for a

set of rep re se nt at iv e examples. This leads to some in te re st in g

con c l u s i o n s about de si g n code me th od o l o g i es , and their

s u i t a b i l i t y for a p p l i c a t i o n to these problems.

37
3.2 The Upper Bo und Method, as A p p l i e d to Thermal Ratchet

Mechanisms.

The upper bou nd and e x t e nd e d up per b ou nd theo rie s have

been explained at length in the pr e v i o u s chapter. In this

sec ti on the th eo ry is su mm ar i s e d briefly with pa r t i c u l a r

reference to the form required for the so l u t i o n of m e c h a n i s m

led equations.

Considering a shell st ru cture s u bje ct ed to a di st r i b u t e d

time in de p e n d e n t load \i/P, wh er e * is a scalar load parameter,

and a cyclic v a r i a t i o n in t e m p er at ur e 0 (x,t) wh e r e x den ot es a

position vector. The thermo-elastic stress history

corresponding to 0 is given by &^j(x,t). The Up per Bound

theorem is expressed in terms of a cycle of i n ela st ic strain

rate e?j, which sa tisfies the crit er ia that the total strain

a c c u m u l a t e d over a cycle of lengt h At ;

TAt
è?.dt 3.1
0

is co mp at i b l e w i t h a d i s p l a c e m e n t i nc re me nt AU. .

The Upper Bo un d S h ak ed ow n th e o r e m ass um es a perfectly

plastic solid w i t h yi e l d surface takes the form

of the inequality;

PAt
PAU^dS <
VJO

wh er e S and V refer to the surface and vo l u m e of the body

respectively and a? j is the state of stress at yi el d

corresponding to e?j. The exact s h ak ed ow n limit is given by

the m e c h a n i s m Au9, and a corresponding h i s t o r y of strain

which minimizes ip. In eq ua l i t y (3.2) only ap pl ie s w h e n reverse

plasticity does not occur, the extension to include this

38
region has been given by Kar adeniz and Ponter [20], as

described in section 2.6.

Thus it can be seen that the solu tio n of in eq u a l i t y (3.2)

requires strain and compat ibl e disp la cem ent s, provided by the

definition of the mechanism, ext er nal force P, and


0
t h e r m o -e la s ti c stress 8^j. The exte rna l force can be

predetermined by d e fi ni n g a me ch an ic al load type. In this

analysis, only i n t er n al /ex te rna l p r es su re and axial load cases

have been considered. Slight modifications to the equat ion s

w o u l d al l o w any load or c om bi n at io n of loads to be examined.

In the pre vi ous si mpl i fi ed a n aly se s discussed in section

2 .8b, th er m o - e l a s t i c stresses have been inc lu de d t h ro ug h the

use of an an alytic so lution of an i d ea li se d t em pe ra tu re

distri bu ti on , such as that produced by A rn au de au , Zarka and

Gerij [29]. However, this limits the a p p l i c a t i o n of solutions

to particular, id ea lised cases. In this a p p r o a c h the e q ua ti on s

are left in terms of pr o p e r t i e s of the t h e r m o - e l a s t i c stress.

This allows either ana ly ti c calculation of the th e r m o - e l a s t i c

stress distribution for simple te mp e r a t u r e profiles, or

nu me r i c a l (perhaps finite element) so lutions for more complex

cases. Co ns i d e r i n g stresses as the cause of ratchet me c h a n i s ms

removes the the unnecessary abstraction of loo ki ng at

tem p e r a t u r e fields.

3.3 The Upper Bo und Sh ak e d o w n Eq u at io n in terms of a

Tr es ca Yi eld Condition.

In the following derivations, the a s su mp ti on s ad op ted by

Ka ra d e n i z and Ponter in the original a p p l i c a t i o n of the upper

b ou nd th eory to this p r o b l e m will be us e d [23]. In particular.

39
this as sumes an elas ti c- perfectly p l a st ic ma t er ia l model,

with a plane stress, Tresca yi eld criterion. A sketc h of the

Tre sc a yi el d surface in space is given in figure 3.1,

wh er e and <r^ are the pr inc ip le stresses in the axial and

c ir cu mf e r e n t i a l di re ct io ns for a tube un der plane stress. The

six surfaces are l ab ell ed k=l to k = 6 .

By ta king a Tresca yi el d surface, the p l a st ic st rain rate

is limi ted by the no r m a l i t y co nd i t io n to six direct ion s. This

all ows the pla sti c strain incr em ent t er m from e q u a t i o n 2.8 to

be rewri tte n as:

c?j * 3.3
If
wh ere is a d ir ec ti ona l tensor normal to yi e l d surface

k and is the ma gn it ud e of the pl as tic strain rate normal to

the k'th yi el d surface.

This can be w r i t t e n explicitly for a plane stress state

as

'0 1 1 0-1 - 1' 3.4

.1 0 - 1-1 0 1>

where êf and s? are the axial and c ir cu mf er en ti al strain rates


X

in a cylinder.
0
The work done by a th er mo -e l as ti c stress h i s t o r y &Yj(t)

such as in fig. 3.2 to pr oduce such pla st ic strains is w r it te n

as :

FAt ft
.(t)dt - dt 3.5
0 J J

40
Su b s t i t u t i n g equ. 3.4 into this gives:

At rAt g g

0 1
0 —
1 0
1 —1
-

0 Î)
1-1 (X.
dt. 3.6

Ug,

or :

FAt
A0 A 0 r A 0 ^0 . A0 A0 rA0 A 0 .
dt. 3.7

Co ns i d e r i n g the first term of equ. 3.7:

FAt
&G(t)Xi dt.

there exists some instant t^ wh ere o^(t) has it's m a x i m u m

po s i t i v e value, i.e:

3.8

SO,

FAt PAt
3.9
0

where
PAt
X- dt 3.10

A p p l y i n g this a r g um e nt to the other terms of equ. 3.7

gi ves the upper bound:

41
FAt 0 _
&f.(t)e?.(t) dt. < 3.11
0 ^J

&®(4),a®(t2),[&^Ô®](t3),-ô®(t4),-â®(t5),-[&®-&®](tg)

Where instants t^-tg are shown s c h e m a t i c a l l y in fig. 3.3

This shows that the use of the Tresca yi e l d cri te ri on

e na ble s a solu tio n of the upper bound i n e q u a li ty 2 .8 , restated

here :

FAt
* PAUV ds < dt dV
Vj

to be found by i n t e gr at io n of stress co mp on e n t s at d i sc re te

po in t s in stress space and X's c o r r e s p o n d i n g to the p o s t u l a t e d

pl a s t i c strain mechanisms.

3.4 D e s c r i p t i o n of R a t c h e t Mechanisms.

This section de sc r i b e s the form of the gr ou p of ratchet

mechanisms se l ected from the results of the linear p r o g r a m m i n g

ana l y s i s and the pr oc es s u se d to derive them. The full

derivation for each mechanism is given in appendix A.

Additionally, the m e c h a n i s m s are su mm ar i s e d in table 3.1. This

is int en de d as a reference guide for the preparation of

i n t e r a c t i o n diag ra ms us i n g this method. It gives the go ve r n i n g

equations, a sketch of each m e c h a n i s m and a d e f i n i t i o n of the

str es se s required for the c a l c u l a t i o n of each mechanism.

42
Mechanism I: This is a generalised form of the so ca ll ed

'Bree' mechanism, caused by pla stic strains on surface 2 or

5 of the tresca yi e l d boundary . A cyclic thermo-elastic

be n d i n g stress results in local axial strains at the point

wh er e the variation of these stresses over a cycle is a

maximum. Eq uat io ns are given in table 3.1 for axial load and

internal pr e s su re me ch an i ca l loads. The first pair of

eq uat io ns are ca l c ul at e d us in g the basic up per bo u n d

equation. This is valid upto the onset of reverse


0
plasticity, when - 2oy. The secon d pair of e q u a t i o n s are

ca lc u l a t e d for t h e r m o - e la st i c stresses above this limit,

u s in g the ext en de d up per bo und equation. These lines

in ter sec t at the thermal stress c o r r e s p o n d i n g to P-0.5.

M e c h a n i s m II: This is a comp osi te mec hanism. A large com po ne nt

of th er mo -e la st ic ho op stress results in a plastic st rain

com po ne nt at two points in the cycle, either sur faces 1 and

3 or 4 and 6 , for example, ta king the stress distribution

from figure 3.10,


Cx
rAe
+ ACg 3.12

O'
As 3.13
lAe,

' Ae'
and Ae 3.14
»■” Ae>

so therefore.
Cx
rAe ' O' ' Ae' 'Ae'
3.15
.Ae> Ae^ V 0.

Thus the net resultant strain inc rement is in the axial

direction, but the mechanism is a c t iv at ed by a hoop stress

43
component.

Mechanisms III and IV: Both of these m e c h a n i s m s are caused by

pl a s t i c hoop strains on surfaces 1,3,4 or 6 of the Tresca

boundary. This requires consideration of a radial ex pan sio n

or contraction of the tube over some length 2a. The

m e c h a n i s m ide nt if i ed in linear p r o g r a m m i n g for this to occur

is th r ou gh the form at io n of three l o c a l i s ed pl as ti c hinges.

This gives a conical deformation pattern when viewed in

se cti on (see figure 3.11). Pl astic energy is d i s s i p a t e d in

two com pon en ts for this 'hinge-cone' mechanism. For ma ti on of

the hinges requires a b e n di ng stress in the ap pr o p r i a t e

direction and the change in vo lume of the cone is gov er ned

by the hoo p stress al ong its length. These two co mp onents

give rise, re sp e ct i v e l y to the two terms on the right hand

side of the e qu at io n s in table 3.1. Me ch a n i s m s m b and IVb

dif fe r in that work is also done due axial strains in the

tube.

Unl ik e Mechanisms I and II w h i c h are localised about the

p o s i t i o n of m a x i m u m a c t i v a t i n g stress, mechanisms III and IV

can be found for any length, 2a and any p o s i t i o n along the

tube. The o p t i m u m bo und from these m e c h a n i s m s will be at the

le ngth and p o s i t i o n w h i c h give rise to the m i n i m u m value of

P. Methods for fi nding this o p t i m u m m e c h a n i s m are given in

the f o ll ow in g section.

It should be no te d that these eq ua t i o n s assume there to be

no v a r i a t i o n of the yi el d stress wi t h temperature. In order to

include the ef fe cts of a v a r y i ng yi el d stress the e q uat io ns of

appendix A needs to be r ef or mu la te d to include the yi eld

stress at the region of p o s tu l a t e d strain growth. This is

straightforward for local me ch an i s m s I and II. It is rather

44
more com pl ex for the other two mechanisms, w h er e strai n growth

occ ur s over a finite length of tube. Ne w terms w o u l d need to

be ad ded to acc oun t for the yie ld stress at each of the

p l a st i c hinges, as well as the m ea n yi el d stress over the

le ng th of the mechanism. This is not co ns is t an t w it h the

s i m p l i f i e d a p p r oa ch and so has been om it te d from the analysis.

Another si mp l i f y i ng a ss um p t i o n ad o p t e d is that only the

b as ic up per bo und has been ca l c u l a t e d for m e c h a n i s m s II, III

and IV. Inc lu di ng the effects of the e x t e nd ed up per bou nd adds

fu rther complexity to the equations. C o m p a r i s o n w i t h exte nd ed

u pp er b ou nd a n al ys is from EEC S h a k ed ow n [27] shows little

difference between basic and e x t e nd ed upper bounds for

mechanisms II, III and IV ex cept in the region close to the at

axis. This assumption of basic up per b ou nd gives a more

co n s e r v a t i v e solution.

3,5 O p t i m i s a t i o n of Bound s for M e c h a n i s m III and IV.

The general result for m e c h a n i s m s III and IV is d e r i v e d in

a p p e n d i x A to be;

P Rh Jl 1 ^ 1
P " FT 2
L — ' Y J ^
W h e r e R and h are the tube radius and t h ic kn es s respectively.

a and 5 are geom et ri c pr o p e r t i e s of the mec hanism, de fi ne d in

figure 3.11. P is the a p p li ed load and P^ the limit load.

The exact definition of hinge forming moment, M^^*

(e xp re ss ed in terms of stress in table 3.1) and m e mb ra ne

component va ries ac co r d i n g to the exact mechanism (see

table 3.1). However, the sol uti on of these equ at io ns follow

the same form. The p o s i t i o n of the ratchet bo u n d c o rr es po nd in g

45
to this mechanism, for any pa r t i c u l ar thermal stress

di str ibu tio n, requires the s pe ci f i c a t i o n of a mechanism of

len gth 2a, wi th a central hinge at & (as shown in figure

3.11). These values should be such that the m i n i m u m value of P

results for that thermal stress.

Calculating the pos it io n of the central hinge

significantly increases the complexity of the so lu tio n to

e qu a t i o n (3.15). From equ at io n (3.15) it can be seen that the

c o m p on en t dep en da nt on the central hinge is

1 ^ 1

wh er e 5 is the relative po si t i o n of the central hi nge,as

defined in figure 3.11. Figure 3.12 is a graph of the

variation of this co mpo nen t w it h &. For a m e c h a n i s m w it h a

con st an t thermal stress along the length, this fun ct io n is

pr op o r t i o n a l to the variation of work requ ire d to acti vat e a

mechan ism . It can be seen that sig ni fi ca nt deviation of the

central hinge from the centre of the m e c h a n i s m inc reases the

e n er gy require d to ac tivate it, thus m ak in g the m e c h a n i s m less

favourable. This has been borne out by e x a m i n a t i o n of op ti mum

m e c h a n i s m s p ro du ce d by E E C _ S h a k e d o w n . In all cases the central

hinge lay close to the centre of the me ch anism. It was found

acc eptable, t he refore to simp lif y eq ua t i o n (3.15) in

ca l c u l a t i o n s to:

V^
by as su m i n g a sym metric mechanism.
4 ' "

C a l c u l a t i o n of the me mb ran e stress requires solu tio n of

the integral

46
,2 a

S+mean* 3.18
0
which must be pe r f o r m e d numerically unl ess an ana ly tic

so lu ti on for the t h er m o- el a st ic ho o p stress 8^ is available.

An a t t em pt was made to find a so l ut io n of this integral us ing

the method of Schwartz inequality. This is described in

ap pe n d i x A. However, the sol ut io ns did not prove acc urate

e no ug h for inclusion.

Un fo rt un at el y, more com plex stress distributions will

require a large volume of repetitive, if simple ca l c u l a t i on s

to o p ti mi se the mechanism. T h e re fo re a comp ute r p r o g r a m called

'Mecalc' has been written to perform the optimis ati on. A

multi-strip Simp son m et ho d was used to solve equation (3.17)

for a gi v e n set of thermal stress, m e c h a n ic al load type, and

geometrical data. By scaling the thermal stresse s to give a

class of thermal stress d is tr ib ut io ns , /j o ^ [ Q , x ) , it was

po ss ib le to give results from this program in terms of

interaction diagrams. The fol lo w in g sec tion gives some

exa mp le s of results p r o du ce d for several different types of

thermal r at ch et ti ng problems, c o v er in g the sort of situ at ion s

encountered in high te mp er atu re design. A list ing of the

co mpu ter p r o g r a m is given in a pp en di x B.

From equation (3.16) it can be seen that ch a n g i n g //, and

thus the ma g n i t u d e of the thermal stress, will change the

o p t i m u m me cha nis m. Since each m e c h a n i s m results in a d i f f e r e n t

line on i n t e ra ct io n d i a g r a m axes, mechanisms III and IV give

rise to cur ve d ratchet bounds. Further examination of

e qu a t i o n s for m e ch an is ms III and IV shows that inc rea si ng the

magnitude of the t h e r m o - e la st i c stress, increases the

significance of (which leads to a dec re as e in the length

47
of the mechanism) relative to l/a^ (which has the ef fect of

i n cr ea s i n g the mechanism length). Thus the m e c h a n i s m s always

d ec r e a s e in length as <y^ increases, givin g bound s w h i c h always

curve tow ards the axis. The ability to find this kind of

observation is a dir ect result of the use of an analytic

tec hn iq ue over a stri ctl y n um eri ca l method.

3.6 Exam ple s of results ac hi e v ed u s in g m e c h a n i s m bounds.

Fi gures (3.14) to (3.19) show some e x am pl es of i nt era ct io n

d ia g r a m s for thermal lo adi ng pr ob le ms A key to these figures

is gi ven in Table 3.2. Th ey were pr od u c e d in the following

ma nn er :

Te mp e r a t u r e extre mes at nodal points, geometry, elastic

m a te ri al pr op er ti es and re strain ts were us ed as inputs to the

Co ni da [40] finite ele men t program. This c a l c u l a t e d va lu es for

the mean axial and hoop co mpo nen ts of the t h er mo -e la st ic

stress and linear va ri at i o n s th ro ugh the wall thickness. These

v al ue s were di re c t l y input to the 'Mecalc' p r o g r a m de sc ri be d

in the pr evi ous section. The only ad dit ion al data required by

this p r o g r a m was the local load type (e.g.. A xi al tension) and

the movement length for non-stationary tem pe ra tu re profiles.

The ou t p u t was in the form of an in te ra ct io n d i a g r a m shown in

figures 3.14 to 3.19.

Also pl ot te d on the figures are the co rr es p o n d i n g

ca l c u l a t i o n s us ing other cu rrent ana ly sis and de s ig n code

methods. The c o m pa ri so n be tw ee n the results from these me thods

will be d i s c us se d later. In this section, the s i g n if ic an ce of

the exa mp l e s chosen, and the general fe atures of the results

p r o d u c e d by the m e c h a n i sm s m e t ho d are described.

48
The first case, figure (3.14) is one of the three

b e n ch ma rk tem per atu re d i s t r i b ut io n s fr eq ue n tl y us ed to look at

thermal stress problems. S che ma t i cs of these three tem perature

distributions are given in figure (3.13). The spike

distribution (see figure 3.13a) is similar in form to the

t e m p er at ur e di st r i b u t i o n ob t a i n e d in a series of thermal

r a t c he tt in g ex p er im en ts d es c r i b e d in the fo ll o w i n g chapter.

The axial ten sion failure mechanism is a local thinning, so,

when co mbi ned wi th local th er mo -e la st ic stresses, the ratchet

mechanisms res ulting were I and II, both local in action.

Mechanism IVb, al tho ugh c a l c ul a t e d for this case, did not

ap pe ar w i t h i n the limits of the axes.

Figure (3.15) gives an example of the second idea li se d

t e m p e r a t u r e di stribution, a step, under internal pressure. The

coll ap se load for a tube under pure internal pr e s s u r e can be

predicted by a hi ng e- c o n e ex te n d in g the length of the tube.

This c o r r e sp on ds to mechanism III at o^-l. As the thermal

stre ss es increased, the mechanism loc al is ed about the region

of m a x i m u m stress.

Figure (3.16) is a t em pe ra tur e variation thr oug h the

thickness, with no axial t e m pe ra tu re gradient. This is the

third idea li se d case, as a n a ly se d originally a n a l y s e d by Bree

[30]. O n l y m e c h a n i s m I was activated, gi vi ng rise to a ratchet

boundary which al lo we d for sh ake do wn or reverse p l a s t i c i t y at

v e r y h ig h thermal stresses.

Figures (3.17) and (3.18) in volved te mp e r a t u r e

distributions mo v i n g c y c l ic a l ly back and forth along a length

of tube. This type of s i tu at io n occurs in some s i tu at io ns

un d e r LM FBR de s i g n conditions. Moving t em per at ur e

distributions can give rise to particularly severe ratchet

49
bou ndaries. When the dista nce mo v e d by the thermal step, 6x_,
^ m
be co me s of the order of the step length, Ax^, then the m a x i m u m

variation of thermal stress over a cycle incr ea ses rapidly

from <j^ to 2a^. Figures (3.17) and (3.18) show this effect. In

figure (3.17), the mo ve m e n t length was r e la ti ve ly short and

the resul tan t in te ra cti on diagram does not di ffer m uc h from

that of the sta ti on ar y equivalent. However, figure (3.18)

shows a longer mo v e m e n t length. In this case the b ou nd was

reduced v i r t u a l l y to the ela sti c line, E. Figure (3.19) shows

that as this m o ve me nt length is incre ase d further, the bound

li mi te d towards ela stic behaviour.

It is si gn ifi can t that in figure 3.19 the ma g n i t u d e of the

thermal stress cycle is not increased. The stresse s are just

moved over a longer length of tube. This leads to the

im po rt an t con cl us io n that s p e c i fi ca ti on of the thermal and

me ch an i c a l stresses alone is not suf fic ie nt to ca lcu lat e the

s ha ke do wn bound. Any m ov em en t of the tem pe ra tu re front must

also be inc luded in the calculations.

This m o ve me nt is pe rf or me d in the m e c h a n i s m s ca lc ula tio ns

by 'dragging' the ca lc ul at ed thermo- elas tic stress maxima

over a spe ci fi ed length of tube.

3.7 Co mp ar i s o n w i t h Other A n a l y s i s Methods.

There are several other me th ods of sol ving thermal

rat ch et ti ng problems, as mentioned in chapter 1. It is

i mp ort ant to deter min e wh i c h is lik ely to yie ld the best

sol ut io ns to the wi d e s t range of loa din g situations. In

general, des ig n met hod s wo u l d be as ses sed by c o mp ar is on wi th a

wid e range of exp er im en ta l data. However, w it h a few

50
ex ce ptions, that data is not ava il ab le for the more severe

thermal stress problems. In order to gain some in sight into

the a c c u r a c y of the me ch a n i s m s method, c o m p a r i s o n will be made

with three other analys is techniques. Namely the up per bound

linear programming method of EEC_Shakedown [26] (plotted on

figures as L), the lower bo und A S ME -I II d es ig n codes [31] (A)

and the empirically ba sed RCC-MR [2] (R), the current French

nu c le ar de s i g n code. This covers the range of te ch ni qu es

currently be ing a p p li ed to severe thermal r a t c he tt ing

problems.

a ) EEC Shakedown.

The me c h a n i s m s ap p r oa c h uses the same the ore tic al

background as EEC Shakedown. Therefore, it would seem

re as on ab le to ex pect a v e r y close correlation between results,

assuming that the same m e c h a n i s m s were se l e c t e d in each case.

In fact, figures 3.14 to 3.19 show that the two ca l c u l a t i o ns

of the bounds differ by no more than 5% at any point. This

conf ir ms both the linear p r o g r a m m i n g m e t h o d and the m e c h a n i s m s

method as tech ni qu es for so lving the up per b o u n d equation. It

also shows that the si mp l i f yi n g assumptions, and l i m it ed class

of e q u a t i o n s has v e r y little ef fect on the solution.

b) A S M E III.

A c o m p a r i s o n be t w e e n A SM E III and the EEC S h ak ed ow n met hod

has been made by Ponter [7]. It is r e -s ta te d here with some

modifications suitable for the cu rrent analysis. In

parti cu la r, a lower case 'p' is used to repr es en t stresses.

51
The A S M E III code rules place re st r i ct io n on the m a x i m u m

al lo w a b l e shear stress at each point in the structure,

ev a l u a t e d as su mi ng linear elastic m a t er ia l behaviour. The

str esses are d i vi de d into five categories:

p_ - p r i m a r y m e m br an e
m
p^ - p r i m a r y local m em br an e

Pg - p r i m a r y b e n di ng

Q - sec on d ar y

F - local peak.

The cl as s i f i c a t i o n of stresses relies on the assumption that

the di f f e r e n t types of stresses have di st i n c t types of

struc tur al p h e n o m en a a ss oc i a t e d with them. In es sence the

primary me m br an e stresses are causes of ma jor load transfer

from one part of the str ucture to another, ig n o r i n g local

through-thickness variations. In thin shell th eo ry they

co r r e s p o n d to the in-plane stress resu lt ant s in uniformly

str es se d regions. The primary local stresses p^ are average

through-thickness stresses near i n t e r s e c t io ns and

discontinuities whereas primary be n d i n g stresses p^ are the

ex t re me va lu es of a l i n e a r i s a t i o n th ro ug h the thickness.

in general the peak stress f pr o v i d e s the residue w h e n the

me m b r a n e and bending stresses, in the form of a linear

through-thickness di st ribution, have been subtracted,

secondary stresses, q arise from displacement con tr ol le d

boundary cond it io ns and include thermo-elastic stress, this

description is a s i m p l i f i c a t i o n of the d e t a i l e d c l a s s i f i c a t i o n

for particular cases given in the code (e.g. table n b -3 21 7- 1

p58 of [31]) but contains the essence of the code philosophy.

the code is d e s i g n e d to ensure that the be h a v i o r remains

entirely el astic after a few c ycles during which residual

52
stresses, which are i n it ial ly present, are ad a p t ed by small

amo unt s of plas tic strain, generally of the order of the

el astic strains, i.e. sh ak e d o w n occurs, the code restr ic tio n

ensure that ne ither pla sti c collapse und er maximum load, nor

incre men tal collapse or reverse plasticity under repeated

l o a di ng can occur, as s u mi n g a perfectly pl as tic mate ria l

model.

The restric tio ns are expressed in terms of S^, a quantity

derived from yie ld pro pe rt ie s with differing formulae for

fe rritic steels, w h i c h c l o sel y a p p r o x i m a t e a p e r f e c t l y pla stic

ma te ri al for small plas tic strains, and austenitic steels,

w h i c h have no dis ti n ct yi e l d stress.

The stress restric tio ns are given, with the c o rr es p o n d i n g

va lue s in terms of y i el d stress, by:

P m ^ ®in ■ I

PL + PB * S* - *y 3.20
A(P l + Pg + Q) < 3 . 20y 3.21

where A refers to the v a r i a t i o n of the stress.

For the cases d i s c us s ed here one can i d en ti fy the me mbrane

st resses due to internal p re ssu re and axial load as pr i m a ry

m em b r a n e stresses. Hence the ap pr op ri at e d es ig n code rule is:

Pm ^ I ®y 3.22

and AQ < 2a 3.23

In the current n o t a ti o n P ^ / f y = P and A Q/ a^- Aâ ^ whe re Ac^

is the m a x i m u m v ar i a t i o n of shear stress in the t h e r m o -e la st ic

solutions. Hence:

53
p < I 3.24

and <2. 3.25

However, it is found that these e q uat io ns are never

co ns er va ti ve and to make sense of the A S M E III rules it is

n ec es sa ry to include the p r i m a r y m e mb ra ne stre sse s in the AQ

equation. This would be a cor rec t interpretation of the rule

if P w a s tak en to be a local m e mb ra ne stress and as su med it

also varied (even if it is on ly one cycle in the plants

lifetime). This in te rp re t a t i o n gives:

P ^ and 3.26

P + Aff^ < 2. 3.27

These e q u a ti on s are p l o tt e d in Figures (3.14) to (3.19).

It can be seen that they prov ide ac o ns er va ti ve and reasonable

result for cases 1 and 2 , i nvo lvi ng s t at io n a r y cycles, wh er e

reverse plasticity (mec han ism II) and Bree-like b e hav io ur

(mechanism I) dominate.

However, for cases where me c h a n i s m s III and IV appear the

results can be co me no n-c on s e rv at iv e, even with the modified

interpretation. If one a s s um ed the m e m b r a ne stress to be

always local and from eq ua t io n (3.24) took P < 1 instead of P

< 2/3 then the code wo ul d be non-conservative for all cases

wiere mechanisms III or IV are optimal. The problem in A SM E

III ar i s e s f ro m the neg le ct of the p o s s i b i l i t y of ge ne r a l i s e d

yielding due to the spread of a si gn ifi can t thermal stress

over a len gt h of tube, per haps caused by a m o v i n g te mp er at ur e

front.

54
C) RCC-MR.

In the RC C- M R codes [2], the str esses are sim il ar ly

cla s s i f i e d in terms of primary and secondary stresses.

However, for the c a l cu la t io n of bo un ds on secondary stress,

the rules are given in terms of two d i m e n s i o n l e s s quantities;

the s e c on da r y ratio (SR) and ef f i c i e n c y index (v). These are

cal cul at e d as follows:

SR - -S___ 3.28
m ax P m

V ■ — -— 3 .29
ma x P m

where AQ is the maximum variation in the t h e r m o - e l a st ic

shear stress over a cycle.

ma xP _ is the 'maximum a l lo wa bl e m e m b r a ne stress.'


m
and P is the 'effective p r i m a r y m e m b ra ne stress', i.e. the

value of P which, in the ab se nce of t h e r m o - e l a st ic

stress, w o u l d pr od uc e the de s i g n p l as ti c strain.

The b ou nd on these value s consists of an empi ri cal

function of v w it h SR, defining a safe design region. This

bound is no r m a l l y pl ot te d on log axes, but for the p u rp os es of

comparison it has been t r a n s l a t e d onto the i n t e r a c t i o n d i ag ra m

axes, su it abl y renormalised, in figures (3.14) to (3.19).

This line bo unds pla sti c strain to an ac c ep ta bl e plastic

strain limit, w h ic h in RC C-M R is taken as 0.45% (In this

comparison it is as su me d that all rules y i e l d the same value

at P =1). However, com pa ri so n wi th the other bou nds shows the

line to be ve ry n o n - c o n s e r v a t i v e in all cases, except for the

Bree case. The code does seem to recognise that pro bl em s can

oc cu r. It states that sec on da ry stresses should be

55
re cl as si fi e d as p r i m a r y unles s '...there is no risk of elastic

follow-up'. However, the de f i n i t i o n of e l a st ic fo llo w-u p is

ambiguous, and no specific crite ria for when to invoke the

clause are given. Boyle and N a k a m u ra [32] di sc us s the concept

of elas tic fo llo w-u p in the con text of A S M E code case N-47, a

very similar work to RCC-MR. It appe ars that n e it he r code

includes a q u a n t it a t i v e a n al ys is of this effect. Boyle and

Nakamura go on to revie w m et ho ds of es t i m a t i n g elastic

fo l l o w - u p for c r e ep in g structures, but there is no d i s c u s s i o n

of eff ec ts b e l o w the creep range.

U n f or tu na te ly , R C C - M R goes on to say of El as ti c fo llow-up

'It should be p oi n t e d out that un le ss the d e s i g n is


re ally clumsy, the harmful effects are in fr eq u e n t if there
is no creep'

This seems to shed d o u b t on the a b il it y of elas tic fol low-up

to a c c ou nt for discrepancies between RCC-MR and the upper

bound solutions.

As a h i s t or ic al note, the R C C - M R code was constructed from

a bound to a set of m a te ria l data ob t a i n e d from structural and

m at eri al tests. Unfortunately none of these tests in volved

moving te mp e r a t u re fronts and ma ny of them inv ol ved through

th ickness te mp era tu re g ra die nt s wh ere mechanism I dominates.

This e x p l ai ne s the sim il ar it y of the RC C - M R bo un d to the

'Bree' line of m e c h a n i s m I. A l t h o u g h an impor tan t problem in

high t e m p er at ur e design, the 'Bree' model is only a special

case of the effects of cyclic thermal stress. M a te ri al tests

were ca rr ie d out on thin w a l l e d tubes s u bj ec te d to a constant

axial load and a rev ersing torque, designed to reproduce the

effects of thermal shocks. The relevance of these ex pe rim ent s

to thermal stress prob lem s is not fully justified.

56
3.8 C o n c l u s i o n s .

The m e c h a n i s m s a p p ro ac h to e v al u a t i n g the ratchet b o un d a r y

for cy li nd ri ca l shells gives results w h i c h c l o s e l y fo ll ow the

full sha ke do wn analysis. This de mo ns t r a t e s that three types of

m e c h a n i s m are im po rtant in the cases we have examined.

The 'Bree Type' mechanism I is g e n e r a l l y accepted by the

w r i te rs of des ig n codes, and is su ppo rte d by a c o ns id er ab le

b od y of exp er i me nt s (for example, the results reporte d in

[33]). Small exc urs io n s bey on d the ratchet bo u n d for this

m e c h a n i s m are known to give rise to the rapid a c c u m u l a t i o n of

pl as tic strain. On the other hand, the 'Reverse Plasticity'

mechanism II may well give a co ns er va ti ve bound. This is due

to igno rin g the ana ly si s of the effects of cyclic ha r d e n i n g on

the yi el d stress. Exp eri me nt al evide nce to this ef fect is

presented in the next chapter. The 'Global Mechan ism s' III and

IV are unfo rtu nat ely , very short of ex pe ri m e n t a l backup.

However, the exp er im en ts of Bell [34] at the UKAEA, reported

by Po nter et. al. [26] does co n f i rm the p r es en ce of these

mecha nis ms. Significantly, for these mec hanisms, the pl as tic

strains only acc umu la te in one di r e c t i o n so the strain will be

l im ite d to a muc h smaller extent by the effe cts of cyclic

hardening.

The ca lc ul at io ns of these m e c h a n i s m bounds should all give

co ns er v a t i v e est im at es of the ratchet bound, but as stated,

the strain growth for mechanism II ma y well be lim it ed by

cyclic ha rd en in g effects. Thus the uns afe bou nd given by

RCC-MR in the first two examp les may not lead to sig ni fi ca nt

accumulated strains. However, where me c h a n i s m s I, III and IV

d o mi na te and give bounds wh i c h are more c o n se rv at iv e than the

57
RCC-MR limit, then g e n e r a l i s ed y i e l di ng can occur be lo w the

RCC-MR line. This is most si gn ifi can t in the mov in g

t em pe ra tu re front cases.

It needs to be emp ha si se d that any set of design rules

shou ld have the go ve r ni ng structural ph en om en a at their heart-

this does not seem to be the case at present. C o mp a r i s o n with

the e m pi ri ca l RCC-MR rules show that ex p e r im en ta l results can

only be g e n e ra li se d to a sit ua ti on where the same mechanism

occurs. It is quite l egi ti mat e to base design rules on

experimental evidence, but some initial a n a ly si s in terms of

ratchet m e c h a n i s m will allow the most si gn i f i c a n t e xp eri me nt s

in the region of high thermal stress to be c o n d u c t ed wit ho ut

any ambiguity. The R CC -M R rule appears to bas ed on a set of

ex pe r i m e n t s of wh i c h none involve the p o s s i b i l i t y of m e c h a n i s m

III or IV.

The m e ch an i s ms m e t ho d yie ld s a simple alternative to

eit her linear programming, or full ine lastic analysis. It can

be u s e d by de s i g n er s to i de nti fy p o ssi bl e danger regions in a

structure, us i n g only the ela stic stresses. This can either be

thr o u g h di re ct use of the formulae, or by the co n s t r u c t i o n of

i n t e ra ct io n diagrams. A l t h o u g h the formulae for m e c h a n i s m s III

and IV are som ewhat more co mplex than those c u r r e n t l y used in

de s i g n codes, the ad va nt ag es of the ap p r o a c h are v er y

significant.

58
Mechanism I - Local Bree Mechanism
Description of Sketch of Mechanism Governing Equations. Definition of Term s.
Mechanism, and Stress cycle.
Upper Bound. AÔ® is the magnitude of
Localised thinning
of the walls (i.e., when Ao ® s 2 0 y )
the variation over a cycle
resulting in a net of themaximum axial
Under axial load, F: therm oelastic stress at
axial strain.
any point in the structure
(see figure 3.4).
i - ' W
Under pressure, P: The subscript L denotes a
limit value.
Activated by axial
therm o-elastic The subscript Y denotes
stresses and a Tresca yield value.
Extended Upper Bound.
occur!ng at a point 'x.O
in the stmcture (i.e., when Aa® a 2 0 y )

corresponding to
Under axial load, F:
the maximum axial
r ® ,
thermal stress
1
com ponent Aa®

Under pressure

Pt I iOyJ

Mechanism I i - Reverse Plasticity Mechanism


Description of Sketch of Mechanism
Governing Equations. Definition of Term s.
M echanism . and Stress cycle.
Under axial load, F:
Localised thinning (AOx is the mean
of the walls through the wall thickness
resulting in a net
J. . 2 -
of the maximum hoop
axial strain. component of the
thermo-elastic stress.
Under pressure, P:
This could occur at any
Activated by point along the length or
reversing hoop through a cycle (see
stresses. Occurs figure 3.5).
at a point in the
structure The subscript L denotes a
corresponding to limit value.
8
T he subscript Y denotes
a Tresca yield value.

T a b le 3 .1 Description of Ratchet Mechanisms.

59
Mechanism Ilia - Outward Hinge Cone
Description of Sketch of Mechanism
Governing Equations. Definition of Terms.
Mechanism. and Stress cycie.

Outward hinge- Hinge forming stress,


cone resulting in
radial expansion.

where 0™* is the


maximum axial stress
acting to form hinge 1
(see figure 3.6).

Membrane stress, 0"$


Activated by axiai
0, .2m
therm o-elastic
w o^„ndx, w here
stresses and
occuring at a point
in the structure
- - fzô"“ dz
corresponding to Gfmman
' hj ‘
the maximum axial
thermal stress (see figures 3.7 and 3.8)
com ponent Aa®
R = Radius,
h = Wall thickness,
2a = Mechanism length,
w = Radial displacement,
normalised to vary between
0 and 1 over a mechanism
of half- length a.

Mechanism lllb - Outward Hinge Cone


Description of Sketch of Mechanism Governing Equations. Definition of Terms.
M echanism . and Stress cycle.

Outward hinge- Pressure, P: Hinge forming stress,


cone resulting in
. ( o r * 2 a : “ * o"“ ) / 4
radial expansion
and axial
where is the
compression.
maximum axial stress
Axial load, F:
acting to form hinge 1
(see figure 3.6).

Membrane stress,
Activated by hoop
.2a
or axial thermo­
0.$
where
elastic stresses
'X'
and internal
- max
pressure or
(a^-ajdz
compressive axiai
loads.
(see figures 3.7 and 3.8)

R = Radius,
h = Wall thickness,
2a = Mechanism length,
w = Radial displacement,
normalised to vary between
0 and 1 over a mechanism
of half- length a.

T a b le 3 .1 Description of Ratchet Mechanisms.

60
Mechanism IVa - Inward Hinge Cone
Description of Sketch of Mechanism
Governing Equations. Definition of Terms.
Mechanism. and Stress cycle.

Inward hinge- Hinge forming stress,


cone resulting in
<r)/4
radial contraction.

where is the
maximum axial stress
acting to form hinge 1
(see figure 3.9).

Membrane stress,
Activated by hoop
stresses and an “ wa^.andx. Where
external pressure
load occuring over
some length of :
O^mean
= - (z
j
ô $r dz
tube.
(see figures 3.7 and 3.8)
'0.-$
R = Radius,
h = W all thickness,
2a = Mechanism length,
w » Radial displacement,
normalised to vary between
0 and 1 over a mechanism
of half- length a.

Mechanism IVb - Inward Hinge Cone


Description of Sketch of Mechanism Governing Equations. Definition of Terms.
M echanism . and Stress cycle.

Inward hinge- cone Pressure, P: Hinge forming stress,


resulting in radial
contraction and
axial extension.
where is the
maximum axial stress
Axial load, F:
acting to form hinge 1
(see figure 3.9).

Membrane stress, 0"$


Activated by hoop
1
.2a
or axial thermo­ 0,
where
elastic stresses
‘ I:
and either external
pressure or tensile
(0, - O x ) , (0^-aJdz
axial loads. '"hi
(see figures 3.7 and 3.8)

R = Radius,
h = W all thickness,
2a = Mechanism length,
w = Radial displacement,
normalised to vary between
0 and 1 over a mechanism
of half- length a.

T a b le 3 .1 Description of Ratchet Mechanisms.

61
F ig u re 3 .1 The Six Surfaces of the Tresca Yield Condition in Stress Space.

^ a

F ig u re 3 .2 A Tresca Yield Surface. Showing a Typical Variation in Maximum


Thermo-Elastic Stress a^jj(t) over a Load Cycle.

Figure 3.3 A Tresca Yield Surface Showing Instants in time tj toT^


Corresponding to Maximum Values of a ^j/t) in each of the Tresca
Surface Directions.

62
Where oi and 02 are the
-a
stresses at any times 1
and 2 during a cyde
which give rise to
maximum and minimum
axial stress
components.
Tube
W all
AOx is the maximum
value of AGx along the
J -c
6 tube
► a

F ig u re 3 .4 Extreme Values over a Cvcle of Thermo-Elastic Stress throuoh the


Tube Wall. Giving rise to the Definition of

Where a^(ti)and
are the maximum hoop
components at instants
ti and fc in the cyde (as
(AÔ). defined in figure 3.3) at
any point along the
Tube structure.
W all

F ig u re 3 .5 Extreme Values over a Cvcle of Thermo-Elastic Stress through the


/\
Tube Wall. Giving rise to the Definition of (A otj)/mean.
.)

63
Where oT" bounds the maximum
and minimum bending stress on the
inside surface of the tube at any at
any time in a cycle.
A conservative assumption is
taken such that only stresses of the
correct sign to activate a mechanism
are included, i.e. stresses are
assumed not to inhibit the formation
of a ratchet mechanism

F ig u re 3 .6 Variation of Thermo-Elastic Bending Stress along a Hinoe-Cone


Mechanism, showing the Definition of Components for
Mechanism III.

2a

Where the area


represents the solution to
the integral
p2 a
I dx =04)
0

replaced by (ô*-ôx) for mechanism Ills


mean

cf4)m«n is replaced by (ox * ct4>^^lean for mechanism IVb

Figure 3.7 Variation of Membrane Stress Along a Hinoe-Cone


Mechanism. Showing the Definition of Gd»Components for
Mechanisms III and IV.

64
Tube
W all

J- 4>

F ig u re 3 .8 Variation of Peak Thermo-Elastic H o o d Stresses through the Tube


Wall. Showing Definitions of ^ "lax and g^^'^for Mechanisms III
and IV.

Where oT" bounds the maximum


and minimum bending stress on the
inside surface of the tube at any at
any time in a cycle.
A conservative assumption is
taken such that only stresses of the
correct sign to activate a mechanism
are included, i.e. stresses are
G assumed not to inhibit the formation
of a ratchet mechanism

F ig u re 3 .9 Variation of Thermo-Elastic Bending Stress along a Hinoe-Cone


Mechanism, showing the Definition of Components for
Mechanism IV.

65
0-0 Ae.

CTx

F ig u re 3 . 1 0 A Tresca Yield Surface showing a Typical Stress Cvcle and Strain


Increments Resulting in a Reverse Plasticity Mechanism.

V
6a

F ig u re 3 . 1 1 A Schematic of an Outward Hinae-Cone Mechanism.

66
10 -

0 0.5 1.0 1.5 2.0


Central Hinge Position, 5,
(as defined in figure 3.11)

F ig u re 3 . 1 2 Graoh showing Variation of the Geometry component of


Mechanism Equation (3.16) with the Position of the Central Hinoe.

67
6

\ “ f Schematic of a Stationary Spike


\ A0
Temperature Front.
AXs
Axs =

00
X

Schematic of a Moving Step


Temperature Front.
AXm
AXm
/ R h

Schematic of a Through-
Thickness Temperature Profile.

00

Where R = Tube Radius


h = Wall Thickness
X = Distance along Tube

9 = Temperature

F ig u re 3 . 1 3 Schematics of Soike. Step and Through-Thickness Temperature


Profiles.

68
I Mechanism IOptimum Solution from
Table 3.1.
II Mechanism IIOptimum Solution from
Table 3.1.
IVb Mechanism IVa Optimum Solution from
Table 3.1.
E Elastic Limit
A ASME-III Lower Bound Design Code
Solution.
L EEC_Shakedown Upper Bound
Linear Programming Computer
Program Solution.
R RCC-MR Design Code Solution.

■D 3.0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0. 0.9 1.0

Mechanical load p

F ig u re 3 . 1 4 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Methods for a Stationary Spike Temperature
Front under Axial Tension. AXg=0.1 (as defined in figure 3.13).

69
Il Mechanism IIOptimum Solution from
Table 3.1.
IIIa Mechanism IIIa Optimum Solution from
Table 3.1.
Ills Mechanism lib Optimum Solution from
Table 3.1.
E Elastic Limit
A ASME-III Lower Bound Design Code
5.0-r Solution.
EEC_Shakedown Upper Bound
Linear Programming Computer
Program Solution.
RCC-MR Design Code Solution.

ikT 3.0 -

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Mechanical load P

Figure 3.15 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Methods for a Stationary Soike Temperature
Front under Internal Pressure. ^«.=0.1 (as defined in figure 3.13).

70
I Mechanism IOptimum Solution from
Table 3.1.
E Elastic Limit
A ASM E-Ill Lower Bound Design Code
Solution.
R RCC-MR Design Code Solution.

5.0-r

4.5-

4.0-

lb 3.0..
"O
ro
o
ro
E
c_
oi
^ 2.0

1.5 --

0.5-

0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Mechanical load P

F ig u re 3 . 1 6 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Methods for a Stationary Through^
Thickness Temperature Front under Internal Pressure. 2^g=0.1 (as
defined in figure 3.13).

71
Il Mechanism IIOptimum Solution from
Table 3.1,
IIIa Mechanism MU Optimum Solution from
Table 3.1.
Ills Mechanism IIleOptimum Solution from
Table 3.1.
E Elastic Limit
A ASME-III Lower Bound Design Code
Solution.
EEC_Shakedown Upper Bound
Linear Programming Computer
Program Solution.
RCC-MR Design Code Solution.

1----- r
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Mechanical load p

F ig u re 3 . 1 7 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Methods for a Moving Step Temperature
Front under Internal Pressure. 2^.=0.6. ^^^=0.4 (as defined in
figure 3.13).

72
Il Mechanism IIOptimum Solution from
Table 3.1.
IIIa Mechanism IIIa Optimum Solution from
Table 3.1.
IVa Mechanism IVa Optimum Solution from
Table 3.1.
E Elastic Limit
A ASME-III Lower Bound Design Code
Solution.
EEC_Shakedown Upper Bound
Linear Programming Computer
Program Solution.
RCC-MR Design Code Solution.

" 2.0 -

T f
0.0 0.1 0.2 0.4 0.5
Mechanical load, p

F ig u re 3 . 1 8 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Methods for a Moving Step Temperature
Front under Internal Pressure. Ax^=0.1. Ax^=2.0 (as defined in
figure 3.13).

73
Il Mechanism IIOptimum Solution from
Table 3.1.
IIIa Mechanism liU Optimum Solution from
Table 3.1.
E Elastic Limit
A ASME-III Lower Bound Design Code
Soiution.
R RCC-MR Design Code Soiution.

Ax„=2.0 (Others removed for clarity)


~a
œ
o

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Mechanical load P

Figure 3.19 Interaction Diagram Comparing Predictions of the Ratchet


Boundary bv Various Mettiods for a Temperature Step under
Internal Pressure Moving Increasinolv Long Distances. Ax^=0,
to 5 (as defined in figure 3.13).

74
Chapte r 4.

Experiments on Thin Walled Tubes Un der Severe Cyclic

Thermal Stress.

4.1 Introduction.

All of the m e t h o d s of thermal stress a n a ly si s d i s c u s s e d in

th e p r ev io us two ch ap ter s must rely on a sound and complete

b a c k g r o u n d of e x pe ri me nt al evidence.

Full elastic-plastic sol utions rely on plas tic

constitutive mo d e l s of the material. The field of p l a st ic

modelling is by no means fully un de rstood, especially as

re ga rds st ro n g l y w o r k h a r d e ni ng ma te ri al s such as au ste nit ic

stainless steels. Th ere for e recourse must be made to

experiment in order to verify and exp and the range of

anal ysi s. B o u n d i n g theory, such as us ed in the sha ke do wn model

of the pr ev i o u s cha pte r relies on muc h simpler and more

thoroughly investigated el astic properties, but if b e ha vi ou r

ab ove the s h a k ed ow n boundary is to be in vestigated, then some

p l a s t i c i t y mo del must be used. Emp irical methods, such as the

R C C - M R de s i g n code rules, rely w h o l l y on the ex pe r i m e n t s used

to form the bounds. Although trends in b e h a v i o ur can o ft en be

accurately id e n t i f i e d from a limited set of experiments,

extrapolating such rules to regions out side those co ve red by

e xp er i m e n t s mu st be a p pr oa ch ed ve ry ca u t i o u s l y w i t h o u t some

the or et ic al basis.

Un for tu na te ly , thermal ra tch ett ing e x p e r im en ts are

notoriously difficult and expen siv e to perform. To induce a

thermal shock such as occurs in the sod iu m circuit of a LMFBR

requires either a liquid sod ium source (both of lim ite d

75
a v a i l a b i l i t y and di ff ic ul t to work with) or some other intense

po w e r source. In addition, the te mp era tur e field p r od u c e d must

be both con tr ol la bl e and re peat able for the results to be

c om pa r a b l e w it h any pr edi ctions. However, des pi te these

problems, several relevant experiments have been performed.

This chapter starts with a re vie w of some of the more

s ig ni f i c a n t ex pe ri me nt s to date, with particular a t t e n t i o n to

the ex pe ri me nt al me th od s u sed and the load types imposed. From

this it became clear that further e x p e r i m e n ta l work was

re qui red before the regions of interes t to de s i g n e r s could be

said to have been investigated. At the same time, insight was

g ai ne d on some of the pr ob lem s in vol ved in the va ri ou s methods

of he at in g and the requir eme nts of a future system.

The rest of the chapter d e s c ri be s the d es ig n of a series

of exp er im en ts to ap pl y a co mb i n a t i o n of cyclic thermal shocks

and axial ten sion to thin walled tubes, and the results

o bt a i n e d from the programme. Finally, the results are compared

w i t h those ob ta ine d us i n g the upp er bo und s h ak ed ow n theorem.

4.2 A Re v i e w of Thermal R a t c h e t t i n g Experiments.

The history of exp er im en ts on the effects of thermal

ra tc he tt in g can be d i v id ed into three groups, ac c o r d i n g to the

sp eci men ge om etr ies tested.

The first group consists of fundam ent al exp er im en ts

d es i g n e d to inv estigate the bu il d i n g blo cks of sha ke do wn and

cr e e p / sh ake dow n int er ac ti on theory. A number of tests were

p er f o r m e d dur in g the 1 9 7 0 ' s and ear ly 1980's on two and three

bar specimens. In these tests, the bars were co ns t r a i n e d to

suffer equal axial strains wh i l s t d i f f er en t thermal stresses

76
w er e applied to the individual bars. In particular, Uga [36]

u se d a three bar g e o m e t r y w it h two of the bars un der identical

thermal cycles, w hi ls t the third remained cold. H e a t i n g was

achieved by w r a p p i n g an insu la te d co n du ct in g wire aro und the

spe c i m e n and p a s s i n g an e l e ct ri c cu rrent th ro ugh it.

The second gr oup of exp er im en ts relate to the shake dow n

model of a nuc le ar reactor fuel can, i n i t i a l ly pr od u c e d by

Bree [30]. An ex pe ri m en t to in ve sti ga te both this model and an

u ppe r bo un d sh ake dow n mo del of the p r o b l e m was pe r f o r m e d by

Ponter et.al. [36]. A copper plate s p ec im en was sub je ct ed to a

combination of axial tens ion and a thr oug h thi ckness

te mp er at ur e gradient. Heating was induced by rad ia ti on from

i nf rar ed ligh tin g elements.

The last group conta ins work more relevant to fast reactor

design; ex pe ri me nt s on thin w al le d tubes. One of the ear li est

tests on this str ucture was by Yam am ot o et.al. [33] in 1976.

The ap pl ie d loads co ns is te d of a thro ugh thic kn ess tem perature

gra di en t co mbi ned w it h an axial load. This is similar in form

to figure 3.9 from the p r ev io us chapter. The t e m pe ra tu re s were

produced by pu mp in g alternately hot and cold liquid sodiu m

th rou gh a test pipe made from type 304 sta inless steel. The

results of three tests were pr e s e n t e d in the form of an

i n t e ra ct io n diagram on which load c o m b i n a t i o ns re su lting in

'sign ifi can t axial strains' were marked. Y am am ot o et.al.

c o mp ar ed these results w i t h the 'Bree line', wh i c h is referred

to as m e c h a n i s m I in this work. The results c o mp ar ed well in

all three tests. Although the measured ratchet strains were

all axial, it was not pos si bl e to de te rm in e the length of the

mechanism, as the tube was hea te d along a c o ns id er ab le length

and so suff ere d strain al on g a length.

77
Another not abl e e xp er im ent was performed by Co us in and

Ju ll ie n [37] at Lyon in France. This us ed a jet of hot gas

directed by baf fle s aro und the ci rc um f e r e n c e of a thin w a l l e d

tube to cause a sta ti on ar y axial te mp era tu re gradient. When

com bi ne d w it h an axial load, the resul tin g e x p e r i m e n t gave an

i n t e ra ct io n diagram similar to the p r e d i c t i o n of figure 3.14

in chapter 3. However, alt ho ug h the tem pe ra tu re d i f f e re nc es

wer e great, upper bo und ana lysis by Ponter et.al. [38] showed

that the resultant thermal stre sse s were such that all

e xp e r i m e n t s lay about the bou nd of m e c h a n i s m I. The com par is on

made by Ponter et. al. be tw ee n these e x p e r i m e n t s and m e c h a n i s m

I aga in showed a good correlation.

The final ex pe r i m e n t co ns i d e r e d is that p e r f o r m e d by Bell

[34] on a m ov in g step tem pe ra tu re front. A very thin tube

(0.381mm thick, 140 mm diameter) was he at e d by me an s of a radio

f r e q ue nc y coil and then lowered at a c o nt ro ll ed rate into cold

water. The exp er im en ts were c o n d u ct ed more w i th the aim of

demonstrating the influence of such thermal cycles, rather

than as a syste mat ic analysi s of the effects. The e x p e r i m e n ts

w ere con du ct ed w it h v er y severe thermal stresses un der no

me ch a n i c a l load. The results sho wed a radial e x p a n s i o n of the

tube co mb ine d w it h axial co nt r a c t i o n (i.e. similar to the

deformation of mechanism m b from table 3.1) de sp ite the

abse nce of mec ha ni ca l load. The report in clu ded p h o t o g r a p h s of

s pe cim ens after testing. These showed gross deformation

res em bl in g some form of hi nge - cone mechanism. However, it was

not poss ibl e from the photograph to positively state the

number of pla s tic hinges present. These results call into

do ubt the con ven tio nal definition of secondary stresses,

states in sec tion 2.2 as "..unable to cause a pl as tic strain

78
growth m e c h a n i s m w i t h o u t the a dd it io n of p r i m a r y load."

This ex p e r i m e n t does appear to support the th e o r y of hinge

cone mechanisms III and IV oc cu r r i n g under some load states.

Unfortunately there are no ex pe ri me nt s wh ic h show the ratchet

bound corresponding to such mec hanisms. Ta ki ng the other

me c h a n i s m s co n s i de re d in the p r ev io us chapter. M e c h a n i s m I has

co ns id e r a b l e support from e x p e r i m en ts under a range of

conditions. However, mechanism II is, as far as can be

determined, untested.

A w id e range of techni que s have been used in the above

experiments. The main di ff ic u l t i e s encountered by the

e x p e ri me nt er s lay in controlling, accurately r e pe at in g and

m ea s u r i n g the severe t em pe ra tu re cycles necessary to ge ner ate

s ig ni fic ant thermal stresses. There were several

c o ns id er at io ns in volved in c ho os ing a suitable method for

further thermal ra tc het tin g experiments. A liquid s o d i um loop,

as u se d by Yam am ot o et.al. was not available. The quenching

m et ho d of Bell was shown to be di ff i c u l t to control, and

suitable only for mo v i n g te mp er at ur e fronts. Ext ernal heat

sources, such as the gas jet used by Co us in and Ju ll ie n suffer

from inh erent d i ff ic u l ti es in c o nt ro ll in g the tem pe ra tu re from

cycle to cycle. A sol ut io n w ou l d be to e mp lo y a feedback

control mechanism to regulate the temperature. However, the

lag invol ved in a sys te m w i t h an external heat source w o u l d

make rapid he at in g impossible.

A n o th er problem for pre vi ou s e x p e r im en te rs was where to

place the tests in order to m ax im is e the in fo rm at io n gained.

It was shown in the pre vi ou s chapter that the ratchet

m e ch an i s m s are dependant, not only on the overall

th er mo -e la st ic stress magnitude, but on the relative size and

79
form of the stress components. Thus the ratchet bound

re su lt in g from an ex pe r im en t can v a r y g r e at ly w i t h the de ta ils

of the ge o m e t r y and a pp lie d temperatures. Experiments such as

those pe r f o r m e d by Cousin and Ju ll ie n could have yi el ded

considerably more inf or ma ti on if a de ta i l e d ana ly si s of the

e xp e c t e d results had been p e r f or me d first.

4.3 Exp eri me nt al Methodology.

The survey of exp er im en ts described in the previ ous

sect ion has revealed that a complete ju st if i c a t i o n of relevant

thermal load in g pro bl em s w o ul d require experiments under

several com bin at io ns of thermal and me ch an i c a l loading

conditions. However, the reverse p l a s t i c i t y mechanism, de fi ned

as m e c h a n i s m II in figure 3.14 has been shown in the exam pl es

g ive n in the pre vi ou s chapter to occur under a wide range of

co nd it io ns of interest to designers. A xi al loads, internal

pressure, st at io na ry and m ov in g tem pe ra tu re have all been

predicted to give rise to ra tc het tin g thro ugh a reverse

p l a s t i c i t y m e c h a n i s m under suitable conditions. As far as can

be determined, there is no ex pe ri men tal verification of the

a pp ea r a n c e of a reverse p l a s t i c i t y mechanism, or the result of

exceeding the boundary. It seemed i m por ta nt therefore, to

in ve sti gat e this m e c h a n i s m experime nta lly .

It is p r e d ic te d that the m e c h a n i s m can be a c t iv at ed by a

severe axial te mpe rat ure gradient, c o m bi ne d w i t h low, or zero

me c h a n i c a l load. To ensure a close co mp a r i s o n w i t h the type of

sit ua ti on s occ ur ri ng in practise, the m a t er ia l tested should

be a thin tube of a reactor grade st ai nless steel. In

addition, a thin wa l l e d section w o u l d al l o w c o m p a r is on wi th

80
thin shell upper bo un d sh ake do wn theory.

The first de c i s i o n to be made in considering the

experiment was the method of heating. A preliminary pro ject

co n d u c t e d by Taylor [39] used the m e t h o d of di r e c t resistance

heating: A high cu rrent pa ss e d t h r ou gh a ma t er ia l results in a

dissipation of el ec tr i c a l power equal to i ^ r . The resistance

of the ma t e r i a l is g o v e r n e d by the relationship

r = pl/A 4.1

wh e r e cross section area, A = 2nRh 4.2

r is the path resistance, p the resistivity,

1 the le ngth of the path, R the tube radius,

h the tube wall thickness.

Th er e f o r e a change of thic kn es s al ong the cu rrent path

wi ll cause a change in resistance, and thus the power

dissipated in the axial direction.

By profiling the thi ck ne ss of the sp e c i m e n a te mp era tu re

gradient can be g e n e r a t e d along the le ngth of the tube. Finite

e l e m e n t heat t ransfer a n al ys is w o u l d a l l o w p r e d i c t i o n of these

te m p e r a t u r e s and thus a c cu ra te de si g n of a t e mp er atu re

gradient to give the d e s i r e d thermo-elastic stresses. Another

ad v a n t a g e of us in g this m e t h o d was that it w o u l d a l l o w direct

fe edback control of t e m p er at ur e th r o u g h the power supply.

On the basis of these advantages, as well as the

co m p a c t n e s s and availability of the equipment, di re ct

r es is ta nc e h e a t i n g was ch osen to a p p l y thermal loads.

S e l e c t i o n of the d e s i r e d t h e r m o - e l a s t i c stress pro fi le and

the de si g n of the sp e c i m e n re quired to achieve it are

described in the f o l l o w i n g section.

Another consideration in the de si g n of the experiment

co n c e r n e d the number of cycles p e r f o r m e d at each load level. A

81
large number of cycles wo u l d ensure that d i s t i n c t i o n be tw een

s h ak ed ow n and ratc he ti ng b eh avi ou r had been made. It wo u l d

also give a good in di ca t i o n of the a s y m p t o t i c rate of pla st ic

strain ac cumulation. However, a large numbe r of cycles w o u l d

entail a longer test time, and so reduce the nu mb er of tests

that could be performed. It was decided that a clear

ind ic a t i o n of the onset of r a t c he tt ing was the mo st im portant

consideration and so the number of cycles would be kept

re la ti v e l y small. This would, of course, be at the ex pense of

val ue s for the as ym pt o t i c strain rate un der some high stress

conditions.

4.4 D e s i g n of Test Specimen.

As with all de si gn procedures, de si gn of the test sp ecimen

in volved the balancing of required cri te ri a with the

l i m i ta ti on s of cost, availability and time. In this case, the

de si g n cri te ri a were obtained from the properties of the

ratchet mechanism un der investigation. It was explained in

section 3.4 that the stress pr ofile requir ed for a c t i v a t i n g a

po s t u l a t e d reverse plasticity mechanism involve a large hoop

stress compo nen t in the t h e rm o- el as ti c stress. The

rel ati on sh ip be tw ee n hoop and axial stress co m po ne nt s due to

ap pl ied tem per atu re di st ri b u t i o n s has been i n ve st i g a t e d by

Carter and Ponter [24]. Figure 4.1 shows this relationship for

an id eal ise d te mpe rat ure spike of length Ax^. The results are

given in terms of di me ns i o n l e s s groups Ax^ and k, de fi ne d in

the figure. Whe re R and h are the tube radius and thickness, a

the co ef fi ci en t of thermal e x p a ns io n and E is the ela stic

modulus. It was clear from figure 4.1 that for the de si re d

82
high ratio of to a^, Ax^ must be small. This in turn

y ie ld ed three des ig n criteria, namely:

A x g should be small,

R, h should be large.

V al ue s of R and h were, to a large extent, g o ve rn ed by

other pra ct ic al criteria. An excessively thick spe ci me n w o u l d

cease to behave as a thin shell, c a us in g pro bl em s wit h

anal ysi s us in g cur rent sh ake dow n methods. On the other hand,

a c c u r a t e l y m a c h i n i n g ve ry thin sections of 316 stai nl es s steel

pr ov ed imp ossible for reasons e x p l ai ne d in se ct ion 4.5. The

total cross sectional area of the w o r k i n g section, 2nRh, was

limited by the av ail abl e heating, and me ch an i c a l loading

capacity. This led to a value of R equal to 75 mm and a

thick nes s h of 1.4mm.

The length of the te mp era tur e spike Ax^ was co nt ro ll ed by

the length of the thi nne d section, 1, as well as the heat

transfer p r op er ti es of the material. Thermo-elastic stress

anal ysi s w it h the CO N I D A [40] finite elem ent program showed

that us in g a long length, 1, gave higher t e mp er at ur es and

te mp era tur e gradients, but also in cre ase d Ax^. A short length

1 had the opp osite effect of lower temperatures, but a

c o r r e s p o n d i n g l y lower Ax^.

In order to de ter min e the effect of this p a r a m e t e r on the

s ha ked own boundary, mod els were pr ep a r e d for the EEC Sha ke do wn

program (see section 2.7) w it h v a r yi ng len gths of thinned

section 1. Figure 4.2 shows the v a r i a t i o n of the tem per atu re

at which ra tch ett ing by reverse plasticity was pr ed ic te d in

the abs ence of mec han ic al load, wi th 1. Also shown on the

graph is the maximum te mpe rat ure in the co rr es p o n d i n g

te mp era tur e profile. This was o b ta in ed u s in g heat transfer

83
analysis routines avai lab le in the A b a q u s F.E. p a ck ag e [41].

The ratchet b o u n d a r y showed little v a r i a t i o n over the range of

1, but the o bt ai na bl e temper at ur e rose sharply. A figure of 1

equal to 14mm was ch osen as the lengt h w h i c h gave the re quired

t e m p er at ur e grad ie nts for a short step length. This resulted

in a AXg va lue of 1.37, mar ke d on figure 4.1.

A s p e ci me n was d e s i g n ed and manufactured us in g these

parame te rs , a photograph of w h i c h is shown in figure 4.3. An

e n g i n e e r i n g dr aw in g for the sp eci me n is given in figure 4.4

4.5 316 Stai nl es s Steel as a S tr uct ur al Material.

316 st ai nless steel is w i d e l y u se d as a ma t e r i a l for pi pes

and p r es su re v e s se ls w i t h i n the power generation industry. It

has pa rt ic ul ar ad van ta ges in high t e mp era tur e ap p l i ca ti on s

be cause it exh ib it s a good resistance to the ef fects of creep

and corrosion. 316 nominally contains 17% Chromium and 12%

Nickel with small amounts of other a l lo yi ng materials.

However, there is a s i g n if ic an t variation in chemical

composition and manufacturing methods for the same mat er ia l

from d i i f f e r e n t sources. This leads to s i gn ifi can t di ff er en ce s

for the mat er ia l pr op er t i es qu ot ed in m a t e r i a l s data sources.

The chemical c o m p o s i t i o n of the sp eci men ma te ria l was sup plied

by the manufacturer (see app en di x C), however, it was still

important to obtai n accurate yi e l d stress data relating to

that pa rt ic u l a r composition. To this end, a series of u n i ax ia l

tensile tests were p e r f o r me d at v a ri ou s te mp er at ur es on

specimens m a c h i n e d from the s p ec im en tube. A graph of 0.1% and

0 .2% proof stress ag ainst te mp er at ur e for the mat er ial is

84
given in figure 4.5 (the c om po ne nt st re ss - s t r a i n curves are

given in ap pe nd ix C). Also pl o t t ed on the figure are the

Br it is h nucl ear de si gn code re co mm en de d val ues [42] for the

yi e l d stress of 316 stainless steel.

316 stai nl es s steel also exh ib it s si gn i f ic an t work

hardening properties. This, com bi ne d with a re l at iv el y low

heat transfer co ef f ic ie n t made the ma t er ia l d i f fi cu lt to

m a c hi ne accurately. In order to ov ercome this p r o b l e m it was

necessary to employ addit ion al manufacturing processes: part

com pl e t e d sp ecimens were a n n e a l e d at 1050®C in ac c o r d a n ce with

B.S. 3100 [43], and the tubes were ri gidly mounted on a

ma nd ri l during machining. The pur po se of the m a n dr il was to

force the tube to run concentrically du r i n g machining by

app l y i n g an internal pressure. This a l lo we d a reasonable

tol erance to be ac h i e v ed on the wall thickness. The price for

this tole ra nc e was that it was impos si bl e to pro fi le the

inside of the wall. This led to the a s y m m m e t r i c wall profile

shown in the photograph, figure 4.3 and e n g i n e e r i n g d r a w i n g in

figure 4.4.

4.6 D e s c r i p t i o n of Test Equipment.

Figures 4.6 and 4.7 show an overall photograph, and a

detail of the e x p e ri me nt al apparatus. These p h o t o g r a p hs are

la be ll ed 1 to 7 corresponding to c om po ne nt s of the apparatus

d e s c r i b e d below.

1. Te nsile te sting m a c hi ne of 25 tonne capacity. U se d to

ap pl y the me ch an i c a l load, and support power tr a n s mi ss io n

equipment. W e i g h t s were sus pe nd ed on the end of a lever arm

wi th a 100:1 me ch an i c a l advantage. The load was app li ed

85
u si n g a m ot or to raise or lower the w ei gh t hanger.

2. M ai n power transformer, rated at 6000 amps and 3 volts

se co nd a r y (Obtained c ou rt esy of the UKAEA). Power was

t r a n sm it te d from the t r a ns fo rm er by 20 paralleled copper

cables, fixed to the ci r c u mf e r e n ti al me c h a n i c a l clamps at

the ends of the specimen.

3. Clamps and wa te r coo ling blocks. 316 stai nl ess clamps

we re de si g n e d and made to ap ply an a x i s y m m e t r i c , even load

to the specimen. In order to m a i n t a i n a sharp temper atu re

gra di en t across the specimen, the ends we re coo led with

wa te r at room te mp e ra tu r e flowing th ro ugh h o l l o w blocks.

4. Te mp er at ur e measur eme nt . Dur in g the exp eri me nt al

de ve l o p m e n t phase, 'k ' type t h e r mo co up le s were we l d e d both

in and outside the specimen. This al lo we d m e a s u r e m e n t s of

th rou gh thick nes s and c i r c u mf er en ti al te mp era tur e

d ev ia t i o n s to be made, in add it io n to the axial profile.

Th ro ug h thick nes s va ri a t i o n s were always less than 1% of

the temperature. C i r c um fe re nt ia l variations did cause

problems, but they were res olved by a simple m et ho d

e x p l ai ne d in section 4.7. Dur in g exp eriments, axial

te mp er at ur es were measured, along w it h extensions, at

several points along the length of the specimen.

5. E x t en si on measure men t. Two clip gauge ex t e n s o m e t e r s were

us ed to me as ure the axial el on g a t i o n of the tube. Figure

4.8 is a ph o t o g r a p h of one of these devices. Th ey used a

res istance strain gauge bridge to me as ur e the be nd in g of a

thin beam f ixed to one end of the gauge length. The be am

bent relative to a rigid b ea m fixed at the other end of the

gauge length. In order to reduce the eff ect s of thermal

e xp an si on wi t h i n the be am material, the dev ice s were

86
ins u l a t e d from the hot sp ec im en by a pair of ceramic

e x t e ns io n arms. The devi ces were also co ol ed wi t h

co m p r e s s e d air to ensure they stayed at a con st ant

te mp er at ur e t hro ug hou t an experiment. It shoul d be noted

that these de vices were found to be most repeatable in this

c o n f i g u r a t i o n w h e n u se d inverted.

6. Data capture. T em pe ra tu re and ex t e n s i o n data was a c qui re d

by a pr o g r a m m a b l e data logger and store d on m a g n e t i c disc

via a B BC -B m ic roc om put er . A com pu ter p r o g r a m was w r i t t e n

to control both the pr ocess and data ac qu isition. This is

liste d in ap pen di x D.

7. Power control. The same comp ute r was u se d to dem an d

te mp er at ur e ramps from a feedback co n tr ol le r w h ic h

c o nt ro ll ed the power sup ply to the transformer. A

th erm oc ou pl e welded to the sp e c i m en provided a t em per at ur e

sensor for the feedback.

4.7 Ex p er im ent al Procedure.

Before each experiment, the sp ec ime n was prepared in the

following manner. The internal diameter, le ngth of thi nn ed

section and wall thic kn es s ar ou nd the c i r c u mf er en ce were

acc ura tel y measured. Variations in wall thic kn es s ar ou n d the

ci rcu mference affect the res istance of the tube, and so cause

cir cum ferential te mp era tur e gradients. A tol er anc e of + 0 . 0 5 m m

in wall thi ckness was accepted, l e ad in g to a po te nt ia l +2%

var iation in tem per atu re around the circumference.

A poten tia l problem as s o c i a t e d with the m e th od of

measuring ex te ns io n was that it gave ave ra ge extension over

the gauge length. Large local strains due to a lo ca li se d

87
mechanism could be m i s i n t e r p r e t e d as smaller av er ag e strains.

In order to in ve sti ga te this po ss ibility, s p eci me ns were

e n g r a v e d on the inside w it h a gr at ic ule of known line spacing.

Fo l l o w i n g this initial checking, the sp ec ime n was ins ta ll ed in

the test machine.

It was impo rt ant to ensure axial sy m me tr y of the

t em pe r a t u r e profile. Although c on si de r a b l e care was taken in

the d e s ig n and a ss em bl y of the a p p ar at us to try to ac hieve

this, the variation of te mpe rat ur e ar o u n d the c i r c u m f e r e n ce

co uld not in it ia ll y be imp roved below +8%. Eventually an

e mp ir ic al solu ti on to the problem was adopted: The

c ir c u m f e r e n t i a l te mpe rat ure g ra di en t was measured, and steel

shims ins erted be tw ee n the sec on da ry side of the tr an sfo rme r

and the cables leading to regions of excessively high

temperature. This had the eff ect of i n cr ea si ng the resistance

of some cables relative to the others in para lle l with them,

and so reducing the current dr awn in a region. By this met hod

the axial te mp er atu re sy mm e tr y was im p ro ve d to within +3%.

Appendix E contai ns more det ails of the co ns id e r a b l e

development requir ed to ob t a in c o n s i s te nt and ax is ym m e t r i c

te mp er a t u r e s and strains.

The ex pe ri m en t then p r o c ee de d as follows. The tube was

heated to the base temperature, Tq (see figure 4.9) and left

in order to al lo w the app ar at us and measurement sys tems to

reach a cons tan t temperature. The tem pe ra tu re was then ramped

up to the peak value T^^^, held for a short time at that

temperature, then al lo we d to cool for several m i nu te s to T q .

This sequence de fi ne d a thermal cycle. The thermal cycle was

re pe at ed a number, N times. On each cycle, readings were taken

at points (A,B,C from figure 4.9) within the cycle. The

88
m e ch an ic al load was then incr eas ed and an other N cycles

performed. The m ec ha n i c a l load was in cr e m en te d in this w a y up

to a load c o r r e s p on di n g to the 0.2% proof stress at room

temper ature, performing N thermal cycles at each stage. On

c o mp le ti on of this series, the spec ime n was removed from the

machine. The final sp eci men ge o m e t r y and sep ar at io n of

gr ati cul e lines was m e a s u r e d and c o mpa re d w i t h initial values.

A second series of e xp er i me nt s were also conducted in

order to ex amine the h i s t o r y d e p e n d e n c e of the loadings. The

spe cimens were p r e p a r e d in the same man ner as for the pr evi ous

experiments. However, rather than a p p l y i n g a c o nst an t thermal

cycle whilst in c r e a s i ng the mec ha ni ca l load, a con st an t

mechanical load was used in each experiment and the

te m p er at ur e T^^^ i n c r e a se d after each set of N cycles. Figure

4.10 is a sc he matic of the two load histories.

R e s ul ts and ana ly si s of these e xp er im en ts are p r e s e n t e d in

the fol lo wi ng two sections.

4.8 Res ul ts of Th ermal R a t ch e t t i n g Experiments.

Six experiments were p er fo rm e d (labelled C-H) wi th the

first type of loa di ng h i s t o r y and two (labelled I and K) with

the second. The pu rp os e of the second set of e x p e r i m e n ts was

to distinguish be t w e e n shak edo wn be ha vi ou r and cyclic

ha r d e n i n g w i t h i n a ratchet region. Cyclic hardening is a load

history dependent effect, unlike s h ak ed ow n w h i c h should y ie ld

the same bo und ir re sp e c t i v e of the load path. Thus a

discrepancy b e t we en the two results would suggest cyclic

hardening. Te mp e r a t u r e pr of il e s measured at point 'B' (see

89
figure 4.9) in the cycle are given for the first set of

experiments in figure 4.11, along w i t h the m a x i m u m var ia ti on s

in these te m p e r a t ur es b et wee n cycles. In the secon d set of

experiments, the thermal cycles were incremented us in g the

tem p e r a t u r e p r of il es from e x p e ri men ts C-H in order of

i n c r e a s i n g T^a*.

The strain accumulated at each mechanical load step in

experiments C to H is p l o tt ed in figure 4.12. It can be seen

from this gr aph that all the tests foll ow ed a similar trend in

strain ac cu mulation. For low strain va lu es i.e. 0.1-0.2%, the

strain gr o w t h rate is small. However, as the me c h a n i c al load

increases, the rate of strain gr owth incr ea se s quite rapidly.

This ef fe ct can be seen more c l ea rl y on figure 4.13. The

gr aph shows the ac cu m u l a t e d strain corresponding to each

thermal cycle imposed dur in g ex p er im en t H. The jumps in

accumulated strain show where the m e ch a n i c a l load has been

i nc re me nt ed and the load at each incr em ent is marked. It can

be seen that until the load reaches 60 KN, no si gni fic ant

strain accumulation is occurring and the total a c c u m u l at ed

strain will stay below 0.1%. Small flu ctu at io ns are pro ba bl y

due to m e a s u r e m e n t errors, or slight te mp e r a t u r e variations.

However, b ey on d this load, the strain rate inc reases markedly,

ra pidly e x c e e d i ng the proof stress. This shows the onset of

ra tc h e t t i n g behaviour.

The full co ll e c t e d data, incl ud in g nu mb er of cycles per step

and cycle times is gi ven in a p pe nd ix F.

Measurements of the th in ne d se ct io n le ng th be fore and

after te sting c o m pa re d well w it h the total accumulated strain

measured by the gauges. Diam et ra l measurements revealed

p l a st ic inward hoop strains in the order of one half the axial

90
pl as ti c strain accumulation. This c o mp ar ed well with finite

e l e me nt hoop strain p r e di c ti on s p e r f o r m e d in ch apter 5.

Res ults ob t a i ne d by m e a s u r i n g the de fo r m e d g ra tic ul e lines

show that no local strains of 0.5% or gre ater (the m i ni m u m

res olu ti on of the method) were developed. This verifies that

the strain a c cu m u l a t i o n was re l a t i v e ly uniform along the

deformed se ction of the specimen. It also confirmed that the

av er a g e strain value over the gauge length, as m e a s u r e d by the

clip gauges gave a good in dic a t i o n of the maximum strain

experienced within the material. However, it is p o ss ib le that

eve n if a local mechanism was respon sib le for initial

ratchetting, a general mechanism in vo lv in g larger strains

be ca me a c t iv at ed at higher m e ch an ic al loads.

4.9 Co mp a r i s o n of Ex p e r i m e n ta l R e su lt s with Up per Bound

Solutions.

Any series of com plex e x p e r im en ts is lim ite d by cost and

a v a il ab le time. The re fo re they can do no more than take a

sna ps ho t of the mat er ia l b e h av io ur under a li mi ted set of load

conditions. The im por tan ce lies in comparing these results

with a piece of the ory de v e l o p e d to en co mp as s the be hav iou r

over a m uc h wider range.

C om pa r i s o n wi th the linear p r o g r a m m i n g up per bo und met hod

discussed in chapter 2 was a c hi ev ed by using temper atu re

p r of i l es measured in e xp er im en ts C-K to cal culate

thermo-elastic stress di str ibu tio ns. These dist rib uti ons ,

produced w it h the thin shell finite el em ent pa ck age CON ID A

[40], were us ed as the basis of the input to the linear

91
programming upper bo und package, EEC Shakedown. Ou tp ut from

EEC_Shakedown was in the form of i n te r a c t i o n d i ag ra ms

consisting of two ratchet mec hanisms. At hi g h thermal

stresses, ra tc he tti ng was p re d i c t e d to occur t h r ou gh reverse

plasticity (mec ha nis m II from figure 4.14), and at high

me ch a n i c a l stresses, the Bree mechanism (me cha ni sm I from

figure 4.14) caused ratchetting. The in te r a c t i o n di a gr am s for

in div idu al e xp er im en ts w ere us ed to bui ld a com posite

in te r a c t i o n diagram for the wh o l e range of experiments by

ta ki ng a point from each co mp on en t in te r a c t i o n diagram

corresponding to the tem pe ra tu re at wh ic h the exp er im en t was

performed. This was made po ss ib le by the fact that the bounds

produced by e l a s ti c - perfectly p l as ti c shakedown ca lc ula tio ns

are not h i s t o r y dependant. Thus, the results of both sets of

experiments should compare w i t h the prediction.

Figure 4.15 shows this ratchet bound, calculated for a

0 .2% pro of stress, along w i t h pl as ti c strain co nto urs from the

experim ent s. It can be seen from this figure that a good

c o r r e l a t i o n b e t we en the ory and exp er im en t was ac hi e v e d for the

regions of lower thermal stress. This c o r r e sp on ds to the Bree

mechanism for e x p e ri me nt s G and H, and the reverse p l a s t i c i t y

mechanism for ex pe r i m e n t C. However, above this thermal

stress, some other mechanism or p h e no me na occ urs and the

ex pe r i m e n t a l contour s increase independently of the a p p li ed

thermal stress.

92
4.10 C o n c l u s i o n s .

These e x p e ri m en ts verify mechanism I and give a

quantitative value to the pla stic strain accumulated by

e xc e e d i n g the sha ke do w n bound. They also show that m e c h a n i s m

II gives a c o ns er va ti v e limit to the on set of cyclic plastic

strain. However, the pla stic strain accumulation above the

predicted bou nd seem to indicate an over c o n s e r va ti ve limit

for small plas tic strains.

An ex pl a n a t i o n for this observation is that cyclic

hardening wi t h i n the ratchet region limits the strain

accumula tio n. This hyp ot he si s is bac ke d by the results from

e x p e ri me nt s I and K, pl ot te d along w it h the sh ak ed ow n bound

prediction in figure 4.16. The results of the two sets of

ex pe r i m e n t s compare well in the vicinity of mechanism I.

However, di f f e r e n t strain rates are found in regions above the

reverse plasticity limit, su gg es ti ng that the phenomenon is

history dependent, unl ike sh ake dow n which is not h i s to ry

dependent.

This cyclic h a r d e n i n g h y po t h es is is e x p o u n d e d in the next

chapter. Di f f e r e n t me th od s are us ed to ca lcu lat e the effects

on ratchet bounds and these me th ods c om pa re d with the

experiments.

93
A)fe
( a s d e f i n e d in

figure 3 . 1 3 )

1.37

.01 -

.001
0.0 0.5 1.0 1.5 2.0 2.5
k=
EaAT

F ig u re 4 .1 Variation of Normalised Thermo-Elastic Stress with AXg (as


defined in figure 3.13) for an Idealised Temperature Spike.

94
600

10 12 14 16 18 20

Thinned Tube Length (mm)

+ Predicted Maximum Obtainable Temperature


A Predicted Ratchet Temperature

F ig u re 4 .2 Graph showing Predicted Variation of Maximum Obtainable Tube


Temperature Required for Ratchetting with the Thinned length of
Tube.

95
Jf

4.4 Photograph of Test Specimen

96
79.0

11.0
25.0 26.7 1.15
6.35. 14.0

//A /^/f^y///A

LO
CO

Section A -A

2 rows of 10 holes 010


equally spaced with 18° o ffs e t

O © © O ©

Half development of specimen surface

Thermal loading specimen

MatT: 316 Stainless tube 6" diameter

F ig u re 4 .4 Drawing of Test Specimen.

97
250
0.1% proof strain from
experiment
Tensile 0.2% proof strain from
Stress experiment
(MNm*^) Recommended proof
stress curve from [42]

200 -

150 -

100 -

0 100 200 300 400 500 600 700

Temperature (®c)

F ig u re 4 .5 Graph of Experimental and Design Code Recommended Values


of Proof Stress for 316 Stainless Steel.

98
4

4.6 Photograph of Thermal Loading Rig Layout

99
4.7 Detail Photograph of specimen and Grip Assemblies.

100
4.8 Photograph of Clip Gauge Extensometer

101
Peak
Specimen
Temperature

Tmax

Time

F ig u re 4 .9 Schematic of a Thermal Load Cvcle. showing Temperature


Measurement Points.

102
Pz

Pi

Time

etc.
To
Time

F ig u re 4 .1 0 a ) Schematic of Cyclic Load and Temperature History for


Experiments C to H.

Time

etc.
To
Time
F ig u re 4 .1 0 b ) Schematic of Cyclic Load and Temperature History for
Experiments I and K.

103
500
Specimen Half- Section

Y '

400

TTC)

300-
D

H
200 -

100 -

I ' I ' I ' r


-2 0 2 4 6 8
Distance along thinned length, L (mm)
(see inset sketch).

(Error bands bound the temperature


extremes measured between cycles).

F ig u re 4 .1 1 Temperature Profiles in Thinned Region of Tube for Experiments


CtoH.

104
140

120 -

100 -

Load

80
(KN)
-

60
Experiment C

Experiment D
40
ExperlmentE

Experiment F
20 -

Experiment G

Experiment H

0.2 0.4 0.6 0 .8 1.0 1.2 1.4

Accumulated Plastic Strain (%)

F ig u re 4 .1 2 Graph Showing IheTotal Accumulated Plastic Strain at Each


Load Step for Experiments C to H.

105
F10
(lOOkNV

I 0.7 -
F9
(90kN)j

"R F8
(80kN;

F7
(70kN)

F3 F6
F2 (30kN) F5 (GOkN)j
0.1 - FI (20kN) F4 (50kN)
(lOkN) (40kN) j i

- 0.1
0 50 100 150 200
Num ber of Therm al Cycles

F ig u re 4 .1 3 Graph showing Accumulation of Plastic Strain with Each Cvcle for


Experiment F.

106
Temp.
Gradient
AT(C)

160
Mechanism Mechanism IV b

140 Mechanism II

120

100

80

60

20

20 40 60 80 100 120
Mechanical Load, P (KN)

F ig u re 4 .1 4 Interaction Diagram showing Ratchet Bounds for Experiment D


using the Mechanisms Equations of Table 3.1.

107
AT (^c)

200
+ 0.1% Accum. Strain
Experiment F ri G -- ° 0.2% Accum. Strain
180 o 0.5% Accum. Strain
A 1.0% Accum. Strain
— 0.2% proof stress
Shakedown bound
160
Experiment E

Experiment D -B---
140

120

Experiment C — © - ------------
100

Experiment H e
80

60 Experiment G -------- A

40

20

20 40 60 80 100 120 140 160

Mechanical Load, P. (KN)

F ig u re 4 .1 5 Interaction Diagram showing Accumulated Strain Points from


Experiments C to H and 0.2% Proof Strain Ratchet Bound
Prediction.

108
AT «c

0.1% Accum. Strain


° 0.2% Accum. Strain
180 - o 0.5% Accum. Strain
0.2% proof stress
Shakedown bound

140 -

120 -

100 -

80 -

60 -

20 40 60 80 100 120 140 160

Mechanical Load, P. (KN)

F ig u re 4 . 1 6 interaction Diagram showing Accumulated Strain Points from


Experiments I and K and 0.2% Proof Strain Ratchet Bound
Prediction.

109
Chapter 5.

Estimation of Plastic Strain Accumulation due to

Ratchetting.

5.1 I n t r o d u c t i o n .

In the pr evi ou s chapter a set of thermal r a t c he tt in g

experiments on thin w a l l e d tubes were presented (see figure

4.20). To summm ari se these results: At low thermal stresses

the exp eri me nt a l contour of (for example) 0 .2 % p l a st ic strain

co m p a r e d well with the corresponding sh ak e d ow n bound

calculated for a 0.2% pr oof stress. However, at high thermal

str esses the contours of pl as ti c strain diverged from the

pre di ctions, becoming i nd ep en da nt of the ap p l i e d thermal

stress. This su gg e s t e d that some other phenomenon was

occurring.

The upper bo un d theorem p r edi ct s that the region of

sha k e d o w n under these high stresses is b o un d e d by a reverse

plasticity mechanism, as p o i nt ed out by Ponter and Ka ra deniz

[20]. It has been further sugg es te d by K a rad en iz [3] and

Megahed [44] that the re versi ng cyclic p l as ti c strains

occurring in such cases can result in s i g n if ic an t cyclic

hardening for some materials. The ma t e r i a l u se d in these

experiments, 316 stai nl es s steel, is k no wn to suffer

co n s i d e r a b l e cyclic hardening effects. This would lead to a

si g n if ic an t re du ction in pla st ic strains in load regimes

bordering the 'F' region. A co n s e r v a t i v e prediction of

accumulated pl astic strain co ntours above the reverse

plasticity bound, li mited by m a te ri al hardening, wo u l d all ow

de signs to be assessed under c o n d it io ns of ex treme thermal

110
stress.

Unl ik e both e l a sti c sh ake do wn and reverse plasticity,

which can be e x p l ai n ed in terms of an el as ti c perfectly

pl as tic ma ter ia l model, this h a r d e n i n g b e h a v i o ur ne c e s si ta te s

some pla st ic analysis. Fortunately, it is still not n e c e s s a r y

to resort to full mu lt i v a r i a t e p l a st ic analysis, such as that

developed by Cha bo ch e [45] et. al. There are simple h a rde ni ng

mod el s w h i c h can be u s e d as a b ou nd to the ma t e r i a l behaviour.

These mo de ls are a d o p t e d in this ana ly si s and d e s c r i b e d in the

fo ll o w i n g section.

There is also a choice of ap p ro ac h in a p pl yi ng such

h a r d e n i n g mod els to cyclic lo ad in g problems. The most rigorous

me t h o d is to app ly h ar d e n i n g e q u a t i o n s d u r i ng ela st ic - plastic

ana ly si s of each lo ad in g cycle. The total p l a st ic strain can

then be ca lc ul at ed by sum ming these strain in cre men ts over an

a p p r op ri at e number of cycles. The co m p l e x i t y of this method

requires a nu mer ica l c o m p u t i ng method. An alternative is to

p o s tu la te a deformation mechanism and assume co mplete cyclic

hardening to a st eady state after an u n d e t e r m i n e d number of

load cycles. K no wi ng the final and initial stress states wo uld

a ll o w a limit va lue for the accumulated strain to be

calculated.

These two me t h o d s are i n ve st ig at ed in this chapter. The

ans wer s are co mp ar e d w i t h e x p e ri me nt al results for two simple

hardening rules use d w i t h each method. From this, a number of

con cl us io ns are dr awn about the specific pro bl em s and

ad v a n t a g e s of the two approaches.

Ill
5.2 A D e s c r i p t i o n of Simple H a r d e n i n g Models.

A truly accura te plastic ma ter ial model would require a

host of va ri ab le s for de scr iption. The ef fects of temperature,

str ain rate, cold creep and Bauschinger ef fect would each

require one or more variables, even a s s u mi ng that the mate ria l

was at a te mpe rat ure b el o w the creep range. This ki nd of

an al ys is is not co mpa tib le w it h the sim pl if ie d the ory em plo yed

in this thesis. However, two simple mo de ls of plastic

beh aviour; Ki nem at ic and Is otropic hardening, are available.

Be t w e e n them, these mod els show some of the more im po rtant

fea tures of a real plas tic material.

Both of these h a r de ni ng mod el s are explained, in general

terms by Ma r t i n [5] and for a num er ic al anal ysi s by Hibbet

et.al. [46]. However, there follows a br ief d e s c r i p t i o n of the

two methods;

i) Kin em at ic Hardening; Plas tic strains cause the yi eld

surface to tran sla te al ong the line of the strain vector.

There is no change in the size of the yield surface, so

repeate d cycles of a stress v a r i a t i o n in excess of 2 ay will

cause some pla sti c strain on each half cycle. Figure 5.1

shows a typical Von Mi ses yield surface. The faint line

surface cor re s p on ds to translation due to ki ne m a ti c

h a r d e n i n g caused by the stress cycle 6 a.

ii) Isotr opi c Hardening; In this model, a p l a st ic strain

incr eme nt causes the yi e l d surface to exp and u n i f o r m l y in

all direct ions. This ex p a n s i o n will cont in ue until the

largest stress extreme has been encompassed. Figure 5.2

shows the ex pa ns io n of a Vo n Mises yi el d surface due to

is otropic ha r d e n i n g caused by the stress cycle 6 a.

112
The accuracy of the estim ate s of p l as ti c strain produ ced

by these two mo del s depe nds on a number of factors, such a

material, strain rate and strain magnitude. Th er ef or e both

models will be a p p lie d and the results c o m p a r ed the

experiments from the prev iou s chapter.

5.3 Si mp l i f i e d Mo dels of B e ha vi ou r Ou ts ide the Sha kedown

Bound.

a ) Introduction.

Several aut hors over the last ten years have co nsi der ed

simple mod el s of the effects of cyclic h a r d e n i n g on shakedown

bounds. Megahed [44], Ka radeniz [3], Ponter and Cocks

[47][48], G o o dm an [49] and Cocks and Ponter [50], have all

contributed to the pr ed i c t i o n of these effe cts for

i n c r e a s i n g l y com plex load cases.

In 1985, Cocks and Ponter [50] ap p l i ed ide al is ed ha rde nin g

rules to a range of pr ob lem s for c o mp ar is on w it h exp eri men tal

results produced by Leb ey et.al [51]. One of the cases

e xa m i n e d a tube under a c om bi na ti on of a tem pe ra tu re step

moving a short d i st an ce along the tube (relative to length of

the step) and an axial me ch an ic al load. It can be seen from

the me c h a n i s m s ba se d anal ysi s of this case (figure 4.14) that

the me c h a n i s m s re sul tin g from such a short travel of the

te mp er at ur e front are the same as those predicted for the

e x p e ri me nt s presented in the pre vi ou s chapter. In particular,

the same local reverse plasticity mechanism occurs at high

thermal stress.

This su gge ste d that the Cocks and Ponter m et ho d could be

113
a p p li e d to the results of an upper b ou nd an aly sis of the

experiments to give an e s ti ma t i on of the pl as tic strain

accumulated wi t h i n the regime of a reverse plasticity

me chanism. In fact the d e r i v a t i o n in [50] was of such general

form that the only as su m pt io n s specific to the g e om e t r y were

that it was a local mechanism, caused by a combination of

re ve rs in g hoop stress and con stant axial stress. However, the

th eo ry did assume that this same m e c h a n i s m was the op ti mu m

ratchet mechanism t h ro ugh ou t the region un der co ns ideration.

If a not her m e c h a n i s m were to occur at a highe r me c h a n i c a l or

thermal load wh ic h did not support these cyclic hardening

effects, then this w o ul d limit the region c a lc ula ble by such

an approach.

The me t h o d employs a Mises yi el d condition, with an

assumption that Poiss on' s ratio is 0.5. M a te ri al ha r d e n i n g is

g ov er ne d by a single modulus, 3E, w h i c h is the gr a di en t of the

p la sti c slope of the s tr es s- s t r ai n curve as shown in figure

5.3. This makes the a s s u m p ti on s that the p l a st ic slope is

inde p e n d a n t of temperature, strain rate or strain history.

Re s ul ts for 316 steel pr o d u ce d by the Oak Ridge Nat io nal

Laboratory [51] show these to be reasonable assumptions over

the range under co ns ideration. For the hardening the ories

adopted, the model assumes either co mplete hardening to an

el astic state (for isotropic hardening) or a st eady pl as tic

cycle (for k i ne ma ti c hardening).

114
b) S i mp li f ie d Kine mat ic H ar d e n i n g Model.

The ki ne m a t ic h a r d e n i n g theory ad op ted by Cocks and Ponter

[50] assumes that the yi el d surface tr ans lat es in the

direction of the strain rate vector over a number of cycles.

This co nti nue s until further cycles result in pure reverse

pl ast icity, with no net ac cu mu l a t i o n of p l a st ic strain . A

precondition for this is that the stress cycle must exceed

twice yield.

The total axial plas tic strain accumulated in reaching

this state can be de t e r m i n e d from the di s ta nc e the centre of

the yi el d surface has been tra ns la te d in the axial direction.

This leads to the equation;

•p - 1
de r i v e d in ap pe ndi x G, where P is the ef fe ct iv e p r i ma ry stress

and gE the pl as tic modulus, d ef i n ed in figure 5.3.

There are three immediate con cl us io ns that can be drawn by

examination of this equation. Firstly, the p l a st ic strain

accumulation is in de pen den t of the thermal stress. This is a

featu re of the results of the e xp er im en ts from the pr evi ous

chapter. However, it should be rep eated that this ass umes that

the same reverse plasticity mechanism still d o m i na te s as the

thermal stress increases. Secondly, the pl as ti c strain is also

not d e p en de nt on the load history. This is cl ea rly in

co n t r a d i c t i o n with the ex pe ri me nt al results. The third

co nc l u s i o n is that since (3 is taken as a constant, the plastic

strain a c c u m u l a t i o n var ies li nea rly w i t h the ap pl ied load, P.

This yie ld s a simple c a l c ul at i on for predicting the pl astic

strains ac cu m u l a t e d in experiments.

115
c) Si mp l i f i e d Is otropic H a r d e n i n g Model.

The ca lc u l a t i o n of pl astic strains u s in g an isotropic

model is sl i g h t l y more complex. The full d e r i v a t i o n from [50]

is gi ven in a p pe nd ix G. However, it can be summarised as

follows. The pl as ti c strai n rate in the axial d i r e c t i o n can be

calculated from the pl ast ic co n s t it ut iv e re la ti ons hip s to be;

1-0 f(3/4)Pa. - (l/2)(jn

"'p ■ ^ ^ ---- -j "'t


wh er e is a Von Mi ses e ff ec t i ve stress corresponding to

first yield. This eq ua t i o n can be in te gra ted across a half

cycle ^ *t ^ ^^t give the strain accumulated. The strain

is accumulated from an initial state, wh ere the size of the

yi el d surface is known, to a final isotropically e n la rg ed

yi el d condition. Assuming com plete iso tropic hardening, this

y ie l d surface is equal to in size to the maximum app lied

eff ec ti ve stress. Half cycle strains can be summed bet ween

these limits (ignoring small second order terms) to give;

- I 5.3

Cy is the initial size of the yi e l d surface and <y^ is the

final y i el d surface, i.e.

f
^e

wh er e 6 <y^ is the ef fe ct iv e thermal stress range.

It can be seen that under ce rt ain circums tan ces , equatio n

5.3 can lead to a h i s t o r y d ep e n d e n t strain ac cu mulation. This

agrees with the results of the e x p e ri me nt s in chapter 4.

116
However, equ at io n 5.3 is d e r i v e d for a co nst ant va lue of P. In

e x p e ri me nt s C to H from chapter 4, P is i n cr em en te d be tw een

stress cycles. In order properly to model this load regime,

eq u a t i o n s 5.3 and 5.4 must be m o d i f i e d by pu t t i ng


n=l
X 5.5

wh e r e s is the number of load steps, and

3(1-6) ((3/4)P^ + (1/4)6,:)


” F"6Ë l°9e((3/4)p:_i + (1/4)6,:) 5.6

The size of each of these steps will effect the total strain

accumulation.

5.4 Finite Elem ent A n a l y s i s

a) Introduction.

As out li ne d at the start of this chapter, pr o bl em s of

cyclically hardening st ruc tu res un der c o mb in ed thermal and

m e ch an ic al loads can be solved by num er ic al co mp ut in g methods.

There are several comme rci al p a c k a g es ava il ab le w h i c h use the

finite el em ent (F.E.) me t h o d to solve such problems. These

v ar y considerably in the range of ge om et ri es and loads that

can be modelled, p a r t i c u l a r l y for thermal stresses.

The ma in pr act ica l ad van tag e of the F.E. method is that

the num ber of as su m p t i o n s req uired to s i mp l i f y the problem

down to soluble pr op o r t i o n s is m uc h smaller than for any

analy tic method. Therefore, if the the ories used are sound and

the ma te ria l data good, then the sol ut ion ach ie va bl e should be

117
be tter than for the sim pli fi ed methods. However, this does

mean that the q u a n t i t y of i n f o r ma ti on that m us t be sup pl ie d is

mu ch greater and, even w ith the fastest computers, each

so lut ion takes a si gni fic ant time. Another problem inherent

w it h F.E. me th ods is that a large volume of output is

produced. This can prove di ff i cu lt to in ter pre t wi th o u t

d e t a i l e d kno wl ed ge of the met ho d of s o lut io n adopted.

The Ab a q u s [41] program is one of the most complete

no n - l i n e a r F.E. p a c ka ges c u r r e nt ly available. Proce du res

in clu ded al l o w cycle by cycle el ast ic- pl as tic anal ysi s with

either ki ne ma ti c or is otropic hardening models. More complex

hardening theorie s are not av ail ab le in Abaqus, or similar

gen eral F.E. packages.

b) Finite ele men t Model.

The data required for finite el em ent c a l cu la ti on s can be

divided into geometry, ma te ri a l p r op e r t i e s and the load

history. In this se ction these co ns t r a i n t s are described,

along w i t h si mp li fy in g a ss um p ti on s used.

A model of the ex p e r i me n ta l spec ime n was pr e pa re d

co n s i s t i n g of 36 ax i s y m m e t r i c solid e l e me nt s (see figure 5.4).

Each elem ent co nt ai ne d 4 nodes and 4 integration points. The

assumption of axial sym me tr y fol lo we d c o n s i d er ab le eff ort to

ensure that this was oc cu rr in g within ex pe riments. The

ex pe r i m e n t was also as su med to be s ym met ri c about the point of

maximum temperature, half way along the len gth of the tube.

Te mp er a t u r e s measured du r i n g e x pe ri me nt s showed this to be

ac cu rat e to wi t h i n 3%.

Ma te ri al data u sed in the model is given in a p pen di x C.

118
Proof stress va lue s were obtained from u n i ax ia l tests on

b a t ch e s of sp ec ime n material performed by the author. The

other data required, sp ec if ic al l y elastic moduli, co ef fi c ie nt s

of thermal e x p a n s i on and Poissons ratio, were o b ta in ed from a

metals reference book [52].

The loa di ng h i s t o r y c o ns is te d of two i n d e p e n d e n t l y applied

loads, one a c y cl ic a l ly v a r y i n g t e mp er atu re distrib uti on, the

ot her an axial mec han ic a l load. Te mp eratures, taken from

ex pe ri m e n t a l mea surements, were ap pl ie d at each node. Thermal

cy cl in g was p r od uc e d usi ng a ramp load fun ct io n supplied

within Abaqus. From this, thermal stresses were c a l c ul at ed

internally and ap pl ied in c o n j un ct io n with mec ha ni ca l

stresses. The mec han ic a l load ap pl i c a t i o n was assumed to be

tak in g place across a rigid bol te d connection, giv in g rise to

the set of res traints shown in figure 5.4. It was a s s um ed that

the comparatively long end stubs in the test sp ec ime n wo ul d

smo oth local load ir reg ula rit ies in acc ord an ce with St.

Ve n a n t s principle.

The use of 4 noded elem ent s li mi ted the anal ysi s to a

linear v a r i a t i o n be tw een nodes. A more a c cur at e an al ysi s wo ul d

be obt ai ne d by us i n g 8 noded elements, a l l o w i n g for qu ad r a t i c

variation be t we e n nodes. However, the 4 no ded results

pr e s e n t e d here still took several mon th s to pr od uce us in g the

V A X 86 0 0 computer time available. Producing a more acc ur at e

model would have required facil it ies w h i c h were not a v ai la bl e

at the time of the analysis. The p ro bl em s experienced in

developing and running an accu ra te F.E. model u n d e r li ne the

need for reliable s im pli fi ed methods.

Figure 5.5 shows a schematic of the lo ad ing history for

e xp er im en ts C-H. Two arbitrary load steps are given.

119
consisting of N cycles at mec han ic al load P^, foll owe d by a

further N cycles at P 2 . For exp er im en ts I and K the thermal

stress was i n c r e m en t e d at each step, in conjunction w it h a

co n s t a n t m e c h a n i ca l load. The de ta ils of the load steps and

nu mb er of cycles us e d for specific e x p e r i m e nt al mo dels are

gi ve n in app e n d i x C.

5.5 R e s ult s of Finite Ele me nt and S i mp l i f i e d H a rd en in g

Model Analyses.

a ) Introduction.

Re s ul ts of the two tech niq ue s for estimating pl astic

strai n accumulation have b ee n plotted onto e x p er im en ta l

results for both is otropic and kin em at ic ha r d e n i n g models.

Re s u l t s are presented in the form of interaction diagrams.

However, the thermal stress axes are left in terms of 6T, the

maximum t e m p er at ur e variation over a cycle, in order to al low

the mo s t di re ct c o m p a ri so n with ex p e r i m e n t al results. The

up per bou nd sh ake do wn b o u n d a r y prediction, p r e p a r e d u si ng the

EEC_Shakedown program is also pr e s e n t e d on the graphs for

reference.

The results from these models are com pa re d w it h those from

the e xp er im en t s of chapter 4 in the fo ll ow in g sections. A set

of h y po th es es are presented to e x p la in the differences that

occur b e t we en the results. Some of these h y p o t h es es are then

investigated, lea di ng to c on cl us io ns about the applicability

of such analysis.

120
b) Ki ne m a t i c Ha r d e n i n g Models.

Re s ul ts from the finite ele me nt analysis, shown in figure

5.6 gave a c o n s e r v at iv e b oun d over the majority of the

i n t e ra ct io n diagram. For low thermal stresses the co rr e l at io n

with e x p e ri me nt s was very good. This was to be expected, as

the mechanism active, type I in the terminology ado pt ed in

chapter 3, in vo lve d p la st i c strain at just one point in the

cycle. This al lo wed considerably simpler ana ly sis of the

effe cts of hardening. At hi gher thermal st resses the

prediction became rather co nservative, overestimating strains

by u pt o a factor of 1 0 .

The si m p l i f i e d kin em at i c hardening model from figure 5.6

gave v e r y co n s e r v at iv e results in c o mp a r i s o n w i t h experiments.

In addition, the predicted st rain accumulation in cr eased

linearly with the applied load. It is po s s i b l e that the

assumption of a st ea dy state cyclic history after each step

and the fact that the result re pre se nts the m a x i m u m strain due

to a local mechanism may have been respo ns ib le for this

overestimate.

In order to i n v e st iga te the second of these hypotheses,

the maximum strain accumulated within any e l e me nt was

extracted in each case from the A b a q u s solution. The strain

contour produced in this w a y was then plotted, al ong wi t h the

si m p l i f i e d s o l ut io n in figure 5.7. It can be seen from this

graph that the result p ro d u c e d by Abaqus tends to co nverge

towards the si m p li fi e d solution. This sug gests that as the

region of peak strain accumulation i n cre as ed relative to the

size of one finite element, the finite e l e me nt solution

approached its th eo ret ica l maximum.

121
It seems clear from these results that a l t h o u gh both

m e t ho ds gave co nse r va ti v e results, nei th er is able to provide

an ac c ur at e p r e d i c t i o n of the e x p e r i m e n t s over the wh o l e range

of loading.

c) Isotropic H a r d e n in g M o d e l s .

For low va lue s of thermal stress the finite e l e m e n t model

gave a good e st im a t i o n of the ratchet b ou nd (see figure 5.8).

However, in the more int er es t i ng regions above the reverse

p l a s t i c i t y bound, the model failed to give a reaso na bl e bound.

The si mp li f i e d isotropic hardening model showe d a very

good c o r r e la t io n with e xp er i me nt s at mid range thermal

st resses (see figure 5.8). Un fo rt una te ly, at highe r levels,

the model sta rted to give an underestimate of the strains.

This is un ac c e p t a b l e for design purposes. This i n t e re st in g

result, that higher thermal st resses pr od uce less strain can

be e x p l ai ne d in terms of the lo ad ing history.

A ft e r the first loa din g step at low m e c h a n ic al load the

ma te ria l is fully i so t r o p i c a l l y hardened to en co m p a s s the

thermal stress. The amount of pl as tic strain ac cu m u l a t e d to

this po in t is small, ir r e s p e c t i v e of the thermal stress. As

the mechanical load increases, the strain a c c u m u l a t i o n


f
incr eas es in p r o p o r t i o n to G^/ay, i.e. the ratio of the final

to the initial y ie l d surface. If that initial yi e l d surface is

larger, due to the larger thermal stress, then the pl as tic

strain as so ci a t e d w it h an incr em ent of m e c h a n ic al load will be

smaller.

This effect il lus tr ate s the impo rt anc e of load h i s t o r y on

the strain a c c u m u l a t e d by this model. In order to in ve stigate

122
the eff ect s of load h i s t o ry on the pl as tic strain

accu mul ati on, a sim pli fi ed isotropic hardening model was

pr ep a r e d for ex pe r im en t s I and K. These e x p e r i m e n ts involved

steps of inc rea si ng thermal shock at a con st an t me c h a n i c a l

load. The results (given in figure 5.9) show considerably

di f f e r e n t pla sti c strain contours than for the e x p e ri me nt s of

figure 5.8. There is a reasonable e s t i m a t i o n of the

ex pe ri m e n t a l strains. However, the pr ed i c t i o n s prove to be

sens iti ve to the choice of initial y i e l d stress un der this

load history.

Overall, the results of this s i m p l i fi ed isotropic

h a r d e n i n g model gave the best a p p r o x i m a t i o n of the a c cu mu la te d

strains in rel ation to the time taken to reach a solution. In

the next se ction a set of m o d i f i c a t i o n s are a d o pt ed in order

to try to improve this model.

5.6 Modifications to the Simplified Isot ro pi c Hardening

Model.

Many of the a ss um p ti on s inherent in the si mp li f ie d

hardening model can be either removed or tested wi th ou t

dramatic change to the model.

The first assumption to be co n s i d e r e d conc er ns the y ie ld

stress at w h i c h the pla sti c strain ac cu mulates. The model used

so far is for a yi e l d stress con st ant with t em per at ure

th ro ug ho ut a cycle. For a reverse p l a s t i c i t y mechanism, w it h

pl as tic st rai nin g at hot and cold ex t re me s of the cycle, this

is o b v i o u s l y untrue. In reality, the ma t e r i a l could yi e l d at

any te mp era tur e be tw een the cycle extremes. However, this

tem pe ra tu re wo u l d va ry from cycle to cycle, so the full effect

123
would be d i f fi c ul t to model. By ass um in g y i e l d i n g just at the

two ex treme t e mp era tu res it wo ul d be possibl e to s ig ni fi ca nt ly

improve the cu rrent model wi th little added complexity. The

derivation of this t em pe ra tu re d e pe nd an t e q ua ti on is as

follows :

As s u m e a thermal cycle b et w e e n two te mp er atu re extremes,

T^ and T^. In the final i s o tr o p i c a l l y hardened state, these

correspond to two Von Mises yi e l d surfaces and <^y c '

shown in figure 5 . 1 0 . The size of these yield surfaces is

related by the d i m en s i o n l e s s con st an t a such that

"y c ■ “ "Yh 5-’


The hoop stresses induced by this t e mp era tur e cycle vary

over the range such that

"t ■ V + 5.8
This gives the effe ct iv e stress at the yi e l d surfaces, in the

pre s e n c e of a con st an t me c h a n i ca l stress to be


P

^ec "p + + *+c'p - "Yc 5.9

and + % h + *+h"p “ ®Yh 5.10

SO, u s i n g eq u a t i o n 5.7

+ '+h*p) - 'p + "+C + % c ’p- 5.11


by substituting for equation 5.8, this can be solved as a

quadratic to give

3
(aS-l)5 ■ ?'p 5.12

Substituting this back into 5.9 gives the ef fe c t i v e yi el d

stress of the fully i s o t r o p i c a l l y h a rd en ed material.

124
3
*ec (ai-ip" ■ T^p 5.13

This eq u a t io n replaces g^ in e q u a t i on 5.4, lim it ing to

equation 5.4 as a a pp ro ach es unity.

Figure 5.11 shows the com pa ri so n be tw e e n this revised

model and the experiments. It can be seen that the effect is

to increase the ac cu m u l a t e d strain as a incr ea ses w i t h larger

te mp er at ur e di ffe rences. This improves the model

si gni ficantly. Un fo rtu na tel y, the strain a c c u m u l a t i o n is still

u n d e r e s t i m a t e d at higher stresses.

There are two other m o d i f i c a t i o n s to the model w h i c h could

be in tr o du c e d to improve the comparison. The current

assumption of com plete cyclic ha r d e n i n g to an el astic state

could be replaced by in complete hardening after a finite

number of cycles. This wo u l d lead to an in cr ea se d strain at

higher therma l stresses, but would require a con sid er ab le

inc rease in the c o m p l e x i t y of the model.

The peak strain c a l c u l a t i on res ulting from this me th o d

co uld also be modified. Currently, only the maximum thermo­

ela st ic stress has been u se d to cal cu lat e the stress at a

point. By ta king the stress at all points w i t h i n the mate ria l

which ex c e e d twice yield, an av erage strain co uld be

calculated. This would tend to de c r e a s e the strain estimate.

However, the problem with this method is that it wo u l d

necessitate a th ermo- elastic stress distribution and a large

num be r of strain calculations. This w o u l d require a computer

to solve for all but the simple st cases.

Consideration of the ef fects of these last m o d i f i c a t i o n s

leads to an alternative method of applying this si mp li f i e d

125
is otr opi c h a r de ni ng model. The model c u r r e nt ly assu mes that

the maximum pla stic strain occurs at the point of maximum

thermo- ela stic stress. This seems reasonable for the first

in cre me nt s in a fully sof te ne d material.

A f te r some pla stic strain has developed, the m a te ri al will

lo ca ll y harden. Further pla sti c strain due to a load increment

will have to d e v e lo p from an isotropically enlarged yield

surface, w h i c h will require p r o g r e s s i v e l y more plas tic work to

accrue the same strain. However, the ma t er ia l adj ac en t to this

point, a l t h o u g h under less stress, is a c c o r d i n g to this model,

still in a fully sof ten ed state. There will reach a point

w he re in work terms, a mechanism in this softer m a te ri al will

become more favourable.

Figure 5.12 shows a h y p o t h e t i c al m e c h a n i s m of this kind.

In 5.12a, a local pla st ic strai n incr em en t occurs due to the

stress distribution shown. This also results in isotropic

hardening at po int A. In 5.12b, an increase of this stress

results in further pla st ic strain at point s B, w h i c h have just

ex c e e d e d yield. These po ints B also undergo isotropic

hardening. Figure 5.12c shows this ef fect continuing, with

pl as ti c strains occurring as new, soft ma t e r i a l ex ce ed s yield

at the edges of this region of i n c r e a s i n g l y high stress.

An approximation to this ef fe ct could be made within the

framework of the Cocks and Ponter th eo ry by assuming that

maximum pl as ti c strain always occurs at a point w he re the

stress is equal to twice yield. This a s su mp t i o n should be

a pp li ca bl e as long as there is a gradient in the stress

profile. The m e c h a n i s m at this po int w o u l d be the same local

mechanism.

The model can be modified to take this into acc ou nt by

126
putting the thermal stress range, equal to 2*^. Eq uation

5.3 then reduces to:

5.14
- I

This h y po th es is is su pp or te d by the o b s e r v a t i o n in chapter

4 that there were no large scale local pl as tic strains in the

exp eriments. It is also su ppo rte d by the Abaqus analysis.

Figure 5.13 shows the pl as ti c strain accumulation al ong a

le ngt h of the se ct ion for mod el s of e x p e r im en ts C and F . In

ex pe ri m e n t C, the m a x i m u m t h e rm o- el as ti c stresses were just in

excess of 2*^. This res ulted in the maximum stress being

developed at the point in the tube corresponding to this

stress. In e x pe ri me nt F , the maximum stress was almost 4<jy.

However, the m a x i m u m strain was p r e di ct ed by Ab a q u s to occur

at a d i f fe re nt location, c o r r e s p o n d i n g to 2oy. C o m p a r i s o n with

e x pe ri me nt al results, as shown in figure 5.14, gave the best

c or r e l a t i o n of the m od el s tried. D i f f er en ce s were no more than

12 %.

A l t h o u g h this model is rather too hy po th e t i c a l to use as a

base for firm con clusions, these results suggest that more

work could be justified. It w o u l d be pos si bl e to prod uce a

w ork bou nd method, sim ilar to, or incl ude d in the upper bound

linear p r o g r a m m i n g a p p r o a c h of Ka rad eni z and Ponter.

127
5.7 In te r a c t io n be tw een m e c h a n i s m strain components.

a ) D e s c r i p t i o n of method.

The cal cul at i on s performed in this sec ti on have been

attempting to pr ed ict the strain a c c u m u l a t e d by e x c e e d i n g the

bo un d of reverse plasticity. However, at po in ts on the

in te r a c t i o n diagram c or re s p o n d i n g to both high thermal stress

and high me ch an ic al stress, the con dit io ns are sa tis fie d both

for rat ch et ti ng by a reverse plasticity mechanism and

mechanism I. This poses the question of w h i c h mechanism is

res pon sib le for a measured pla st ic strain. The choice of

mechanism will det er mi ne the plas ti c strai n accumulation.

Without full ine lastic analysis, it is im po ssi ble to state

w h e th er mechanism I, II, or some com po st it e mechanism will

yi el d the optimal solu tio n for a stress state in this region.

However, it is po ssi ble to give an estimation of the

strain accumulation u si ng previously calculated results for

mechanisms I and II. The mechanism equ at io ns of chapter 3

(table 3.1) show that these two m e c h a n i s m s are a c t i v a t ed by

different co m p o n e nt s of the t h e r m o - el as ti c stress. For most

thermal stress pr o f il e s these will occur at d i f f e r e n t points

in the material, or at d i f f e r e n t times in a cycle. This allows

the a s s u m p t i o n that, for small ex c u r si on s b e yo nd the shake dow n

boundary, the strains accumulated by these mechanisms are

in de p e n d a n t and cumulative. This assumption should give an

o v e r e s t i m a t i o n of the strain if the m e c h a n i s m s do occur in the

same location.

An estimation of the strain a c c u m u l a t i o n due to m e c h a n i s m

II has been made in se ct io n 5.6. A method of calculating the

128
plastic strain due to m e c h a n i s m I is de r i v ed in the following

section. A c o m b i n a t i on of these results will al l o w the

prediction of pla stic strain a c c u m u l a t i o n at any point on the

interaction diagram corresponding to the experiments of

chapter 4.

b ) Plas tic strain a c c u m u l a t i o n due to m e c h a n i s m I.

The accumulation of plastic strain ca us ed by this

mechanism is most c le ar ly d e m o n s t r a t e d us in g the lower bound

theorem.

Cons ide r an axial t h er mo - e la st ic stress v a r y i n g between


0 0
+G^ and -a^ thro ugh the wall of a thin shell over a thermal

load cycle. Figure 5.15a shows a sec tion th ro ug h a thin shell

wall, w it h the stress th ro ug h the thi ck ne ss shown at the two

ex t r e m e s of the cycle (solid and d a s h ed lines). A residual

stress field can be c o n s tr u ct ed such that w h en s u pe ri m p o s e d on

the the th e r m o- e l a s t i c stress of figure 5.15a, <y^ has a

m i n i m u m value, for an equilibrium stress field. This is the

o p t im um residual stress field for one cycle e x t re me and is

shown as a solid line on figure 5.15b. By a similar argument,

the other ex treme of the cycle results in the residual stress

field shown as a da s h e d line in figure 5.15b. On the ad dition

of a primary stress g^, it can be seen that p l as ti c strain

will occur when

Gp + g ®/4 > Gy 5.15

This gives the lower bo und sha ke do wn limit for this

m e c h a n i s m as

Gp + g®/4 < Gy 5.16

wh ic h is identical to the upper bo u n d result of chapter 3.

129
Returning to figure 5.15b it can be seen that once this

bo un d has been reached, the add it io n of a primary stress

in cr em en t will lead to a pro por ti on al incre men t of axial

pl as ti c strain. The ma gn it ud e of this in cre men t is equal to

AG
Ae 5.17
P
wh e r e is the pl as tic modulus, which is the slope of the

pl as ti c region of the stress- strain curve.

c ) Res ult s of C om bi ne d Strain A c c u m u l a t i o n Calculations.

V al u es of pl ast ic strain accumulated by in cr ements of

stress above the onset of m e c h a n i s m I were ca lc u l a t e d us ing

e qu a t i o n 5.17. The p r e di c t i o n of ratchet strains due to

mechanism II was also cal cul at ed using the method from

e qu a t i o n 5.13, as p r e s e n te d in figure 5.11.

T ak i ng in i ti a l ly the case of e x p e r i me nt C, the results

from these two eq ua t i o ns were pl o t t ed in figure 5.16 on top of

the ex p e r i m e n t a l load- strain data. For low strains, the

mechanism II equation gives a good c o rr e l a t i o n with

experiment, but at higher strains the predicted and

experimental lines start to diverge. However, if the effect of

m e c h a n i s m I is taken into ac count then a be tt er c o rr e l a t i o n is

achieved. This comp os it e line, w it h the strai n c on tr ib ut io ns

from mechanisms I and II summed, is also plotted on figure

5.16.

Figure 5.17 shows an in te r a c t i o n d i a g r a m w it h the results

of this analys is s u p e r i m p os ed on top of the e x p er im en ta l

results. Lines corresponding to the 0.5% strain contour

predicted by mechanisms I and II are also shown to give an

130
in di ca t i o n of relative strain contributions of the two

mec han ism s. It can be seen that for the e x p e ri me nt s w it h a low

te mp er at ur e gr adi en t (AT) mechanism I is the sole pr ed i c t e d

cause of ratchet strain ac cum ulation. At moderate value s of

AT, mechanisms I and II both c o n t ri but e significantly to the

strain. This was shown for e x pe r i me nt C in figure 5.16. In the

region of h i g he st AT, m e c h a n i s m I ag ain starts to dominate.

Comparing the co mb ine d mechanism model with the 0.5%

pl as ti c strain points from e x p e r i m e n t s C to H (of chapter 4)

shows a good correlation. The predictions overestimate the

pl as ti c strain at all points, which is co ns i s t e n t w it h the

proposed con ser va ti ve model. The strains are o v e r e s t i m a t e d by

up to 34%, alt h o ug h more typically, strains are w i t h i n 15% of

the ex pe ri me nt a l results. Considering both the degree of

error s in the e x pe ri me nt and the a p p r ox im at e nature of the

analysis, this is a good result.

The results of this anal ysi s are cu r r e n t l y no more than a

theory. It has been tested ag ai ns t one e x p e r i m e n t and given a

good result. In order to use this approach for de si gn

purposes, both further e x pe ri me nt and a more d e t a i l e d analysis

of the mechanisms would be required. However, if this theory

can be validated, then a simple me t h o d of predicting the

strain ac cu m u l a t e d outside the sh ak ed ow n boundary wo ul d be

available. A simple eq ua tio n for p l as ti c strain c or re sp on di ng

to each of the me ch a n i s m s de r i ve d in chapter 3 wo u l d extend

the sha ke do wn bounds to give limi ted p l a st ic strain

a c c u m u l a t i o n s over a wide range of problems.

131
5.8 C o n c l u s i o n s .

The finite el em en t mod els gave a reasonable a p pr ox i m a t i o n

to the strain accumulated, at the expense of a large amount of

co mp ut er time. The model s could be imp ro ved by u s i n g quad ra tic

e l em en ts and smaller time steps. However, this w o u l d require

c o n s i d e r a b l y more co mputer time than was available.

The sim pli fi ed mod els gave results at considerably less

e xp ens e in terms of computer time. However, the q u a l i t y of the

results va r i e d greatly b et w e e n methods. In addition, the

advantages of the sim pli fi e d method became less clear when

h i s t o r y d e pe n de nc e req uired each model to be m o d i f i e d to suit

the experiment. The e x t e n s i o n to simple d e s i g n code rules also

be co me s more complex. However, the final iso tr op ic result of

figure 5.17 does show a very good c o r r e l a t i on with the

experiments. This gives c o n f i d e n ce that us i n g a simple

hardening rule tied to each mechanism has p o te nt ia l to give

simple, accur ate p r e d i c t i on s of p l as ti c strain.

Be fore this me t h o d can be applied to d e s i g n problems, a

larger number of c om pa ri so ns with ex pe ri me nt al data must be

performed. Whether this is done will depend on the will and

resources to p e r f o r m d e s i g n o u t si de the s h ak ed ow n region.

132
Final Yield

S u r f a c e

Aa

Initial Yield

S u r f a c e

A p p l i e d S t r e s s

Cycie, Aa

F ig u re 5 .1 Von Mises Yield Surface showing a Typical Yield Surface


Translation due to Kinematic Hardening.

F inal Yield

S u r f a c e

Aa

Initial Yield

S u r f a c e
A p p l i e d S t r e s s

Cycle, A a

F ig u re 5 .2 Von Mises Yield Surface showing a Typical Yield Surface


Expansion due to Isotropic Hardening.

133
Axial
Stress

T a n 'E

Axial Strain, z

F ig u re 5 .3 Schematic of a Uniaxial Stress-Strain Curve, showing the


Definition of p.

134
i<B

^ = rolling restraint on a node

F ig u re 5 .4 Sketch of Finite Element Model.

Temp.

etc.

Time
Mechanical
Load

Step two etc.


Step one

Time

F ig u re 5 .5 Schematic of Typical Finite Element Load History for Experiments


Ct o H .

135
Temp.
Gradient Upper Bound Shakedown
AT(C) Prediction, 0.2% Yield.

□ Experimental Points,
180 0.2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

Abaqus Model, 0.2%


160 - 1
Accumulated Strain.

140 -

120 -
V.

100 -

80 -

60 -

40 —

20 -

n --------- 1---------1--------- 1--------- T"


20 • 40 60 80 100 120
Mechanical Load, P (KN)

Figure 5.6 Interaction Diagram showing Comparison between 0.2% Strain


Points from Experiments C to H and Kinematic Hardening Models.

136
Temp.
Gradient Upper Bound Shakedown
A T (°C ) Prediction, 0.2% Yield.

Experimental Points,
180 - 0.2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

Abaqus Model, 0.2%


160 -
Accumulated Peak Strain.

s. □
140 -

120 -

100 -

80 -

60 -

40 —

20 -

T" “I I I I
20 40 60 80 100 120
Mechanical Load, P (KN)

Figure 5.7 Interaction Diagram showing Comparison between 0.2% Strain


Points from Experiments C to H and Peak Strain Values from
Kinematic Hardening Models.

137
Temp.
Gradient Upper Bound Shakedown
A T (°C ) Prediction, 0.2% Yield.

Experimental Points,
180 0,2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

Abaqus Model, 0.2%


160 Accumulated Strain.

140

120

100

20 40 60 80 100 120
Mechanical Load, P (KN)

Figure 5 .8 Interaction Diagram showing Comparison between 0.2% Strain


Points from Experiments C to H and Isotropic Hardening Models.

138
Temp.
Gradient Upper Bound Shakedown
A TTC) Prediction, 0.2% Yield.

Experimental Points,
180 0.2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

160

140

V.
120

100

40

20 40 60 80 100 120
Mechanical Load, P (KN)

Figure 5 .9 Interaction Diagram showing Comparison between 0.2% Strain


Points from Experiments I and K and isotropic Hardening Models.

139
iiOi

Yc

F ig u re 5 .1 0 Typical Von Mises Yield Surfaces at Extremes of a Thermal Cvde.

140
Temp.
Gradient Upper Bound Shakedown
A T (°C ) Prediction, 0.2% Yield,

Experimental Points,
180
0.2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

160

140

120

100

60

60

20

20 40 60 80 100 120
Mechanical Load, P (KN)
Figure 5.11 Interaction Diagram showing Comparison between 0.2% Strain
Points from Experiments C to H and a Temperature Dependent
isotropic Hardening Model.

141
X

H ardened
m aterial.

Omax i i

H ardened
m aterial.

Gmax i i

Hardened
m aterial.

F ig u re 5 . 1 2 Sketches showing a Hypothetical Model of the Influence of Cyclic


Hardening on the Details of Plastic Strain Accumulation.

142
Local
Strain
(%)

1
Experiment F
under 60 KN Load
1
Specimen Half- Section

Experiment C
0 under 60 KN Load

1 2 3 4 5 6 7

Distance along thinned length, L (mm)


(see inset sketch).

F ig u re 5 . 1 3 Abaqus Predictions of Local Plastic Strain Accumulation for


Experiments C and F.

143
Temp.
Gradient Upper Bound Shakedown
A TCO Prediction, 0.2% Yield.

Experimental Points,
180 -
0.2% Accum. Strain.

Simplified Model, 0.2%


Accumulated Strain.

160 -

140 -

120 -

100 -

80 -

60 -

40 —

20 -

n” I I
20 40 60 80 100 120
Mechanical Load, P (KN)

Figure 5 .1 4 Interaction Diagram showing Comparison between 0.2% Strain


Points from Experiments C to H and a Modified Isotropic
Hardening Model, assuming Complete Local Hardening, as
Proposed in Section 5.6.

144
-a,

Tube
W all

J-o
0 8
-a 0 a

Tube
W all

J-a
0

F ig u re 5 . 1 5 Variation of Bending Stresses through a Wall showing Residual


Stress Distributions Corresponding to Mechanism I.

145
120

100 -

Mechanical
Load (KN)

Mechanism IIstrain
prediction, using equ.5.13.
Mechanism Istrain
prediction, using equ.5.17.
Combined strain prediction
for mechanisms I+ II.
Strain accumulation from
experiment C.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Accumulated Strain (%)

F ig u re 5 . 1 6 Prediction of Plastic Strain Accumulation for Experiment C,


Showing Components from Mechanisms I and II.

146
Temp.
Gradient Upper Bound Shakedown
Prediction, 0.2% Yield. AÀ
A T (°C )
Experimental Points,
180 - 0.5% Accum. Strain.

0.5% Strain Predicted


Due to Mechanism i
0.5% Strain Predicted
Due to Mechanism ii
160 -
0.5% Strain Predicted due
to Mechanisms i and ii

140 -

120 -

100 -

80 -

60 -

40 —

20 -

n I I I------- 1—
20 40 60 80 100 120

Mechanical Load, P (KN)

F ig u re 5 . 1 7 Interaction Diagram showing Comparison between 0.5% Strain


Points from Experiments C to H and Combined Predicted Strains
due to Ratchet Mechanisms I and II. as Proposed in Section 5.7.

147
Chapter 6.

S i m p l i f i e d M e t ho ds for the S o lut io n of Thermal Ra tc h e tt in g

Problems; A Summary.

6.1 Introduction.

Each chapter in this thesis addr es ses a different aspect

of the thermal r a t c he tt ing problem. Chap ter s 2 and 3 consider

methods of p r e d i c t i n g the s ha ke do wn boundary, chapter 5 looks

at the ef fects of e xc ee di n g this boundary and chapter 4

de s c r i b e s e xp er i me n t s p e r f o r me d to eva lu at e these theories.

However, the results of each section has i m p l ic at ion s on the

others. It is the pu rpose of this chapter to outline the wider

implications of the work. In particular, the aim of p ro du ci ng

a simple set of de si g n rules.

6 .2 A S i m p l i fi ed Method.

The development of a si mp li f i e d method for solving the

p r o b l e m of thermal r at ch et ti ng in tubes was successful in the

sense that a set of eq ua t i o n s were given in chapter 3, table

3.1 w h i c h u se d p u r e l y ela st ic p r o p e r t i e s of the thermal stress

distribution to calc ul at e a bo und to sh ak e d o w n behaviour. In

order to ob ta in so lu tions to it is si mp ly necessary to

calcu la te the ext re me s of the thermo-elastic stresses

co r r e s p o n d i n g to the a p p li ed t em pe rat ure profile, then take

the re quired c o m p on en ts of these st resses and apply them to

the eq ua t i o n s of table 3.1. This has b ee n shown to require

v e r y simple ca l c u l a t i o n s for m e c h a n is ms I and II wh ic h yie ld a

148
linear re l at io ns hip b et we e n thermal stress and maximum

allowable load. However, for m e ch an i s m s III and IV, the

s ol u t i o n of the equat ion s is somewhat more complex. The

optimum length of the h i ng e -c on e mechanism changes w it h the

ratio of thermal to me ch an i ca l loads. This required that the

pr oc e s s of mechanism optimisation needed to be repeated a

num be r of times in order to b ui ld up an i n te ra cti on diagram

containing one of these mechanisms. This led to the

development of the computer program 'Mecalc', li sted in

a p p e n d i x B in order to p e r f o r m the optimisation.

Thus the m e th od is not no w as simple as ori gi na ll y

intended. However, it is still c o n s i d e r a b l y faster than either

F.E. or linear pr og r a m m i n g anal ys is w it h E E C _ S h a k e d o w n . More

importantly, it allows insight into me th od s of improving

de s i g n s by showing the c on tr ib ut io ns made by stress and

g eo me t r i c a l co mpo nen ts to the onset of ratchetting. As such it

will ho p e f u l l y serve as a tool in the de ve l o p m e n t of designs

which are more ta il ore d to the prevention of thermal

r a t c he tt in g problems.

6 .3 Co mp ar iso ns W i t h De sig n Codes.

A con sid er ab le part of chapter 3 was co nce rne d with

comparing the p r op o se d me c h a n i s m s w it h e x i s t i n g d e s i g n codes.

However, to gain proper p e r s pe c t iv e on these c o mp ari son s it is

i m p o rt an t to con sider the a p pl ic at io ns of the des ig n code.

It was shown that there we re c o n si de ra bl e di ff er en ce s

between the m e c h a n i s m m e t h o d results and for example, RCC-MR.

It was exp la in ed in chapter 3 that RCC-MR is bas ed on

ex pe ri m e n t a l methods, and the bo und generally allowed for

149
considerably higher stresses than either the m e c h a n i s m me th od

or other e l a st ic stress based des ign codes. Howe ver it must be

remembered that RC C - M R defines the onset of ra tc het tin g in

terms of 'allowable pl astic strain' w h e re as the mec ha ni sm s

method defines the limit of sha ke do wn w he re the history of

stress lies within yield and the limit of pl as tic strain

accumulation is small. If the amount of pla st ic strain

accumulated in excess of the sha ke do wn limit is small then

experimentally measured onset of rat che tti ng behaviour, as

characterised by plas tic strain accumulation ma y vary

c o n s i d e r a b l y be tw ee n the two approaches.

The e x p e ri me nt s of chapter 4 gave the opportunity to

i nv e s t i g a t e this effect. Figure 6.1 shows the pla stic strain

predictions from RC C- M R and the m e c h a n is ms method, along with

the e x p e r i m e n t a l results. E x a m i n a t i o n of the co mp a r is on s wo ul d

suggest that the empir ica l m et ho d of RC C - M R gives a better

prediction of e x p e ri m en ts C to H than the ov e r l y con ser vat iv e

mechanism result.

However, w he n the results of ex pe r i me nt s I and K are

con sidered, in whi ch a di ff er en t load h i s t o r y was used, but

with the same loads, then a di ff er en t rate of strain

accumulation results. The reason for this is that the strain

accumulation in the ra tch et tin g region is gov er ne d by work

hardening eff ects in the region under co nsideration. It was a

conclusion of chapter 5 that p re di c t i n g the ef fects of work

hardening is not straightforward and can vary co n s i d e r a b ly

w it h de ta il s of the lo ading cycle.

Now, RCC-MR, or any empirically base code, can make no

d i s t i n c t i o n b e t we en regions of el astic or sh ak e d ow n behaviour,

(such as the po int ma r k e d 'A' on figure 6.1) and regions of

150
work h a r de ni ng c o nt r ol le d r atc het ti ng (such as the point on

figure 6.1 ma rk e d 'B'). There must be some co nc ern that

predicting a safe design zone close to point 'B' will leave

open the possibility of the structure encountering a load

cycle which allows little or no work hardening. This is

tho ug ht to be especially likel y in the case of mo vi ng

t em pe ra tu re fronts. The li mited range of m o v i n g t em pe ra tu re

front experiments, such as those performed by Bell [34]

(repo rte d by Karad eni z [3]) have shown that sig nif ica nt

ratchet strain can occur in regions just in excess of the

predicted sh ake dow n bound. It is sig ni fi ca nt that the RCC -MR

code does not include any exa mples of m ov in g temper atu re

fronts.

The r e c om me nd at io n from this must be that des ig n codes

sh oul d have at least some indic ati on of the dif fe re nc es

between ela stic and plas tic beh av io ur if they are to be

ac ce p t a b l e for general application.

6 .4 Exp er im en ts on Thin W a l l e d Tubes.

The ex pe ri me nt s of chapter 4 y i e ld ed some nota ble results,

b o t h in terms of ex pe ri me nt al m e t h o d o l o g y and data.

The novel ele ctr ic al resistance he at in g me t h o d used

eventually gave reliable results. Howe ver the d e t a i l e d design

of the rig was such that it pro ve d d i f fi cu lt to obtain

circumferentially symm etr ic te mpe rat ure profiles. There are

inheren t pr ob lem s in us i n g high current, low vo lt ag e power

sup plies due to the prob lem s a s so ci at ed w it h contact

resistance. Future e x p e ri me nt s p l an n i n g on u s i n g this h e a ti ng

m e t h o d will need to cons ide r ca re fu ll y ways of ov er c o m i n g this

151
problem.

The success of the e xp er ime nts lay la rg ely in the careful

prediction and de si g n of tempe ra tur e p r ofi le s and the

res ul ti ng thermal stresses. This al lo wed a small number of

ex pe r i m e n t s to cover a range of regions of beh av io ur us ing a

single spec ime n design.

The exp eri me nt al results co nf ir me d the pr es enc e of

s h a ke do wn bounds c or r e s p o n d i n g to both mechanisms I and II,

but also of another region of ver y low pla st ic strain,

attributed to the eff ects of work hardening. This extra region

is not in con flict wi th the p re di ct io ns of shake dow n theory.

The up per bound sh ake dow n th eo rem does not n e c e s s a r i l y predict

the onset of pla sti c strain accumulation, it only pr e d i c t s the

limit to wh i c h el astic residual stresses can control yielding.

6 .5 The Pr ed i c t i o n of R a t ch et Strains.

In chapter 5, a number of m e t ho ds were app li ed to predict

the ratchet strains m ea s u r e d in the experiments from chapter

4. The final prediction from figure 5.16 shows good

c o r r e l a t i o n wi th ex p e ri me n ta l strains. However, the m e t h o d of

producing this prediction was by no me ans as general in

application as the m e ch an i s ms me th o d for p r ed ic t i n g the

s h a k e d o w n boundary. It was shown to be n e c e s s a r y to tailor the

method of pr e d i c t i o n to suit the two d i f f e r e n t load histo rie s

in order to ach ieve s a t i s f a c t o r y results.

This history dependence of ratchet strains suggests that

it will prove d if f i c u l t to pr oduce a us ef ul general desig n

code for the l im it a t i o n of ratchet strains. In m a n y desig n

situ at io n s the exact load hi s t o r y seen by a comp one nt cannot

152
accurately be predicted. The de s i g n cri te ria con sist of an

'envelope' of tem pe r at ur e s and stresses. Although this wo uld

be suf fic ie nt to perform a sha k e do wn analysis, it w o u l d be

d if f i c u l t to pr ed ict the ratchet strains w i t h o u t some accurat e

load histories.

Desp ite this problem, some general p r in ci pl es have been

established or re-confirmed. The FE an aly sis p r o d uc ed some

reaso nab le pr ed i ct io n s of the exp er im en ta l results, even

though the model was rather crude. This increases con fidence

in the p l a s t i c i t y mo de ls used in Abaqus. It could be exp ec te d

that better results will be a ch ie ve d as com puter m o d e l li ng of

p l a s t i c i t y improves.

It was also shown that a good pl as tic strain p r e d i c ti on

can be obt ai ne d us in g a tail ore d but simple model. This model

wo r k e d on the h y po th es is that the pl as tic strain pre di ct io ns

from any active ratchet m e ch an is ms are in de pen den t and

cumulative. There is the poten tia l for predicting general

ratchet strains by us in g an equ at io n for ratchet strain

a ss oc i a t e d w it h each of the mechanism eq ua ti on s from table

3.1. If this w ere to be c o mb ine d w it h a pr e d i c t i o n of the

creep strains us in g the m et ho d p r o p os ed by Ponter and Cocks

[1 ], then an overall simple m et ho d for a n a ly si ng structures

su bje cte d to severe thermal stresses w o u l d be available.

153
6-6 Proposals for Future Work.

It was no t e d in the review of ra tc he tt in g exper im ent s

presented in se ction 4.2 that there is con sid era bl e

ex p e r i m e n t a l verification for the a pp ea ra nc e of mechanism I

and some evidence for mechanism II (including this work).

However, mechanisms III and IV, as p r o p o s e d in chapter 3, have

little or no confirma tio n. It would be useful in the

d e v e l o p m e n t of this m e t h o d to have some ex pe ri me nt s wi th which

to ei ther c o n f i r m or d e n y their presence. More importantly, it

has b ee n st at ed both in this thesis and elsewhere, that (from

the results of Bell [34]) ratchet strains due to this

me ch anism, e s p e c i a l l y w h e n caused by m o v i ng te mpe rat ure fronts

may not be controlled by work hardening ef fects and so could

be m uc h greater. This w o u l d have serious im pl ic at io ns on the

re l i a b i l i t y of the R C C - M R de si g n code, wh i c h assu mes that work

h a r d e n i n g occurs, lim it in g strain ac cu mulation.

An experimental programme c o n s i s t i n g of c a r e f u l ly designe d

moving temperature shocks would resolve this problem. A

derivative of the heating and quenching method used by Bell

[34] would be su itable for such a programme, as s u m i n g the

temperature gradients co uld be p r o p e r l y controlled.

Another field for further development is in the ex tension

of the mechanism equations of chapter 3. The eq ua t i o n s given

in table 3.1 are derived for tube geometries. However, the

thermal ratchetting cases of interest to LM FB R design

frequently in volve other geometries, such as to ri spherical

ends and be nds in pipework. In order to solve these kinds of

problems, a more ge neral form of the m e c h a n i s m equat ion s would

be required, ta ki ng into acc oun t variations in radius along

154
the length of the shell.

Finally, if it proves important to d e t e r m i n e the limits on

ratchet strain, perhaps in order to assess the exp ec te d life

of a th er m a l l y shocked component, then the m e th od s p r o p o s e d in

chapter 5 should be extended. As a first step, the me t h o d for

c al cu la ti ng the ratchet strain due to each m e c h a n i s m should be

a pp li ed to other ex pe rim en tal results. If this prove s to be

g en e r a l l y applicable, then the e q ui va le nt equ at io ns for the

other ratchet m e ch an is ms can be derived.

155
Temp.
Gradieni Mechanisms Method
AT(°C) Prediction, 0.2% Yield.

Experimental Points,
180
0.2% Accum. Strain.

RCC-MR Code Design


Limit.

160

140

120

100

80

60

40

20

20 40 60 80 100 120
Mechanical Load, P (KN)

F ig u re 6 .1 Interaction Diagram comparing Predicted Safe Working Design


Regions with Experimental Results,

156
A p p e n d i x A.

D e r i v a t i o n of M e c h a n i s m Equations.

Mechanism I .

In m e c h a n i s m I we are c o n s i d e r i ng ra tc het tin g on the

yield surface (see fig. 3.1) caused by a v a r i a t i o n in the


0
thermo-elastic stress o\j. From the ar g um en t of se ction 3.3,

the p l a st ic strain ve cto r resul tin g is:

A.l
(ASx'O
So l u t i o n of the Up per B oun d S h ak ed ow n eq ua t i o n (Equ.3.2)

req uires a com pat ib le t h e r m o -e l a st ic stress, which from

s ec tio n 3.3:

A.2

This is r ep re se nt ed in fig. 3.3.


0
As the axial stresses are pu r e l y bending.

A.3

A. 4
»x(t2)max
1 2
we can call + <j^ àa. A.5

A p p l y i n g this to a tube of the g e o m e t r y in fig.Al

un der axial load F, wi th axial strains ta king place over the

short length a. The Upper Bo und Sh ak ed ow n E qu at io n can now be

r e w r it te n as:

PAUY dS * F2 nR hAU A .6
(*x - *x(t2))A:x dV

This integ rat es to:

a ^( t 2 )Ae^ dz dx A.7
Oj

r h crj h
= 2jiRaAe^ <ayh - A .8
I 2 2 2 2j

157
S ub s t i t u t i n g in Ae^ = AU/a, and u s i n g = 2nRh(jy as the

lim it load for axial te nsi on gives:

or

F ^^X

However, this is only v a li d wh er e A<x^ < 2oy. Ab ov e this

line, some p l a s t i c i t y must occur du r i n g a cycle. However,

this does not n e c e s s a r i l y result in ratchetting. The Ex tended

Up p e r Bo un d t h e o r u m of Kara de niz and Ponter al lows a sh ak edown

c a l c u l a t i o n to be made in some cases. The e q u a t i o n is w r it te n

as :

* P . AU? dS .mean^g d v A . 11
V^x X F
S ^ ^

The d e r i v a t i o n of this eq u a t i o n is gi ven in s e ct io n 2.6.

W h e r e Vp is the part of the volume, V wh er e stresses exceed

2cTy at some po int d u r i n g the cycle. A s c hem at ic of this

distribution is shown in Fig. A . 2.

Where, by symmetry, g =
-1---^ A . 12
4 + "t

U po n su bstitution, the terms at the two ex tremes

of the ma t e r i a l cancel, le av ing the re duced vol um e Vg to carry

the loads.

So for axial load:

F AU » aA8^2nRh^ ffyp -
( ( * t^ + *St ) 1 A . 13

158
which, by su b s t i t u t i n g for 3 gives;

1 for axial load A . 14

or by s u b s t i t u t i n g in A . 5

1 P i.IS
'Y

M e c h a n i s m II.

This m e c h a n i s m occurs w he n a thermo- ela sti c stress

history is d o m i n a t e d by a large v a r i a t i o n in ho o p stress. It

co ns is ts of two co mp on en ts of pl as tic strain on the and Xg

surfaces, as shown in fig. A . 3.

wh er e = (0,Ae^) A . 16

ASg = (Ae^,-Ac^) A . 17

Assuming Ac^ - Ae^ and summing

Ac = Ae^ + ACg »(Ac^,0) A . 18

This results in a loc al is ed m e c h a n i s m of the same form as

in fig. A.I.

S u b s t i t u t i n g the val ues into the Up per Bo und

E q u a t i o n gives:

F AU - of .(t. ))AeJ.dV + (*i]3 )Ac?jdV


•‘• J l ^ J ■*'J
V ‘ -'I
A . 19
Inte gra te equ. A . 19 around the c ir cu mf er en ce and along the

m e c h a n i s m length a to to give:

rh
FAU = 2nRaAc^< 2(jyh - &®(t3)dz|
A . 20
the G ^ ( t 2 ) term, being a pure be nd i n g stress, in te grates

159
to zero thr ough the thickness. The hoop stress term

co rr e s p o n d s to the mea n thr oug h the thic kn ess of the m a x i m u m

h o o p stress, (û&^)„ean

h Q Q
where, (A&^)„ean= K ^(ffj(t^) - aJCtg)) dz. A . 21

This allows e q ua t io n A . 21 to reduce to;

5 . , .

Mechanism III(A).

Me ch an i s m s I and II have res ul te d in p u r e l y axial

strains, ass um in g that all strain takes place at the po int of

m a x i m u m stress. To derive me ch a n i s m s w i t h a radial strain

co mp on en t requires co n s i d e r a t i o n of st resses over a finite

length of tube.

From the result of equ. 3.13, for a m e c h a n i s m on ,

the fol lo wi ng pl ast ic strains result:

AG..C = (0,Ae. ) A . 23
IJ 9
as shown in fig. A . 4

T a ki ng a 3 hi ng e- c o n e mechanism, as shown in fig. A . 5, of

he ig ht 1/a such that the area of the e nc lo se d cone is unity.

Fr om the Upper Bo und Sha ke d ow n e q u a t i o n (3.2), and taking

an int ernal pre ss ur e load ing (axial loads do no work on this

mechanism), we find:

'2a ' A

.0

wh er e
w dx =
Vh
(My - M y ) e dv,, +

is the volume
J
of the hinges,
(*Y - »% (ti))e+ d V *
m
and is the volume
A . 24

of the cone formed by the mechanism.

160
This gives two c o m p o n e n t s to the solution. A hinge

te rm from plas tic d e f o r m a t i o n at hinge points and a me m b r a n e

te rm due to the change in volume.

Ta ki ng the hinge term first.

2nR i(M y - + (My - M^ieg + (My - M+IBg A . 25

By sym me tr y from fig. A . 5, for small angles:

A . 26
1 1 . 1
0. 0.
®2 + ®3 8 + 0 A . 27
A . 28

Su bst it ut e these val ues into A . 25 to give:

8 JiRM,
1 . 1 A . 29
aJ

Now, My = <Tyh /4, so A . 29 can be re wri tte n as:

2 iiRh 2 (T, M,
1 ^ 1 A . 30
/2 -
1 ^

wh er e 1 . 1 2 for a sy mme tri c mechanism.

For the m e mb ra ne term, the d e f o r m a t i o n of the tube along

the m e c h a n i s m must be found. Fr om f i g . A . 5:

w * .— for 0 4 x 4 Sa A . 31

" - (2 a - &a < X < 2a A . 32

This gives the hoop strains » w/R

so the m e mb ra ne term in A . 24 can be w r i t t e n as

rh P2a
2 HR ( (jy - dx dz A . 33
OJ
or
rh P2a
2nl dyh - a® (ti )w dx dz.j A . 34
OJ 0

161
fh
D e fi ni ng A. 35

reduces this to
r%2a A
O’
= 2 nh Oy < 1 — dx A . 36
0
If the so lut ion is be ing found on a computer, as w i t h the

e xa m p l e s provided, then it is rea sonable to solve

P2a
A . 37
0 *+mean*

nu me rically, using for example, Simp so n's rule. However,

to a l l o w hand ca lc u l a t i o n an a p p r o x i m a t i o n can be made us ing

the Sch wartz I n eq u al it y [53]

(Jfg dx)2 4 ( J f 2 dx)(Jg2 dx) A . 38

and p u t ti ng f = o^"^®^^(x), g - w(x) gives

P2a 2 2a
1
dx or A . 39
0 *+ me an" ^ /I Q *mean

wh e r e o^ is the root m ea n square of *^mean"

Thus the m e mb ra n e work term becomes;

2nhOyjl A . 40

C o m b i n i n g the hi nge term, A . 30 w i t h this m e m b r a n e term


and u s i n g

P^U? dS » 2nRP for internal p r ess ur e

gives the total bound:

Rh ll 1 ^ 1
A. 41

Where is d e f i n e d from either Equ. A . 37 or A . 39

In practise, the m e c h a n i s m s for the pr o bl em s so far

162
e n c o u n t e r e d are always v e r y ne a rl y symmetric. By a s s u mi ng a

sy mm et ri c mechanism, a c o n si de ra ble saving in c a lc u l a t i o n can

be a c h i e v e d by cha ng in g A . 41 to:

V S ‘ 4
M e c h a n i s m I I I (B ).

Although similar in form to the p re vi ou s case,

m e c h a n i s m III(B) has a pl astic strain vector:

= (-Ae^,Ae^) A . 43

This results in a sh o r t en in g of the tube com bi ne d wi th

expansion, thus the func tio n v(x) must also be found.

By p u t t i n g -e^ » and » S v / 6x

and u s i n g the results A . 31 and A . 32 we find that

v(x) = - ^ A . 44

In addition, the left hand of A . 24 is modified,

because the axial strains do wo rk a g a in st the internal

pressure, i.e.

m e m b r a n e work » 2pnR A . 44

end cap work = - p n R A . 45

net work = PiiR A . 46

The resu lt in g bo und for this m e c h a n i s m is in the

same form as A . 42, but the eq ua ti on A . 35 is cha nge d to:

. rh
O’ dz A . 47
*mean H

163
M e c h a n i s m IV.

The d e r i v a t i o n for m e c h a n i s m IV(A) and (B) is the same as

for the e q u i v a l e n t m e c h a n i s m III. However, the d i r e c t i o n of

hi nge m o m e n t s and m em br ane stresses are reversed.

164
......- —

Ae

F ig u re A1 Schematic of a Thin W alled Tube showing the Geometry of a


Local Ratchet Mechanism.

“CTl

-cr.
o’^mean

F ig u re A 2 Variations of Thermo-Elastic Stress through a Tube Wall, showing


Stress Components for Extended Upper Bound Shakedown
Analysis.

165
(T, A e,

CTx

F ig u re A 3 A Tresca Yield Surface, showing a Typical Stress Cycle and


Strain Increments Resulting in a Reverse Plasticity Mechanism.

166
^cr.

F ig u re A 4 A Tresca Yield Surface showing a Typical Stress C vd e and Strain


Increments resulting in Mechanism Ilia.

F ig u re A 5 Schematic of a Three Hinae-Cone Mechanism.

167
Appendix B .

Li st i n g of 'Mecalc' Comp ut er Program.

c P r o g r a m MEC AL C
c R.J.M.V ene ss , 3— 4— 88
c Takes a C O N I D A format stress file, and ca lc ul a t e s the
c c o r r e s p o n d i n g ratchet mechanisms,
c m o d i f i c a t i o n 18-4-88
c re -wr itt en bend routine to a c co un t for the sign of the
c b e n d i n g stresses
c m o d i f i c a t i o n 19-4-88
c max st r c o m p l e t e l y rewritten, h o p e f u l l y correctly,
c m o d i f i c a t i o n 21-4-88
c sig me m chan ge d from h p m e m + / - h p b e n to hp m e m
c m o d i f i c a t i o n s 27-4-88
c - no lon ger looks at 1 ele me nt me ch a n i s m s
c - hinge angles a c c o u n t e d for in o p t i m i s a t i o n
c ** ME C AL C3 **
c sta rte d 4-5-88
c o p t i m re -w rit te n to integrate over the shape of the
c mechanism, rather than the rms value,
c m o d i f i e d 23-4-88
c - integ rat es w *s t r e s s over m e c h a n i s m
c - uses me an hoop stress rather than max
c ** ME CA LC4 **
c sta rte d 23-6-88
c change input file to spe ci fy loa di ng type
c a l l o w for in clu si o n of mo v i n g te mp er at ur e fronts
c LO AD TY P E- 0 axial
c 1 i n t e r n a l / e xt e r n al pr ess ure
c LO AD SI G N- 1 t e n s i l e / i n t e r na l pr ess ure
c 0 c o m p r e s s i v e / e x t e r n a l p r es su re
c co rr ec t ed 7-10-88 s i g t o t ,si gm em mod. for m o v i n g fronts
c M o d i f i e d 21-10-88
c -O nl y looks at hinge cone me c h a n i s m s eg. or shorter than
c the pre vi ou s iteration,
c ME CA L C 5 27-10-88
c -fi xes p o s i t i o n of central hinge in the mi dd le of the
c mechanism
c M E C A L C 6 9-2-89
c combi nes getstr & getln to ensure stress va lu e s are
c co ns i s t e n t w i t h those in ma xstr
CO M M O N / d a t a / s t r e s s ( 1 0 0 0 , 5 ) , a xb en a( 1 0 0 0 ) , a lo ng x( 1000)
DIMENSION bree(2,10),stosy(0:8),popl(0;8)
1 ,e u b ( 2 ,1 0 ),r p a l ( 2 ,1 0 ),r pip (2 ,1 0 ), hcm(2 ,1 0 )

CALL g e t s t r (r a d i u s ,n o e l ,t h i c k ,l o a d t p , l o a d s n , d i s m o v )
W R I T E ( 7 , 5 ) r a d i u s ,n o e l ,t h i c k ,l o a d t p , l o a d s n , d i s m o v
5 F O R M A T (E l 4 . 7 , I 4,E l 4 . 7 , 2 1 3 , F 7 . 4)

IF ( d i s m o v . G T . 0.0) THEN
CALL m o v e r ( n o e l, di s mo v)
END IF

CALL m a x s t r (n o e l ,sig tot ,s ig me m, si gb en )

DO 102 l o o p = l ,(noel+ 1 ) ,1
WRITE(7,6)loop,alongx(loop),
1 (s t r e s s (l o o p , n o ), n o = l ,4,1 )

168
6 F0R MA T( I4,5E14.7)
102 CONT INU E

W R I T E (7 ,7)s ig me m ,s i g b e n , s i g t o t
7 F O R M A T ( 3 E 1 4 .7)

c set up graphics

CALL PAPER(l)
CALL M A P (0.0,1.1,0.0,5.0)
CALL AXES

c calc, hinge cone mech for the c h o s e n lo ad type and sign

c axial load me c h an i sm s

IF (loadtp.EQ.O) THEN
W R I T E ( 7 , * ) ' A x i a l load'
IF (loadsn.EQ.l) THEN
W R I T E ( 7 , * ) 'Lambda 3'
it y p e=2
jtype=3
ELSE
W R I T E ( 7 , * ) 'Lambda 6 '
itype=l
jtype-4
END IF
END IF

c i n t e r na l/ ex te rn a l pre ssu re m e c h a n i s m s (2 for each case)

l o o p -0
IF (loadtp.EQ.l) THEN
l oo p-2
W R I T E (7, *) ' In te rn a l/ Ex t er na l pressure'
IF (loadsn.EQ.l) THEN
W R I T E ( 7 , * ) 'Lambda 1'
it yp e-1
jt yp e-0
ELSE
W R I T E ( 7 , * ) 'Lambda 4'
itype-3
j t y p e-0
END IF
END IF

c ** in iti al ise s length of m e c h a n i s m search


8 in o d e -2
l e n g t h - n o e l -2

DO 170 m e c h l p -0 ,8 , 1
s t o s y ( m e c h l p ) - m e c h l p / 2 .0
CALL o p t i m ( n o e l ,i t y p e ,j t y p e ,s i g t o t ,s t o s y ( m e c h l p ) ,
1 r a d i u s ,t h i c k ,o p t s t ,b e n s t r ,l e n g t h , i n o d e , m n o d e ,
1 popl(mechlp))
WRITE( 7 , 11 ) p o p l ( m e c h l p ) ,s t o s y ( m e c h l p ) ,length,
1 i n o d e ,m n o d e ,o p t s t ,b e n s t r ,sigtot
11 F0R MA T( 2 F9 .4 , 3I 4, 3 E 14 .7 )
170 CO NT INU E

169
plot a smooth curve th rou gh the points

CALL C U R V E O ( p o p l ,s t o s y ,1,8 )

second in te r na l/ e xt er n al pres su re m e c h a n i s m

IF (loop.EQ.2) THEN
l o o p =0
IF (loadsn.EQ.l) THEN
W R I T E ( 7 , * ) 'Lambda 6 '
itype=l
jtype-4
GOT O 8
ELSE
W R I T E ( 7 , * ) 'Lambda 3'
it y p e -2
jtype-3
GOTO 8
END IF
END IF

calcu la te thi nn in g me c h an i s ms

CALL t m e c h ( s i g t o t , s i g m e m , s i g b e n , b r e e , e u b , r p a l , r p i p )

IF (loadtp.EQ.O) THEN
W R I T E (7 ,10)(b r e e (1,n o ),bree(2 , n o ), n o - l ,10,1)
10 F O R M A T ( 2 F 7 .4)
CALL P O S I T N ( b r e e ( l , l ) , b r e e ( 2 , l ) )
DO 174 J-2,10,1
CALL J O I N ( b r e e ( l , J ) , b r e e ( 2 , J ) )
174 C O NT IN UE

W R I T E ( 7 , 1 0 ) ( e u b ( l , n o ) , e u b ( 2 , n o ) ,n o - 1 , 10,1)
CALL P O S I T N ( e u b ( l , l ) , e u b ( 2 , l ) )
DO 175 J-2,10,1
CALL J O I N ( e u b ( l , J ) , e u b ( 2 , J ) )
175 CO N T I N U E

W R I T E ( 7 , 1 0 ) (r p a l (1,n o ) ,r p a l (2,n o ),n o - 1 ,10,1)


CALL P O S I T N ( r p a l ( l , l ) , r p a l ( 2 , l ) )
DO 176 J-2,10,1
CALL J 0 I N ( r p a l ( l , J ) , r p a l ( 2 , J ) )
176 CO N T I N U E

ELSE

W R I T E ( 7 , 1 0 ) ( r p i p ( 1 , n o ),rp i p ( 2 , n o ), n o - l ,10,1)
CALL P O S I T N ( r p i p ( l , l ) , r p i p ( 2 , l ) )
DO 177 J-2,10,1
CALL JOIN( r p i p d , J) , rpip(2, J) )
177 C O N TI NU E
END IF

ter mi na te the gr aphics

CALL FRAME
CALL GR END

STOP

170
END

C
***********************************************************

SU B R O U T I N E g e t s t r ( r a d i u s , n o e l ,thick,
lloadtp,loadsn,dismov)

c Uses the subro uti ne getln to co llect a w ho le


c file of stress data in C O N I DA format,
c also returns x c o o rd in ate s of the nodes

COM MO N / d a t a / s t r e s s (1 00 0 , 5 ) , a x b e n a (10 00 ), a l o n g x (1000)

OPEN(1,status»'unknown')
OPEN(7,status='unknown')

c read header lines

R E A D ( 1 , * ) n o e l ,s i g m a y ,thick

READ(1,*)loadtp,loadsn,dismov

c read stress data lines

DO 100 J = l , ( n o e l + 1 ) ,1

R E A D ( 1 ,2 )n o u s e , n o u s e , r a d i u s ,a l o n g x (J )
1 ,a x m e m , h p m e m , a x b e n ,hp be n, sh ea r
2 F0R MAT (I 4 , I5 ,7 E 14 .7 )
IF ( ( J . L E . n o e l ) . A N D . ( J . G T . l ) ) THEN
R E A D (1,2)n o u s e ,nouse
1 ,s c r a p , s c r a p , a x m , h p m , a x b , h p b , s h e a r
a x m e m » (a x m e m + a x m ) / 2.0
a x b e n » (a x b e n + a x b ) / 2 .0
h p b e n » (h p b e n + h p b ) / 2.0
h p m e m » (h p m e m + h p m ) / 2.0
END IF

c calc, po sit ive and neg ati ve va lu es of hoop and axial


c stresses
c s t r e s s ( *, 1 )=positive hoop
c s t r e s s ( * , 2 )»positive axial
c s t r e s s ( * , 3 ) = n e g a t i v e hoop
c s t r e s s ( * , 4 )=necative axial
c st re s s ( * , 5 ) = s h e a r

s t r e s s ( J , 5)“ Shear

s t r e s s ( J , 1 )=0.0
s t r e s s (J , 3 )=0 .0

IF ( A B S (h p m e m ) . L T . A B S ( h p b e n )) THEN
s t r e s s ( J , l ) » ( h p m e m + A B S ( h p b e n ) ) / 2 .0
s t r e s s (J ,3)=(h p m e m - A B S (h p b e n ) ) / 2 .0

ELSE IF (h p m e m . G T . 0.0) THEN

171
stress(J,l)=hpmem
ELSE
s t r e s s ( J , 3 )=hpmem
END IF

c Sets m a x i m u m p o si ti ve axial stress


c Sets m a x i m u m n e ga ti ve axial stress

s t r e s s ( J , 2 )=0.0
s t r e s s ( J , 4 ) = 0 .0

vplus=axmem+axben
vminus=axmem-axben

IF (vplus.GT.0.0) THEN
s t r e s s (J ,2 )=v plus
ELSE IF (vplus.LT.0.0) THEN
s t r e s s ( J ,4 )= vp l us
ENDIF
IF (v m i n u s .G T .s t r e s s (J ,2)) THEN
st r e s s ( J , 2 ) = v m i n u s
ENDIF
IF (v m i n u s .L T . s t r e s s (J ,4)) THEN
st re s s ( J , 4 ) = v m i n u s
ENDIF

100 CON TI NU E

RETURN
END

C
**********************************************************

SU BR O U T I N E m a x s t r (no el ,s ig t, s i g h p , s i g b e n )

c reads a C O N I D A disc file


c ca lc ul at es the m a x i m u m total stress (sigt)
c m a x i m u m h o o p me m br an e stress (sighp)
c and m a x i m u m b e n d i n g stress (sigben)
c w i t h i n the th er mo -e l a s t i c stress pr ofile

C O M M O N / d a t a / s t r e s s ( 1 0 00 , 5 ), a x b e n a (10 00) ,a l o n g x (1000)


D I M E N S I O N amax(5)

s i g t = 0 .0
s i g h p « 0 .0
s i g b e n » 0 .0

OPEN(1,status='unknown')
REWIND(l)

R E A D ( 1 , * ) n o e l ,sc ra p, sc ra p
R E A D (1,*)n o u s e , n o u s e , s c r a p

c reads lines of stress data

DO 130 J = l , ( n o e l + 1 ) ,1
R E A D (1,21)n o u s e ,n o u s e ,s c r a p , s c r ap ,a xm em ,h pm em ,a x be n,

172
1 h p b e n , shear
21 F0R MAT (I 4,I 5, 7 E1 4. 7)
IF ( ( J . L E . n o e l ) . A N D . ( J . G T . l )) THEN
R E A D ( 1 , 2 1 ) n o u s e , n o u s e ,s c r a p , s c r a p , a x m , h p m
1 ,a x b , h p b , shear
a x m e m » (a x m e m + a x m ) / 2.0
a x b e n » ( a x b e n + a x b ) / 2 .0
h p b e n » (h p b e n + h p b ) / 2 .0
h p m e m » (h p m e m + h p m ) / 2.0
END IF

c store the axial be nd ing for su bro uti ne bend

a x b e n a (J )» - a x b e n

c store p os si ble va lue s for sigt

a m a x (1)» A B S (h p m e m - h p b e n + a x b e n )
a m a x (2)» A B S (h p m e m + h p b e n - a x b e n )
a m a x (3)» A B S (a x b e n )
a m a x (4)» A B S (h p m e m + h p b e n )
a m a x (5)» A B S (h p m e m - h p b e n )

c pick the m a x i m u m thermal stress at this node

t o t m a x » 0 .0
DO 131 K»l,5,l
IF ( a m a x ( K ) .GT.totmax) THEN
to tmax»amax(K)
END IF
131 C O NT IN UE

c look for the m a x i m u m be nd in g stress so far

b i g b e n » a m a x ( 3)

IF (b igben.GT.sigben) THEN
s i g b en »b i g be n
END IF

c look for the m a x i m u m hoop stress so far

b i g h p » s t r e s s ( J , 1 )- s t r e s s (J ,3)
IF (bighp.GT.sighp) THEN
s i gh p» b ig hp
END IF

c look for the m a x i m u m total stress so far

IF (t o t m a x . G T . s i g t ) THEN
s i gt »to tm ax
END IF

130 CON T I N U E

CLOSE(l)

RETURN

173
END

Q ****************************************************

SUB RO UT IN E o p t i m ( n o e l ,i t y p e ,j t y p e ,s i g t o t ,s t o s y ,
1 r a d i u s ,t h i c k ,o p t s t r , b e n s t r , l e n g t h , i n o d e ,
Imn,popl)

c gives o p t i m u m m e c h a n i s m for stress itype


c if jtype is gi ven (ie jtype not * 0 ) then the op ti m u m
c m e c h a n i s m for (i t y p e - j t y p e ) is c a lc ul at ed
c A value of p/pl is returned for a given Sig y/ sig t
c Inte gr at es (stress*w)dx a lo ng m e c h a n i s m

C O M M O N / d a t a / s t r e s s (1 00 0 , 5 ) , a x b e n a ( 1 00 0) ,a l o n g x (1000)
DIMENSION str(lOOO)

c Sets m a x i m u m length of m e c h a n i s m search from last


lit era ti o n
l e n s e t- le ng t h
nodset=inode

c me ch is set to 1 for an o u t w ar d o u tw ar d m e c h a n i s m
c and 0 for an inward one.

IF ((i t y p e .E Q .1 ).O R .(i t y p e .E Q .6 ).O R .(j t y p e .E Q .1 ).


1 O R . ( j t y p e . E Q . 6 ) ) THEN
mech»l
ELSE
m e c h »0
END IF

c stores suitable stress val ues in the ar ra y str()

DO 110 M= l, no e l, l
st r ( M ) = s t res s( M,i typ e)
IF (jtype.GT.O) THEN
str(M)=str(M)+stress(M,jtype)
IF (mech.EQ.l) THEN
IF (str(M).LT.0.0) THEN
s t r ( M ) = 0 .0
END IF
ELSE
IF (str(M).GT.0.0) THEN
s t r ( M ) » 0 .0
END IF
END IF
END IF
Str(M)=ABS(Str(M))
110 CON TI NU E

c calc ula te the o p t i m u m m e c h a n i s m over all strint()

popl = 1.0E15
DO 120 J=2 ,le ns et ,l
DO 121 K = n o d s e t ,(n o e l + l - J ),1

a = (alongx(K+J) - a l o n g x ( K ) ) / 2 .0
CALL b e n d ( K , J , m e c h , s i g t o t , b e n m e c , m n o d e )

174
c work out stress * disp. at each point w i t h i n
c the mechanism, and sum them up.

X = a l o n g x (m n o d e )- a l o n g x (K )
t o t m e m = 0 .0
DO 122 L = K , ( m n o d e - 1 ) ,1
s t e p = a l o n g x (L + 1 )-al ongx (L )
wa=(alongx(L)-alongx(K))/X
wb=(alongx(L+1)-alongx(K))/X
a m e m = (s t r (L )* w a + s t r ( L + 1 )* w b )* s t e p / 2 .0
totmem-totmem+amem
122 CON TI NU E
X = a l o n g x (K + J )- a l o n g x (m n o d e )
DO 123 L » m n o d e ,(K + J — 1),1
s t e p * a l o n g x ( L + 1 )-alongx(L)
w a » ( a l o n g x (K + J ) - a l o n g x ( L ) ) / X
w b * ( a l o n g x (K + J )- a l o n g x (L + 1 ) )/X
a m e m = ( s t r ( L ) * w a + s t r (L + 1 )* w b )*s t e p / 2 .0
totmem=totmem+amem
123 CO NTI NUE
t o t m e m = t o tm em / a

c work out a value of P/Pl (amech)

t e r m a « ( r a d i u s * t h i c k )/(a * a )
t e r m b » (t e r m a * b e n m e c * 2 .0 )/(s i g t o t * 3 . 0 )
termc=totmem/sigtot
a m e c h » (t e r m a + 1 .0 )-(s t o s y * (t e r m b + t e r m c ))

IF
(a mec h. L E . popl) THEN
popl»amech
be ns tr » b e n m e c
optstr»totmem
le ngth«J
m n» mn ode
inode»K
END IF
121 CON TI NU E
120 CON TI NU E

R ET UR N
END

Q ****************************************************

SU BR OU T I N E b e n d ( i n o d e ,l e n g t h , m e c h , s i g t o t ,b e n m e c ,m n o d e )

c Calcs, b e n di ng stress for the hinge c o n t r i b u t i o n to the


c hi ng e - c o n e mechanisms.
c Uses b e n di ng stresses on ly over the m e c h a n i s m length,
c in the su bro ut ine optim.
c m o d i f i e d 18-4-88 to take ac count of the d i r e c t i o n of
c be nd in g
c si mp li fi ed 27-10-88 to fix hinge in centre of m e c h a n i s m

CO M M O N / d a t a / s t r e s s ( 1 0 0 0 , 5 ), a x b e n a (1 00 0), a l o n g x (1000)

c ca lcu lat e stresses at ends for o u t w a r d m e c h a n i s m

s i g b a » 0 .0

175
s i g b c = 0 .0
s i g b b = 0.0

mnode-inode+(length/2 )

IF (mech.EQ.l) THEN
IF (a x b e n a (i n o d e ).L T . 0.0) THEN
sigba- A B S (a x b e n a (i n o d e ))
END IF
IF (a x b e n a (m n o d e ).G T .0.0) THEN
sigbb»axbena(mnode)
END IF
IF (a x b e n a ( i n o d e + l e n g t h ) . L T . 0.0) THEN
sigbc= A B S (a x b e n a (i n o d e + l e n g t h ))
END IF

c calcu lat e stress at ends for inward m e c h a n i s m

ELSE IF (mech.EQ.O) THEN


IF (a x b e n a (i n o d e ) .GT.0.0) THEN
sigba-axbena(inode)
END IF
IF (a x b e n a ( m n o d e ).L T . 0.0) THEN
s i g b b - A B S ( a x b e n a (m n o d e ))
END IF
IF ( a x b e n a ( i n o d e + l e n g t h ) . G T . 0.0) TH EN
s i g b c « a x b e n a( in o de +l e n gt h)
END IF
ELSE
STOP
END IF

c sets b e n di ng stress

b e n m e c » (s i g b a + ( 2 .0 * s i g b b ) + s i g b c ) / 4 .0

R ET UR N
END

Q ******************************************************

S UB RO U T I N E
t m e c h ( s i g t o t , s i g m e m , s i g b e n , b r e e ,e u b , r p a l ,rpip)

c cal cul at es val ue s for the three l o c a l is ed thi nn in g


c me ch a n i s m s
c (under axial load and internal p r es su re for reverse
c plasticity)

D I M E N S I O N b r e e ( 2 , 1 0 ) , e u b ( 2 , 1 0 ) ,r p a l ( 2 , 1 0 ) ,rpip(2,10)

a»sigben/sigtot
b»sigmem/sigtot

DO 140 J»l,10,l

c bre e- ty pe m e c h a n i c m

bree(2,J)»(J/2.0)-0.5
br ee ( l , J ) » l - ( a * b r e e ( 2 , J ) / 4 )

176
c ex te n d e d upper b ou nd m e c h a n i s m

eub(2,J)»(J/3.0)+1.8
e u b ( 1 ,J )=!/(a * e u b ( 2 ,J ))

c reverse p l a s t i c i t y me ch anism, axial load

r p a l ( l , J ) = J / 1 0 . 0-0.1
r p a l (2 ,J )*(2 .0- r p a l (1 ,J ) )/b

c reverse p l a s t i c i t y me ch anism, internal p r e s s u r e

r p i p ( l , J ) - J / 1 0 . 0-0.1
r p i p (2,J )»(4 . 0- rpi p( 1, J) )/ (2 .0 *b )

140 CO N T I N U E

RE T U R N
END
Q **************************************************

SUBROUTINE mover(noel,dismov)

c revised m o v i ng te mp era tur e front p r o g r a m


c sta rt ed 16-6-88

CO M M O N / d a t a / s t r e s s ( 1 0 0 0 , 5 ) , a x b e n a (1 00 0) ,a l o n g x (1000)
D I M E N S I O N strmov(1000,4)

c copy stress onto strmov

DO 220 1 = 1 , ( n o e l + 1 ) ,1
DO 221 J-1,4,1
s t r m o v ( I ,J ) » s t r e s s ( I ,J )
221 C O NT IN UE
220 C ON TI NU E

c find the number of nodes c o r r e s p o n d i n g to the move men t


c length
c and move the front for ward by half that ammount.

DO 200 1 =1 ,n o e l , 1
nod mov »!
201 IF ((I + n o d m o v ).L E .n o e l ) THEN
IF ((a l o n g x (I + n o d m o v )- a l o n g x (I ) ) .LT.
1 ( ( d i s m o v - 1 . 0 E - 6 ) / 2 . 0 ) ) TH E N
no dm ov =n o d m o v + l
GO TO 201
END IF
END IF

DO 202 J = l , n o d m o v , 1
DO 203 kty pe* l, 2,l
IF ( s t r e s s ( ( I + J ) ,k t y p e ) . G T . s t r m o v (I ,k t y p e ))
1 THEN
s t r m o v ( I ,k t y p e )« s t r e s s ((I + J ) ,ktype)
END IF
IF ( s t r e s s ( ( I + J ) , ( k t y p e + 2 ) ) .LT.
1 s t r m o v ( I ,(k t y p e + 2 ))) THEN

177
s t r m o v ( I , ( k t y p e + 2 ) )= s t r e s s ((I + J ) , (k t y p e+ 2 ))
END IF
203 CO NT INU E
202 CONT INU E
200 CONTI NUE

c repeat for the reverse half of the m o v e m e nt


c (this ma y prove unnecessary)

DO 210 I » ( n o e l + l ) ,2,-1
nod mov»l
211 IF ( (I-nodmov).GT.l) THEN
IF ((a l o n g x (I )- a l o n g x (l - n o d m o v ) ).LT.
1 ( ( d i s m o v - 1 . 0 E - 6 ) / 2 . 0 ) ) THEN
n od mo v= n o d m o v + l
G OTO 211
END IF
END IF

DO 212 J » l , n o d m o v , 1
DO 213 kt ype -1 ,2, 1
IF ( s t r e s s ( ( I - J ) , k t y p e ) . G T . s t r m o v (I ,k t y p e ))
TH EN
s t r m o v ( I ,k t y p e )- s t r e s s ((I - J ),k t y p e )
END IF
IF (s t r e s s ((I - J ),(k t y p e + 2 )).LT.
1 s t r m o v ( I ,(k t y p e + 2 ))) THEN
s t r m o v ( I ,(k t y p e + 2 ))- s t r e s s ((I - J ) , ( k t yp e+ 2 ))
END IF
213 CO NTI NUE
212 CO NT INU E
210 CON TI NU E

c copy strmov onto stress for the p r o g r a m

DO 30 l o o p - 1 , ( n o e l +1 ), 1
DO 31 no-1 ,4, 1
s t r e s s (l o o p , n o ) - s t r mo v( lo op ,n o)
31 CON TI NU E
30 CON TI NU E

R ET UR N
END

178
Appendix C .

S p e c i m e n and Materia l Data.

S p e c i m e n Mat erial Specification.

Source: M a n u f a c u r e r s Mill Certificate.

316 Stainless Steel to A S T M A511

S e a m l e s s tube, hot formed, heat tr eated (solution annealed),


pickled.

As-bought size: 15 2. 4mm d ia me ter 6. 3 5m m wall (nominal sizes)

El ements (%)
Cr 16.7
Ni 11.04
Mo 2.03
Mn 1.42
C 0.049
Si 0.046
P 0.026
S 0.006

179
Material Data Used in F i n i t e Element Models.

Temp. E V a P r f .Stress P r f .Strain


°C xEll Nm 2 xE-5 xE8 Nm 2 %

20 1.78 0.3 1.64 1.78 0.0


20 - — — 1.95 0.1
20 - — - 2.11 0.2
20 — — — 2.47 1.0
100 - - — 1.32 0.0
100 - - — 1.47 0.1
100 — - - 1.63 0.2
100 - — — 1.87 1.0
200 — - — 1.14 0.0
200 — - — 1.23 0.1
200 — - - 1.39 0.2
200 — — — 1.56 1.0
600 1.50 0.3 1.87 0.821 0.0
600 — — — 0.905 0.1
600 - — — 0.994 0.2
600 - — — 1.15 1.0

180
Cyclic Data Used in Plastic Strain Mo dels

Experiment H (Kinematic and Isotropic)

Step N o .Cycles M e c h .Load Max.Temp.


xlE7 N ®C

1 20 1.744 211
2 20 2.626 211
3 20 3.488 211
4 20 4.360 211
5 20 5.232 211

Experiment C (Kinematic and Isotropic)

Step N o .Cycles M e c h .Load Max.Temp.


xlE7 N ®C

1 20 0.872 260
2 20 1.744 260
3 20 2.626 260
4 20 3.488 260
5 20 4.360 260

Experiment D (Kinematic and Isotropic)

Step N o .Cycles Mech.Load Max.Temp.


xlE7 N ®C

1 20 0.872 289
2 20 1.744 289
3 20 2.626 289
4 20 3.488 289
5 20 4.360 289

Experiment F (Kinematic and Isotropic)

Step N o .Cycles M e c h .Load Max.Temp.


xlE7 N ®C

1 20 0.436 416
2 20 0.872 416
3 20 1.744 416
4 20 2.626 416
5 20 3.488 416

181
C \J
s s

CVJ
d

8 s
CVJ

III lO

F ig u re C 1 Experimental Stress- Strain Curves for Specimen Material at


Various Temperatures (Traced from Chart Recorder Plots).

182
A p p e n d i x P.

E xp er i m e n t Control and Data C o l l e c t i o n Program s

50 REM thermal lo ading rig


60 RE M control and data logger
70 RE M 2 8 6 8 8
80 REM Ray .Ve ne s s
81 RE M M O D I F I E D 7-12-88
82 RE M Extens reads on F2 due to noise
90 REM
100 DIM dat(8,180)
110 chout$- "O0 9"
113 sca le«14.2
116 zer o-491.0
117 ON ER RO R P R O C cr a sh
120 REM
130 REM
140 R E PE AT
145 CLS
147 RE ST OR E
150 PR IN T"B ase T e m p ......
155 basetemp-FNdefault
160 PR I N T " R a m p up time...";
165 r a m pu p- FN d ef au l t
170 P R IN T" Pe ak T e m p .....
175 peaktemp-FNdefault
180 PR I N T " H o l d time up...";
185 hothold-FNdefault
190 P R I N T " R a m p down time..";
195 ra mp do wn - F N d e f a u l t
200 PR I N T " H o l d time down..";
205 coldhold-FNdefault
210 PRINT"No. C y c l e s ";
215 cy c l e s - F N d e f a u l t
220 INPU T"A re these cor re ct "; r e pl y$
230 UN TI L (r e p l y $ - " Y " )0R(r e p l y $ - " y " )
240 REM
250 REM
260 PR OC i n i t i a l i s e
270 *FX3,4
280 P R I N T " A n y key to set up"
290 scrap$ =GE T$
300 *FX3,6
310 REM
320 P R O C r a m p ( 0 .0 ,b a s e t e m p , 1 0 .0 ,c h o u t $ ,1.0)
330 REM
340 RE M
345 *FX3,4
350 CLS
360 P R I N T " A n y key to start ex pe riment"
370 PRI NT "Q - Quits at end of cycle"
380 scrap$ »G ET $
390 *FX3,6
400 REM
410 REM
420 FOR c ur cy cl e=l TO cycles
425 *FX3,4
430 PRI NT "c yc l e no. ";curcycle

183
435 *FX3,6
440 P R O C l o g ( (c u r c y c l e * 3 ) - 2 )
450 P R O C r a m p ( b a s e t e m p , p e a k t e m p , r a m p u p , c h o u t $ ,1.0)
455 P R O C l o g ( (c u r c y c l e * 3 ) - l )
460 PROChold(peaktemp,hothold,chout$)
465 P R O C l o g ( (c u r c y c l e * 3 ))
480 P R O C r a m p ( p e a k t e m p , b a s e t e m p , r a m p d o w n , c h o u t $ , - l .0)
490 PROChold(basetemp,coldhold,chout$)
500 REM
510 REM
520 qu it$ =IN KEY $(0 )
530 IF ( q u i t $ = " Q " ) O R ( q u i t $ - " q " ) THEN G OT O 550
540 N EX T cur cycle
550 P R O C r a m p ( b a s e t e m p , 0 . 0 , 5 . 0 , c h o u t $ , - l .0)
560 REM
570 REM
575 *FX3,4
580 INPUT "Sa ve to disc ";reply$
590 IF (r e p l y $ » " Y " ) 0 R ( r e p l y $ - " y " ) THEN
P R O C d i s c f i l e (c y c l e s ,d a t )
620 STOP
960 REM
970 REM
980 REM
990 REM
1000 DEF PR OC i n i t i a l i s e
1010 REM sets up the comark
1020 *FX2,2
1030 *FX7,7
1040 *FX8,7
1050 *FX3,7
1060 PR INT "Fl"
1070 REM 32 readings 20ms
1080 PR INT "GO"
1090 RE M +/- Ivolt scale
1100 PR INT "009 0"
1110 RE M zeros out put line
1120 *FX3,6
1130 EN DP ROC
1960 REM
1970 REM
1980 REM
1990 REM
2000 DEF P R O C r a m p ( t e m p i ,t e m p 2 ,r a m p t i m e ,c h o u t $ ,s t e p l e n )
2010 RE M Ramps up or down
2020 RE M d e p e n d i n g on sign of steplen
2030 RE M be tw e e n tempi and temp2
2040 s t e p t i m e = r a m p t i m e * 1 0 0 / A B S (t e m p l - t e m p 2 )
2050 FOR cou n t -t em pl TO temp2 STEP steplen
2060 t-TIME
2070 v o l t a g e = ( c o u n t * s c a l e ) + z e r o
2080 *FX3,7
2090 PRINT c h o u t $ ;v ol tag e
2100 *FX3,6
2110 REP EA T
2120 UN TI L (TI ME -t )> =s te pt im e
2130 NE XT count
2140 ENDP ROC
2960 REM
2970 REM

184
2980 RE M
2990 RE M
3000 DEF P R O C h o l d ( t e m p , h o l d t i m e ,c h o u t $ )
3010 REM holds at temp for holdtime
3020 t=TIME
3030 v o l t a g e - (t e m p * s c a l e )+zero
3040 *FX3,7
3050 PR IN T ch ou t S ,v ol t ag e
3060 *FX3,6
3070 RE P E A T
3080 U N T I L (TIME- t) >- (ho ld tim e* 100 )
3090 EN DP RO C
3960 REM
3970 REM
3980 RE M
3990 REM
4000 DEF P R O C d i s c f i l e ( n o c y c l e s , d at )
4010 RE M stores data on disc file
4020 *FX3,4
4030 CLS
4040 INPUT "Filename ";filename$
4050 INPUT "Comments ";comment$
4060 X - O P E N O U T filename$
4070 PRINT£X,comment$
4080 PRINTEX,nocycles
4090 FOR cyc leno=l TO (nocycles*3)
4100 FOR no=l TO 8
4110 P R I N T £ X , d a t (n o , c y c l e n o )
4120 N E X T no
4130 N E X T cycleno
4140 RE M
4150 CLOSEEX
4160 E N D PR OC
4960 R EM
4970 R EM
4980 REM
4990 REM
5000 DEF PROClog(pt)
5010 REM co llects one set of
5020 RE M temp and d i s p l a c e m e n t data
5040 *FX3,7
5050 P R I N T "T05"
5060 P R I N T "TO6"
5070 P R I N T "T07"
5080 P R I N T "TO8"
5100 *FX3,6
5110 *FX2,1
5120 FOR loop-1 TO 4
5130 IN PUT dat(loop,pt)
5140 N E X T loop
5142 *FX2,2
5143 *FX3,7
5144 P R I N T "F2"
5146 P R I N T "101"
5147 P R I N T "Fl"
5150 *FX3,6
5152 *FX2,1
5154 FOR loop-5 TO 5
5156 INPUT dat(loop,pt)
5158 N E X T loop

185
5159 *FX2,2
5160 *FX3,4
5180 PRI NT
T A B ( 1 ) ; d a t (1,p t );TAB(9 ) ; d a t (2 , p t ) ; T A B ( 1 7 ) ; d a t ( 3 , p t );
T AB (2 5); dat (4, pt)
5190 PRINT
T A B ( 1 ) ; d a t (5 , p t ) ; T A B ( 9 ) ; d a t (6 , p t ) ; T A B ( 1 7 ) ; d a t ( 7 , p t );
T AB (2 5); dat (8, pt)
5200 *FX3,6
5210 EN DP RO C
6000 DEF FNd efault
6010 R EA D def
6020 PR IN T ” (";def;")";
6030 INPUT value
6035 IF val ue<0 TH EN PRINT "Don't be s i l l y " : G O TO 6030
6040 IF va lu e - 0. 0 THEN -def ELSE -value
6050 END
6060 D A T A 6 0 , 1 0 , 2 5 0 , 1 5 , 1 0 , 1 8 0 , 1 5
6980 REM
6990 REM
7000 DEF PRO Cc r as h
7010 RE M Takes over in the event of a
7020 RE M p r o g r a m fault and saves data.
7030 *FX2,2
7040 *FX3,4
7050 PRI NT" ** ** FATAL ER ROR ****"
7060 INPUT "Try to save d a t a ( Y / N ) "; reply$
7070 IF (r e p l y $ - " Y " ) 0 R ( r e p l y $ - " y " ) THEN
P R O C d i s c f i l e (cycles,&at)
7080 STOP
7090 EN D PR OC

186
Data R e c o v e r y Program.

lODIM STORE(4,900)
20CLS
2 5 *CAT
30 lN PU T" Na m e of data file ";NAME$
40 X- O P E N I N NAME$
41 INP UT £X ,N
42 FOR C=1 TO N
43 FOR L=0 TO 4
5 0 I N P U T £ X ,S T O R E (L ,C )
60 NE XT L
70 NE XT C
8 0 C LO SE £X
9 0 l N P U T " C o p y to p r i n t e r (Y / N ) " ;ANS$
95 CLS
100 IF ANS $= "Y " TH EN VDU2
390 PRINT
T A B ( 2 , 1 ) ;" T I M E " ; S P C ( 5 ) ;" T 5 " ; S PC ( 6 ) ; "T 6" ;S PC (6 ); " T7 ";
SP C(6 );"T8"
400 FOR PL-1 TO N
410 P R I N T T A B ( 3 ) ;S T O R E (0,P L ) ;T A B (9);F N T E M P (S T O R E (1,P L ));
T A B (1 7); F N T E M P (S T O R E (2,P L ) );
415 PRI NT T A B ( 2 5 ) ; F N T E M P ( S T 0 R E ( 3 , P L ) ); T A B ( 3 3 ) ;
F N T E M P (S T O R E (4,PL))
420 N E X T PL
4 30 VDU3
440 STOP
500 DEF F N T E M P ( A ) - ( (( 24 .0 3 8 4 6 * A + 2 4 . 4 2 3 0 8 ) * 1 0 0 0 )DIV D / I O O O

187
Appendix E.

D e v e lo pm en t of Ex p e ri me n ta l A p p a r a t u s

Introduction.

The e x pe ri me nt described in chapter 4 was the result of

several mon ths of d e v e l o p m e n t on all as pects of the load and

measurement systems. The e x pe ri me nt al development section

o u tl in ed some of the more im por tan t changes req uired to the

apparatus. The pu rp ose of this sec tio n is to add some more

detail to the com ments of cha pter 4 and des cr ib e some other

pr obl ems en co u n t e r e d in the ex pe ri m e n t a l work. It is hoped

this would be of use to future experimenters us i n g similar

techniques.

Initial Des ig n and Proof Testing.

Three changes from the in it ia ll y designed sp ec ime n and

app ar at us were forced ear ly in the development of the

experiment.

The spec ime ns we re designed to be c l am pe d d i r e c t l y to the

ho ll o w e d co ol ing blo cks to ens ure the most di re ct heat

c on du c t i o n from the specimen. However, the sp eci men tube

sup pli ed by the manufacturers pr o v e d to have a signif ica nt

de gre e of ovality. It was decided that it wo u l d prove

d if f i c u l t to reli abl y manufacture a perfectly round specimen

from this tube. In order to p r e ve nt residual stresses being

set up due to the cla mp in g a third block was d e s i g n e d on which

to clamp the specimen. This block was split into 10 segments,

each c o nt ai ni ng a sp eci men mounting bolt and at tac hed

188
se p a r a t e l y to the co oli ng block.

Proof tes ti ng of the he at ing and cool ing supplies were

most satisfactory. T h e r m o c o u pl es and lo gg ing devi ces were

first ca li b r a t e d with b oi lin g w at er and melting ice. Then,

running the ma ch in e at full power, te mp e r at ur e g r a di en ts of

450®C were a c h i ev e d across the s p eci me n after just 2.5 seconds

of heating.

Testing of the m e c h a n i c a l in te gr it y of the s ys te m was not

so successful. The load train was c a l i b r a t e d w i t h a load cell

in the place of a specimen. A spe ci men was then pl a ce d in the

machine for pr oof testing. Un fo rtu nat ely , wa te r leaks were

discovered from b e t w e e n the 'O' ring seals at loads ab ove 0.8

(7y. In order to remedy this si tu at io n a s e c o n da ry wa te r seal

was designed (not shown in drawings) consisting of a

th e r m o p l a s t i c ring and further pair of 'O' rings. This can be

seen in the photograph of the s p ec im en gi ven in figure 4.7 A

re peated pr oof test to 1. 5*^ was p e r f o r m e d sati sf act ori ly .

Axial st rain in the spec im en was measured initially by

me ans of Linear Variable Differential Transducers (LVDTs).

Four LVDTs were mounted on a pair of rings c l a mp ed to the

sp ec im en on knife edges. After co n s i d e r a b l e tes ti ng it was

decided that the data from these d e vi ce s was not su ff ic i e n t l y

reliable. T he y were t h er ef or e re placed with the clip gauge

system described in se ct io n 4.6. The calibration of these

d e v i c e s was c h e ck ed pr ior to each e x p e r i m e n t by m e a s u r i n g the

elastic strains produced by applying a known m e c h a n i c a l load

to the specimen.

189
D e v e lo pm en t of A x i s y m m e t r i c Te mp e r a t u r e Profiles.

It was e x p l a in ed in s ec tio n 4.7 that ci r c u m f e r e n t ia l

te mp er at u r e s y m me try was a c h i ev ed by the variation of the

r es is tan ces of the power lines. The m e t h od was ad op ted after

co ns id e r a b l e in ve st i g a t i o n of the causes of the as y m m e t r y in

the te mp e rat ur e profile. This se ction contai ns a su mm ary of

this investigation.

An initial surve y of the t em per at ur e aro un d the

circumference of the tube sho wed v a ri at io ns of ±8% over a

range of mean temperatures. The first po ssi ble cause to be

co ns i d e r e d was the cool ing supply. However, repe at in g the

m e a s u r e m e n t s w ith the coo li ng sw it che d off resulted in similar

te mp er at ur e var iations.

The second co n s i d e r a t i o n wa s the power supply. A hall

ef fec t amme ter was use d to m ea sur e the cur rent p a s s i n g through

e ach cable. Variations of upto 50% were measured be tw ee n the

20 cables. As all the cables w er e fed d i r e c t l y from the same

tr an sf or me r blocks, the p r o b l e m was na rr o w e d down to either

the cables or the specimen.

The first cause to be c o ns i d er ed was inductive cou pling

between cables. El ect ric fields aro und the cables and

t r a n sf or me r were m e a s u r e d w it h a Gauss meter. These were found

to be significant, es p e c i a l l y ar ou nd the back of the

t r a n sf or me r and so two steps were taken to m i n i m i se the

effects. A large steel plate was ins erted be h in d the

tra nsformer, grounded to the t r a n sf orm er earth, as shown in

figure 4.6. This had the eff ect of c o ns t r i c t i n g the electric

field produced in the t ra ns fo rm er away from the cables and

190
specimen. In addition, the cables were routed such that they

ran pa ra lle l to each other for most of their length. This was

in or der to m i n i m i se flux linkage b e t we en the cables. These

modifications y ie l d e d an i m p r o ve me nt of ±1% in the

circumferential te mp era tur e gradients.

The next pos si bl e cause of v a r i a t i o n s in cable current to

be co n s i d e r e d was the res istance of the cur rent paths. This

was attributed to three pos si bl e causes:

i) Cable resistance.

Considerable care was taken in the manufacture of the

cab les and cri mpe d co nn ect ors to ensure un if ormity. Each

cable was the same length and cri mped to the same pressure.

In order to c o n fi rm the quality of the cables, five were

tes te d on a h i gh ly sen sitive resistance bridge. As all five

cab les showed a res istance of 81±l//2, the cables were

d i s c o u n t e d as a cause of the problem.

ii) V a r i a t i o n of res istance w i t h i n the specimen.

Small va r i a t i o n s in the wall t h ick ne ss of the specimen

would lead to a co ns id e r a b l e variation in the power

dissipated aro un d the c i r c u m f e r en ce of the specimen. The

res ist an ce of the speci men and thus the power dissipated,

is in ve rs el y p r op or ti on al to the wall thickness. On this

basis, a wall th ick nes s tole ran ce of ± 0 . 0 5 m m was accepted,

allowing for a maximum res istance variation of ±2%. This

would ac co unt for a proportion of the cir cu mf er en ti al

te mp e r a t u r e di ff e r e n c e discovered.

iii) Co nt ac t resistances.

E ac h of the 20 cable c on ne ct io ns con ta in ed 3 bol ted

contacts. Although care was taken to ensure that all

c on ta ct s were clean and the bolts tig ht en ed to the same

191
torque, there was still poten tia l for si gn if ic an t contact

resistance. Due to the v er y low resis ta nce s involved, it

was not po s si bl e to take di re ct me as ure men ts. An at tempt

was made howe ver to s ys te m a t i c a l l y i n v e st ig at e the contacts

by swap pi ng c o n n e c t io n points. This led to a se em in gl y

illog ic al change in the cable cu rrents each time a

c o n n e c t i o n was changed. The co nc lu s io n dr aw n from this was

that the path re sistance was dependant to a large, if

co mp le x ex tent on the contact resistances.

At this stage it was d ec id ed to adopt the emp irical

solution e x p l a in e d in se ction 4.7 rather than become involved

in fu rther analysis.

192
Appendix F.

E x pe ri me n t a l Data.

Refer to figure 4.11 for te mpe rat ure pro fi les

Test No. Load(KN) No. Cycles S t r a i n ( %) Accu. Strain(%)

C3 40 13 0.016 0.016
C4 60 14 0.141 0.157
C5/C6 80 30 0.296 0.557
C7/C8 90 40 0.209 0.661
C9/C10 (100) (L o s t ) (0.230) 0.891
Cll 110 40 0.243 1.134
C12 120 40 0.309 1.443

Test No. Load(KN) No. Cycles S t r a i n ( %) Accu. Strain(%)

D2 10 20 0.017 0.017
D3 20 15 -0.021 -0.004
D4 30 15 -0.016 -0.020
D5 40 15 0.037 0.017
D6 50 20 0.019 0.036
D7 60 20 0.101 0.137
D8 70 20 0.123 0.260
D9 80 20 0.154 0.414
DIO 90 20 0.187 0.601
Dll 100 20 0.264 0.865

Test No. Load(KN) No. Cycles S t r a i n ( %) Accu. Strain(%)

E2 10 12 -0.004 -0.004
E3 20 20 -0.024 -0.028
E4 30 20 0.001 -0.027
E5 40 20 0.036 0.008
E6 50 20 0.060 0.068
E7 60 20 0.076 0.144
E8 70 20 0.106 0.250
E9 80 20 0.134 0.384
ElO 90 20 0.169 0.558
Ell 100 20 0.217 0.775

193
Test No. L o a d ( K N ) No. Cycles S t r a i n ( %) Accu. Strain(%)

F3 30 15 0.000 0.000
F4 40 20 0.030 0.030
F5 50 20 0.037 0.067
F6 60 20 0.046 0.113
F7 80 20 0.321 0.434
F8 80 20 0.099 0.533
F9 90 20 0.267 0.800
FlO 100 20 0.406 1.206

Test No. Load(KN) No. Cycles S t r a i n ( %) Accu. Strain(%)

G1 20 15 -0.017 -0.017
G2 30 15 -0.007 -0.024
G3 40 20 0.007 -0.017
G4 50 15 0.004 -0.013
G5 60 20 0.016 0.003
G6 70 20 0.02 0.023
G7 80 20 0.039 0.062
G8 90 18 0.067 0.129
G9 100 20 0.119 0.248
GIO 110 20 0.200 0.448
Gll 120 20 0.194 0.642

Test No. Load(KN) No. Cycles S t r a i n ( %) Accu. Strain(%)

HI 10 10 -0.010 -0.01
H2 20 10 0.001 -0.009
H3 30 10 -0.007 -0.016
H4 40 10 0.004 -0.012
H5 50 20 0.011 -0.001
H6 60 20 0.033 0.032
H7 70 20 0.020 0.052
H8 80 20 0.063 0.115
H9 90 20 0.084 0.199
HlO 100 20 0.120 0.319
Hll 110 20 0.181 0.500
H12 120 20 0.309 0.809

194 .
Mechanical Load: 40 KN

Test No. AT(°C) No. Cycles S t r a i n ( %) Accu. Strain(%)

11 55 20 -0.031 -0.031
12 91 20 -0.009 -0.039
13 117 20 0.03 -0.009
14 127 20 — 0.008 -0.017
15 139 20 0.02 0.003
16 143 20 0.11 0.113
17 185 20 0.056 0.169

Mec ha ni ca l Load: 70 KN.

Test No. AT(°C) No. Cycles S t r a i n ( %) Accu. Strain(%)

Kl 68 11 0.022 0.022
K2 92 20 0.068 0.09
K3 105 20 0.072 0.162
K4 117 10 0.05 0.212
K5 154 20 0.16 0.372
K6 193 20 0.144 0.516

195
Appendix G.

D e r i v a t i o n of Wor k H a r d e n i n g Equations.

(After Cocks and Ponter [51]

No m e n c l a t u r e

P Con st an t ap p l ie d p r i m a r y stress.

p as a sub- or superscript. Pl astic stress or strain.

D e vi at o r ic stress component,

e as a subscript. E f fe ct ive stress or strain.

Initial yi el d stress.

E Elastic modulus.

Ki nem a ti c H ar d e n i n g Model.

The mat er i a l is a ss um ed to be e l ast ic - linear h a r d e n i n g as

sh own for axial ten sio n in figure G.l, gi v in g rise to the

constitutive relationships:

ÿ - dsP. - I d e P ^ G1,G2
8
For a Von Mi ses material, initial y i e l d i n g occurs wh e n

"y " "e " I G3

For a kinematically hardening material, an increase in cr^

above leads to a rigid t r a n s l a ti on of the yi eld surface

such that

~ 1 " “ ij>

where is the n e w centre of the yi e l d surface.

196
Assuming the surface tr ans l a te s in the direction of the

st rain rate allows the resultant pla st ic strain in cr e m e n t to

be w r i t t e n as:

dCij - G5

Figure G.2 shows the yi el d surface in d e v i a t o r i c stress

space

"x
wh er e ^ G6,G7
“ I - 3-

are the d e v i a t o r i c stress components.

An axial stress, P is applied, re sulting in a primary

stress v e c t o r A, from figure G.2. The a p p l i c a t i o n of a cyclic

stress in excess of 2a^ results in a t r a n s l a ti on of the yield

surface and a c c u m u l a t i o n of pla st ic strain until the d i r e c t i o n

of t r a n s l a t i o n is along the line of the a p p l i e d stress vector,

BC from figure G.2. The net a c c u m u l a t i o n of p l a st ic st ra in per

cycle is then zero. From this fact, and the sym me tr y of the

yi e l d surface, it can be seen that - P/2. This gives the

plastic st ra in a c c u c m u l a t e d to be:

- 1

It sh ou ld be no ted that this pl as tic st rain is the mean

value ab ou t which the actual st rain at any instance in the

cycle oscill ate s.

I so tr o p i c H a r d e n i n g Model.

Using the same model of p la sti c strain (see figure G.l)

the m a t e r i a l ag ain u n d e r g o e s plastic deformation a c c o r d i n g to

197
e q u a t i o n s G1 and G 2 . The thermo- el astic stress is d e f i n e d as:

a. — G9

wh er e <y^ varie s be t w ee n and over a thermal load ing

cycle. T he n initial yi el d i n g occurs when:

(j2 + (3/4)P2 GIO

S u b s t i t u t i n g e q ua tio ns G1 and G2 gives the rate of pr im ar y

pl a st ic strain ac cumulation:

de da. Gil

I n t e r g r a t i n g Gil b e t w e e n the limits

eP = 0 w he n a^ = a^ and

“ ^^x w ^ e n a^ = a^ + Aa^

gives the strain a c c u m u l a t e d over a half cycle:

+ A»t): + (3/4),:)
Ae^ ( 1 - 9 ) |3
9E ||piog, {aj + (3/4)a^}

-1 -1 G12
Aa^ - tan + tan
/ ( 3 / 4 )a /(3/4)a
x; x;

wh er e the sign app lies in the first half of the cycle and

the sign in the second.

When G12 is summed over a number of cycles, the +/- terms

tend to cancel, leav ing an approximation of the strain

accumulation (for small (3) as:


f
G13
- I

where

f
(3/4)p2 + Aa2/4
*e

198
Axial
Stress

Tan'^E

Axial Strain, z

F ig u re G 1 Schematic of a Uniaxial Stress- Strain Curve showing the


Definition of B.

Initial
T ranslation

Translated Yield
Surface

Initial Yield
S urface

F ig u re G 2 Von Mises Yield Diagram in Deviatoric Stress Space showing


Kinematic Hardening.

199
REFERENCES

1. Ponter, A . R.S •Rachetting in th e Creep Range.

Cocks, A.G.F. Part III: S i m p l i f i e d T h e o r e t i c a l Me t h o d s

of A s s e s s i n g the C r e e p D e f o r m a t i o n of

Stru c t u r e s S u b j e c t e d to C o n s t a n t

M e c h a n i c a l L oad and Cyclic T h e r m a l

Load'. European Commission Report

EUR9876EN, 1985.

2. Design and C o n s t r u c t i o n Rules for

M e c h a n i c a l C o m p o n e n t s of FBR N u c l e a r

Islands. R C C - M R S e c t i o n 1 S u b - s e c t i o n B:

Class I Components'. Association

F r a nçaise p o u r les Regies de C o n c e p t i o n

et de C o n s t r u c t i o n de M a t é r i e l s des

C ha u d i è r e s E l e c t r o - N u c l e a i r e s (AFCEN).

3. Kar a d e n i z , S 'The D e v e l o p m e n t of Upper B o u n d and

A s s o c i a t e d F i nite Element T e c h n i q u e s for

the P l a s t i c S h a k e d o w n of T h e r m a l l y

Loaded Structures'. Ph.D. Thesis,

L e i c e s t e r University, F e b ruary 1983.

4. Koiter, W .T 'General T h e o r e m s for E l a s t i c - P l a s t i c

Solids'. Prog. S o lid Mech., eds. R. Hill

and I. S n e d d o n (North-Holland,

Amsterdam), 1964, V o l . l pp 167-219.

200
5. Ma r t i n , J.B. 'Plasticity: F u n d a m e n t a l s and G e n e r a l

Results'. M.I.T. Press, Boston, 1975.

6. Clement, G. 'A D e sign Rule for Th e r m a l R a t c h e t t i n g '.

Levey, J. J. Press. Ves. Tech. (ASME), 1986,

Roche, R.L. Vol. 108 pp 188-196.

7. Ponter, A.R.S 'Shakedown and R a t c h e t t i n g B e l o w the

Creep Range, Par t 1: Design Codes,

M a t e r i a l R a t c h e t t i n g and the B e h a v i o u r

of Some S i mple S t r u c t u r a l C o n f i g u r a ­

tions'. D e p a r t m e n t of Engineering,

L e i c e s t e r U niversity, M ay 1981.

8. Bleich, H 'Uber d ie B e m e s s u n g statisch u n b e s t i m m -

ter S t a h l t r a g s e r t u n t e r B e r u c k s i c h t i g u n g

des e l a s t i s c h - p l a s t i s c h e n V e r h a l t e n s des

Baustoffes'. Bauingenieur, 1932,

V o l . 19/20 p 261 on.

9. Mela n , E. 'Teorie S t a t i s c h U n b e s t i m m t e r S y s t è m e

aus I d e a l - p l a s t i s c h e m B a u s t o f f .

S i t z u n g s t e r i c h t e d e r A k a d e m i e der

W i s senschaften, Wien, Abt Ila, 1936,

V o l . 145 pp 195-218.

201
10. Prager, W. 'Shakedown in Elastic, Plastic M e d i a

S u b j e c t e d to C y c l e s of Load an d T e m p e r a ­

ture'. S y m p o s i u m Su la Plas t i c i t a N e l l a

Scienza D e l l e C o n s t r u z i o n i . Varenna,

Italy, Sept. 1956. Ed. N. Zanichelli.

11. Leckie, F.A, 'Shakedown as a G u i d e to the D e s i g n of

Penny, R.K, Pr e s s u r e Vessels'. J. Eng. Ind. Trans.

ASME, S e r i e s B, 1969, V o l . 91 pp 799-807.

12 . Specifications for U n fired F u s i o n W e l d e d

P r e s s u r e Vessels. BS5500, 1988. Br i t i s h

S t a ndards Institution, London.

13. Koiter, W.T 'A Ne w G e n e r a l T h e o r e m on S h a k e d o w n of

E l a s t i c - P l a s t i c Structures'. K. Ned.

Akad. Wet., 1956, Pt.B, V o l . 59 pp 24-34.

14. Maier, G. 'S h a k e d o w n T h e o r y in Perfect E l a s t o -

P l a s t i c i t y w i t h A s s o c i a t e d and N o n -

A s s o c i a t e d F l o w Laws: A Finite E l e m e n t

Linear P r o g r a m m i n g Approach'. Meccanica,

1969, V o l . 4, N o . 3, pp 250-260.

15. Ma r t i n , J.B 'Plasticity: F u n d a m e n t a l s and G e n e r a l

Results'. M.I.T. Press, Boston, 1975,

1st Ed. p 658.

202
16. Horne, M.R. 'Plastic T h e o r y of Structures'. T.

N e l s o n a n d Sons, 1971.

17. Cocks, A.C.F 'Lower B o u n d S h a k e d o w n A n a l y s i s of a

Simply S u p p o r t e d P l a t e C a r r y i n g a

U n i f o r m l y D i s t r i b u t e d Load and S u b ject

to C y c l i c T h e r m a l Load'. Int. J. Mech.

Sci., 1984, V o l . 26, N o . 9 pp 471-475.

18. We b s t e r , J.J. 'Elasti c - P l a s t i c B e h a v i o u r of a Simply

Sahari, B.B. S u p p o r t e d C i r c u l a r Plate S u b j e c t e d to

H y d e , T.H. S teady T r a n s v e r s e Pr e s s u r e and Cyclic

Linear R a d i a l T e m p e r a t u r e Variation'.

Int. J. Mech. Sci., 1987, V o l . 29, No.8

pp 533-544.

19. Carter, K.F. 'Interaction D i a g r a m s for C y l i n d r i c a l

Ponter, A . R.S Tubes S u b j e c t e d to Axial T e m p e r a t u r e

Gradients, R e p o r t No.l. U K A E A Co n t r a c t

IR55721, L e i c e s t e r University, UK.

20. Ponter, A.R.S 'An E x t e n d e d S h a k e d o w n T h e o r y for

Karadeniz, S. S t r u c t u r e s t h a t S u f f e r Cyclic T h e r m a l

Loading. P a r t I - Theory'. J. App. Mech

(ASME), 1985, V o l . 52 pp 877-882.

203
21. Ponter, A.R.S 'An E x t e n d e d S h a k e d o w n Theory for

Karadeniz, S. S t r u c t u r e s that S u f f e r Cyclic T h e r m a l

Loading. Par t II - Ap p l i c a t i o n s ' . J.

App. Mech. (ASME), 1985, V o l . 52 pp 883-

889.

22. Karadeniz, S. 'The I n f l u e n c e of T r a n s i e n t T h e r m a l

Ponter, A.R.S L o a d i n g on the Bre e Plate; a Simplified

Me t h o d of A nalysis'. Nuc. Eng. Des.,

1984, V o l . 80 pp 359-374.

23. Ponter, A.R.S 'A L i n e a r P r o g r a m m i n g Upper B o u n d

K aradeniz, S. A p p r o a c h to th e S h a k e d o w n Limit of Thin

Shells S u b j e c t e d to V a r i a b l e T h e r m a l

Loading'. J. Strain. Anal., 1984, V o l . 19

pp 221-230.

24. Carter, K.F. 'Interaction D i a g r a m s for C y l i n d r i c a l

Ponter, A.R.S Tubes S u b j e c t e d to A x i a l T e m p e r a t u r e

Gradients'. R e p o r t N o . l to th e U K A E A

u n der C o n t r a c t No. IR55721.

25. Carter, K.F. 'Interaction D i a g r a m s for A x i s y m m e t r i c

Ponter, A.R.S Geometries', R e p o r t N o . 2 to t he U K A E A

u n der C o n t r a c t No. IR55721.

^04
26. Ponter, A.R.S. 'Upper B o und M e t h o d s for Use in t he

Carter, K.F. De sign and A s s e s s m e n t of A x i s y m m e t r i c

Thin Shel l s S u b j e c t e d to Cyclic T h e r m a l

Loading'. Nuc. Eng. Des., 1989, V o l . 116

pp 239-254.

27. Carter, K.F. 'Interaction D i a g r a m s for Use in the

Porter, A.R.S. C o n s t r u c t i o n of D e s i g n Codes a n d the

Ellington, J.P A s s e s s m e n t of A x i s y m m e t r i c T h i n S h e l l s

Porter, A.G. S u b j e c t e d to C y c l i c T h e rmal Loading'.

Recent Advances in D e s i g n P r o c e d u r e s for

High T e m p e r a t u r e Plant. I. Mech. E.

Seminar, Risley, 1988, pp 7-11.

28. Franco, J.R.Q. 'Bounding T e c h n i q u e s in Sh a k e d o w n a nd

Ratchetting'. P h . D Thesis, Leicester

University, 1987.

29. Arnandeau, J. 'Thin C i r c u l a r C y l i n d e r Under A x i s y m

Zarka, J. m e t r i c T h e r m a l and M e c h a n i c a l Loading'

Gerij, J. Proc. F o urth Int. Conf. on Struct.

Mechs. in R e a c t o r Tech., 1977, V ol L,

Paper L6/5.

205
30. Bree, J. ' Elasti c - P l a s t i c B e h a v i o u r of T h i n T u bes

S u b j e c t e d to I n t e r n a l Pressure and

In t e r m i t t e n t H i g h H e a t Fluxes w i t h

A p p l i c a t i o n to F a s t - N u c l e a r - R e a c t o r Fuel

Elements'. J. Strain. Anal., 1967,

V o l . 2, N o . 3, pp 226-238.

31. C r i teria for D e s i g n of Elevated T e m p e r a ­

ture C l a s s 1 Components in S e c t i o n III,

Di v i s i o n 1 of the A S M E Boiler and

P r e s s u r e V e s s e l Code. The A m e r i c a n

S o c iety of M e c h a n i c a l Engineers, N ew

York, USA.

32. Boyle, J.T. 'The A s s e s s m e n t of Elastic F o l l o w - u p in

Na k amura, K. Hig h T e m p e r a t u r e P i p i n g Systems -

O v e rall S u r v e y a nd T h e o r e t i c a l Aspects'

Int. J. Pres. Ves. & Piping, 1987,

V o l . 29, pp 167-194.

33. Yamamoto, S. 'Thermal R a t c h e t t i n g E x p e r i m e n t of T ype

Kano, T 3 04 S t a i n l e s s S t eel Pipes U n d e r A l t e r n a

Yosh i t o s h i , A ting C o l d and H o t T h e r m a l Shocks w i t h

V a r y i n g A x i a l Loads'. ASME E l e v a t e d

T e m p e r a t u r e D e s i g n S y m p o s i u m , e w York,

USA, 1976, pp 25-32.

206
34. Bell, R.T. 'Ratchetting E x p e r i m e n t s on T h i n

C yl i n d e r s S u b ject to A x i a l l y M o v i n g

T e m p e r a t u r e F r o n t s '. U K AEA N o r t h e r n

D i v i s i o n R e p o r t No. ND-R-835(R), Sept.

1982 .

35. Uga, T. 'An E x p e r i m e n t a l S t u d y of P r o g r e s s i v e

Strain G r o w t h due to T h e r m a l - S t r e s s

Ratchetting'. J. Eng. Mat. & Tech.,

1978, V o l . 100 pp 150-156.

36. Ponter, A . R.S 'An E x p e r i m e n t a l S t u d y of S i m p l i f i e d

Jakeman, R.R. M e t h o d s for the P r e d i c t i o n of the

Morrison, C.J D e f o r m a t i o n of S t r u c t u r e s Su b j e c t to

Severe C y clic T h e r m a l Loading'. J.

Strain. Anal., 1985, V o l . 20, N o . 4, pp

225-240.

37. Cousin, M. 'Specifications de l'essai p o u r Step II

Julien, J.F. B e n c h m a r k C a l c u lations'. Institut

N a t i o n a l des S c i e n c e s A p p l i q u e de Lyon,

France, M a y 1983.

38. Ponter, A . R.S 'The C o m p u t a t i o n of S h a k e d o w n L i m i t s f o r

Karadeniz, S. S t r u ctural C o m p o n e n t s S u b j e c t e d to

Carter, K.F.C V a r i a b l e T h e rmal L o a d i n g - B r u s s e l s

Diagrams'. C o m m i s s i o n of E u r o p e a n

Communities, R e p o r t EUR12686EN. 1989.

207
39. Taylor, P.M. 'Development of a T h e r m a l L o a d i n g Rig'.

Leicester University Engineering

D e p a r t m e n t Final Y e a r Report, 1986.

40. Clarke, P.W, 'CONIDA - A F i n i t e E l e m e n t P r o g r a m for

the Stre s s A n a l y s i s of A x i s y m m e t r i c Thin

Shells'. U K A E A R e p o r t No. TRG2382(R),

Central T e c h n i c a l Services, Risley.

41 Abaq u s U s e r M a n u a l - V e r s i o n 4.6.

Hibbitt, K a r l s s o n and Sor e n s e n Inc.,

USA, 1985.

42 In t erim DCW G r e c o m m e n d a t i o n n o t e on

A l l o w a b l e D e s i g n L i m i t s for T y p e 316

Sta i n l e s s Steel in the T r e a t e d S o l ution

Condition. U K A E A R i s l e y N u c l e a r Power

D e v e l o p m e n t Establishment, UK. Report

No. C F R / D C W E / P ( 8 0 ) 269.

43. S p e c i f i c a t i o n for Steel C a s t i n g s for

Ge n e r a l E n g i n e e r i n g Purposes. BS3100:

1976(1984) British Standards Institu­

tion, London.

44. Megahed, M.M 'Influence of H a r d e n i n g Rule on the

E l a s t o - P l a s t i c B e h a v i o u r of a Simp l e

S t r u c t u r e U n der C y c l i c Loading'. Int. J.

Mech. Sci., 1981, V o l . 23 pp 169-182.

208
45. Chaboche, J.L ’On the P l a s t i c a nd V i s c o p l a s t i c

C o n s t i t u t i v e Equations. Part 1- Rules

Developed with Integral Variable

Concept.' J. Press. Vess. Tech. (ASME),

1983, Vol. 10 N o . 2 pp 135-164.

46. Abaqus Theory Manual - V e r s i o n 4.6.

Hibbitt, K a r l s s o n a n d Sor e n s o n Inc.,

USA, 1985.

47. Ponter, A.R.S 'The I n c r e m e n t a l S t r a i n G r owth of an

Cocks, A.C.F. E l a s t i c - P l a s t i c B o d y L o a d e d in Excess of

the S h a k e d o w n Limit'. A S M E J. App. Mech.

Paper No. 8 4 - W A / A P M - l O , 1984.

48. Ponter, A.R.S 'The i n c r e m e n t a l S t r a i n G r o w t h of

Cocks, A.C.F. E l a s t i c - P l a s t i c B o d i e s S u b j e c t e d to High

Levels of Cycl i c T h e r m a l Loading'. ASME

J. App. Mech. P a p e r No. 84-WA/APM-ll.

49. Goodman, A.M. 'Incremental P l a s t i c D e f o r m a t i o n of a

C y l i n d e r S u b j e c t e d to C y clic T h e r m a l

Loading'. Non L i n e a r Pr o b l e m s in Stress

Analysis. Ed. P. Staley. (Applied

S c ience P u b l i s h e r s Ltd.), 1978 pp 317-

344.

209
50. Cocks, A.C.F. 'The P l a s t i c B e h a v i o u r of C o m p o n e n t s

Ponter, A.R.S. S u b j e c t e d to C o n s t a n t Primary S t ress and

Cyclic S e c o n d a r y Strains'. J. Strain.

Anal., 1985, V o l . 20, No.l pp 7-14.

51. Lebey, J. 'E x p e r i m e n t a l T e s t s on R a t chet of

Roche, R.L. S t r u c t u r a l Elements. Diagrams for

Consseran, P. P r i m a r y T e n s i o n and S e c o n d a r y T w i s t . ' I

Mech. E. C o n f e r e n c e on Creep, 1980,

Paper C99/80.

52. Smithells, C.J. 'Metals R e f e r e n c e Book'. Butterwoth, 6th

Edition (revised), 1983.

53. Hardy, G.H. 'Inequalities'. Cambridge University

Littlewood, J.E. Press, 2nd Edition, 1952, 1973 printing.

Polya, G.

210

Vous aimerez peut-être aussi