Caso Rotura

8/21/2019 Caso Rotura

http://slidepdf.com/reader/full/caso-rotura 1/200

* *ISSN 1018-5593

European Commission

phys ica l sc ien ces

High -temp era ture c rack growthin s team turbin e mate ria ls

Report

EUR 14678 EN







' „ ^ s j ; ' * - .•»i?»'*

j »



European Commission

phys ica l sc iences

High -temp era ture c rack growthin s team turbin e materia ls

J . EwaldSiemens Power Genera t ion Group KWU

Mülheim-Ruhr — Germany

T. HollsteinFraunhofer Inst i tut für Werkstoffmechanik

Freiburg — Germany

G. A. Webster and F. DjavanroodiImper ia l Co l lege of Sc ience and Techno logy

London — United Kingdom

S. R. HoldsworthGEC-Alsthom Turbine Generators Ltd

Rugby — United Kingdom

Edited by:

J . B. Marriott, Secretariat for COST 501 and COST 505,JRC, Petten, The Netherlands

Supported by the

European Commissionthrough Contract Nos

COST 0032-D.. .1COST 0014-D.. .2.3

COST 0015-UK...4 / July 1992

5

AÍ çhot AltZbr

Science, Research and u]e>ftMtøe£UR0P. Biblìotfl.D ¡ rectorate-G e n IfãrTTT""

1994 ( S j ç EUR 14678 Erv

cl ~



Publish ed by theEUROPEAN COMMISSION

Directorate-Genera l XI I ITe lecommunica t ions , In format ion Market and Exp lo i ta t ion of Research

L-2920 Luxembourg

LEGAL NOTICENeither the European Commission nor any person acting on behalf

of the Commission is responsible for the use which might be made of thefollowing information

Cataloguing data can be found at the end of this publication

Luxembourg: Office for Official Publications of the European Communities, 1994

ISBN 92-826-7536-X

© ECSC-EC-EAEC, Brussels • Luxembourg, 1994

Printed in Italy



Abstract

Modern steam turbines must retain a very high reliability throughout their service lifeof typically 200 000 hours, which in p ractice extends over m ore than 25 years. One of thefeatures which must be considered at the design and manufacturing stages and during theassessment of fitness carried ou t periodically during the service life is the growth of themanufacturing type defects at temperatures up to about 550°C.

Within the concerted action research programme CO ST 505, Materials for SteamTurbines one coordination group studied this problem with a view to enabling a moreaccurate evaluation of defect acceptability on the basis of data gathered using laboratorytest-pieces. The work of the group was structured under three headings which form thethree Parts of this Monograph:

* Part I - Cree p Crack Initiation and Growth in terms of K* Part n - Creep Crack Growth in 1 % C T M O V steel and Alloy 800H -

an evaluation of the results of the COST 505 and an EGFRound Robin

* Part HI - High Tem perature Fatigue Crack Growth in Steam TurbineMaterials.

In each part solutions are given according to the current state-of-art. None can beregarded, however, as giving a well established methodology for practical application.Further data will have to be determined together with results from complex, simulativebenchmark tests which remain to be conducted, before th ere will be sufficient critical

evidence upon which to base general rules for practical applications.





HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS

~ Preface

Materials for power engineering applications have been important for various concertedaction research programmes throughout the 20 year history of COST. Within this field,one of the areas to which specific attention has been devoted in recent years has beenMaterials for Steam Turb ines; COST 505 . The countries represented in this programm e

were Austria, Belgium, Denmark, Finland, Germany, Italy, Sweden, Switzerland andUnited Kingdom together with the Joint Research Centre of the Commission of theEuropean Communities. Jointly, organisations from these countries tackled a range ofproblems concerned with the improvement and reliability of steam turbines.

One of the coordination groups containing the 7 organisations Usted in Table A on page7 was involved with high temperature crack growth. They had the overall objective to

enable a more accurate evaluation of defect acceptability in full scale power plantcomponents on the basis of data gathered using laboratory test-pieces. The defects wereunderstood to be those associated with the manufacturing process rather than defectsinitiated by creep exhaustion.

The work of this group was structured under three headings:* creep crack initiation and (early) growth* creep crack growth* high tem pera ture fatigue crack growth

These subjects formed the basis of reports in which the work was reviewed and evaluated

in the light of presen t day literatu re and experience in studies which were conducted withsome financial assistance from the Commission of the European Communities. Theround robin work in the second heading was also integrated into a task being undertakenby the European Group for Fracture (now ESIS) and reported by them in 1990.

The three constituent pa rts which make u p this Monograph a re independ ent in terms ofchapters, literature references, table and figure numbering. They are , however, precededby a common Executive Summary and an integrated list of the symbols and abbreviationsused throughout

Acknowledgement must be given to the input which all of the participants made

throughout the investigations and also when the reports were being prepared. Theassistance of Dr. S.R. Holdsworth with the preparation of the list of symbols is alsogratefully acknowledged.

J.B. Marriott

Secretariat for COST Projects 501/505EUCO/MST/01/92JBM/tp/1/1397Petten, July 1992

V -





Contents

Executive summary 1

List of symbols used throughout 11

Part I Creep Crack Initiation and Growth in Terms of K 21

1. Basic considerations upon the use of K 23

2. Crack tip/farfield models 27

3. Test results on creep crack growth 33

A . Discussion of results 38

5. Conclusions 43

References 44

Table 46

Figures 47

Pa rt II Creep Crack Growth in 1% CrMoV Steel and Alloy 800H -an evaluation of the results of the COST 505and an EGF Round Robin 69

1. Introduction 71

2. Material 73

3. Specimens and test procedure 76

4. Param eter determination 78

5. Num erical investigations 79

6. Results 83

VII



7. Discussion 84

8. Con clusions 91

References 92

Tables 96

Figures 104

Pa r t H I High Tem peratu re Crac k Growth in S team Turb in e M ater ia l s 129

1. Introdu ction 131

2. Fatigu e crack growth regimes 131

3. High tem pe ratu re fat igue crack growth correlat ion pa ram ete rs 132

4. Low strain fatigue 134

5. H igh stra in fatigue 138

6. High tem pe ratu re crack growth in we ldmen t micro structures 139

7. G en era l observat ions 140

8. Con clusions 141

References 143

Tables 145

Figures 147

VIII



COST 505

HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS

Executive SummaryJ. Ewald. Coordinator

Introduction

Within COST 505 Materials for Steam Turbines , the High Temp eratu re Crack GrowthWorking Group contained 7 research groups, Tab le A. They investigated 1% CrMoV,2 1/4 Cr 1 Mo and 1 Cr 0,5 M o steel forgings, castings, and p ipes with respect to staticcreep crack initiation and growth and to cyclic crack growth, between 530 and 550°C,Table B. In addition to the mechanical testing, 2 groups (D22 and UK18) performedfinite element analyses to control the validity range of C application and to investigatewhether cracked specimens creep under plain strain or plain stress conditions.

Principal Findings

The m ain conclusions which can be drawn from the results and evaluations of the workof the group can be summarised as follows:

In spite of the undoubted plane strain situation which exists near to the crack tip,the overall structure is exposed to creep under a plane stress condition, whichmeans that the load line displacement rate of the specimens - which is decisivefor the amount of C* - is dominated by plain stress behaviour.

The behaviour of the specimen within the range of the tails and during the earlycreep crack stage may be described in terms of K with additional geometry factorssuch as a crack tip/farfield ratio.

The real creep crack behaviour with higher creep crack rates can best bedescribed by means of C*2 which is a version of C* including the load linedisplacement due to creep. However, it is probable that these displacement ratescan rarely be measured in service for low stress, long time loadedspecimens/components.

Cyclic crack growth is accelerated by hold time. The hold tim e cyclic crack growthrates lie between the cyclic crack growth and the static crack growth curves.Numerical accumulation rates

da/dN total = d a / d N ^ + da/dN ,creep

based on an equivalent AK, considering the cyclic crack closure effects and C forhold time effects were investigated and used to describe the cyclic crack growthbehaviour.

- 1 -



The solutions given in of this Monograph have been obtained according to the currentstate of the art, but none can be regarded as giving a well established methodology forpractical application. In the area of static high tempera ture crack growth further data willhave to be determined for specimens with different shapes and sizes under loading

conditions which lead to long crack initiation times and low crack growth ra tes for m oreor less creep ductile materials. A broad data base will have to exist before the finaldecision can be made about the most economic and appropriate evaluation method.

With respect to high temperatu re fatigue crack growth the situation is similar. Th ere aremany data, a considerable level of mechanistic understanding and various proposals forevaluation methods, but even here there remains a lack of critical evidence upon whichto base general rules for practical applications. Furthermore, it is necessary to have agood data base for creep crack growth because according to the proposed accumulationrules for load controlled cyclic crack growth with hold times both static and dynamictypes of data are required.

However, there still remains the complex loading condition of strain controlled cyclesarising from therm al stresses with superimposed stress relaxation. This type of behaviouris frequently met in service and so far it has not been possible to model the behaviourusing relatively simple laboratory tests. At p resent the only way to obtain information isby performing expensive tests which closely follow the actual load cycles.

These conclusions have been established on the basis of three studies, which wereconducted with partial financial assistance from CEC, DG XH-Gl, to review the workof the Group and to examine separate aspects of the field. The summaries of thesereports follow in the order:

Part I - Creep crack initiation and growth in term of K by J. Ewald;

Part n - Creep crack growth in 1 % CrMoV steel and Alloy 800 H - an evaluationof the results of the OCST 505 and E G F round robin by T. Hollstein,G.A Webster & F. Djavanroodi;

Part HI - High tem pera ture fatigue crack growth in steam turbine materials by S.Holdsworth



P a r t i

Creep Crack Initiation and Growth in Terms of KJ. Ewald

Siemens, Power Generation Group KWU, Mülheim-Ruhr, Germany

It is clear that two schools of thought exist about the selection of the appropriateparameter for creep crack initiation and growth. There are the protagonists of C*, whofeel supported by the fact that C* is the loading param eter which is reasona ble from thephysical point of view. On the other hand there are the people who try to apply creepcrack initiation and growth data in terms of K for the description of components whichhave thick walls, long loading times, and low loading stresses. With respect to crackinitiation it can be concluded:

Ky,, (the fictitious or ideal elastic stress intensity factor) seems to be basicallyusable to describe the crack tip situation for crack initiation.

The time for crack initiation increases with decreasing crack tip driving force(K lid), Fig. A.

Specimen size and shape determine the damage m ode and the related specimenbehaviour. Therefore, it is necessary to use Kj¡d together with the parameter Gn

(nominal stress or net section stress) to describe the farfield loading situation,because within the creep range time dependent changes in stress distribution,exhaustion and damage occur both at the crack tip and in the farfield (ligament).

Consequently, crack initiation can be described basically by means of a twocriteria diagram for creep crack initiation which covers the rang e of the tailsfrom å = f (K Iid) plots and which is able to demonstrate the influence of differentcrack tip/farfield ratios K E d/G n.

Creep ductility dramatically influences the creep crack initiation and growthbehaviour. Notch weakening m aterials should be avoided, since they tolerate onlyextremely short defects. The related data base for these materials is not yetavailable.

The creep crack growth behaviour when there is only a limited crack increment may bedescribed by plots of K,¡d over t with param eter lines for a, similar to Fig. A. However,with such diagrams, which are only valid for CTl-specimens, it is hardly possible toestimate the influence of specimen shape and thickness. Thus, crack growth with smallincrements should be rated with diagrams like Fig. B which shows the influence ofspecimen shape and specimen thickness, e.g. lateral constraint If it is necessary todescribe the behaviour of cracks of greater depth, the crack tip/farfield method with V0

= Kjjd/G,, or some kind of description by means of C* may be used. This aspect isconsidered in greater detail in Part II of this Monograph.

- 3 -



The most important point for the future is to generate further data on specimens withdifferent shapes and sizes with long crack initiation times and low crack growth rates.This would allow a final decision to be made about the most economic and appropriateevaluation method.

P a r t i i

Creep Crack Growth in 1 CrMoV Steel and Alloy 800H - an eva luation of the resultsof the COST 505 and an EGF Round Robin

T. Hollstein*, GA . Webster**, F. Djavanroodi**

* Fraunhofer Institut für Werkstoffmechanik, Freiburg, Germ any

** Imperial College of Science and Technology, London , UK

Experiments and analyses have been performed on a 1% CrMoV steel and on Alloy800H which have shown that creep crack growth in these materials is described mostsatisfactorily by the creep fracture mechanics parameter C*. Recommendations havebeen made about how to obtain the most reliable estimates of C* from experimentalmeasurements. These have been supported by numerical computations, Fig. C. It hasbeen found that the use of 20% side grooves (10% of the total thickness each side) incompact tension specimens, and a seven-point polynomial fît to obtain crack growth anddisplacement rates, produces the most consistent correlations.

It has been dem onstrated for the 1% CrMoV steel that all the cracking took place underplane stress conditions, Figs. C and D . Increased scatter, due to 'tails' in the early stagesof cracking, Fig. D , has bee n shown to be caused mainly by the progressive build up ofdamage at the crack tip until a steady state distribution is reached. This can take up to30% of the life of a specimen and can be important in practical applications. However,little evidence of a 'tail' was noticed for Alloy 800H.

Comparisons have been made with data obtained in other test programmes on the 1%CrMoV steel. These have reinforced the findings of this investigation.

Part III

High Temperature Fatigue Crack Growth in Steam Turbine MaterialsS.R. Holdsworth

GEC Alsthom Turbine Generators Ltd., Rugby, UK

Th e high temperature fatigue crack growth (HTFC G) p roperties of a number of steamturbine m aterials have be en reviewed. Many of the results were gathered by membersof the COST 505 High Temperature Crack Growth Working Group who were active

- 4



during the period 1985 to 1988, but additional information has also been collated fromthe published literature to complete the overview.

High temperature low and high strain fatigue crack growth rates may be considered interms of two components, one due to cyclic loading and the second due to creep, ie

da/dN total = d a / d N ^ + d a / d N ^

The da/dNjydj,. term is a power law function of A K ^ (the equivalen t cyclic stressintensity) and may be influenced by crack tip damage due to oxidation and prior creeploading. ăa dNmep is expressed in terms of the C* parameter.

At 530/550°C and for a range of steam turbine forging, casting and pipe steels, there is no apparent effect of creep on low strain fatigue crack growth rates for frequencies in excess of 1Hz.

HTFCG threshold AK levels increase with decreasing frequency due to oxide blocking and crack closure whereas at somewhat higher AKs, low strain fatigue crack growth rates increase with decreasing frequency due to enhanced oxide growth.

For frequencies less than 1Hz, da/dn,otai is still mainly influenced by oxide assisted growth effects on d a / d N , ^ at relatively low AKs. However, with increasing AK, the role of oxidation diminishes and creep becomes more important through its effect on da/dN^y,. and its contribution in terms of d a / dN^ ^ .

In a simple engineering model d a / d N ^ ^ for a given AKeq, increases to a maximum as the crack tip damage zone develops to the critical condition necessary for th e onset of creep crack growth. The prior hold time necessary to achieve this peak acceleration increases with increasing creep ductility.

At frequencies below 13Hz when the cycle is in load control, fatigue crack growth rates

are determined by the rate of creep crack growth (ie the magnitude of da/dN^y,. is small relative to that of åa/äSmcp). This is not always the case when HTFCG is due to a strain controlled cycle. In these circumstances, stress relaxation occurs and da /dN^^ , is no t the dominant crack growth component until still lower frequencies.

The resistence of a material to creep-fatigue crack growth is strongly influenced by creep ductility. The magnitudes of both cyclic and creep crack growth rate components are lower for steels with high creep ductility.

Prior long term thermal exposure (without load) has no influence on high temperature fatigue crack growth rates, at least while th e ageing treatment is no t responsible for a significant reduction in creep ductility.

In load controlled tests, high temperature fatigue crack growth rates through the weld heat affected zone (HAZ) of a cast lCrMoV steel are faster than those through the parent material at frequencies of around 0.1Hz. The difference in cracking rates increases dramatically with the introduction of a hold time at peak load, coinciding with

5 -



a change in fracture path from the partially transformed intercritical HAZ to the fullytransformed coarse grain HAZ immediately adjacent to the fusion boundary.

High temperature fatigue crack growth rates through the weld HAZ and parentstructures of a cast 12CrMoV steel are similar in both low frequency continuous cycleand peak load hold time tests. For the test conditions investigated, all fracture paths inweldment tests are contained within the soft sub-critical HAZ.



Table A

Research Projects

No.D 20/21

D 22

D 35

1 3

SF 2

UK 5

UK 18/22

OrganisationSiemens

IWM Frbg.

AEG

ENEL

V i l Espoo

GEC-A

Imp. Coll.

Project ResponsibleBerger/Ewald

Hollstein

Kanbach

Ragazzoni

Rintamaa

Holdsworth

Webster/ Djavanroodi

- 7



Table B

Materials/Tests

ProjectNumber

D 20/21

D 22

D 35

1 3

SF 2

U K 5

UK 18/22

1% CrMoV550°C

RoundRobin

X

X

X

X

X

X

X

Additional Materials

1% CrM oV; 530°C

1% CrMoV ; 550°C

1% CrMoV cast; 530°C

1% CrMoV (used rotor;540°C)

1% CrMo V (used rotor;550°C)pipe mat; 550°C

2 1/4 G 1 Mo cast; 538°C

1% CrM oV; 550°C

crack growthstatic cyclic

X X

x (+ analyticalwork)

X X

X X

X

X X

X X

(analyticalwork)

8 -



scatterband, makroscop. Creep Crack Initiation Siemens 1989

100 1000 10000 loading t ime ( h )

100000

Fig. A Creep Crack Initiation, 1 CrMoNiV 530°C, CT25S

Specim.Thickn. (mm) D X

+ V Y D A O o O

CT 25/50S , a/W = 0.55 CT 50/1 OOS , a/W = 0.55

CT100/200 , a/W = 0.55 0 9 , a/W = 0.40 D60 , a/W = 0.40 D60 , a/W = 0.20 D60 , a/W = 0.10

CT12.5 , a/W = 0.50 CCP , a/W = 0.50 SENT , a/W = 0.20

1%CrMO(Ni)V,T = 550'C

D.60

A A 6

$ 1 0 0

0 X ¡ Q \ Hollstein

' 2 5 / C T 1 2 . S -D^-^Jfom a ,

50 [ 3 ]

10 10 0 1000 loading t ime t¡ ( h )

10000

Maile/Tscheuschner |Siemens|Ragazzoni| Hollstein |Holdsworth|Rintamaa c * + A Y O A

0.004W(CT),0.01W(OENT) spark eroded

B •0,5mm

a O • 0,5mm

CT25 - , CT5 0 X

■ 0,3mm

D - O '0,2mm

Fig. B fatigue precracked

a o - 0,5mm

Creep Crack Initiation for different specimen thickness and size



2ICrMoNiV57. T = 550°C CT12.5M0 specimen AB2B

tH R

8 -

§ °

pi strain (à 10) pi . stress (á=0) / pi. strain(àsO)

y ¿. s

8 - | — , — | — i — i — i — , — i — i — i — i — i — i — i — i — I I I I I I T~

°0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 B.00 9.00 10.00 11.00

t . se c CE+06)

F i g . C Exp e r i m e n t a l a n d n u m e r i c a l e va l u a t i o n o f C * f o r C T 12 .5 / 40 spec imen A B 2 B , w i t h a n d w i t h o u t c r a c k g r o w t h

Q B D 3 , B D » , B D S . B W 1 , B W 2 , B W 3 , A H J . A H * , A O S , A Q 5 , B C l , B C 2 , A N Í , A E l , A E 2

Q A F Î . A B 3 . A B R A M I , A H I , A l l , A I 2 , A B S , A B Í , A E S , A C 1 - 6 ( C T I 5 / S 0 )

S A N 1 . A N 3 . A N « , A 0 3 ( C T 2 0 / 4 0 ) • RR 5 (SENB 9 .5 /191

O AQ1 .AQ2 ( C N I 2 . S / S 0 )

+ ABZA .AB1B ( C P ï . S / « a )

M A P I . A P 2 . A P 3 ( C T 5 0 / 1 0 0 )

D AM1 .AM2 ( S E N T 1 2 / 2 0 )

X BE31.BE32 ( C T 1 0 / 2 0 ) ' *

♦ BB 1 ( C T 6 Ï . S/137)

o C C P 5 .C C P Í (CN2S /S0 )

. A A1G.A17 ( C Z 1 5 / Ï 0 )

■ RR 7 ( C T 13 /16 )

*t / S

-• <■

CT j / f c n 2« ]

F i g . D C r a c k g r o w t h r a t e â i n 1 C r M o V s t ee l a t 550°C as a f u n c t i o n o f C * .

10 -



List of symbols and abbreviations used throughout the Monograph

symbol or abbreviation

a

a0

Aa

Aa¡

á

â.

A

A,

(bk

b(t„, er)

B

B .

B .

C

C(f, t„, er)

C '

meaning

crack length

initial crack depth

crack extension

crack initiation criterion

crack growth rate per unit time

initial creep crack growth rate

steady state creep crack growthrate .

uniaxial elongation at fracture

impact energy

remaining ligament (w-a)constant in creep crack growthEquation.

function of hold time and ductilityin creep crack growth expression

specimen thickness

effective B = B - (B-BJ2/B

net section thickness

constant in Paris FCG law

function of frequency, hold timeand creep ductility in d a / d N , ^law

constant in Norton minimum creeprate law

Part where used

i;n;mI;H

n

i

I;II; da/dt usedin IIIn

n

II ; El also usedb y n

n

n \m l

m

I;II;m

n

i;n

m

ra

n;m

- 11




C*

C*

c*2

CCP .CN \-

CNT J

CT

CTOD

ACTOD

ACTOD0

da/dN

d a / d N ^

d a / d N ^

da/dN,,^

da/dt

dr

ds

DENT

meaning

parameter characterising stress and

strain rate fields at crack tip increeping solid

C* determined using Kumar-Shinbased solution

experimentally determined C*

centre notched specimen

compact tension specimen

crack tip opening displacement

cyclic CTOD

threshold ACTOD

crack growth rate per unit cycle

da/dN due to creep loading

da/dN due to cyclic loading

total da/dN due to both creep andcyclic loading

crack growth rate per unit time

crack tip ligament

arc length on r

double edge notched tensionspecimen

Part where used

I;II;m

I

I; included inC byü

i;n

i;n;m

m

m

m

m

mm

m

lu; å used ini;nn

nr,n

12 -




E

E '

El

f

F

FCG

G

gi(7w,ß)

gi(7W,n)

H

HAZ

HSF

HTFCG

In

ICHAZ

meaning

elastic modulus

elastic modulus includinglateral contraction (i.e.including pi. e)

uniaxial elongation atfracture

frequency; dN/dt

load

fatigue crack growth

elastic strain energyrelease rate

geometry function inKumar-Shih J solution

geometry function inKumar-Shih based C* (orC*i) solution

constant in cyclichardening law

heat affected zone

high strain fatigue

high temperature fatiguecrack growth

non dimensional functionof n

intercritical HAZ

Parts where used

m

i

II; A also used by

m

I;H; P used in m

mn

m

I;ffl

m

m

m

m

n

m

i;n

13 -




J

AJ

k,

K

AK

Kc

K *

A K ^

Kt

K M

K W o

Kno

Knm

Kniin

A K C

A K «

meaning

path independent lineintegral characterising

stress and strain fields atcrack tip in plasticallydeforming solid

cyclic J

theoretical elastic stressconcentration factor

elastic stress intensityfactor

cyclic K

apparent elastic K

critical value of K leadingto instability

effective K

cyclic K ^

initial K

fictitious (ideal) K

K M at a0

creep crack initiation K

maximum K

m i n i m u m K

equivalent A K

total AK

Parts where used

i ;m

m

I

I;II;m

mIH; referred to as KM

i n i

m

I

m

H; referred to as K ^ in I

I; referred to as KA in UI

I

I; referred to as K¡ in II

i n

m

m

m

14




AK„

LSF

M

WJt» O

n

N

P

* D E

* D S

^ m a x

min

Po

q

q»

r c

R

R '

meaning

fatigue crack growththreshold A K

low strain fatigue

exponent in Paris law

exponent in da/dN cycliclaw

stress exponent inNortons (minimum creeprate) law

number of cycles

load

load at end of dwellperiod

load at start of dwellperiod

maximum load

minimum load

crack opening load

exponent in creep crackgrowth rate law

effective load range

creep damage zone size

load ratio (R=KmJKaJ

geometrical variablerelated to crack tip -farfield condition ie.R' =( K 2 / 0

Parts where used

m

m

m

m

I;II;m

n i

HI; F used in I;II

m

m

m

m

m

m

m

n

HI; also used for radialdistance from centre ofbar in H

I

- 15 -



sym b ol or ab b r e v i a t i on

R "

RA

Rpo.2

Rpi

R »

Rff

RK

R«,«

S E N B

S E N T

S C H A Z

t

»d

tf

»h

ti

t .

' i

m e a n i n g

ac t i vat i on e ne r gy t e r m

uni ax i a l r e duc t i on o f ar e aat fracture

0 . 2% pr oof s t r e ngth

1 % pr oof s t r e ngth

te ns i l e s t r e ngth

stress rat io (o^JR^

stress intensi ty rat io(Kftjo/Km))

c r e e p r uptur e s t r e ngth

s i n g l e e d g e n o t c h e d b e n ds p e c i m e n

s i ng l e e dge notc he dt e n s i o n s p e c i m e n

subcr i t icai HAZ

t i m e

t ime at which crack t ipc r e e p dam age f i r s t for m s

t ime to fai lure

hold t ime

time at w hic h crack tipc ondi t i ons ar e r e spons i b l efor creep crack ini t iat ion

t ime to rupture

tr ans i t i on t i m e

P a r t w h e r e u s e d

n

H; Z also used by II fors a m e p u r p o s e

i;n

n

n

i

i

I; re ferred to as a 0 in II

I;II;m

n

m

I;ü;in

m

II; referred to as tr in I

m

I;ÜI

I; referred to as t f in II

i;n

- 16




T

T¡

ú ¡

U*

5U*/ ía

V

v£

vL

AVL

v.

v.

w

w*

X

Y

Z

ß

e

meaning

temperature

traction vector (T¡ = a^np

displacement rate vector

energy term

energy dissipation rate

load line displacement

crack openingdisplacement


cyclic VL

crack tip farfield ratio^ o = ^ D d o / f f i i o a e b e l


rate due to creep

specimen width

stress work ra te

distance

compliance/geometryfunction in K solution

uniaxial reduction of areaat fracture

exponent in cyclichardening law

strain

part where used

n

nn

n

n

I; H; referred to as VL

m

m

Iu ; referred to as V in

m

I

i;n

i;n;ra

n

n

m

H; RA also used by H

m

i;n;m

in

I

17




e c r

« o

e r

« * f

«*.

A e p

A e T ( 0 P )

é

¿ c r

¿ S

'min

« « f

de/dt

D

Oc

a

meaning

creep strain

strain parameter inprimary term of creepstrain expression

uniaxial creep ruptureductility

multiaxial creep ductility

ductility exhausted in theligament prior to the

arrival of the crack

cyclic plastic strain

nominal total outer fibrestrain range

creep strain rate

creep strain rate

strain rate tensor

minimum creep rate

creep strain rate atreference stress inuniaxial specimen

creep strain rate

factor from limit loadanalysis

factor depending ontestpiece geometry

stress

Part where used

n

n

n;m

n

n

m

m

I;II; referred to as ¿„inn and de/dt in m

II;see comments above

n

n

I

III; referred to as £ ini;n

n

i;n

i;n;m

18




La

ff STM

» .

»«*

a n ã c b c l

» u n d i c i

°0

°ra

aY

r

meaning

cyclic stress

nominal stress as definedby ASTM

nominal stress forcomponents, net sectionstress for specimens

net section stress

nominal stress accordingto Siebel

a„ úax¡ for initial crackdepth a„

plain specimen rupturestrength

reference stress

yield strength

integration path aroundcrack tip connecting lower& upper crack face in acounter clockwisedirection

Part where used

m

I

I; for tensile specimens,alternatively referred toas ona in IÜ

HI; see comment above

I

I

U; referred to as R,,,, in I

I

m

n

- 19





HIGH TEMPERATURE CRACK GROWTHIN STEAM TURBINE MATERIALS

Parti

Creep Crack Initiation and Growth in Terms of K

J. Ewald

Siemens, Power Generation Group KWU,

Mülheim-Ruhr, Germany

21





1.r Basic considerations upon the use of K

Creep crack initiation and growth has been broadly described by

means of C*, which is a path independent line integral round

the crack tip and/or the energy dissipation rate for a specimen

with a fatigue crack, creeping under steady-state creep conditions

[1,2]. This means that the specimen is assumed to have a completely

redistributed stress condition which enables it to have a steady-

state creep crack growth rate.

To determine C*, most researchers use the version of C*, which

incorporates the load line displacement rate due to creep, V ,

(this C* is designated below as C* 2) [3].

This gives the following formula:

C* ' = t e B (W-a) <')n

where F is load, B is net section thickness, n is a factor,

depending on test piece geometry and on the creep exponent n

of the Norton creep law.

Details of the numerical and physical background of C* and of

the determination o f ^ are described in [ 3] .

From the basic evaluations in [1] and from recent finite element

calculations [5] it is obvious that there exists a distinct

time at which a specimen reaches the steady-state condition,

this being the lower bound validity limit for C*. The equation for

fc1 = (n+1) C* Ë1" ( 2 )

includes the power exponent n of the Norton steady state creep

law in addition to two parameters describing the stress state

at the crack tip - K and C*. It is clear that the use of theNorton law for the description of the creep behaviour is a rough

simplificaction, because it is not able to describe the primary

creep range. A further problem which has also to be taken into

consideration is that the Norton exponent - basically defined

for the secondary creep range - changes with both stress and

test time and it is not possible to describe the increasing

stress redistribution portions which occur with ongoing time

within the secondary creep range by the Norton creep law.

- 23 -



Nevertheless, t ] is an estimate of the time from which C* can

be used. For t i m e s * t., the basic mathematical formulation

shows that C* is path dependent and then the use of C [4] or

simply K is allowed.

The practical application of the C* 2 integral is limited bythe current availability of'7 solutions for different geometries

[7,8]. This means that additional expensive analysis is required

for non standard configurations and complex components. In addition,

the measurement of V for real existing cracks in a component

or specimen loaded to a low stress and for a long duration is

rather difficult or even impossible because of the small displace

ment rates involved. It therefore makes sense to try to use

the fictitious stress intensity factor K T. . to describe thelid

conditions at the crack tip. This is done in spite of the knowledge

that its use is incorrect in physical terms. Nevertheless, K... ,

serves as a vehicle to describe the geometrical situation near

the crack tip for a specimen/component. However, because we

are considering creeping bodies, it is not sufficient to use

K-., as the only parameter to describe the crack tip behaviour

of the component.

A single parametric approach to describe the crack tip such

as K or J can only be used in the sub-creep regime where the

yield strength, which depends on temperature but not on time,

is the decisive material characteristic. Within the creep range

the situation is different because - as mentioned before - time

dependent changes in stress distribution occur both at the crack

tip and in the farfield (ligament). Consequently, exhaustion

and damage of the material have to be considered not only at

the crack tip but also in the farfield as time goes on.

This means that a crack tip/farfield concept must be employedto assess crack initiation and growth in the creep range. Models

covering the crack tip/farfield situation have been developed

and proposed step by step since 1979 [9-13]. A "two criteria

diagram of creep crack initiation" was developed in 1984 [10,11],

which can be used to assess the crack initiation behaviour of

ductile CrMoV grades of steel and which can also be adapted

to other steel grades.

24



Later in 1987 [12,13] attempts were made to describe the creep

crack growth behaviour in a consistent manner by a crack tip/far-

field ratio, K Ţ. J / G too. In [14] similar concepts to describe

the specimen behaviour in terms of K were developed.

It was clear from the very beginning that the use of K for speci-mens and components with large ligaments such as castings with

thick walls and rotors wich are subjected to long loading times

at relatively low stresses does not seriously affect the validity

conditions of the K-parameter. This was because, according to

[1], K may be used if the material is creep ductile, but the

creep zone, in spite of the long loading times, is small in

relation to the crack length and/or the specimen width (as a

result of low net section stress).

The other validity conditions for K which exclude the use of

C* are:

- the material is creep brittle (low creep deformation capacity),

- the material is creep ductile, but the creep zone is still

small due to short loading times ( t < t . ) .

For ductile materials the latter condition may also support

the trend to apply K, at least in the range of ongoing stress

redistribution and crack initiation.

The need to use a 2nd parameter for the farfield, which is the

nominal stress, is supported in total by 3 facts [11]:

1. The time dependence of the material characteristics, as mentioned

before.

2. The possible existence of different farfield stresses in

specimens of different sizes and shapes in spite of a constant

crack tip parameter K (Fig. 1) .

3. The finding from finite element calculations that the experi-

mental results were best fitted by assuming plane stress

conditions [3,5] (Fig. 2 ) . This means that in spite of the

undoubted existance of a plane strain condition within the

near crack tip region, the farfield with its plane stress

condition dominates the behaviour of the specimen.

The last finding gives an additional hint, why the C*2-version

with V (load line displacement due to creep) describes the

specimen behaviour best and why the other C*-version based on

the Norton's power exponent n:

- 25 -



C*, = a- n

n\ ,3)

does not fit the experimental results in a satisfactory manner.

The reason is that V describes the overall specimen behaviour

by combining the influences of stress and time at the crack

tip and in the farfield. Unfortunately, the C*2 = f (V )-version

is less practicable in long time/low stress situations, for

the reasons-outlined above.

In the following chapters the different crack tip/farfield models

for crack initiation and growth will be described and specific

data which were evaluated in the course of the work of the COST

505 High Temperature Crack Growth Group [15-20] or are available

in the literature will be summarised and their usefulness ex-

plained.

Basically it is the aim of these models to use K in a manner

similar to the employment of the elastically determined stress

concentration factors, K , in the plastic and creep range in the German design rules [9,26,27].

In order to distinguish this K used in the creep range from

LEFM K, it is called K T. , [19]. lid

With the use of K Ţ . , as the crack tip geometry describing factor

and the farfield nominal stress it is believed that the basic

stress features of a specimen/component are fixed so that it

should be possible to describe the creep behaviour of a specimen

or a component using the data of other specimens or components

which have the same stress features. In this case, it is un-

important that the use of K Ţ. , may not be correct in physical

terms over the whole loading range.

26



2.) Crack tip/farfield models

2.1 Basic considerations

As explained above, in addition to the application of K . , as

the parameter describing the crack tip situation, the nominal

stress within the farfield has to be taken into consideration.

According to ASTM E 616-82, the nominal stress within the ligament,

i.e. the farfield, of the usual fracture mechanics specimen

is determined as follows: ■

For symmetrically loaded specimens (DENT, C N T ) ,

ÂSTM = F / B ( W- a )- ( 4 )

For specimens with bending fractions, (e.g. C T ) , ÂSTM = F / ( B ( W _ a ) ) ( 1 3 (W + a)/(W-a)). (5)

In specimens with tensile loading and bending, the linear elastic

bending stress fraction is added to the tensile stress in accor-

dance with ASTM, Fig. 3. In tensile specimens, however, the

completely balanced (redistributed) tensile stress in the ligament

is used, Fig. 4; this is not consistent. As introduced by Siebel

[16], the bending stress should also be used as a fully redistri-

buted stress, which means that the linear bending stress fraction

must be divided by 1.5 (maximum general yield coefficient for

bending, Fig. 3 ) . The nominal stresses determined in this way

for specimens with bending fractions are referred to as O „. , ,, ^3 n Siebel

contrary to stresses according to ASTM.

At this point, reference must also be made to two peculiarities

associated with the use of K T. , and G~ „. , . as crack tip and lid n Siebel r

farfield parameters:

(1) With increasing specimen size, the K T. ,/G~ „. , , ratio ^ ^ lid n Siebel

increases for the same K I. , (similar relationships arise lid r

for the C*/S . ratio) Fig. 1. This means that large n Siebel — ^ 3

CT-specimens have lower farfield stresses for the same K T. ,, lid'

i.e., these specimens are more likely to exhibit crack tip

damage.

27



(2) Attention must also be drawn to the influence of crack growth,

A a, on the shape/geometry function for specimens with

differing sizes. It is clear that if, in geometrically similar

CT-specimens, the crack growth rate is constant, the increase

in a/W as a function of time is different for each specimen

size. Of course, different values of the shape function

result from different a/W ratios. If the change in shape

function with respect to time for given constant increments

in A a for three CT-specimens with different sizes is determined,

the curves shown in Fig. 5 result. The diagram also shows

that the increase of the shape function is approximately

the same for the different CT-specimens up to an a/W ratio

of 0.57, while it increases more rapidly for the smallerspecimens beyond this value. This is logical, as a crack

growth increment of 1 mm in the ligament of the large speci

men is "hardly noticed", while the ligament of a small specimen

is altered profoundly by such a change. To show the influence

of the shape function, the experimentally determined crack

growth behaviour of a real CT100-specimen was transferred

to CT50- and CT25-specimens, Fig. 6. All specimens were

assumed to have the same initial fictitious elastic stress

intensity factor K-. ,. Due to the different changes of theshape function, different slopes of the a = f (K_..) curves

are obtained for the different specimens.

2.2 Thinking model [9,10]

In the following, the significance of the two parameters K-. ,

and is explained by means of a model distinguishing between

ligament damage (related to Ç" ) and crack tip damage (related

to K...). In Fig. 7, both the creep strength of smooth specimens

and the creep crack initiation resistance in terms of K_., are

plotted independently as a function of the loading time.

If a component with a large flaw (high K... ,) was approaching

the end of its life (say at 10 h) due to a net section stress C" ,

then Fig. 7a shows that creep crack initiation would be caused

relatively early with subsequent creep crack growth into a low

pre-stressed farfield. This high K T. ,/ situation would lead3 lid n

to a failure mode of crack tip damage, promoting leak before

fracture.- 28 -



If the same component had a relatively small flaw (low K T. ,)

(Fig. 7 b ) , creep crack initiation with subsequent crack growth

into a highly pre-stressed farfield would only occur at a very

late stage in life. This would lead to a ligament damage mode

with low K . ,/Œ ratios. When this mode of ligament damage is dominant, the first indication of failure of the component would

be given by the overall component deformation. This final failure

could, however, occur in a sudden manner.

2.3 Two criteria diagram [10,11]

The correlations which are based on the model explained above,

may be summarized analogously in a two criteria diagram for

creep crack initiation . Such a diagram is shown in Fig. 8.

It describes the decisive damage modes for a material with

sufficiently high creep deformation capacity. (The creep defor-

mation capacity should be adequately high if the material suffers

no notch weakening). The diagram has normalized axes with the

respective time dependent materials characteristics (R(j- = b n c /Rm t

and R_, = K T . , / K T T r J . C indicates the nominal stress in the K lido IID no farfield of a precracked component, Fig. 9, or the net section

stress in a specimen. R is the creep rupture strength of smooth

tensile specimens. K T. , is the fictitious elastic K-value at lido the crack tip within the component, Fig. 9, or a specimen.

K Ţ Ţ denotes the creep crack initiation value of the material,

which is a kind of material characteristic. It depends, however,

on temperature, shape and size of the specimens. Because K--

describes the crack tip damage behaviour, it has to be evaluated

from specimens with deep cracks, which means specimens with high ratios KT. , / C , such as is found in CT 1-specimens.

^ lido no r

The value 1 on the abscissa, Fig. 8, indicates that the fictitious

elastic stress intensity K.. ,, which is obtained in a component,

corresponds to a materials macroscopic creep crack initiation 4

value, K T T , for a certain time, e.g. for 10 h. The respective IID 4

net section stress normalized with the creep strength for 10 h

is plotted on the ordinate.

The diagram distinguishes between three fields of damage, which

are separated by ratio-lines K . ,/G . The meaning of the ratios

has been explained by the thinking model (Section 2.2) .

- 29 -



The damage modes are:

- ligament damage;

- crack tip damage;

- and a mixed damage.

The ratio lines of K T. ,/5" in Fig. 8 were estimated on the basis oflid n — 3

available results. Taking account of the different fields of the

diagram it is possible to estimate the failure mechanism to be

expected each time. Crack initiation can only be expected outside

of the hatched boundary for the respective damage mechanism.

Inside different stages of exhaustion and increasing microscopic

damage are expected, Fig. 10. For such estimation, the time

dependent materials data R . and K_Tr> are required.mt IID

Special attention has to be paid to the creep deformation capacity,

because it influences strongly the K T. ,/5" -ratio-lines and thelid n

whole structure of the diagram. Fig. 11 shows as an example

a schematic "two-criteria diagram" for a notched weakening 1 %

CrMoV material. The necessary data are not yet available to

support and quantify it. We can learn from this diagram, however,

that ligament damage is not to be expected in notch weakeningmaterials except for crack free ( K T J -> 0) components. The most

important damage mode is crack tip damage but with earlier crack

initiation than for creep ductile materials (the latter cannot

be seen from the diagram, but it results from the fact thatKTTr. , .... < K T , _ , ... for a certain time).IID creep brittle IID creep ductileAnother important fact is the influence on creep crack initiation

of the lateral constraint for specimens, resp. components of

large thickness/size. This can greatly influence the crack initi

ation time of specimens with equal K x. , and K T. ./G" -values.

^ ^ lid lid nThus, the effect of the constraint should be taken into account

by differing materials K T---values (see chapter 4, below).

2.4 Description of crack growth by a crack tip/farfield concept [12,131

Although the description of crack growth by means of the C*-

integral results in narrower scatterbands than the comparable

description with K . , [ 4] , the description in terms of C* 2 will

not be further discussed in this paper because of its inferior

practicability, as outlined above.

30



According to Fig. 12 [14], small scale specimens show a tendency

to faster crack growth than larger ones, with the crack tip

farfield ratio, KT . ,/<S c u i < increasing with specimen size. lid n Siebel = c

In accordance with the finding [9,10] that different specimen

shapes must also have different crack tip/farfield stresses,

Fig. 1, the crack growth results were normalized by the crack

tip/farfield ratio, V = Kx. , /(T „. . ,, of the initial crack o lido' no Siebel'

[12]. For this purpose, the stress intensity factor was plotted

versus the crack tip/farfield ratio of different specimens for

a constant crack growth rate of, say á = 1 x 10 mm/h.

Fig. 13 [12,13] shows such a plot for a creep ductile 1 % CrMoV

steel at 600 C. The diagram shows the stress intensity factor

over the crack tip/farfield ratio V for crack growth rates, • -3 • -3 ° a = 1 x 10 mm/h and a = 5 x 10 mm/h. Joining the data points of this diagram results in a horizontal line for low crack tip/far-field ratios V ; thus there is a constant K T. , for specimens

o' lid ^

with different (but low) crack tip/farfield ratios. The value

of K Ţ . , then rises with growing V ratio and is expected to

reach another plateau for high V ratios.

It might be expected that the mechanisms of damage defined in

the two-criteria diagram, i.e. crack tip damage, ligament damage, mixed mode damage, could be identified again in this diagram.

This is indeed the case.

Creep crack growth at a crack-like defect is the next step in

the sequence of events after the creep crack is initiated. Thus, a specimen with a distinct K T . , /G ratio wich gives rise to ^ lido no = the appropriate damage mode will continue to have crack growth

with the same specific mode. Therefore, the crack tip/farfield

ratio should be the dominating quantity in the crack growth

diagrams depicted in Figs. 13-15. According to Figs. 14 and 15, for the ligament damage and crack tip damage plateau regions,

creep crack growth is found to be independent of geometry and

specimen width ( K Ţ - J = const.). An influence of specimen width

is only found in the regime of mixed mode damage, which can

be described in the example of Fig. 13 by the relation of K_. ,

= 100 V + const. The transition point between ligament damage

and mixed mode damage appears to be at V 2.5, that of mixed

mode damage to farfield damage at V *- 6.5.

- 31 -



As further evaluations of this type show [12,13], the slope

of the relation, K.. , = m V + const., is a function of

- the material conditions, i.e., whether it is creep ductile

or not,

- the temperature, and

- the class of materials (1 % CrMoV; 12 % CrMoV).

It can be derived from the evaluations shown above that the

two parameters K-. , and G may be applied to describe the crackgrowth behaviour of laboratory specimens of various sizes and,

ultimately, also of components on the basis of the ratio V =

KIido/'''ño Siebel' I n ' 1 3 ' such a method which needs further

confirmation is described.

However, some boundary conditions have to be observed:

Specimens and/or components

(a) have to be in the same range of applicability of K T., Æ " „. , ,

? ft- i lido no Siebel

with regard to the main modes of damage. While the modes

of failure are probably clear in the areas of ligament damage

and crack tip damage, ductile large specimens (DENT with

65 x 100 mm cross section, Fig. 15) lying within the mixed

mode regime can show normal cracking behaviour side by side

with crack branching, which indicates superposition of the

two failure modes in this case [22,23]. Crack branching,

which means "good natured" cracking behaviour, does not

necessarily have to occur. As can be seen from Fig. 15,

for the two specimens with crack branching, the test results

lie in the general scatterband when the stress intensity

factor for branched cracks, K , f = K/ ] 2 , is substituted.

Another DENT-specimen, "D" in Fig. 15, had no crack branching.

32



Moreover, it surely is not permitted to describe a behaviour

in the regime of dominant ligament damage with data from

specimens exhibiting crack tip damage or mixed mode damage

with high K I i d / Ç ; s i e b e l ratios,

(b) must have identical thickness B, i.e., have the same lateral

constraint,

(c) must show roughly the same sufficiently high creep rupture

ductility (no notch weakening).

(d) may not have an excessive crack increment in order not to

change the V = K T. , / 5 ~ „. ,^ o lido no Siebis based on initial crack size

change the V = KT. , / 5 ~ . , value too much, since this

o lido no Siebel

3. ) Test results on creep crack growth

After having proposed and explained the two-parametrical models

using K... as crack-tip paramter, the creep crack initiation

and growth results evaluated in the COST 505 working group on

HT crack growth will be summarized and evaluated in relationto the proposed models. Figs. 16 and 17 show basic results for

creep crack initiation and growth as plots of "a" over loading

time, for CT-specimens [5,15,18] and CCP-specimens [18] at 550

2 TIÒ 530°C. Both materials are of 1 % CrMoV grade, but all the

GEC-results [18] and the IWM-results [3,5] were performed with

the Round Robin material, while the Siemens/KWU tests, Fig. 17,

were made with another cast, described in [ 15 ] , and with the

Round Robin material (specimen AE1, Fig. 16 ). The conditions

of the tests and their evaluations are described in detail in

the respective final reports [5,15-20].

33 -



3.1 Creep crack initiation results

From the plots, Figs. 18 and 19, it can be seen that the time

for crack initiation depends on the criteria which is used to

define the crack initiation time. A broad scatterband for crack

initiation originates from the different initiation criteria,

Fig. 18. A rough evaluation of these results and of other

available results from the literature [20-24] shows the following

tendencies:

1. The tighter theAa. limit for crack initiation, the lower

the t./t ratio,i r

2. The time of crack initiation depends on the creep ductility.

For very creep ductile materials, the majority of life is

occupied by the initiation process: high t./t ratio. For

creep brittle materials it is proven that crack propagation

is very fast and crack initiation occurs rather early but

- because of the limited number of test results - it is unclear

whether this results in a higher or a lower t./t ratio.

An evaluation of some tests on 1 % CrMoV material ends up with

the following relations t./t , shown in Table 1.

34



Fig. 19 shows all Siemens/KWU data at 530°C together with 2 data

points of COST 1 % CrMoV cast material [ 16 ] , as a plot of K .,

over loading time with crack depth parameter points of da. =

0,5; 1 and 2 mm. If a crack depth of ¿la. = 0,5 mm gives the

engineering crack initiation criteria, then the lower bound

line represents the crack initiation times of CTI-specimens

at 530 C. This lower bound curve is supported by the great number

of data with<d a. = 1 and 2 mm, and here especially by one point

of a 1 % CrMoV forged material [9,10] which reached A a. = 1 mm

after about 30,000 h. It would, therefore, seem to be allowable

to extrapolate the scatterband boundary lines up to 10 h.

Fig. 20 depicts a summary of all crack initiation data fromCOST 505 partners and also the results of a further German research

group [21-23] for 1 % CrMoV at 550°C.

All partly solid points are from COST 505 Round Robin small

scale specimens. The specimens were fatigue precracked with

an intended crack initiation criteria of 0.2 to 0.5 mm crack

extension.

In addition, the picture includes specimens with differing size

and shape [21-23], detailed in Fig. 2 1, which had spark erodedcrack tips. Some preliminary tests, Fig. 19, [15] showed that

spark eroded crack tips delay the crack initiation noticeable,

but not too much compared with the usual scatter, if a creep

ductile material is applied.

For technical applications, a creep crack initiation criteria

of A a. = 0,5 mm is adequate, because it represents a smaller

crack growth increment than is detectable by nondestructive

evaluation methods, for example by UT for thick walled casings

or rotors. From the diagram in Fig. 21 it is obvious that specimens

with different shapes and sizes result in specific crack initiation

times.

35



From Figs. 20 and 22 which are plots for K . . versus time for

crack initiation it can be derived that the large scale specimens

with deeper cracks, a , and larger thickness, in other words

with predominant crack tip damage (ratios V = K .,/G >4,5)

and higher lateral constraint (thick specimens, 50-100 nun, Fig. 22),

create the upper boundary. The small scale specimens (V = 1 7-3

with short cracks and a smaller specimen thickness (- 25 mm)

form the lower bound of the data band. This is consistent with

the expected behaviour of the different damage modes, mentioned

above, according to which specimens with ligament damage fail

with faster crack growth than specimens with crack tip damage

(deeper cracks) in the case of equal Kj..,-values (Fig. 12 ■

Medium size specimens lie in the middle of the crack initiation

range. In addition an attempt was made to include in this diagram

those medium size specimens [3,5] for which initial cracking

rates,á a. were defined and which were correlated in terms

of K_.. (because stress redistribution may not yet have been

completed). Here, the crack initiation times were calculated

with the assumption thati a.~0.3 mm was accumulated with the

initial cracking rates, a.. The initiation curve evaluated by

this method falls into the lower middle zone of the band.

According to Fig. 20 it seems to be possible to rank the specimens

by the KT. ,/(T ratios. Of course, there is some scatter, but this I i d n is not unexpected in considering the different kinds of crack tips,

the different heats of 1 % CrMoV material, the specimen thick-

ness B, the different crack initiation criteria, and methods* used.

From the test results described above it can be concluded that:

- KT•j seems to be basically usable to describe the crack tip lid J r

situation for crack initiation, Fig. 19,

- the time for crack initiation increases with decreasing crack

tip driving force K J - J J ) » Figs. 18-22,

- specimen size and shape determine the damage mode and the

related specimen behaviour, which appears to be described

basically by means of the two criteria diagram for creep

crack initiation, Fig. 7 and 10, with its crack tip/farfield

ratios and with some additional notice of the specimen thickness,

Figs. 20 and 22,

♦Evaluation of crack initiation by: - interrupted tests and metallographic evaluation [14]

- evaluation of potential drop and/or COD curves [5,16-19] - compliance measurement at RT with test interruption [15] - 36 -



- the crack initiation times and the ratio of crack initiation

to rupture times depend on the creep ductility,

- due to the differing methods and criteria used to determine

crack initiation, the scatter of data is considerable, Fig. 18

[18], Table 1, but the data are rather reproducible when usingthe same evaluation methods and material heats, Fig. 19.

3.2 Creep crack growth results

Fig. 23 includes most COST 505 Group 6 data on crack growth

at 550°C [5,15-20]. In spite of the number of specimens and

the different specimen sizes, the crack growth scatterband in

terms of KT. , is not too wide. It shows that for most of thelid

specimen types the curve has a crack initiation tail.

Fig. 24 shows all Siemens/KWU data on CTI-specimens at 530 Ctogether with some D35 data [16] from cast 1 % CrMoV and from

the heat affected zone of a weldment made with the same steelgrade*. At first view the base material of the cast 1 % CrMoV

version fits well into the crack growth bands of the forged

version. However, it should be remembered (Fig. 19) that these

data points represent crack depths,¿a, of 3.5 and 6 mm. This

is relatively deep compared with the bulk of the data points

from the forged material which for similar K-. .-values have

crack depths in the range of A a = 2 mm (Fig. 19). The heat affectedzone data of the casting weldment is weaker and shows still

higher crack growth rates.

The Siemens/KWU data within Fig. 25 include specimens in the

as-received condition as well as specimens with long term annealing

(560°C/10 h which is equivalent on the basis of Larson-Miller

to 530°C and 10 h) and with fatigue precracks which were initiated

at 530°C. Both pretreatments did not influence the crack growthrate. This means that creep crack growth or fatigue crack initi

ation after long term annealing (or low accumulated creep strains)at service temperature do not increase the subsequent crack

growth rate compared to the as-delivered condition [15].

Fig. 26 depicts the crack growth scatterbands based on the data

points of Fig. 25 together with parameter lines of certain crack

growth de pt hs¿a = 0.5; 1 and 2 mm (for individual data points see

Fig. 27). This figure clarifies once again tnat the tails are

part of the k - t (K T. J-curves which represent the stress re

distribution and creep crack initiation phase within the life

of a precracked creeping specimen.

* Crack tip within the HAZ [16]



Conclusions from the'crack growth data are:

- Tails represent stress redistribution and crack initiation

portions of the specimen life, which lasts 30-50 % of the

rupture life for a creep ductile material.

- In a similar manner to crack initiation, the crack growth

behaviour with low crack growth rates and smaller resulting

cracks (¿a = 2-3 mm) may be described in terms of K-. ,.

- The different crack growth rates due to specimen shape and

size seem to be normalized by the crack tip/farfield ratios V .

- Long term annealing and fatigue crack growth at 5 30 C does

not accelerate crack initiation and growth rate compared to

the as-delivered condition.

- Cast 1 % CrMoV tends to have slightly earlier crack initiationand higher growth rates than similar forged materials.

- The heat affected zone of weldments in castings shows faster

crack growth than the respective base material.

- Crack growth at 550°C is faster than at 530°C, Fig. 23 and 25.

4 -) Discussion of results

For the application of crack initiation and growth data some

basic questions arise:

1 . Is it sufficient to use a one parameter model?

2. Which mathematical formalism is adequate to describe crack

initiation and growth?

3. How can the specimen shape and size dependent constraint

be taken into consideration?

4. How can the influence of creep ductility be introduced?

These factors will now be considered.

4 .. 1 The parametric model

From the explanations given above it has to be concluded that

it is necessary to use a two parameter model and to pay attention

to the different damage modes.

- 38



In addition it seems possible to describe the influence of the

different damage modes by means of the crack tip/farfield ratio,

V = KT. , /CT _. , ,. Therefore, the models for use of theo lido no Siebel

two parameters are:

- the two criteria diagram for creep crack initiation

- and the normalisation of crack growth data with the V ratio.^ o

4.2 The load parameter

At least for creep ductile specimens with higher crack tip/farfield

ratios (V = 3.0) it seems to be possible to use the fictitiouso

elastic stress intensity factor, K . ,, for the description of

the crack tip situation. The use of K.., may be allowed up to

4a = 2-3 mm crack growth at deeper notched specimens. (For CT-spe-cimens the size effect due to the shape/geometry factor is negli

gible up to a/WPs0.57). For specimens with shallow cracks, the

net section stress G~ of the ligament becomes the dominating

parameter because the damage mechanism changes to ligament damage.

But for the transformation of test results from one specimen

to another specimen or to a component it is necessary to useboth loading parameters KT. , and G" _. . ,. Furthermore, it

^ ^ lid n Siebel

is important to compare only specimens with broadly equal damage

modes and related creep crack initiation and growth behaviour.For creep crack initiation this can be achieved by using the

two criteria diagram.

For creep crack growth, many researchers prefer to use C* instead

of K .,, but the use of C* has the disadvantages mentioned abovelid

which are due to the inexactness of the solutions based on the

Norton's law, including the exponent n, on the one hand, and

on the other hand there is the fact that it is rarely possible

in practicable applications to evaluate the load line displacementrate V .c

Another attempt, which uses a C*-value for the description of

crack behaviour at high temperature, comes from the CEGB [25]

and is referred to as the R5 procedure.

This procedure distinguishes, in a similar manner to the "two

criteria diagram for creep crack initiation", between

- the overall structure behaviour which is defined above as

ligament damage and

- local events at the crack tip (crack tip damage).



The overall structure behaviour is controlled by means of the

reference stress,GT., the local events (crack tip damage)

are proposed to be assessed by C*, which is expressed in the

following form:

<c* = 5~ , 2 . R' (6)

ref ref

Where £ , is the strain rate that occurs in a uniaxial creep

test at the reference stress 5" ,.

ref

The quantity R' is "a geometrical variable which relates the

stress/strain conditions near to the crack tip, to the nominal

conditions in the structure". This C* is proposed in particularfor use for materials showing creep brittle behaviour. For creep

ductile materials the use of 5" c is recommended.

ret

This R5-version of C* consequently includes with the quantity R'

a parameter which relates to the crack tip/farfield situation

of the specimen/component. From another CEGB-paper [7] it transpires

that a good approximation of R Ì S obtained by using the stress

intensity factor K, so R is defined as

R ' = K 2 /5"ref 2 ( 7 )

which tends in the direction of the square value of V ( R ~ V 2 )o o

of the crack tip/farfield ratio. Thus, R tends in the same

direction as V but gives higher differences (square values)

between the different specimen shapes than V .

It should be noticed that reference [7] gives the same reasons

for not using the C*.«-*ţjn - g . (3) and C*,~V (1) versions

of C* when it is stated that unfortunately, displacement rates

can rarely be measured in service and estimates of C* cannot

be readily derived from equation (1) . The equation for C*.

(3) is also said to be inconvenient as detailed finite element

solutions are requested to generate g-i - Thus, the opinion of

the CEGB authors with respect to the use of the usual C*-version

is completely in line with this author's opinion.

40 -



4.3 Specimen shape and size effects

The graphs of a versus K i d or C* 1, which are normally used

to describe the crack growth behaviour, have the disadvantage

that for large deeply precracked specimens such as CT50 or CT100

the inclination of the crack growth curve is very steep as a

result of the small change of K Ţ. , or of C* when there is only

a slight increase of 4 a. Only the C* version which includes

V gives appreciable inclined slopes of a over C*~.

To overcome these difficulties, it is proposed to use K_. , versus

loading time plots with parameter lines of Constantsa, see

Fig. 19. These curves can then be applied in a similar manner

as creep curves for smooth specimens, by means of which the

total strain of the specimen or of the component is described.

In addition, these kinds of curves can be limited to those crack

depths for which K Ţ. , may be adapted. They are also applicable

for the determination of the influence of specimen size and

shape either for creep crack initiation. Fig. 20 and 22, or -

when the data are available - for different accumulated creep

crack depths in reading the times for the different crack depths

from the diagram.

The data of the type as shown in Fig. 20 supply also the necessary

values, K Ţ Ţ_, for creep crack initiation, which may be introduced

as materials characters for specimens with different size and

lateral constraint into the two criteria diagram for creep crack

initiation.

41 -



4.4 Creep ductility

From the statements in Section 4.2 and from Part 2 of this brochure

[5] it becomes clear that creep ductility dramatically influences

the creep crack initiation and growth behaviour.

Creep brittle materials are able to tolerate only extremely

short defects, whereas creep ductile materials withstand a

reasonable defect size without a decrease in rupture time. This

service experience teaches us to avoid the use of notch weakening

materials.

The best method to evaluate a potential notch weakening behaviour

is to test smooth and notched specimens (K. «"4.5).

If a material shows no notch weakening (rupture times for notched

specimens equal to or longer than those for smooth samples)

the above mentioned models and design curves in terms of K may

be used. The application of notch weakening materials should

be avoided. If, in extreme situations, such materials have to

be used, a detailed investigation of the creep crack initiation

and growth behaviour is necessary.

42



5. ) Conclusion

It becomes obvious that two schools of thought exist about the

selection of the appropriate parameter for creep crack initiation

and growth. There are the protagonists of C*, who feel supported

by the fact that C* is the loading parameter which from the

physical point of view is reasonable . On the other hand there

are the people who try to apply creep crack initiation and growth

data in term of K Ţ . , for the description of components which

have thick w a l l s , long loading t i m e s , and low loading stresses .

For such applications, the use of K . , together with the parameter

C to describe the farfield is recommended. Crack initiation n

should be handled with the two criteria diagram for creep crack initiation which covers the range of the tails from the a = f

(K Ţ . ,) plots . The creep crack growth behaviour with limited

crack increment may be described by plots of K Ţ - J over t with

parameter lines for A a, similar to Fig. 19 . With such d i a g r a m s ,

which are only valid for CTI-specimens, however, it is hardly

possible to estimate the influence of specimen shape and thick-

n e s s B . Thus, crack growth with small increments should be rated

with diagrams like Fig. 20 (specimen shape) and / or Fig. 22 (speci-

men thickness, e.g. lateral constraint ) . If it is necessary

to describe the behaviour of cracks of greater depth, the crack

tip/farfield method with V = K T. , / G , Figs . 13-15, or some o lid n '

kind of description by means of C* [5] may be used. The most

important point for the future is to generate further data on

specimens with different shapes and sizes with long crack initi-

ation times and low crack growth rates, upon which a final decision

for the most economic and appropriate evaluation method can

be made.

- 43 -



References

[1] H. Riedel and J.R. Rice, ASTM STP 700, pp. 112-130, 1980.

[2] K.M. Nikbin, D.J. Smith and G.A. Webster, Proc. ASME Int. Conf. on Advances in Life Prediction Methods, Albany, pp. 249-258, 1983.

[3] R. Hollstein and R. Kienzler, Numerical simulation of creep crack growth experiments, COST 505-D22, Institut für Werk-stoffmechanik, Freiburg, FRG, January 1987.

[4] A. Saxena, EPRI Meeting, Sess. 6, Raleigh, North Carolina, Sept. 12-14, 1984.

[5] T. Hollstein, G.A. Webster and F. Djavanroodi, Creep Crack Growth in 1 % CrMoV Steel, an Evaluation of COST 505 Creep Crack Growth Round Robin, Prepared for the EEC, August 1989.

[6] J. Ewald, K.-H. Keienburg and K. Maile, Estimation of manu-facturing defects in the creep range, Nucl. Engrg. Des. 87, pp. 389-398, 1985.

[7] J.W. Goodall and R.A. Ainsworth, Structural Assessment of Metal Components, Paper No. 6, Conference of the Institute of Metals on Materials and Engineering Design , London, 1988.

[8] K.H. Kloos, K. Kußmaul, J. Granacher, K. Maile, R. Tscheuschner und W. Eckert, Kriechrißeinleitungs- und Kriechrißwachstums-verhalten unter Berücksichtigung des Größeneinflusses, Ab-

schlußbericht AIF-Vorhaben

Nr.

6038, 1988.

[9] J. Ewald, K.-H. Keienburg und K. Kußmaul, Hinweise auf Mechanismen und Einflußgrößen zur Beurteilung des Bauteil-verhaltens im Kriechbereich anhand von Kleinproben, VDI-Bericht Nr. 354, pp. 39-57, 1979.

[10] J. Ewald and K.-H. Keienburg, A two criteria diagram for creep crack initiation, Int. Conf. on Creep, Tokyo, pp. 173-178, 14-18 April 1986.

[11] J. Tscheuschner, Anriß- und Rißfortschrittsverhalten zeit-standbeanspruchter warmfester Schmiedewerkstoffe, Dr.-Ing.

Dissertation DI 7, Darmstadt, 1988. [12] J. Ewald, Evaluation of the creep crack growth behaviour

by means of a K Ţ-concept, presented at the European Croup on Fracture, Task Group, Freiburg, May 21/22, 1987.

[13] J. Ewald, K. Maile and R. Tscheuschner, Creep crack growth assessment by means of a crack tip/farfield concept, Nucl. Eng. and Design 117, pp. 185-195, 1989.

[14] K.H. Kloos, J. Granacher und R. Tscheuschner, Kriechrißfort-schrittsverhalten des Stahles 28 CrMoNiV 49, Z. Werkstoff-techn. 18, pp. 390-398, 1987.

- 44 -



[15] J. Ewald, C. Berger and H. Brachvogel, Investigation oncrack initiation and propagation under static, cyclic andcombined loading conditions of 1 CrMoNiV steels at 530 C,COST 505 D20/D21 Final Report, Siemens Report No. TW 1187/89,June, 1989.

[16] H. Kanbach, Crack growth in welded turbine materials atelevated temperatures, COST 505 D35 Final Report, AEG,April, 1989.

[17] I. Ragazzoni, COST 505 13 Final Report, ENEL, to be issued.

[18] S.R. Holdsworth, High temperature crack growth in turbinesteels, COST 505 UK5 Final Report, CEG Alsthom Report No.RM 8 72/8 9, November, 1989.

[19] G.A. Webster and F. Djavanroodi, Elevated temperature crackgrowth in steam turbine materials, COST 505 UK18 Final

Report, Imperial College (Mech. Eng. Dept.), January, 1989.[20] G.A. Webster and F. Djavanroodi, Determination of the crack

growth behaviour and failure mode of pre-exposed material,COST 505 UK26 Final Report, Imperial College (Mech. Eng.Dept.), January, 1989.

[21] R. Tscheuschner, W. Eckert und J. Ewald, Bewertung vonBruchmechanik-Parametern zur Beschreibung des Kriechrißwachstums, Arbeitsgemeinschaft für Warmfeste Stähle, 11.Vortragsveranstaltung "Langzeitverhalten warmfester Stähle",25.11.88, Herausgeber: VDEh.

[22 ] J. Granacher, R. Tscheuschner, K. Maile, W. Eckert undJ. Ewald, Ermittlung und Beschreibung des Rißausbreitungsund Rißwachstumsverhaltens bei hohen Temperaturen, DVM,Vorträge 21. Arbeitskreis Bruchvorgänge, Bad Nauheim, 1989.

[23] R. Tscheuschner und W. Eckert, Kriechrißverhalten von Klein-und Großproben bei betriebsnaher Langzeitbeanspruchung,DVM, Vorträge 21. Arbeitskreis Bruchvorgänge, Bad Nauheim,1989.

[24] S.M. Beech, J.W. Selway and A.D. Batte, Factors influencingcrack development in 1 % CrMoV steam turbine rotor forgingsteels, Proc. Internat. Conf. on Creep and Fracture of

Engineering Materials and Structures, Swansea, 1984.

[25] J. Milne and J.W. Goodall, Defects: can we live with them?,CEGB Research, pp. 48-59, May, 1988.

[26] K. Wellinger and H. Dietmann, Determination of "Formdehngrenzen", Materialprüfung 4, pp. 41-43, 1962.

[2 7] Hütte I, Ingenieur-Taschenbuch, 2 8. Auflage 6, Abschnitt:Festigkeitslehre, pp. 847-851.

45



Table 1

Author

Holdsworth

Slemens/KWU

^ Malle°> Tscheuschneri

Tscheuschner

Beech

Literature

18

15

22

11

24

Material Condition

partly creep ductile

creep ductile


ductilepartly creep ductilecreep brittlecreep brittle


Temp. °C

550

550550

550

550

Type of Specimen

CCPl fatigueCT ) precrack

CT) fatigueCT) precrack

CT eroded crack tip

CT eroded crack tip

CCP short cracks

aQ = 1-3 mm

tl/tr

0.50.3

0.4-0.750.3-0.4

0.3

0.11-0.180.2-0.50.15-0.50.22-0.48

0.1

a j * mm

~ 0.2^ 0.2

'v 0.5'v 0.5

¿ 0.5

¿ 0.1á 0.1

•v 0

*¿iaA: criteria for crack Initiation



F i g .

2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0

n o m i n a l ( n e t - s e c t i o n ) s t r e s s [ N / m m 2 ]

Stress Intensity and Nominal Stress as Function ofSpecimens Size

47



21CrMoNiV57. T=550°CCT12.5/40 spec men AB2Bé m i n = l 3 - I O - 2 0 < J 6 5 [ h - , l , O inMPa

Fig . 2a .

-1 1 1 1 1 1 1 1 1 r~

'o.CO 1.00 2.00 3.00 4 .00 5.00 6.00 7.00 8.00 9.00 10.00 11.00

i . see (E+06ÌExperimental and numerical evaluation of C forCT 12.5/40 specimen AB2B, with and without crackgrowth, creep law [3] .

o -c'

3*"

o -r

o -o í

21CrMoNiV57, 550°CCT 12.5/40, AB2 B

x : Experiment" /

— : E q ( 1 5 )

f-*tt-¿V '

3000

Fig. 2b. Experimental and numerical values of crack growth [3]- 48 -



F

ò ° ASTI

Q JfnK;

h

0\ f n Siebel T T T I

Q

ab at

I l u

F F

F' , . . ., w + a , F (7ASTM ~ r— ; - ( 1 + 3 ) (7nSiebel= r —

B ■ (w - a w — a B ( w — a ) w —a

, a - _ _fc. c ASTM for— —0.5 ^

W f » Siebel

'1 .43

Fig. 3. Nominal Stresses in the Farfield of Specimens with Bending Portions.

F On = ^ = ÂSTIVI

Ã

comple te ly redistr ibuted

general yield condi t ion

Fig. 4. Nominal Stress in symmetrically loaded Specimens.

49



4 0 0

a

aia.aci/ì

10'

K ' ^ " a / W l

1

I /W.157 J° _,

1

CT 25 / f CT 50

(C T 100

>icfTr t ie t / h

i>\d

Fig . 5. Shape Function f(a/w) as a function of time for differentspecimen sizes in case of equal crack growth rate.

F m"'-r "-1 -£\

4)

8.r.

o¿3

8 irf1-P KJ -o8- I

ò3 . TO"-

i

1ji

1r(i

f

¿V

7

— 30

c/w

. V

=

... -V -

357

... - v -

—

A CT KW. e«p .

F i g . 6 .

.10' KJ1

Stres s Intens ity Foctor K^ / MPoV ñT

Theoretical creep growth characteristic of differentspecimens with identical crack increase.

- 50



K.gf On ™ h igh rat io — p redominan t crack t i p damage

^ S

Ig loading t ime t -

Fig. 7a. Scheme of Creep Loading Conditions with PredominantCrack Tip Damage, ^n¿/ n̂ - Ratio is high.

Ig loading t ime t -

Fig. 7b. Scheme of Creep Loading Conditions with PredominantLigament Damage, Kii(j/on - Ratio is low.

51



c r a c k t ip d a m a g e ^ >

— ^ —t h o u t j ; w i t h

( l e a k a g e )

— T -

0.5

w i t h o

a c r . c r e e p c r a c k1.0 1.5 2.4 3.9

_ K j ,de« K

" UD

Fig. 8. Two-Criteria-Diagram for Creep Crack Initiation,Creep Ductile %CrMoV-steel, 530 °C.

Fig. 9. Explanation of the Normalized Values R for Stressand Stress Intensity.

- 52



wi t h wi t hou t

micro crack init.

□

O

■

macro crack init.

o I • eng. crack ¡nit. A 1

V

SENT, duct i le britt le (Beech . Se lway . Bane )

smal l manufactur ing defects in smooth spec

(Taylor, Bat te l

castings w i t h manufactur ing defects (duct i le)

D E N T ' . 30x 112. a/w - 0.45

OENT *. 40 x 65 . a/w - 0.1 and 0.4

*) wi t h spark-eroded crack- t ip

I crack tip damage ~> i io ~i\jfZT without J5 with

acr. creep crack

>12 Ki*/o„M0 O

Ç | leakage)

0.5 1.0 1.5 2.4 3.9

KiidoU)

Fig. 10. Two-Criteria-Diagram for Creep Crack Initiation

Creep Ductile l CrMoV-steel 530 °C.

R -5=.

0,5

burst

1 CrMoV-Steel; 530 CC

mixed damage

ñv crack tip damage J>

—1 rw i t h creep crack

0,5 K no

Fig. 11. Draft for a Two-Criteria-Diagram for Creep Crack

Initiation for Notch Weakening Materials.

- 53



Fig. 12.

250 500 K,« (N/mm3/2| 1000 1500

Creep crack growth rate, J, as a function of the

stress intensity factor, K I ¡ J , for different specimenshapes and specimen sizes (IfCrMoV steel, CT compact

tension specimen, K notched round bar specimen) [14].

1000

soo

200

Fig. 13.

AGA

AGB 28CrMoNIV49, 600 °C

Ligament

Damage Crack-Tip Damg.

a — 5 • 10s mm/h

i - 1 • 10'mm/h

60 0 "C

10 101 10

s 10 ' (h)

K8 1 Notched Round Bar

K1 7 j (T.-rVAit

CT with Side Grooves2 3 4 6 6 7 8 9

Creck.Tip/Farf.eld Ratio V0 - Kuo/o«, »« (mm1) ((Tnosj#|>#| _ ,,„ „̂.43,

K,,J for constant crack growth rate, å versus the,-d for constant crack growth rate, a vi

initial crack tip far field ratio, Vg = Kj;j/a

for creep ductile l CrMoNV material, 600 *CnO'

- 54



1 5 0 0

bE

2

1 0 0 0

5 0 0

1

m r

2 0 0 -

1 \ C r M o N i V . 5 5 0 °C . c r e e p d u c t i l e

Lig. D.

I -™

1

111111

11 .

« y^

1

1 0

1n / hi KB

— M —

M i x e d M o d e D a m a g e C ra c

K, -Al i d

• / ^* ^ Í »

* /X

CTO. 5

-4—

s 120- V

x ,X*

jr

>•

<*s¿^^^ ^ *

C T 0. 5 ( W - 4 0 )

HC T I C T 2

- ^ + n r - L - T -

k-T ip D¡

^

m a g e

«

C T 4

-4-

0 5 - I O 3 m m / h A G A

9 1 1 0 ' m m / h (T s c h e u s c h n e r ) / 8 /

• 1 • I O " * m m / h

550 " C

1

I O ' | h | 1 0 «

<^ à - 1 • 1 0 J m m / h / 9 /

A M A ( M a i l e )

A » — 6 • I O ' 5 m m / h

A - C T I

£ - C T 0 . 5 ( W - 4 0 m m )D 2 2 ( H o l l s t e i n / K i a r u i e r ) / 10 /

2 3 4 5 6 7 8 9

C r a c k - T i p / F a r f i e l d R a t i o V 0 — K u o i O n O b M i ( m m , / 2 )

Fig. 14. K ^ d for constant crack growth rate, å, versus theinitial crack tip/farfield ratio, V Q

for l%CrMoNiV, 550 °C.Klid /crnO'

- 55 -



2000

»EE^̂zs

s?

1500

1000

500

T1 2 3 4 5 6 7 8 9

C'ock-T ip/Farl ie ld Ratio Vo ~ K<„jo <7>io s» i» i (m m" 2 )

Fig. 15. Kîd f ° r c o n s t a n t growth rate à versus the initialcrack tip far field ratio, V Q = K l i d / a n 0 , for 12% Crsteel, 550 °C, creep ductile.

56



35

— 30

EE

O) ¿

°c0)

ü(D

o 20

15 '

CCP

» «« r » " * = «

100

•4

X A

CT25/50S [15]

CT25/50S [18]

CCP25 [18]

ROUND ROBINCT25/50S [ 5 ]

1%CrMo(Ni)V , T = 550°C

1000 10000

loading time ( h )

—i 1 t 1——

100000

Fig. 16. COST 505 - Res u lts



Su mma r i z e d Da t a

Ol00

35

DAY

X

+

Seg.3 as rece ived condit ionSeg.4 as rece ived condit ionSeg.6 aged 560°C, 10000hSeg.9 as rece ived condit ion precrac ked a t 530°CSeg.10 as rec e ived con ditionSeg.10 as rec e ived con dition f rom p ot, drop me as m.

1000 10000loading t ime ( h )

Fig. 17 . Cre e p Cra c k Growth , 28CrMoNiV49(KWU 1508), 530°C, CT25S



KA - ^Pa/m

50

20

10

10

10M( iV \ Rotor steel

v

■ ♦

d

-i •

- rv

a / w = 0 . 3 6

-TCT251 -

N S. s

■-

^

• V _

a«

i fSfâo-craã -= -1

±_.-

=•7

•>. s

♦

^ D E N T 1 : i 5

7 * •

=02mm

ì : 1

i

1

S. i N .

v ^ / ^ O O m m ^ - ^ - -. :

\Âa=0-5mm (Ref.lSl. -:::

^^^rcT25,530oC-: :

JM L

^ AasO-Olmni i lRefZy lb ] : - - - . -^ ~- . — _ : _ ; ; _ : ^ _ . L =±..~~

. j u-y.íí . : - ~Ş fes--

• ::==.:;• . . i ¡=1-5-.

: ■ ::

)\i -:. = i r = '

1 : • • ~~ -~ :

10* 103 10^ 105 Time fo i n i t i a t i on -hou rs

Example

Depend i ng on I n i t i a t i o n c r i t e r i o n s e l e c t e d . I n i t i a t i o n t ime may v a r y be tween 20h (Aai=0.01mm) and 2 0 , 0 0 0 h ( ai=2ram) a t K«. = 20MPa/m .

F i g . 1 8 . C r e e p c r a c k i n i t i a t i o n t i m e s i n l C rMoV r o t o r s t e e l s [ 18 ]

59



Oí

o

60

50

c o

Q .

1 40

30 -l

20

10

0.5 mm

D

A

X

A a

0.5 mm

1.0 mm

2.0 mm

s p a r k e r o d e d

c ra c k ti p

cas t mat . [16]1 mm forged [6]

s c a t t e r b a n d , m a k r o s c o p .Cre e p Cra c k I n i t i a t i o nS i e m e n s 1 9 8 9

1001 1 1 1 1—i—i—i

-

1000 10000load in g t ime ( h )

T — T — I — I -

100000

Fig. 19 . Cre e p Cra ck In i ti a t ion, 1 % CrMoNiV530°C, CT25S



6 0 -

50

ra Q .

_ 4 0 -

30

20

10

1 CrMo Ni)V,T = 550°C

a/W = 0.55 a/W = 0.55

, aAV = 0.55 a/W = 0.40 aAV = 0.40 aAV = 0.20 aAV = 0.10

, aAV = 0.50 , aAV = 0.50 , aAV = 0.20

Hollstein C T 1 2 . 5 - 5 0

J r o m à , [ t ]

10000 loading time

Creep Crack Initiation a Criteria

Fig. 20

■}

M a i I e/Ts c h euschner D. x + A Y □ A

0.004W CT),0.01W DENT)

spark eroded

Siemens H

-0,5mm

Ragazzoni a <►

- 0,5mm

Hollstein CT25 - , C T 5 0 X

0,3mm

Holdsworth D o 0,2mm

Creep Crack Initiation with Parameter K Ur i I fatigue precracked

Rintamaa

u e u 0,5mm



K.

iH/mrrO

tf -

3*10*

30 CrMoNiV 4 11, AMA, 550 °C A - Cs25. q/tø=0.55

7- 0 9 . q/W=0,4

a = CT100.q/W=0.55

O - Cs50. q/ty=0.55

O - D60. (v^W=0.*

B -0 6 0 . q^W=02 • = D60.(VVi=O.1

10

F i g . 2 1 . S t r e s s I n t e n s i t y v e r s u s c r a c k i n i t i a t o r t i m e fo r a R = 0,01 W , l C r M o N i V , 5 5 5 ° C [ 2 1 ]

6 2 -



u>

60

5U

ra o. Z . 40

30

2 0 -

10

CT 25/50S CT 50/1 OOs CT100/200 D g D60 D60 D60

CrMo(Ni) V,T = SSO C

10

a/W = 0.55 a/W = 0.55 a/W = 0.55 a/W = 0.40 a/W = 0.40 a/W = 0.20 a/W = 0.10

a/W = 0.50 a/W = 0.50 a/W = 0.20

/Ho l ls te in CT12 . 5 - 50 J r om à , [ 3 ]

10000

loading t ime

Creep Crack Initiation Criteria

Fig. 2 2

• >

Mai le/Tscheuschner □ x + ▲ Y D A

0.004W CT ,0.01W DENT

spark eroded

Siemens B

~ 5mm

Ragazzoni

a ~ 5mm

Hollstein CT25 - ,CT50 X

3mm

Holdsworth

2mm fat igue precracked

Rintamaa

y e ~ 5mm

Creep Crack Initiation with Parameter Spec. Thickness



10°

Oí

10"

E

reT3

10"

10"

.XDA

Y+

•

•

CT 12CT 25

CT 50C C P 2 5SENT 12

1%CrMoV,T = 550<'

:

Y BB B

T B

S u aB

Y B

' CT U

m B B

BB A V

B j y k « ^ B BB

^ s P3 a a B

QìÉ i - iW B

ET aB B B

0 + B+

B

10 20 K|id ( MP a Y m) 50 100

Fig. 23. COST 505-Group 6, Summa rized Data



O ) UI

10

10 -2

E E

ra ■D

10

10 -4

1 0

DA+XY D 20, T = 53 0 °C (forged new material) D 35 , T = 530 °C (casting)

1 CrMoV

10

welded,HAZ

^ 4. y + +

*

¿ti

Y Y

20 K, id (MPa\/m) 50 100

Fig. 24. COST 505-Group 6, Summarized Data, CT25



O) O)

•o

1

1

E 10-

1

1

Summa r i z e d a ta

D A

Y X +

Seg.3 as received condition Seg.4 as received condition Seg.6 aged 560°C, 10000h Seg.9 as received condition fatigue precracked at 530°C Seg.10 as received condition

/ /

4

*

* /

9 X 2/

I

■ ^

v

I

I

1 20 K||d MP a t f m ) 50 100

Fig. 25. Creep Crack Growth, 28CrMoNiV49 KWU 1508), 530 °C, CT25S



10

cr>

RI ■o

1 CrMoNiV-Steel 530 °C , CT 25s

¿a * o t

2 .0 m m

1.0 m m 0 .5 m m

r a m e t e r l i nes fo r a c c u m u l a t e d c r a c k d e p t h ¿fa to t

crack g ro w t h curves

2 0 Kild ( MPa i/rfT) 5 0 ■ i i

1 0 0

Fig. 26. Creep Crack Initiation and Growth



10

O) 00

10 -2

j e

E

7o T3

10

10-

10

c*/\

1

SP

♦

I

Aaqe s

. 2,0 mm

1,0 mm

0,5 mm

i i i

♦ D

■

i i

10 20 K||d ( M P a ^ m ) 50 100

Fig. 27. Creep Crack Initiation, 28CrMoNiV49 KWU (1508), 530°C, CT25S




Parti i

Creep Crack Growth in 1% CrMoV Steel and Alloy 800H

an evaluation of the results of the COST 505 and an EGF Round Robin

T. Hollstein*, G.A. Webster**, F. Djavanroodi**

* Fraunhofer Institut für Werkstoffmechanik, Freiburg, D

* * Imperial College of Science and Technology, London, UK

69 -





ABSTRACT

The report presents the results of a Creep Crack Growth Round

Robin conducted by the COST 505 Group 6 "High-Temperature

Crack Growth" and an EGF Working Party within Task Group 1

"Elastic Plastic Mechanics". Data from twenty-five

laboratories are compared, 'as-determined' by the participants

and following a unified evaluation. For the two materials

tested, Alloy 800 H at 800°C and 1% CrMoV steel at 550°C, it

is shown that the most satisfactory correlations of creep

crack growth rate are achieved with the creep fracture

mechanics parameter C* when a unified evaluation procedure is

adopted. Precautions for obtaining reliable experimental data

and interpretations are discussed and supported by numerical

analysis and models of the cracking process. Little influence

of specimen shape and size is observed and it is demonstrated

that all the cracking occurs under plane-stress conditions. In

the 1% CrMoV steel, transient effects in the early stages of

cracking are shown to be caused by a build-up of creep damage

at the crack tip.

INTRODUCTION

In recent years, an increasing interest has developed in being

able to predict the behaviour of high-temperature components

containing flaws. Such analyses are required to assess defect

acceptability at the design stage, to predict remaining life

and in failure diagnosis.

Depending on the circumstances under consideration, various

approaches are available which may be based on net section

rupture or fracture mechanics methodology. When crack

propagation occupies a significant portion of the lifetime,

creep crack growth rates have been characterised in terms of

the stress intensity factor K, the J integral, the C* integral

71



and the C. and C(t) functions, see for example Ref. [ 1] . There

are advantages and disadvantages with the use of each of these

parameters. For example, K is easy to calculate but is notably

geometry dependent in situations where the crack tip stress

field is redistributing at a faster rate than the defect is

propagating. In contrast, C* appears to be able to correlate

creep crack growth rates under steady-state conditions

relatively independently of geometry, but it is more difficult

to apply to the analysis of real components.

With this background, twenty-five laboratories [2-26],

proceeded to conduct a Creep Crack Growth Round Robin to make

an inter-comparison of data generated in differentlaboratories according to agreed procedures. The aims of the

programme were:

- to compare different methods of measuring crack initiation

and growth,

- to evaluate appropriate fracture mechanics field para

meters such as K and C*, and

- to determine the limits of acceptability of fracture

mechanics concepts for high-temperature crack growth

characterization.

The overall objective of the Round Robin was to evaluate the

consistency of the results gathered from different

laboratories and to establish the ability of the field

parameters to correlate creep crack growth rates in a range of

testpiece geometries and ultimately in service components. To

assist in this exercise, two laboratories [3,7] performed

analytical and numerical investigations to produce an

appropriate unified evaluation procedure.

Also included in the report are data which were collected on

the same material under similar testing conditions in other

Round Robin investigations that were organized by the American

Society for Testing and Materials (ASTM) [27] and the Japanese

- 72



Society for the Promotion of Science (JSPS) [ 2 8 ] .

Collaboration was achieved through the Versailles Agreement on

Advanced Materials and Standards (VAMAS) [ 2 9] . The data were

analysed together with the EGF results to provide a rigorous

check of the evaluation procedures being developed and a

comprehensive database.

MATERIAL

1% CrMoV steel

The test material was a round bar of a 1% Cr steel(21 CrMoV 5 7) with a diameter of 28 6 mm. The bar was supplied

by Buderus Edelstahlwerke, Wetzlar, identification number

3/22282-Z. Details of the test material given by the

steelmaker are compiled in Table 1.

To check the homogeneity of the material with respect to 0.2%

proof stress R . _ and ultimate tensile stress R ,po .2 m

conventional tensile tests were conducted at room temperature

and 550°C at two different positions along the bar and at four

radial locations (see Fig. 1) . The results are compiled in

Table 2 and Fig. 2. They show no signifcant difference between

the values for the two axial positions but a fairly strong

decrease of the R . _ and R values towards the centre of thepO. 2 m

bar. This is especially conspicuous for the values in the

region AA. It is expected that the strong decrease of the

0.2%-proof stress and ultimate tensile stress near the centre

of the bar (below 0.23 R) will not influence the creep crackgrowth test results, because all the fracture mechanics

specimens were taken from the bar with the crack tips at

approximately 0.5 R.

The impact energy of Charpy specimens tested in COST 505 -

Project D 20 [2] on the same material showed the same

tendency. The results are compiled in Figs. 3 - 6 . The FATT is

- 73 -



25°C for the half radius position of both regions AA and DA,

55° C for the centre region of AA and 35° C for the centre

region of DA (Figs. 4, 5 ) . The impact energy over the cross

section for R/2 shows no significant variation (Fig. 6 ) .

A check of the chemical composition over cross sections at the

top and position AA indicated no significant variation (Table

3, from [2] .

The bainitic microstructure of the material is shown in Fig. 7

for region AN. The ferritic portion increases towards the

centre of the bar which explains the deterioration of the

mechanical properties.

The results of creep rupture tests conducted in two

laboratories [6,30] are summarized in Tables 4 and 5 and are

plotted in Figs. 8 - 1 0 . The creep rupture tests at three

different axial positions show a marginal reduction of creep

strength towards the top of the bar [6] which is comparable to

the reduction in tensile strength and Charpy energy [2].

The minimum creep strain rate e . determined from the plots min

in Fig. 9 are shown in Fig. 11 as a function of stress. For stresses between 150 and 250 MPa c . can be described at

min 550°C to a good approximation by a Norton law

è . = 1.3 • I O - 2 0 ■ <r6'5 [ h - 1], <r in MPa. (1) m i n L j r

Over the whole stress range measured, the exponent in Eq. (1)

varies between 4 and 8. Also because of the variation in n the

stress dependence of c . is better described by an min

exponential law (see Fig. 1 2 ) :

I . = 7.06 - 1 0 - 9 - e ° -0 3 6 9 «■ [ h - 1 ] f v in MPa. (2)

min

When the creep strain and strain rate are represented by a law

according to Garofalo, to include primary creep then,

- 74



—■RI • G c r = eo t 1 6 I + «min* a n d ( 3 )

i r r = e V e A + l . [h _ 1] cr o min ,

where from Fig. 13 and 14 e and R' can be described by

o - o ic m - 4 Q0.0088 <r e_ = 2.16 • 10 • e o

and R = 3.08 • I O - 3 • o- 0' 5183 [h - 1] (¡r in MPa) .

Alloy 800 H

The test material was a 32% Ni 20% Cr alloy (Alloy 800 H)

supplied by Vereinigte Edelstahlwerke, Kapfenberg. The

production procedure is shown in Table 6 together with the

chemical composition and some basic mechanical data given by

the steelmaker. The corresponding microstructure is shown in

Fig. 15.

The specimens for the Round Robin investigations have been 2 taken from the 28 x 103 mm plate.

The minimum uniaxial creep strain rate è . for the Alloy min

800 H at 800°C may be described by Norton's law with stress in

MPa:

? .„ = 5.7 - I O - 1 6 o-6'5 [h _ 1] (4) min L J

The constants were determined on the basis of results from Kempf et. al [31], see Fig. 16. A more detailed description of

the material and its mechanical properties can be found in

[31,32].

75



SPECIMENS

1% CrMoV steel

Most of the results were gathered using 25 mm thick and 50 mm

wide compact tension testpieces (CT25/50), but other specimen

types (i.e. single-edge notched three-point bend - SENB3,

single-edge notched tension - SENT and centre-notched

tension-CN) and sizes (i.e. thicknesses from 5 to 63 mm) were

also tested. Parts of the material cut-up plans for the test

specimens are given in Fig. 17. The other specimens were cut

similarly.

All the cracks in the specimens were oriented in a radial

direction to give crack extension towards the centre of the

bar where possible. The crack tips of the CT specimens were

positioned to be at about R/2. Some basic data of the tested

specimens are compiled in Table 7. Specific details of all the

specimen dimensions and test conditions are given in the

individual participants' reports.

Alloy 800 H

Most of the results were gathered using 25 mm thick

compact-tension specimens, but some smaller compact-tension

specimens and centre-notched tension specimens were also

tested. The material cut-up plan and some basic data of the

tested specimens are compiled in Fig. 18 and Table 8,

respectively.

TEST P ROCEDURE

It was required that all the specimens should be fatigue

precracked at room temperature according to the procedures

recommended by ASTM [33,34] to a crack length to width ratio

76



a/W = 0.5, approximately, although shorter initial crack sizes

were allowed. It was stated that the final maximum load during

precracking was not to exceed that to be used for creep crack

growth testing.

After precracking most specimens were provided with 20% deep

side grooves (10% each side). A few were retained for testing

without side grooves. No guidelines were given for other

specimen geometries.

All the tests in the COST and EGF creep crack growth

investigations were conducted at 550°C. Loads were chosen

mainly to give approximately 4 mm of crack extension inapproximately 4 to 10 weeks although some testing was

performed at higher and lower loads. It was specified that all

specimens should be held at the test temperature for 16 hours

before the load was applied.

Throughout all tests, it was required that load line

displacement and crack length, using electrical potential

methods, should be recorded continuously. Other methods of

measuring the crack length were allowed in addition.

Each participant was allowed to determine the crack growth

rate, K and C* using their own preferred route. It was

instructed that all the raw data were to be transferred to a

central point to enable the results to be processed by a

single analysis route.

The ASTM and JSPS tests were carried out at 538°C and 594°C

mainly on CT specimens with 25% side grooves.

- 77



PARAMETER DETERMINATION

Crack Growth Rate à

The creep crack growth rate à was determined from the cracklength versus time records. In general, these were con

structed using the output from direct- or alternating- current

potential drop instrumentation. However, two participants

[3,26] also used the single specimen partial unloading

compliance method and the multiple specimen unloading

technique.

There was some disagreement between participants concerningthe existence of an incubation period prior to the onset of

crack growth, which led to further testing to resolve the

situation. However, the difference in the growth rates

calculated assuming either the presence or absence of an

incubation period did not vary by more than a factor of two.

In the unified approach à was determined from a seven point

polynomial fit to the crack length versus time recordings

[35]. In some cases, only one value could be evaluated by

linear interpolation due to insufficient experimental pointsor relatively short testing times.

Stress Intensity Factor K

The stress intensity factor K was determined according to the

formulae in the ASTM Standard [33].

C* Integral

To determine C*, participants were encouraged to use a formula

based on the relationship:

F vC* = *c F I T ' <5>

nwhere F is load, v_ is the load line displacement rate due to

- 78



creep alone, B is the net section thickness and b is then

uncracked ligament (i.e. (W-a) where W is the specimen width).

T) is a factor that depends on the testpiece geometry and on

the creep exponent n, but which may in general also be

influenced to a small extent by the crack length and the

stress state. The form of Equation (5) is consistent with that

used in the J Estimation Procedure recommended in ASTM-E 813

[34], but with T) V replacing 7)V in the determination of J.

Different formulae are available for i) for the more commonc

specimen types (see e.g. Refs. [34, 36, 37 ] . However, for a

given geometry, the values do not differ by very much. In the

Round Robin exercises it has been found that:

''c = Tn W V (6)

where TJ is obtained from limit-load analysis. The values used

in the unified approach for TJ are [34,36,37]:

for CT specimens: TI = 2+0.52 b/W

for SENB specimens: TJ = 2

for SENT , CN specimens: ij = 1.

In the unified analysis $ was obtained from a seven-point

polynomial fit to the load-line displacement versus time

records. Displacements due to creep were determined first by

subtracting the elastic contribution due to crack growth from

the total load line displacement.

NUMERICAL INVESTIGATIONS

The values of TJ obtained using the expressions listed above

have been compared in Fig. 19 with those calculated from

alternative derivations (see Refs. [38-42]), including those

based on numerical simulations incorporating the material law

given by Eq. (1) . The numerical analyses were performed for

79



different specimen types and stress states using line integral

and/or energy dissipation rate equations for C* [43]:

f 9ui 1

C * = ƒ [ W* dy - T ± gir ds J (?)

The equivalence between (7) and (8) is only valid for a

viscous material under steady-state creep conditions, when

elastic strains are negligible.

•emn

W* is the stress work rate W* = T <r. .dê. .

oJ 13 ^

V n

and U* is defined as U* = ƒ F dv ,o

where the following notations are used:

r integration path around the crack tipconnecting the lower and the upper crackface in a counter-clockwise direction,

ds arc length on r,o-ij components of the stre ss ten sor ,Ti = o-ij nj components of the t raction vect or,eij components of the s trai n rat e ten sor ,ui components of the displac ement rate

vector.

The summation convention is implied for repea ted indi ces .

Subroutines to calculate J and C* integr als ha ve bee n

implemented into the finite-elemen t program me ADINA [44].

Since all quant ities entering Eq. (7) are known in the cour se

of an FEM calculation, the line integra l can readily be

evaluated. More effective and easily extendab le to

three-dimensional problems is the calculation of rele ase rate s

by the virtual crack ext ension method [ 45 ] . Both method s have

been employed and lead to the same results withi n the

machine-dependent accuracy, see [ 3 ] .

- 80



Several numerical simulations have been performed for

stationary and growing cracks in a material which deforms

according to the uniaxial creep laws (1) and (3).

In the short-time regime (t < t- , where t is a redistributiontime defined in [1]), C* has been found to be path dependent

as discussed in [3]. In the long-time regime (t » t ) , C* is

path independent and for a stationary crack, it is constant in

time, see Fig. 2 0, when secondary creep is reached. Here, the

creep laws (1) and (3) were used separately to demonstrate

also the influence of primary creep in the C* calculation.

Taking primary creep into consideration leads to higher

initial C* values and slightly greater redistribution times.

If, for the steady-state conditions in the centre-notched (CN)

specimen AQ1 shown in Fig. 20, ij is determined numerically by

comparing Eqs. (5) with (7) or (8 ), then TJ values of 0.81

(0.78) and 0.71 (0.64) are evaluated for the creep laws (1)

and (3), respectively. The TJ values without brackets hold for

plane stress conditions and those in brackets for plane strain

conditions. These values are in fairly good agreement with TI

= n/(n+l) = 0.87 determined from Eq. (6) and which were usedin the unified evaluation. In Fig. 20, the finite-element

results are compared to experimental results, which were

evaluated using an average TI = 0.75. The experimental results

are fitted best by the finite-element calculations assuming

plane stress situations. Both creep laws, Eqs. (1) and (3) ,

yield essentially the same results.

Little difference in predictions would be achieved with the

range of TI values mentioned above.

In Fig. 2 1, the results of the finite-element simulations of a

particular experiment AB2B (CT 12.5/40) are shown. For

comparison of the 2-dimensional finite element calculations

with the experiments the force is related to the effective2thickness B = B-(B-B ) /B [46]. Calculations, with and

81



(1). It was assumed, as determined in [3] , that after an

incubation period of 1 hour

à = 7.41 • 10~ 3 • c*0'72 (9)

with à in mm/s and C* in N/(mm>s).

These finite-element calculations gave ij values of 1.86 andc

1.82 for plane stress and plane strain, respectively, which

are in good agreement with TJ = 1.94 according to Eq. (6).

Whereas the calculations for C* without crack growth are

similar to those of CN 12.5/50 specimens AQ2 (see Fig. 20) and

decay to a constant value, those which include crack growth

rise and give good agreement between the plane-stresscalculations and the experimental results. Also, the

experimental crack growth versus time curve corresponds best

with the plane-stress predictions, see Fig. 22.

The ability to perform numerical calculations of this type is

particularly important when there is a requirement to

determine C* for non-standard testpiece geometries or real

components. Some examples of special cases which have been

examined using these methods, for n = 6.5, are:

T) = 0.17 - elliptical surface crack loaded in tension at

infinity [40],

T) = 1.35 - circumferential crack in a tube subjected to

combined tension and internal pressure loading

[40],

ij = 1.75 - square section SENB3 testpiece, assuming only

horizontal displacement of specimen at roller

supports [47],V = 1.54 - square section SENB3 testpiece, assuming free

rotation of specimen at roller supports [47].

In all these circumstances, the values obtained for plane

stress and plane strain deformation were effectively the same.

82



Consequently, the finite-element analyses confirm the values

of ii adoptei

approximations.

of ii adopted in the unified approach as reasonable

RESULTS

1% CrMoV steel

Initially, creep crack growth rates were determined in terms

of K and C* by the individual participants, see Figs. 23 and

24. The degree of scatter associated with both loading

parameters is considerable. It can be attributed, partly, to

different interpretations of the raw data and evaluation

procedures and partly to effects of geometry, size and

pronounced crack growth rate "tails" in the early stages of

cracking. In order to determine the significance of each of

these factors, all participants were asked to submit their raw

data to one laboratory [7] in a form suitable for a unified

evaluation.

The results of applying a single standard analysis are givenfor K and C* in Figs. 25 and 2 6, respectively. It is apparent

that the unified approach has only little influence on the

extent of the à versus K correlation. However, the overall

scatter of the C* data is reduced significantly by adopting

the recommended procedure for obtaining C* from Eq. (4) .

However, there is still appreciable scatter associated with

this loading parameter, particularly in the early stages of

cracking.

Alloy 800 H

The degree of scatter displayed by the initial creep crack

growth rate plots for the Alloy 800 H was lower than that

exhibited by the 1% CrMoV steel (the original C* data

collation for Alloy 8 00 H are shown in Fig. 2 7 ) , covering about

- 83



one order of magnitude on a. In part, this was due to the fact

that the database for the alloy was less extensive and that

the results were gathered using fewer specimen geometries and

sizes. It was also due to the fact that the crack growth rate

'tails' were less pronounced for this material.

The results of applying a single standard analysis are given

for K and C* in Figs. 28 and 2 9, respectively. Using a uni fed

approach had no influence on the extent of the à versus K

databand (and hence the K results are only given once in Fig.

3 0) . The overall scatter of the C* data is reduced by about

half a decade by adopting the procedures described above.

The correlation between à and C* is about the same as

determined by [38] for the same batch of material and for

different geometries and sizes of specimens (see Fig. 3 0 ) .

That is shown by the power law correlation

à = 0.0034 • c*0,733 (10)

with k in mm/s and C* in N/(mm-s), which is represented by the

straight line in Figs. 29 and 39.

DISCUSSION

Unified Correlations

The collations of the creep crack growth rate results from the

Round Robin programme, 'as-determined' by the individual

partners, give poor correlations with K and, at least for the

1% CrMoV steel, with the C* parameter. The situation is not

much improved for the à versus K correlations when a single

assessment procedure is adopted, suggesting that the linear

elastic expression is not a satisfactory parameter for

describing creep crack growth data over a wide range of

cracking rates in 1% CrMoV at 550°C and Alloy 800 H at 800°C.

84 -



The observations, that a correlation with C* is improved when

a standard analysis of the data is carried out, implies that

an appreciable cause of the initial scatter was due to the

application of different methods of data assessment. For

example in the present study, à and ^ were derived from the crack length and displacement versus time records using a

range of techniques. These included manual, cubic spline and

seven-point polynomial curve fitting routines. In the

calculation of C*, some participants used analytical

estimates, which are very sensitive to the choice of n in the

creep law, and others utilised total load line displacement

rate rather than that due to creep alone (i.e. ) . Similarly,

some used gross rather than net section thickness. The degree

of scatter was particularly exaggerated with the C* estimates derived using theoretical representations of ţ according to

Ref. [42] in Eq. (5). The most consistent interpretations are

obtained using the preferred standard evaluation route of the

unified analysis.

The scatter of the unified analysis can be reduced by

separating out data on specimens of different size and

geometry. The effect of specimen size is shown in Fig. 31 for

the 1% CrMoV steel, compact tension results.

It can be seen that the total spread in the data is much less

than is depicted in Fig. 26. However, there is still

significant scatter at the lower cracking rates observed in

the initial stages of a test. This is the so-called 'tail'

region where a small increase in C* (or K) is responsible for

a large increase in a. Later a linear relationship is obtained

in Fig. 31 with little scatter consistent with the form of Eq. (9) assumed in the numerical predictions.

A detailed examination of the experimental data has revealed

that the tails represent a period during which a decreasing or

an approximately constant displacement rate prevails. It can

occupy up to about 3 of the overall lifetime [6] of a test.

- 85 -



The linear region on Fig. 31 corresponds with a progressively

accelerating displacement rate and is associated with having

achieved a steady-state distribution of stress and damage

ahead of a crack tip. An approximate expression for describing

this behaviour has been given by Nikbin, Smith and Webster[48] as

à = 3 • C*°'85/e£ (11)

with â in mm/h, C* in MJ/m h and e4 is creep ductility

appropriate to the state of stress at the crack tip. This is

taken as the uniaxial creep ductility c_ for plane stress

conditions and ef/50 for plane strain. The predictions of this

expression, for an average uni-axial creep ductility from Fig.10 of e f = 0.15 for the 1% CrMoV steel, are shown in Fig.

31. It is apparent that good agreement is obtained when plane

stress is assumed consistent with the numerical calculations

shown in Fig. 20 to 22. In addition there is no evidence of

size effects in Fig. 31 over a range of specimen thicknesses

from 10 to 63.5 mm.

Initiation of Crack Growth in 1 % CrMoV steel

An example, which sheds light on the first part of an

experiment with the 1 % CrMoV-steel specimens after loading is

shown in Fig. 32. Plotted are the evolution of the load-line

crack opening displacement V and of the final crack extension

Aa in the CT 50/100 specimens AP2 and AF3 (thickness B = 50

mm, width W = 100 mm, 20% side grooves). The crack growth was

measured also by the direct-current potential differencemethod, see Fig. 33. From these data, it can be seen that the

initial crack growth is approximately linear with time. The

microstructure in the crack tip area in the middle of specimen

AP2 is shown in Fig. 34.

In addition, extrapolation of the crack growth and potential

values, respectively, in Figs. 32 and 33 to the beginning of

86



the tests indicates that for these two CT 50/100 specimens an

initiation time , i.e. an incubation time to the onset of crack

growth - if any - is negligibly small. This conclusion can

also be drawn from Fig. 35, where results of three CT 25/50

specimens are plotted.

A comparison has been made by [49] between the crack growth

observed in fatigue precracked and spark-eroded notched

identical CT 25/50 specimens with 20% side grooves under the

same loading conditions. The microstructure in the crack tip

area of two of these specimens is shown in Figs. 36 and 37.

It is apparent, from Fig. 36, that little crack growth has

occurred after 500 hours ahead of the spark-eroded notch,

indicating a significant initiation/incubation period, whereas

appreciable crack growth has taken place from the fatigue

precrack in 200 hours (Fig. 37) suggesting little evidence of

an incubation period consistent with Figs. 32-35.

Transient Crack Growth in 1 % CrMoV steel

It is claimed that the early cracking behaviour can be

attributed to the combined effects of primary creepdeformation, the development of a creep damage zone aground

the crack tip and a redistribution of stress during the

transition from the initial elastic to the steady-state creep

conditions. An indication of the redistribution time can be

obtained from [1]

fc i = T ïS îyc * <12>

where G is the elastic strain energy release rate. Since this

formula is considered to provide an upper estimate of t.,

stress redistribution should be essentially complete for t >

t 1. In the case of the 1% CrMoV steel, t. is tyically around

10 h. Strictly speaking C* is only valid for values of t > t..

It has been found that elimination of data points with t < t.

still leaves most of the 'tail' so that some other explanation

- 87 -



is required of this behaviour. This can be obtained as

follows.

Consider Fig. 38 which shows a creep damage zone ahead of a

crack. When a steady state distribution of damage has developed in this region it has been shown [48] that the

steady state creep crack growth rate å is given by

as = 1 - 1 ) [ Ş ^ ]n / n + 1 )

c r c ) V n + D (13)

where I is a non-dimensional function of n and state of n

stress and C is the proportionality factor in the Norton creep law. In deriving this relation, it was assumed that crack

advance takes place when the creep ductility is exhausted at

the crack tip. Under steady state conditions progressively

more damage exists as the crack tip is approached and little

extra strain is required to break a ligament dr at the crack

tip since it will be almost broken before the crack reaches

it.

This situation does not exist on first loading. The small ligament dr will not have suffered any creep strain and

failure will not occur until a time dt has elapsed, given by

e_ = e dt

This leads to an initial creep crack growth rate a of

_ . C*l n/(n+1),„'_ /< o=-b-[?] / V ' ^dr)V(n +l) (14)

which is similar in form to the steady-state relation Eq. (13) . Because the dr and r are raised to a small fractional ' c power in Eqs. (13) and (14),

s WIT as < 1 5>

For most materials therefore the initial crack growth rate is

88



expected to be approximately an order of magnitude less than

that predicted from the steady state analysis. This is

consistent with the experimental results shown in Fig. 31. For

each crack advance dr, each successive ligament in Fig. 38

will progressively accumulate more damage prior to fracture.The cracking rate will increase correspondingly to

(€{"€*)[-r

/(n+1)(c'dr)VCn+D (16)

L n J

where e* is the creep ductility used up in the ligament prior

to the arrival of the crack. Numerical integration is required

to evaluate Eq. (16). Its prediction of some tests is shown in

Fig. 39. The satisfactory correlation indicates that the

majority of the 'tail' can be attributed to the build up of

damage at the crack tip during the early stages of cracking.

On similar physical grounds, but in a more mathematical

formulation, the 'tails' have been described by Kubo et al

[50], Riedel [ 51], and Bassani [52].

For most tests it has been found that the build up of damageoccupies about the first 0.5 mm of crack extension. When this

cracking is eliminated from Fig. 39, as shown in Fig. 40, the

scatter in the data is further reduced.

Geometry Effects

In order to examine these effects for the 1 % CrMoV-steel, the

'tails' have been eliminated from the data for clarity. Theresults are shown in Fig. 41. It is evident that there is

little influence of geometry on the cracking rate although the

CT data tend to be distributed towards the top of the scatter

band. The use of CT data in design should therefore be

preferred as they will result in the safest predictions. In

Alloy 800 H, a geometry effect could not be found, see Figs.

29 and 30

89



Comparisons with ASTM and JSPS data on 1 % CrMoV-steels

These data are summarized together with the EGF results in

Fig. 42. It can be seen that, although the test conditions

were different, broadly similar correlations are achieved.

This would be expected from Egs . (11) and (13) unless a change

in temperature causes a significant change in creep ductility.

It has been advocated that the C(t) and C. [53] parameters can

be employed to interpret creep crack growth data when t < t..

Both of these parameters tend to C* at long times. Since it

has been found that t. a 10 hours for the test conditions

imposed, it is expected that the data should mostly correlatewith C* and the use of C(t) and C. is not necessary.

Initial Cracking Rates

It can be argued that the inital cracking rates for the

1 % CrMoV stee l, with a transition time of typically 10 h or

more before stress redistribution has had time to occur,

should be described by K. The correlations for standard size

CT geometries and the other geometries are shown in Fig. 43.

It can be seen that all the data can be described

'satisfactorily by the same equation [3 ] .

90 -



CONCLUSIONS

Experiments and analyses have been performed on a 1% CrMoV

steel and on Alloy 800 H which have shown that creep crack

growth in these materials is described most satisfactorily bythe creep fracture mechanics parameter C*. Recommendations

have been made about how to obtain the most reliable estimates

of C* from experimental measurements. These have been

supported by numerical computations. It has been found that

the use of 20% side grooves (10% of the total thickness each

side) in compact tension specimens, and a seven-point

polynomial fit to obtain crack growth and displacement rates,

produces the most consistent correlations.

It has been demonstrated for the 1% CrMoV steel that all the

cracking took place under plane stress conditions. Increased

scatter, due to 'tails' in the early stages of cracking, has

been shown to be caused mainly by the progressive build up of

damage at the crack tip until a steady state distribution is

reached. This can take up to 30% of the life of a specimen and

can be important in practical applications. However, little

evidence of a 'tail' was noticed for Alloy 800 H.

Comparisons have been made with data obtained in other test

programmes on the 1% CrMoV steel. These have reinforced the

findings of this investigation.

91



References

[I] Riedel, H., Fracture at High Temperatures, SpringerVerlag, Berlin, Heidelberg, New York, 1987.

[2] Ewald, J., Berger, C , Brachvogel, H., Investigations onCrack Initiation and Propagation under Static, Cyclicand Combined Loading Conditions of 1% CrMoV Steels at530'C; COST 505-D20/D21, Annual Progress Reports andFinal Report, Siemens-KWU, Mülheim, June 1989.

[3] Hollstein, T., Kienzier, R., Numerical Simulation ofCreep Crack Growth Experiments; COST 505-D22, AnnualProgress Reports and Final Report, IWM Freiburg,December 1988.

[4] Kanbach, H., Crack Growth in Welded Turbine Materialsand Elevated Temperatures; COST 505-D35, Annual Progress

Reports and Final Report, AEG Frankfurt, March 1989.

[5] Raggazzoni, S., High Temperature Crack Growth in SteamTurbine Rotor Material under Static and Cyclic Loading;COST 505-13, Annual Progress Reports and Final Report,ENEL, Milano, March 1989.

[6] Holdsworth, S.R., High Temperature Crack Growth inTurbine Steels; COST 505-UK5, Annual Progress Reportsand Final Report, GEC Turbine Generators Ltd., Rugby,1989.

[7] Webster, G.A., Djavanroodi, F., Elevated TemperatureCrack Growth in Steam Turbine Materials, COST 505-UK18,Annual Progress Reports and Final Report, ImperialCollege, London, January, 1989.

[8] Rintamaa, R., Salonen, J., Auerkari, P., Residual Lifeand Strength of Steam Pipings and Turbines, VTT, Espoo,Annual Progress Reports and Final Report, 1989.

[9] Bressers, J., JRC, Petten, priv. comm.

[10] Curbishley, J., UKEA, Risley, priv. comm.

[II] Fesneau-Falbriard, P., Héritier, J, UNIREC, Firminy,priv. comm.

[12] Guedou, J.-Y, SNECMA, Evry, priv. comm.

[13] Gooch, D.J., CEGB/CERL, Leatherhead, priv. comm.

[14] Hay, E., NEI, Newcastle upon Tyne, priv. comm.

[15] Hippsley, C.A., UKEA, Harwell, priv. comm.

[16] Huthmann, H., Interatom, Berg.-Gladbach, priv. comm.

92



[17] Krasovsky, A., Baumstein, M., Institute for Problems ofStrength, Kiev, priv. comm.

[18] Maile, K., MPA , Stuttgart, priv. comm.

[19] Mandorini, V., IR B, Milano, priv. comm.

[20] Nazmy, M., AB B, Baden, priv. comm.

[21] Pique s, R., ENSMP, Evry, priv. comm.

[22] Rant ala, J., Imatran Voima, Vantaa, priv. comm.

[23] Remke, M., RWTÜV, Essen, priv. comm.

[24] Rõdig, M., IRW-KFA, Jülich, priv. comm.

[25] Saxena, A., Ha n, J., Georgia Tech., Atlanta,priv. comm.

[26] Tscheuschner, R., Granacher, J., IfW, Darmstadt,priv. comm.

[27] American Society for Testing and M ateri als, ASTMCommittee E 24.04: Cooperating Program on Creep CrackGrowth, Chairman: A. Saxena.

[28 ] Japane se Society for the Promotion of Science (JSPS),Committee 129 and National Research Institute for Metals(NRIM), Chairman: A.T. Yokobori.

[29] VAMAS : Creep Crack Growth - A State of the Art Repor t,Issue 1, February 19 89, Ed. T.B. Gibb ons,NPL, Teddington.ton.

[30] Tscheuschner, R., IfW, TH Darmstadt, priv. comm.

[31] Kempf, B., Bothe, K., Gerold, V., Study of Fatigue andCreep-Fatigue Interaction in a High-Temperature Alloy, COST501-D11, Final Report , Stuttgart, 1987.

[32] Drossier, E., Danzer, R., Aigner, H., Mitter, W.,Lebensdauer von Alloy 800H unter kriechnahen Bedingungen,COST 5 01-Al, Endbericht, Leoben, 1987.

[33] ASTM-E 399-78 , Standard Method of Test for Plane StrainFracture Toughness of Metallic Materials, Annual Book ofASTM Standards, Section 3, Vol. 03.01.

[34] ASTM-E 8 13-8 1, Standard Test for Jic, A Measure ofFracture Toughness, Annual Book of ASTM Standards,Section 3, Vol . 03.01.

- 93



[35] Saxena A., Han, J., Evaluation of Crack Tip P arametersfor Characterising Crack Growth Behaviour in CreepingMaterials, ASTM Task Group Report, Joint Task Group:E24.08.07/E24.04.08, 1987.

[36] Webster, G.A., Crack Growth at High Temperatures, inEngineering Approaches to High Temperature Design, Eds.B. Hilshire and D.R.J. Owen, Pineridge Press, 1983.

[37] Hollstein, T., Djavanroodl, F., Webster, G.A.,Holdsworth, S.R., High Temperature Crack Growth in Alloy800 H and 1% CrMoV Steel - The Results of an EGF RoundRobin; in: Failure Analysis - Theory and Practice, ECF7,Ed. E. Czoboly, EMAS (1988) Vol. II, 656-668.668.

[38] Hollstein, T. and Kienzier, R., Fracture MechanicsCharacterization of Crack Growth Under Creep and FatigueConditions, IWM Report W 2/87, Freiburg, February 1987.

[39] Kienzier, R. and Hollstein, T., Experimental andNumerical Investigations of Creep Crack Growth, Proc.3rd Intern. Conf. Creep and Fracture of EngineeringMaterials and Structures, Swansea, The Institute ofMetals, 1987, 563-576.

[40] Hollstein, T. and Kienzier, R., "Numerical Simulation ofCreep Crack Growth Experiments", IWM Report Z13/87,Freiburg, December 1987.

[41] Koterazawa, R. and Mori, T., Applicability of FractureMechanics Parameters to Crack Propagation under Creep

Conditions", Trans. ASME, J. Eng. Mat. Tech. 99, 1977,298-305.

[42] Kumar, V., German, M.D . and Shih, C F . , An EngineeringApproach for Elastic-Plastic Fracture MechanicsAnalysis, Topical Report No. EPRI NP-1931, ResearchP roject 1237-1, General Electric Co., Schenectady, July1981.

[43] Landes, J.D., Begley, J.A., A Fracture MechanicsApproach to Creep Crack Growth, ASTM STP 590, AmericanSociety of Testing and Materials (1976) 128-148.

[44] Bathe, K,.-J., ADINA, A Finite-Element Program forAutomatic Dynamic Incremental Nonlinear Analysis. ReportAE 84-1, Massachusetts Institute of Technolgy, Cambridge,Mass., USA (1984).

[45] P arks, D.M., The Virtual Crack Extension Method forNonlinear Material Behavior. Comp. Methods Appi. Mech.Eng. 12 (1977) 353-364.

94



[46] Shih, C F . , DeLorenzi, H.G., Andrews, W.R., ElasticCompliance and Stress Intensity Factors for Side-GroovedCompact Specimens. Int. Journ. of Fracture 13 (1977)544-548.

[47] Siegele D., Ockewi tz, A., Hollstein T., Berechnung des

ij-Faktors für eine 3-Punkt-Biegeprobe, IWM-BerichtV 25/87, Freiburg, July 1987.

[48] Nikbin, K.M., Smith, D.J. and Webster, G.A., AnEngineering Approach to the Prediction of Creep CrackGrowth, J. Eng. Mat. and Tech., ASME, 108, 1986,186-161.

[49] Tscheuschner, R., Ma ile, K., Stichprobenartige Untersuchung des Kriechrißverhaltens von Proben mit angeschwungener und von Proben mit erodierter Rißstartfront,26. Oktober 1989

[50] S. Kubo, K. Ohji, and K. Ogura, An Analysis of CreepCrack Propagation on the Basis of the Plastic SingularStress Field, Engineering Fracture Mechanics 11, (1979)315-329.

[51] H. R ied el, The Extension of a Macroscopic Crack atElevated Temperature by the Growth and Coalescence ofMicrovoids, in Creep in Structures, (A.R.S. Ponter andD.R. Hayhurst, eds.) Springer-Verlag, Berlin Heidelberg(1981) 504-519.

[52] J.L. Bassani, Creep Crack Extension by Grain BoundaryCavitation, in Creep and Fracture of EngineeringMaterials and Structures (B. Wilshire and D.R.J. Owen,eds.), Pineridge Press, Swansea (1981) 329-344.

[53] Saxen a, A., Creep Crack Growth under Non-Steady-StateConditions, Fracture Mechanics, Vol. 1 7, ASTM STP 905 ,1986, 185-20 1.

[54] Rödi g, M., Kienzier, R., Nickel, H., Schubert, F.,Ermüdungs- und Kriechrißwachstum in Rohren einesRöhrenspaltofens bei Temperaturen oberhalb 700"C,13. MPA-Seminar, 8 .-9.10.1987.

95



General

Material supplied to program by:

Steelmaker :

Component :

Nominal composition/specification :

Composition, 7.

Fraunhofer Institut, Freiburg

Buderus Edelstahlwerke/Wetzlar

Bar, 386 mm dia

21CrMoNiV57

C

0 . 2 2

S i

0 . 2 4

Mn

0 . 6 4

P

0.009

S

0.003

Cr

1 . 2 9

Mo

0 . 6 6

Ni

0 . 6 6

V

0 . 2 8

Al

0 . 0 1 4

Cu

0 . 1 2

Sn

0.009

Heat Treatment : 6 h 930 C, oil

10 h 690°C, air

Room temperature mechanical properties:

Lo ng i tud . , removal p o s i t io n: near sur face

R p o z , MPa

613

Rm, MPa

727

A, 7.

1 9 . 8

Z , l

72

Av,J

1 3 9 , 158, 163

R p 0 . 2 , MPa

623

Rm, MPa

744

A, 7.

19

Z.7.

6 9

A v , J

9 2 , 84, 85

near KA1,

left side

see f i g . 1)

near DX,right side

(see fig. 1)

Table 1: D etails of the 1% CrMoV test materia l give n by the

steelmaker

96 -



Spec imen P o s i t io n R „ R Z Ac A, , n

po .2 m 5 L=60[MP a] [MP a] [%] [%] [%]

1-

< U

a.El1—EO

oOí

0oLT)L D

II

1—

AB 7

AA8A

AA63

AA65

CG61

CG63

CG65

AB8

AA8B

AA62

AA64

CG60

CG62CG64

center

0.23 R

R/2

surface

center

R/2

surface

center

0.23 R

R/2

surface

center

R/2surface

483

536

594

608

561

600

615

316

355

372

388

380

375397

635

676

705

712

683

714

721

345

389

400

410

398

400416

70

72

68

71

72

72

86

89

89

88

8989

21.0

22.0

20.5

/21.5

/

27.5

24.5

23.5

/

25.522.0

19.0

19.0

18.0

19.0

19.0

16.5

22.5

17.0

16.5

20.5

20.520.5

Table 2 : Re su l t s o f th e te n s i l e t e s t s on 1% CrMoV s t e e l (Rra d i us o f b ar ) , from [3 ]

- 97



Bezeichnung

der Probe

Kohlen toll *;.

Silizium ' I .

Mangan *' .

Phosphat * •

Schwein •/.

Stichïlolt 'f .

Aluminium ' losl Aluminium

Chrom '' .

Molybdän *'•

Vanadium 'f.

Nichel *'•

Kobalt V.

Wollram *'•

Tilan *'•

Niob *' .

Eisen ■'•

Kupier '•

Zink V.

Zinn *' .

Blei •'•

Antimon '■

Arsen */ .

/.

X /K = 0 . 1

. 2 1

. 2 5

. 6 1

. 0 1 1

. 0 0 2

. 0 0 9

1 . 2 9

. 6 H

. 2 6

.61»

. 1 1

. 0 0 8

. 0 0 ?

. 0 0 6

X /H - 0 . 2 5

. 2 0

. 2 5

. 6 2

. 0 1 1

. 0 0 2

. 0 0 9

1 . 2 9

.6 5

. 2 7

. 6 s

. 1 1

. 0 0 ?

. 0 0 2

. 0 0 6

X /R . 0 . 5

. 2 2

. 2 5

. 6 M

. 0 1 2

. 0 0 3

. 0 0 8

1 . 3 2

. 6 7

. 2 7

. 6 6

. 1 1

. 0 0 8

. 0 0 2

. 0 0 5

X /R . 0 . 9

. 2 2

. 2 5

. 6 3

. 0 1 2

. 0 0 3

. 0 0 8

1 . 3 1

.65

. 2 7

.66

. 1 1

. 0 0 9

. 0 0 2

. 0 0 7

Round Robin Mate r i a l

X / R - D i s t a n c e from cen te r

21C rMoN iV57

Po s i t i o n ; A A

Table 3 : Chemical composition over cross s e c t i o n , from [ 2 ]

- 98



(OCO

AxialPosition

BA

DA

AA

Stress(MPa)

263247

216

185

154

139

216

216

0.1

22

7

17

42105

5

4

0.2

89

29

75

290565

24

17

Time to X Strair

0.5

3436

142

428

1,600

3,089

117

100

1.0

6992

3661,102

4,170

7,700

320

302

(h)

2.0

118

185

770

2,250

7,900

690

635

5.0

184

333

1,383

4,000

1,275

1,150

Rupture

time(h)

224

429

1,595

4,49512,486

17,908

1,577

1,362

EL(X)

22.3

18.7

9.06.85.7

3.9

14.3

13.2

RA(X)

78.4

75.8

39.9

12.95.0

5.7

43.9

40.8

Table 4: Creep rupture test results for 1% CrMoV steel at

550"C, from [ 6] ; see Fig. 1 for key to axial pos ition



Spec.-No.

AK 6

AK 7

AK 4

AK 5

AK 2

AK 3

AK 8AK 1

Stress[MPa]

285

260

235

190

155

135

10580

Time

0.2

3.5

4

8

100

530

1425

300012000

to % strå[h]

0.5

22

22

80

460

2000

3900

13500

in

1.0

36

65

200

1010

4400

9500

Rupture

time[h] El[%]

88 22.5

237 25.7

900 19.3

4193 10.8

10661 5.9

RA[%]

81

77

59

18

10

Table 5: Creep rupture test results on 1% CrMoV steel [26],

axial position AK, see Fig. 1.

100



Production Procedure:

Melting in an electric arc furnace andVOD Process +)

Hot forming to ø 250 mm

Hot forming toa 90 mm

Rolling to a 19,5 mm

Solution annealing1125 - 1130* 30'/Water quenching

Hot forming to P roduction of seamless28 x 103 mm tubes by extruding

Solution annealing Cold tube rolling1130* 30'/

Water quenching

Solution annealing1130" 30'/Water quenching

.+) VOD ... vacuum oxygen decaburisation

Chemical composition of the Incoloy 800 H investigated:

Si Mn Cr Ni Al Ti Fe

Weight 0.07 0.46 0.68 0.020 0.004 20.26 31.11 0.34 0.31 bal

(%)

Plate dimensions : 28 x 103 mm

Heat treatment: 1130' + 30'/Water

Mechanical data:

Testing conditions: Round bar, d = 8.0 mm, L = 40 mm

Temperature

[•C]

25

800

or =

%

573577

282270

5 MPa/s

[N/mm2

RP 1

289293

190192

RP0.2

250254

155161

[i

A

4646

5252

Z

7474

7474

Charpy energy (ISO-V-specimen, 25"C): 297 J

Table 6: Production procedure, chemical composition and some

mechanical data of the Alloy 800 H



No.

1

2

3

4

S

6

7

S

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

Specimenno.

AF 2

BW 1

BW 2

BW 3

AH 3

AH 4

AM 1

AH 2

AO 3

AO 5

AQ 5

BC 1

BC 2

BB 1

AB 2A

AB 2B

AN 6

AE 1

AE 2

AP 2

AP 3

AQ 1

AQ 2

RR 5

RR 7

Specimentype

CT 25/50

CT 20/50

CT 20/50

CT 20/50

CT 25/50

CT 25/50

SENT12/20SENT12/20

CT 20/40

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT63.5/127CT12.5/40CT12.5/40

CT 25/50

CT 25/50

CT 25/50

CT50/100CT50/100CN12.5/50CN12.5/50SENB9.5/19

CT 13/26

S i d e -groove[*]

20

20

20

20

20

20

0

0

20

20

20

20

20

20

20

20

20

20

20

20

20

0

0

20

20

F[kNJ

16.5

10.2

13.5

17.0

11.1

19.1

30.5

40.0

15.1

16.2

24.1

16.5

12.1

70.0

7.6

5.1

8.5

17.8

20.0

32.1

32.0

50.0

60.7

6.0

5.6

a/W

0.498

0.547

0.569

0.548

0.539

0.540

0.214

0.188

0.450

0.505

0.516

0.532

0.535

0.535

0.513

0.513

0.538

0.536

0.538

0.539

0.540

0.426

0.547

0.449

0.556

NO.

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

Specimenno.

BF1/CC P 5

B F 2/CC P 6

AN 1

AN 3

AN 4

AA 69

AB 3

AB 4

AP 1

AH 1

AH 2

AG 1

AG 2

AG 3

AI 1

AI 2

AB 5

AB 6

AI 6

AI 7

AE 5

AC 1-6/AD 1-6

BD 3

BD 5

AK 4

Specimentype

CN 26/51

CN 26/51

CT 20/40

CT 20/40

CT 20/40

SENB 6/6

CT 2 5/50

CT 25/50

CT50/100

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT 15/30

CT 15/30

CT 25/50

CT 25/50

CT 25/50

CT 25/50

CT12.5/40

S i d e -groove[*]

20

20

20

20

20

0

20

20

20

20

20

20

20

20

20

20

20

20

20

20

20

20

20

20

20

FfkN]

84.7

107.6

12.5

12.5

10.0

0.88

11.5

18.5

26.7

11.1

19.1

10.5

17.9

16.1

16.0

10.8

10.8

26.7

10.9

14.1

20.0

20.0

a/W

0.509

0.504

0.548

0.563

0.550

0.528

0.566

0.555

0.535

0.628

0.545

0.557

0.561

0.536

0.536

0.515

0.506

0.491

0.550

0.564

0.470

0.375

Table 7: Data of fracture mechanics specimens; 21 CrMoNiV 5 7

T = 550 °C; constant force

- 102



LaborayIMPIMP

RWTUV

RWT UV

1 ATOM1 ATOM

KFA

KFAUNIREC

UNIRECUNIREC

F HIWMF HIWM

ENELENEL

P e l t e n

Petten

No1

234

56

78g

1011

1213141516

17

Spec imen NeA76

A77A89

A90

A69A70A64

A66

A93A94A95

A13A 6 0 4A91

A92A72

A73

S p e c im e n t y p eCT 25/50CT 25/50CT 20/40CT 25/50

CT 12.5 /25CT 12.5/25

CT 25/50CT 25/50

CT 20/40CT 20/40CT 20/40

CT 25/80CT 25/50CT 25/50CT 25/50

CN 4.8/12

CN 4.8/12

Side qroove %

202 02 02 0

2 02 020

2020

2 02 0

2 0

2 02 02 0

F rkNl11.56.364.3

7

1.71.6

65.5

9

45.214 .38

a /W0.60.60.6

0.60.60.5

0.5

Ki [MPaVml11.5

1515 .62 0 . 2

10 .89.7

12 .3210 .88

1114

8

14.49.9

11 .169 .35

a i imm/hr l3.89E-07

6.67E-06

3.46E-05

7.61 E-05

2.05E-065.28E-07

3.58E-06

8.83E-06

5.83E-06

2.78E-05

8.33E-07

7.5E-06

4.86E-07

2.2E-06

4.05E-07

Table 8: Data of fracture mechanics specimens; Alloy 800 H;

T = 800°C; constant force

103



EE

toA

. J _A A

- 2 * J -S A

1000

^ c h a r p y s p e c i m e n s

L A

2000

D"X

I3000 mm

^ c h a r p y s p e c i m e n s

Fig. 1: Specinen layout for the tension tests

Rptt2

IM P Q]

700-

6 0 0 -

500

400-

300

2 1 C r M o N i V 5 7

25 °C

25°C

550°C

550°C

~i 1 1 r0

I I I I I ~0.5 1

d i s t a n c e f ro m c e n t r e x /R

Fig. 2: 0.2% proof stress RPo.2 and ultimate tensilestrength Rn for different radial and axial positions(see Fig. 1, from [3])

104



« 150

D

- 3

T=20"C

2 U r M o N i V 5 7

O O

D AA. tang / rad

O DA, t an g /rad

' 0,9 0,8 0,7 0,6 0,5 0,4 0.3 0,2 0,1

Distance fram cen ter x/R

Fig. 3: Charpy energy variation with radius , from [2]

» I S O

50

O OD

aa a

a_ o

ego

CDtb

8o

. 4G

- 1 6 0 -120 -8 0 -tO

21CrMoNiV57

D AA, x/R= 0 55, ta n g/ra d, FATT=2S°C

O DA. x/R= 0.55. ton g/rad, FATT=25°C

*0 60 120 160 Temp era ti«¡n »C

Fig. 4: Charpy energy versus temperature, R/2 - position,from [2]

105 -



•E 150 -

100 - '

50 -•

OO o

D ü

ooDB

O DD

-160 -120 -80 -40 to 80

21CrMoN¡V57

D AA,x/R=0.rang/rod,FATT= 55 "C

O DA.x/R=0, tang / rad, FATT= 35 'C

120 160 Temperature ¡n C

Fig. 5.:Charpy energy versus temperature, centre position,from [2]

21CrMoNiV57

Av in J

20°C 80°CAA, ra n / ro dDA. " "

Fig. 6: Impact energy over cross section [2]

- 106 -



: -. fjft.̂ tv,jv..:.- _\ i 9 8 2 3

CJS^-TTT^ i f p i - / - w p i » nn**i * -«(»> ^ «• »**»■

■^ ̂ ^¿¿¿âfc^'i, ̂ ^ '::A »J?* 9827

y OP*-« *v.

0.1 mm

k«..« 9831

Fig. 7: Microstructure of 1% CrMoV steel, longitudinal sections



Creep stra in %7

6

5

U

3

2

1

0

216MPaf-550°C AA/ /

/ -0A,-

-

.4?^

/ // ' /' ' /

•' '' // ' ' /

''''/

.si''Cs^ 9 *?° ,3P°^¿¿¿^

,

AA BA

RA

2600 mm

DA

T I IAK

•I BCG

500 1000 1500Time - hours

Fig. 8: The effect of axial position on the creep propertiesof 1% CrMoV at 550"C, from [6]

e [%]

100O 2000 3000 4000 50Ò0t [h]

Fig. 9: Creep properties at 550'C, ( : Region BA [6], :Region AK [26]; for axial position see Fig. 1)

108



500

200

100

20

A sec t i on A A 1

• s e c t i o n B A > H o l dsw of t h ( 1 9 8 6 ) [ G ]f s e c t i on O A J

O sec t i on A K . G r a n a ch e r . Tsch e usch n e r 1 13 88 } 1 2 6 1

10

o -& I

_ W

« 20

010 102 W3 10' Ws

Time to Rupture [h )

Fig. 10: The creep rupture properties of 1% CrMoV steel at550'C

21 CrMoNiVS7, 550°C

100 200

a [ M P Q ]

F i g . 1 1 : T he e f f e c t o f s t r e s s o n m inim um c r e e p r a t e a t550*C

- 109



Erran [ h"1

10

10 5

10

10"

-6 _

21CrMoN¡57. 550°C

è m n =7,06-10- 9e W 3 6 9 d

_L j L100 150 200 250 tf[M Pa]

F i g . 1 2 : S t r e s s d e p e n d e n c e on t h e m inim um c r e e p r a t e

' 0 100 200 300 dWPa]

F i g . 1 3 : D e t e r m i n a t i o n o f t h e p a r a m e t e r c o, c r e e p la w ( 3 )

110



log R

-1,5

- 2

21 CrMoNiV57, 550 °C

R=0,00308-cr0-5183

80 100 150 200 250 300 d [MPa]

2,1T

1,9 2,1 2,3 2,5 log d

Fig. 14: Determination of the parameter R creep law (3)

YfXVW/ <J

Fig. 15: Microstructure of Alloy 800 H

- 111



l ø - 3

10

10

10 -

10

10

•

:

:

i a^d^

s ^Q

L^

V ***

£» in _

^

xx S

- 1 5 . 2 4 + 6 ,4 Log v [h ~ , i

10 □ [ M p a ]

Fig. 16a: Ţhe stress dependence of êmin (full symbols) and of e at 1% plastic strain (open symbols); round bar results (19.5 mm ø ) , see Table 6, from [31]

[ M p a ]

Fig. 16b: Relationship between the life time tt and the applied stress o- from [31]

112



55

A

A

1 4 -

27

A

B

14

i

A

C

.27

A

D

.27 ■ ,

A

E

27.

A

F

2i

A

G

.27.

A

H

2 7i

A

I

2:

A

K

,27

A

L

,27

A

M

27,

A

N

27

-

A

0

.5 3

A

P

, 53 22 5 5

A t

Q

\ A

? S

■800

m m

Schn i t t A-A

io i

mm m m

27 A I B 27 A T E

Fig. 17: Typical layout for 1% CrMoV specimens; CT 25/50 specimens, ABl, AB3 - AB6 and CT 12.4/40 specimens AB2A and AB2B and CT 25/50 specimens AE1 - AE6.

113



108

500

471 ţ 1

A 61

o |o

A6 2

o|o

475 [ 1 T 1

A6 3

o |o

A64

o |o

« • [ — 1 — 1

AGS A66

o |o o |o

A67

o |o

A70

o|o A69

o|o AGS o | o

O' o o jo o jo o o

O O

o |o o |o o j o

o jo

o jo

o jo

o|o

AB7

o|o A8G

o lo

A89

o|o ABS

o|o

A 90

o|o A91

o|o A92

o|o A93

o|o A9 Í

o |o A95

o |o

F i g . 1 8 : A l l o y 800 H c u t - u p p l a n

2.0

15

1.0 -

0.5

A

A

-ffrlV-f). Ref. (35)

^ ( 2 + 0 . 5 2 b / W ) , Ref.(3M

FE-Results:

Reference

plane stress

plane strain Specimen

(Í2)

V

CN

A

A

CT

(38-41) O

X

CN

□ + CT

2n -1

n

Fig. 19: Values of uc, a comparison of different approaches; CT: a/W » 0.54, CN: a/W » 0.43, 7 and ß are functions of a/W, from [37]

- 114 -



^ J

21CrMoNiV57, 550 °C

CN 12.5/50, AQ1

-Rt

E c r=e 0 {1 -e -R ,

} +

emin-t, Eq.(3)é m i n = 1 . 3 -10 " 2 O a 6 5 [h - , ] l d i n MPa

plane stress

-i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r

0 .00 1 .00 2 .00 3 .00 4 .00 5 .00 6 .00 7 . 00 8 .0 0 9 .00 10 .00 11 .00

i . sec CE+06)Fig. 20: Experimental and numerical values of C* for CN

12.5/50 specimen AQ1, from [3]

Fig. 21: Experimental and numerical evaluation of C* for CT12.5/40 specimen AB2B, with and without crackgrowth, creep law (1), from [3]

115



Fig. 22: Experimental and numerical values of crack growth, crack growth law: â = 7.41 • IO-3 • C* °'72 (a in mm/s, C* in N/(urn-s)), from [3]

• . in -

U

T « S

s - : E

m ■G)

m ■

10 KS

in -

: D 0

0

*

D

D

U D D

DO

l/ir

D o #

- i? i \ 1 :

V

X *

ín RB

« å

Wf i f l

ko* J . . 1 +

I 4

^

V

* •u e dk *

»fei ¡s fB

£ 4

B

B V B

'B

T u k„ H

S

#

' ♦

M

o

5 G

5 0 Z

x

o

» B

H 1

' l 1

E

?

O 1

1

I K

F

1

^

[ K

1 1 9

K IMPaÆ]

Fig. 23: Crack growth rate å in 1% CrMoV steel at 550*C as a function of stress intensity factor K -participants' evaluation

116



C IkJ/fmelili

5 -A ¡ J 0

C [ N / i n m / scc l

Fig. 24 Crack growth rate å in 1% CrMoV steel at 550'C as a function of C* integral - participants' evaluation

D UÜ3.B[HBD3.,BWI.BW2IBW3.AII3.AH4.A03.AB5.AB6.BCI,BC2.AN6.AEI.AE2

a AN41AF2,AB3.AB4.AHl.AH2.AGI.AG2,AG3,All,Al2.AE5.ACI-6 (CT 25/50)

■ ANI,AN3.AN4,AO3(CT20/40)

* AB2A.AB2B.AK4 (CT I2JM0)

■ APl.AP2j\PJ(CTSttlOO)

X BE3I.BE32(CTIO20)

. BBI1CT63J/I37)

A A16, AP (CT 13/30)

■ RR7 (CT 1 3/2È)

a CCP3.CCPÍICN257S»

g R U (SENB I2J/50)

K IMPaVm)

Fig. 25: Crack growth rate à in 1% CrMoV steel at 550'C as a function of K, unified evaluations

117 -



Q *qoMH COILHDI * ■■tlCTOi'1»)

+ A * J A » U | C T IU - ^

C* (J/m 2hr )

Fig. 26: Crack growth rate â in 1 CrMoV at 550'C as a function of C*, unified evaluations

1 10 1000

RI to y 6BBH , T-B00*C

a Ml ■ Ul O

A M 9, » M . □ U4 , 0 U ) ,

, U l ; CT H /W ; CT M/40

, ui-, cr H/sa , Ul ; CD 4 . 8 ƒ I I; CT » / S Ol; CT IS/SO, «0 ; CT II. S/ISAI. U/ l . U/2; VM/*, M S , M t ; CT H/SO, A M . U S ; CT n / 4 0

° M

.fo a

»•«H '

î r ^^ «

Fig. 27: Crack growth rate å in Alloy 800 H at 800*C as afunction of C* integral - participants' evaluation

118



10 '

IO '

I O 4 '

■ A T M 7 7 . A » 0 J W I . A W > 6 4 , A 6 6 . A « M Í C T Í Í ' » )

♦ A».A»3.AW.A«(CT2(Wtl)

M AW.A70 (CT UJ /2 5 )

M A IUCTH /W )

10° 101

K ( M P A VW )

1 0 *

Fig. 28: Crack growth rate à in Alloy 800 H at 800 C as a function of stress intensity factor K, participants' analysis.

Æ

È E

iu ■

I O ' :

i o 0 -

l o -

i o : -

l O ' -

i o J -

o

♦

X

o

H

A76. A77,A64.A66.A90.A<)1,A9

Afc9.A93.A94.A95 (C T 20/40)

Af.<>.A70(CT 12.5/25)

A 7 2 .A 7 3 (C I * H I / | 2 )

A13(CT2S/>0)

'A?

A' /i

DO t

, A 6 W | C T : 5 - S 0 )

tp u

o

D

♦ - /

oo -o o o / ♦

få /9

i ¥/t

á = 0003¿ - C * 0 - 7 3 3

. r mu l i a [ — J

c* M M L ImmsJ

■i t— i IO 2 IO 3 IO4

C* (J/m7hr)

Fig. 29: Dependence of creep crack growth rate on experimental C* for Alloy 800 H at 800'C (unified evaluation)

- 119



SL

å = 0.0034 C

0.733

Incoloy 800H , T « 800 t ;

Specimen (LT)

PTC

•

125/50

CN

X

12.5 /50

CT

+

2 5 / 5 0

CT

• ■

25/BO

CT

o

D

12 5 « 0

Looding

F = const V = const ï = cons t

B /W

207* side grooves

2 3 4 3

C* , N/mm / s e c

Fig. 30: Creep crack growth rate å as a function of C* for Incoloy 800 H (Z= constant rate of cross head displacement). From [38].

o CT63J/I37

a CT5CH00

+ CT25I50

a CT2IV40

■ cr IISMO

m CT 13/36

4 CT 13/30

D (Titrai

v. /y

'S êi:

>mP

*? — E q ( 1 1 )

C« ( J /m ' h r )

Fig. 31: Effect of specimen size on crack growth rate in 1% CrMoV steel at 550 C for compact tension (CT) specimens

120



v Imml

0.6 -

ûa

[mm]

0.4

0 2

21CrMoNiV57, T = 55 0 C CT 50/100, 20V. SG F = 32.1 k N , ao /W = 054

- 2

Fig. 32: Crack opening displacement at the load line and final values of creep crack growth as a function of time; CT 50/100 specimens

PROBE RP3

. 262

. 2 G

. 2 5 8

. 2 5 G

J 254

« . 2 5 2

0

\ 25 u E . 2 4 8

a. 2 4 4 -

. 2 4 2

. 2 4

CT50/100-AP3. 20V. SG

- j 1 1 1 1 1 1 1 1 1 1 1 i i i i ■ » ■ ■

2 0 0 400 G 0 0 B 0 0 1000

time [h]

Fig. 33: Normalized potential difference as a function of time; specimen AP3, see Fig. 32

121



Fig. 34: Microstructure in the crack tip area of CT 50/100specimen AP2

i — i — i — i — i — i — i — i — i — i i ' r

Fig. 35: Crack opening displacement at the load line and final

creep crack growth values as a function of time; CT25/50 specimens



*V- r- X

.■Í » T

Spark eroded crack l i

,A

T ^ ^ W ^

l

^ « reep crack H *ţ-f > ;£,/, ^ ^ V : ? & 3 eroded cracks ,-* ̂ _.\ \ V -w^^kjtr*'~ 3 0 5 * « ^ , X

Fig. 36: Creep crack initiation in the middle of the crack front, 524 h after loading, from [49]

^¡fesífí Fa t i g u e c r a ck Creep c r a ck T 2 - ^ í ^ V ^ U l l r l i . wr-- - ^ T aw ' S uK r t ß SB v ■?>■ < V ^

Fig. 37: Creep crack initiation in the middle of the crack front, 209 h after loading, from [49]

123 -



Fig. 38: Development of damage in creep process zone atcrack tip

t' (J/m'hr)

Fig. 39: P rediction of transient crack growth at 550'C in1% CrMoV steel for some compact tension specimens

124



o CT6Î-VH7

( CT » 1 0 0

♦ CTJS50 /

a CT 20140 , . » *

B CT 12 J/40

■ CT1VI6

aï: / ?

¿¿F-A CT 15/30

a CTiaao

♦ v * : * * * //;:vy

Eq.111)

C* ( J / m *h r l

Fig. 40: Effect of specimen size on creep crack growth rate in CT specimens with 'tails' omitted

E I O 2

E

CT

CN

S E N T

S END

A * ♦ •

* # *

?r.y

C» M/m>hr )

Fig. 41: Effect of specinen geometry on creep crack growth rate in 1% CrMoV steel at 550'C with 'tails' omitted

125



EGF 550°C

EGF550*C

JSPS S38°C

— — JSPS S38-C

ASTM538°C

— — ASTMS38«C

C» (J /m 2 h r )

~ I""1-

•S 10"21

EGF 550*0

EGFS50"C

JSPS 594-C

JSPS 594»C

ASTM594°C

ASTM 5940C

10' 10° 10

C* (J/m2hr)

Fig. 42: Comparison between EGF data at 550*C with ASTM [27]

and JSPS [28] results at

a) 538'C and b) 594*C

126



1.0E-

1 .0E-

06-

07-

08-

21 Cr Mo Ni V 57

T = 550 °C

x Ä

•

X

X

XX .

o s*°

•

X jS Q

jf X

x X x / \

; jy^ a N à i = 6t4-10"13-K*'

*X * Ref 1 3]

x / x/ *x/ ° X/ D

XX

X

X

• CN 26/51

x CT 25/50 ; CT 20/50 ; CT 20/40

• CN 12.5/50

n CT 12.5/40 : CT 13/26 ; CT 15/30» SENB 6/6 ; SENB 9.5/19

a CT 50/100 : CT 63.S/127

o SENT 12/20

30

K¡ [MPa*fm ]

40 60 70 80

Fig. 43: Initial crack growth rate vs. initial stressintensity factor

127 -






Part III

High Temperature Fatigue Crack Growth

S.R. Holdsworth

GEC - Alsthom Turbine Generators Ltd, Rugby, UK

129 -





INTRODUCTION

High temperature fatigue crack growth (HTFCG) data may be used in the power plantindustry to assist in the specification of acceptable defect sizes, in remaininglife assessment and in failure diagnosis. It is also used as an aid to materialselection. Fracture mechanics parameters have been used to model crack

propagation at elevated temperatures for over 20 years. However, for much of thisperiod, a major difficulty has been the inherent size and geometry dependence ofthe available crack growth correlating parameters when creep processes havedominated the fracture process. In practical terms, this has meant that it hasnot always been possible to accurately assess defect tolerance in large powerplant components on the basis of results from relatively small laboratorytestpieces. The main aim of the COST 505 High Temperature Crack Growth WorkingGroup (Table 1) was therefore to pool the resources of a number of Europeanindustrial organisations and academic institutions in a joint consideration of theproblem. The following Review examines the results of the Group investigationsrelating to High Temperature Fatigue Crack Growth in Steam Turbine Materials[ 1 - 7 ] , with reference to other published findings.

The term high temperature fatigue crack growth covers a range of crackingmechanisms resulting from a wide spectrum of loading cycle types. Stress-straincycles may be linear elastic or elastic-plastic and have been applied in load orstrain control. HTFCG may be due to one or a combination of pure cyclic, primarystress creep or secondary stress creep loading, and is dependent on such factorsas material properties, temperature, strain rate, cycle type and hold time. Theservice cycles experienced by high temperature steam turbine components such asrotors, valve chests and inner casings are complex and comprise a variety ofloading transients conceivably spanning the entire range of behaviour.

In laboratory tests mounted to generate HTFCG data for the above components, it isoften convenient to adopt an idealised isothermal cycle which closely models themost damaging service loading transient at the specific location being assessed.

For example, the operational cycle experienced by a rotor is composed of acombination of rotational and thermal transients ( F i g . l ) . The particularcombination varies with position in the rotor, and may be dominated by either theprimary stress or secondary stress cycle. Hence load control cycle tests may beadopted to model the rotational stress cycle at the rotor bore, whereas straincontrol cycle tests may be selected to model the thermal stress transient at a rimposition. A similar situation exists for valve chests, inner casings and mainsteam pipework, which experience operational cycles comprising superimposedpressure and thermal transients. Thermal stress variations tend to provide themost likely driving force for cracking at critical locations in valve chests andcasings whereas pressure stress cycles are likely to be more influential in thecase of pipework. The load and strain control cycles selected by the COST 505HTFCG investigators are typical of idealised isothermal cycles adopted to model

service loading transients in such components ( F i g . 2 ) .

FATIGUE CRACK GROWTH REGIMES

Fatigue crack growth behaviour is conveniently considered in terms of threeregimes (Fig.3) [8]. These are a lowAK regime close to the fatigue crackthreshold, A , K ° . a mid AK regime in which propagation rates are modelled by a powerlaw [9] ( E q n . l ) , and a high AK regime in which K M » approaches K c

At low AK levels close to A K ° , the magnitude of da/dN is very sensitive to smallincreases in AK and dependent on the same factors which influence ̂ Ko. These are

- 131 -



material microstructure and yield strength, temperature, environment and R ratio(ie. mean stress). Various expressions have been devised to model \Ko[eg. 10,11], although none are specific to low alloy creep resistant steels atelevated temperatures.

In the mid AK regime, crack growth rates are effectively modelled by an expressionof the form:

da/dN = C (AK)m (1)

where C and m are constants dependent on material, temperature and environment,and m is typically in the range 2 to 4. Propagation rates in this regime are lesssensitive to microstructure and mean stress effects. There have been a number ofmodifications to the Paris Law which, for example, minimise its dependency onmaterial properties and temperature [10] or extend its range of applicability intothe low and high AK regimes [12,13].

In the highAK regime, da/dN becomes increasingly sensitive to the level of AK,

and particularly to K M J U C

as critical Kc or plastic collapse is approached.Depending on the deformation and fracture characteristics of the material, crackgrowth rates can be strongly influenced by size and geometry. In thesecircumstances, AK is not the most effective correlating parameter, and alternativecyclic load functions have to be employed to minimise any dependence on size andgeometry (see below). In addition to the factors already listed, da/dN in thisregime is strongly dependent on microstructure, temperature, environment andfrequency (ie. strain rate). Tearing fatigue [14] does not fall within the scopeof this Review.

In the following text, the term low strain fatigue (LSF) is used to refer to theload/strain transients resulting in linear elastic loading cycles and crack growthin the low and mid AK regimes (Fig.3). Load/strain transients responsible forcyclic plastic loading involving some degree of general yield in tension and/orcompression are referred to as high strain fatigue (HSF) cycles.

HSF crack growth rates are due to higher AKs and may be influenced by theload/displacement control mode, particularly when there is superimposed creeploading. Since HSF cycles can involve yield in compression but not tension, thereare circumstances when the effective AK responsible for crack opening may belinear elastic. Consequently, it is not possible to rigidly fix the lower boundof the HSF crack growth rate regime in Fig.3, and caution with the terminologyused in this area is necessary.

HTFCG CORRELATION PARAMETERS

Low strain fatigue crack growth rates expressed as a function of AK areindependent of size and geometry for a wide spectrum of engineering materials attemperatures below the creep range. This means that crack growth rates determinedusing laboratory specimens may be reliably applied to large components. AK isalso an effective correlating parameter in the LSF regime at elevated temperatureswhen frequencies are relatively high (ie. f>lHz). Cyclic stress intensity factoris a function of cyclic stress, A0\ and crack size, 'a' (Eqn.2), and K solutionsare now available for a wide range of laboratory testpiece and componentgeometries [15].

AK = Y A < ra°- s (2)

132



where Y is a compliance/geometry function.

It is sometimes convenient to consider the mechanics of fatigue crack growth interms of cyclic CTOD [ 1 6 , 1 7 ] . For example, da/dN = ¿CTOD/2 is regarded as auseful upper bound fatigue crack growth law for non-work hardening materials [17].

The crack tip opening displacement is generally taken to be [ 18]:CTOD = 0.44 K 2 (3a)

E OV

Assuming that local crack tip flow stresses are increased by hardening to 2ÖV, anestimate of maximum cyclic CTOD is given by:

¿CTOD = 0.22 ( A K T Q T ) 3 (3b)(1 - R) E CÑ-

where R = K M I N / K M A X .

In practice, fatigue crack growth is the result of cyclic crack tip opening anddoes not occur when the crack is closed [19]. Eqn.3b is a reasonableapproximation of the situation when the R ratio is positive (ie. when K M I N andK M A X are both due to tensile loading) and when there is no premature crackclosure, for example due to crack face oxidation. However, this is notnecessarily the case when the crack tip is loaded in compression for part of thecycle (ie. when R is negative). It is then more appropriate to think in terms ofcrack propagation being due to the effective A K responsible for crack opening, ie.

ACTOD = 0.22 Í A K g . ^ ) * (3c)

E ÖV

To derive A K E F F it is necessary to know the crack opening load ratio, ie.

qo = (PMAJC - P O )/( P M A X - P M I N) (4a)

where P o is the crack opening load determined experimentally from P-Vc hysteresisloops (Fig.4) [ 4 , 2 0 ] . This approach is not a simple matter in practicalsituations, but there are now a number of empirical formulations to estimate qo[ 2 1 , 2 2 ] , eg.

qo = (1 - R/2)/(l - R) (4b)

AKeff is then simply defined by Eqn.5, and the well documented K solutions for awide range of laboratory testpiece and component geometries are still applicable[ e g . 1 5 ] .

AKarr = qo AKTOT ( 5 )

As the magnitude of A K increases, the crack tip plastic zone increases to a sizewhich is no longer small relative to other significant dimensions. Fatigue crackgrowth rates expressed in terms of A K= r r a r e no longer size and geometryindependent when general yield occurs in tension, and alternative parameters arerequired to describe HSF crack growth rates.

Cyclic J integral has been shown to be independent of size and geometry in bothlow and high strain fatigue regimes [23], and there are now 3 solutions availablefor a large number of standard geometries [ 2 3 , 2 4 ] . For example, for a simple

133



power law hardening m ater ial ( i e . Afe = H (¿O)*1):

kJ = H gi( a/W ,p ) (AOS**-)'*1*1» a (6)

where gi(a/w,/0 is a function representing the crack tip path independent lineintegral

for the geometry

of interest.

The physical basis

for using& 3

tocorrelate fatigue crack growth rates is considered in some detail elsewhere [17],

It is more usual to correlate HSF crack growth rates in terms of AK=o since itthen becomes possible to adopt a single parameter throughout the whole range of AK( F i g . 3 ) . Equivalent AK is simply AK=pr corrected for plasticity, and thecorrection may be applied either via a ¿J or an equivalent energy calculation[ 2 0 , 2 5 ] .

AK*o = A K ^ r Tl « AfeP E 1 (7)l «• A f e p E 1

where Afep is cyclic plastic strain. HSF crack growth rates expressed in terms ofA K E O

are independent of size and geometry [26],At low load/strain cycling rates and/or with the introduction of hold times, creepprocesses can influence the rate of crack propagation at high temperatures. Crackextension due to creep is not effectively correlated in terms of \K (or \K=o)because of crack tip stress relaxation effects. In these circumstances, creepcrack growth rates are described using the C* energy rate line integral which hasbeen shown to characterise stress and strain rates in the vicinity of a crack tipsubject to steady state creep conditions [27], The practical application of thisparameter is still relatively limited. Nevertheless, C* solutions are becomingavailable for an increasing number of specimen and component geometries [ 2 7 - 2 9 ] .For example, for a material obeying the secondary creep rate law de/dt = C'o"":

C* =C'gi(a/W,n) (5»«x'"*

l

> a (8)where gi(a/W,n) is a function representing the crack tip line integral for thegeometry of interest [24] , Creep crack growth rates for the COST 505 lCrMoV rotortype steel are independent of specimen thickness and geometry when expressed interms of C* [29].

LOW STRAIN FATIGUE

LSF without Creep

The term low strain fatigue was introduced above to refer to the load/straintransients resulting in linear elastic AK*rr and crack growth in the low to mid A Kregimes ( F i g . 3 ) . LSF crack growth rates due to cyclic stresses applied in eitherload or strain control are usually consistent in the absence of creep.

Threshold A K o levels increase with increasing temperature (particularly at low R),and the trend is illustrated for lCrMoV rotor steels in Fig.5 [ 1 , 1 6 ] . Since at550°C the higher frequency 10Hz/air AKo values are similar to those determined ata much lower frequency in vacuum, the increase in threshold is at least partly dueto the reductions in elastic modulus and yield strength incurred at the highertemperature. This trend is anticipated by reference to Eqn.3c (ie. ACTODo isinversely proportional to E and Ö V ) . In air, AKo increases further withdecreasing frequency as a direct consequence of premature crack closure due tocrack face oxidation [16] .

- 134



LSF crack growth rates increase with increasing temperature. In vacuum, crackpropagation rates are inversely proportional to elastic modulus and yield strength( i e . da/dN o«. ACTOD I * ( A K ) 2 / E O Y ) both E and OV decreasing with increasingtemperature. Modulus normalised crack propagation laws have been proposed toaccount for the effect of temperature [10], but their application should really belimited to the consideration of internal defects or to materials which do not

oxidise in the environment of interest.

With increasing temperature and decreasing frequency in air, oxidation becomesincreasingly influential in controlling fatigue crack growth rates [16,30,31].Even at relatively high frequencies, LSF crack propagation rates are generally notaccurately predicted on the basis of room temperature behaviour and a modulusnormalised growth law. Crack propagation rates are notably accelerated by oxideassisted growth at low AK levels in excess of AKo, and the effect is enhanced withdecreasing frequency ( F i g . 6 ) .

High frequency LSF crack growth rates for lCrMoV rotor steel at 530/550°C (f>lHz)are shown in Fig.7 [ 1 , 3 , 5 ] . This data was collected as part of the COST 505 HTFCGprogramme, and the chemical compositions and mechanical properties of all the

steels forming part of this collaborative activity are summarised in Table 2.

The use of high frequency high temperature LSF crack growth data for steam turbineapplications is limited since defects are generally not tolerated in componentssubject to this type of loading, eg. blading. Those parts in which certaindefects may be acceptable tend to operate at cyclic frequencies much lower than1Hz and in these circumstances, oxidation and creep become increasingly importantconsiderations. The bores of rotors and main steam pipe welaments are examples ofhigh temperature component locations for which low frequency LSF crack growth datadetermined in load control may be applicable ( F i g . l ) .

LSF-Creep

As frequencies are reduced by either lowering the cyclic load/strain rate and/orby extending hold time durations, crack face oxidation and creep strainaccumulation at the crack tip increasingly influence LSF crack propagation rates.In addition, as time dependent processes become more influential, the cyclecontrol mode also becomes important. For example, the load is sustained duringthe hold time in a load controlled cycle whereas it relaxes when the cycle iscontrolled within strain limits ( F i g . 2 ) . Consequently, lower frequency LSF crackgrowth rates are usually faster when the cycle is applied in load control. Theeffect is not great at low AK but becomes pronounced in the HSF regime (seeF i g . 2 1 ) .

Providing that the magnitude of the peak tensile stress is sufficiently low toavoid general yield in tension or creep deformation, high temperature fatigue

crack growth rates may still be described in terms of A K = » - F , using a power law ofthe type given by Eqn.l, but with C being very dependent on frequency. It is alsonoticeable that the m exponent tends towards 2 with increasing temperature. Thevalidity limit to using linear elastic fracture mechanics is provided by acritical &CTOD, above which creep deformation causes cracking [30].

At very low AK, crack face oxidation is primarily responsible for retardingfatigue crack growth rates due to crack closure, as seen by the effect on AKo inFigs.5 and 6 [16]. There is a marked beneficial effect of reducing frequency onthe apparent AKo for lCrMoV rotor steel at 550°C in air.

135



At somewhat higher &K, the combined effects of enhanced oxide growth and creep oncrack extension more than outweigh the influence of crack closure ( F i g . 6 ) .Initially, strain assisted oxidation is the primary driving force for the observedincreases in crack growth rate but with increasing & K (ie. &K>20MPa/l for lCrMoVat 5 5 0 ° C ) , the relative effect of environment diminishes and creep increasinglybecomes the controlling factor [ 7 , 3 1 ] .

Hence for ûK> & Ko, the main influence of reducing frequency at elevatedtemperatures is to increase LSF crack growth rates. The effect is shown forlCrMoV rotor steel (Fig.Ba) [5], 2.25CrMo pipe steel (Fig.8b) [6], 2.25CrMoweldment microstructures (Fig.8c) [7] and two turbine casting steels, ie. lCrMoV(GS-17CrMoV 5 11) (Fig.8d) [2] and 12CrMoV (G-X22CrMoV 12 1) (Fig.8e) [2] . Thedata in Fig.8a also indicate the tendency for Kc to reduce with decreasingfrequency.

The effect of frequency on LSF crack growth rates for the COST 505 lCrMoV rotorsteel cycled in load control at 530/550°C is conveniently summarised for&K=35MPa/JT in Fig.9. For f>lHz, da/dNxoxxi. is fatigue dominated and frequencyindependent (ie. da/dNror* . = da/dNcvcuc). Static load creep crack growth ratesare also plotted in Fig.9 and it is clear that da/dN is creep dominated forf<10_ 3Hz (ie. da/dNroTAi. = da/DNcRKp).

The point is also demonstrated in Fig.10 [5] , in which low frequency LSF crackgrowth rates per unit time (ie. da/dt = da/dN x dN/dt) are consistent with staticload creep crack growth rates for CT10/20 and CTas/so specimens. Crack growthrates are plotted here in terms of C* determined by substituting P M A X into O N * T inEqn.8, in addition to using the appropriate value for gi(a/w,n) [24]. A similarconclusion has been drawn with load control test data for a O.SCrMoV pipe steel[ 3 2 ] .

At intermediate frequencies (ie. 10*3<f<lHz, F i g . 9 ) , fatigue crack growth ratescomprise cyclic and creep components, ie.

da/dN-roTAi, = da/dNc-rct-ic + d a / d N c n u * (9)

In these circumstances, cyclic crack growth rates are influenced by oxidation andcrack tip creep processes. For example, da/dNcvctic may be accelerated by thepresence of creep damage established at the crack tip during prior dwell periods(Fig.11), and the C and m terms in Eqn.l become functions of frequency, hold timeand creep ductility, ie.

da/dNcvcL.ic = C(f,th,6 .) (AK«j)-"- ,: h-*' ,> (10)

where en is uniaxial creep rupture ductility.

For a given AKio, da/dNcv«.ic typically increases to a plateau level withincreasing hold time (Fig.12) [33,34,35]. The behaviour is associated with anincrease in the magnitude of C(f,t»,&it) and some reduction in m(f,tt»,ei») due tothe development of a creep damage zone at the crack tip during the hold time(Fig.13). For hold times out to to (ie. the time at which crack tip creep damageis first formed), any changes to the values of C(f,th,6i») and m(f,tn,fen) aremainly due to oxidation. The damage intensity and zone size increase withincreasing hold time to ti (ie. the time at which crack tip creep damage achievesa critical condition resulting in the onset of creep crack growth due to localductility exhaustion). As a generality, da/dNcvcnc is not accelerated by further

136



increases in hold time (ie. for tn>ti), except perhaps in extremely creep brittlesteels [ 36]. Both the extent to which da/dNc-rcnc is accelerated and the holdtime dependence of C(f,t>»,eR) & m(f,tt»,en) are determined by 6R, since creepductility influences ta, ti and the intensity and size of the damage zone. Forexample, ta, ti and the critical crack tip damage density for creep crackinitiation increase with increasing creep ductility.

Creep crack growth rates are conveniently expressed in the form of Eqn.lla [37],ie.

da/dtcREEP = b_ (C *) " (11a)6 R

and alternatively as

th

da/dNcREEP = f b(th,6R) (C*)- dt (lib)b(th,6R) (C*)

where b and q are constants dependent on material and temperature, which may beinfluenced by prior cyclic damage. The exponent q is typically in the range 0.8to 1. The term b(t»,eR) is a function of tn and 6R, since creep crack growth doesnot occur until t> ti.

In the intermediate frequency regime of Fig.9, creep-fatigue crack growth ratesare the summation of Eqn.10 and Eqn.llb (ie. as given by Eqn.9). The magnitudesof both crack growth rate components decrease with increasing 6R, and henceturbine steels with high creep ductilities are more resistant to high temperaturefatigue crack growth than those with low ductilities.

LSF with Prior Thermal Exposure

It has been shown above that creep crack growth may dominate the HTFCG processwhen cyclic primary loads are linear elastic, when loading rates are slow and/orwhen hold times (steady running periods) are long. One COST 505 projectinvestigated the situation for a lCrMoV (28CrMoNiV 4 9) rotor steel when periodsof creep crack growth (ie. 1,000 to 4,000h @ 530°C/Kx-30MPa/m) are followed by aperiod of low AK cyclic loading at 530°C [1] . In these circumstances, cycliccrack growth rates are initially retarded until the crack has extended beyond theinfluence of the prior creep damage (Fig.14).

The crack tip damage in the creep pre-cracked CTis/so specimens was typicallydiscontinuous grain boundary microcracking (Fig.15) [1] . The condition is notdissimilar to that observed ahead of creep-fatigue cracks in lCrMoV [35] and other

turbine and power plant steels [20,33,34,36] (cf. Fig.11), although in the latterexamples the damage is associated with an acceleration in cyclic cracking rate(ie. da/dNcvcLic). It has already been noted that, with decreasing frequency,high temperature ÀKo levels increase due to oxide blocking and crack closure whileda/dNxoTAi. increases for somewhat higher & Ks when oxide assisted crack growthbecomes more influential [16,31] (Fig.6). The behaviour displayed in Fig.14 isconsistent with cyclic crack growth rates being retarded by crack closure atrelatively low initial levels of AK.

Prior ageing for up to 5,000 hours at 530°C (with no applied load) has noinfluence on high temperature LSF crack growth rates for lCrMoV type steels [1,2]

137



(Fig.L4). Similarly, LSF crack growth rates for a 2.25CrMo pipe steel areapparently unaffected by 120,000 hours service at 565°C (Fig.16) [6] . It is notedthat reductions in creep ductility were not reported for either the lCrMoV rotorsteel after 5,000 hours at 530°C or the 2.25CrMo pipe steel after 120,000 hours at565°C. The possibility of a reduction in HTFCG resistance should not bediscounted in steels which suffer a deterioration in creep ductility due to

isothermal/service exposure.

HIGH STRAIN FATIGUE

In high temperature turbine components, HSF crack growth may occur due to cyclicprimary stress loading in regions of stress concentration, but is more likely tobe due to thermal loading. The thermal transients experienced at, for example,steam inlet locations in valve chests and inner casings include on-load periods atrelatively constant temperature when creep deformation is most likely to be due tosecondary stresses ( F i g . l ) . Consequently, with reference to steam turbineapplications, most effort has been directed to the measurement of HSF crack growthrates resulting from strain controlled cycles with hold times [4,20,22,26,33-36].

HSF-Creeo

It has already been noted that the thermal strain cycles experienced by realcomponents are invariably complex, and that engineering assessments usually assumeidealised cycle types. For example, at turbine start-up, through section thermalgradients can be responsible for the generation of large compressive strains inexcess of the cyclic yield strain (OA, Fig. 17). During steady running,temperature gradients are low and thermally induced strain levels are effectivelyzero. Nevertheless residual tensile stresses can be initially high due toreversed plasticity following compressive yield during start-up (B, F i g . 1 7 ) .Creep strain accumulation occurs as these stresses relax while the turbine is onload (BC, F i g . 1 7 ) . On shut-down, components may cool down slowly or rapidlydepending on their location in the unit. Thermally induced strain transients are

minor on slow cooling (CO, Fig.17a), whereas more rapid cooling is responsible fora tensile peak strain (CDO, Fig.17b). These two service cycle types are referredto as Type I and Type II respectively, and form the basis of the UKCEGB/Turbinemakers' HSF endurance database [38] .

HSF crack growth rates have been measured for a 2.25CrMo turbine casting steel at540°C using the two cycle types described above [ 4] . The approach adopted in thisCOST 505 study was novel in that large SENB-rs/ioo specimens with short pre-crackswere used to model the constraint existing in the wall section of a turbinecasting. Typical test records are shown in Fig.18, and these show how peak andeffective load ranges and the loads at the start and end of the dwell periodreduce with cycle number in this type of test. The result is that crack growthrates are initially relatively constant and then reduce to zero as the crack

length approaches -75mm (Fig.19) (ie. a/W-0.75). The behaviour is qualitativelysimilar to that displayed by thermal fatigue cracks in service. This contrastswith crack growth rate records determined from fatigue tests conducted in loadcontrol which continuously increase with increasing crack length.

The effect of dwell period on HSF crack growth rates is shown in Fig.20. For aType I cycle, da/dNTorju. increases significantly with increasing hold time. Crackgrowth rates resulting from a cycle with t»=4h are 4 to 5 times faster than thosefor tn=0. In contrast, there is a relatively small influence of increasing holdtime for a Type II cycle (Fig.20). The higher residual stress level at the startof the hold time of the Type I cycle is responsible for a greater degree of creep

- 138 -



strain accumulation at the crack tip. This can lead to creep crack growth duringthe dwell period for tn>t<=, although the evidence indicates that crack extensionby this mechanism was limited in the highly ductile cast 2.25CrMo steel, even fora Type I cycle with a 4 hour hold time. The observed differences in growth ratedue to cycle type and hold time were mainly related to the extent of the crack tipcreep damage zone and its effect on da/dNcvci.ic. In less creep ductile steels,

the da/dNcREEP component is more significant after 4 hour hold time durations[ 3 4 , 3 6 ] . Furthermore, the effect on da/dNcycLic and the contribution ofda/dNcREEp becomes increasingly more notable in turbine steels at longer holdtimes (ie. 16<t»<200h) [ 3 3 - 3 6 ] .

There is a marked effect of load/strain control mode on HSF crack growth rates in2.25CrMo at 540/550°C, particularly at low frequency (Fig.21). With decreasingfrequency, propagation rates due to a Type I cycle shape applied in strain controlare increasingly lower than the da/dNTOTAL. values resulting from the same cycleshape and KE O applied in load control. This is because the magnitude ofda/dNcREEP reduces during the hold time of a strain control cycle due to stressrelaxation and a reducing C*. In contrast, da/dNcREEP increases during the holdtime of a load control cycle.

HIGH TEMPERATURE FATIGUE CRACK GROWTH IN WELDMENT MICROSTRUCTURES

The HTFCG properties of a number of parent steels have been reviewed. However inpractice, parts such as steam pipework, valve chests and turbine casings arejoined by welded connections and it is therefore important to know the propertiesof weldment microstructures. The incidence of defects in weld metal and heataffected zone regions is not uncommon either during manufacture or in service, andit may be necessary to assess their acceptability using fracture mechanicsmethodology. In many cases, the properties of the weld metal and the HAZ can beinferior to those of the parent steel and therefore need to be taken into accountin design assessments. HTFCG properties have been determined for both simulated[32,39] and real [2] weldment microstructures.

The metallurgical structure of multi-pass weldments in turbine components iscomplex [40] . For example, each weld bead may comprise a region of as-weldedcolumnar grains adjacent to a region of fully transformed fine equiaxed grains,the latter being the result of re-austenisation during a subsequent weld pass.The HAZ may similarly consist of pockets of coarse grain HAZ contained by thefusion boundary on one side and regions of fine grain HAZ elsewhere. Modernwelding procedures aim to limit the extent of coarse grain HAZ regions such thatthey are not continuous and can, in the limit and with extreme care, be whollyrefined by re-austenisation during a subsequent weld pass. Typically, steam pipeweld heat affected zones extend no more than 2 to 3 mms from the fusion line andthis band contains not only the fully transformed HAZ microstuctures referred toabove, but also a transition region comprising a similarly complex mixture of

partially transformed intercritical tempered ICHAZ and untransformed subcriticaitempered SCHAZ microstructures.

There are two schools of thought as to how the properties of weldmentmicrostructures should be established. The first is to identify the metallurgicalregion responsible for minimum properties and then to simulate the appropriatemicrostructure as a homogeneous matrix in a block of material of sufficient sizeto yield the requisite number of laboratory specimens. This approach is adoptedto minimise data scatter and to ensure that the properties determined have notbeen influenced by surrounding microstructures. The alternative opinionacknowledges that it is very difficult to precisely simulate individual HAZ

139



microstructures and that in practice the performance of these regions is almostcertainly influenced by the properties of the surrounding matrix. Theprotagonists of this view support the use of specimens machined from real welds,even with the associated difficulties in producing a weldment which is trulyrepresentative of the actual welded joint to be assessed, and experimentally in

conducting the laboratory test.High temperature fatigue crack growth rates have been determined for realweldments based on cast lCrMoV (GS-17CrMoV 5 11) and 12CrMoV (G-X22CrMoV 12 1)parent steels [2] . HSF crack growth rates have been measured using both CTao/«oand CÎ29/SO specimens cycled in load control. The two testpiece types weremanufactured with pre-cracks running adjacent and parallel to weldment fusionboundaries (Fig.22). During testing, propagating cracks followed the HAZmicrostructure offering minimum resistance to cracking.

The fatigue crack growth rates determined for the HAZ structures of twoGS-17CrMoV 5 11 steels are 2 to 3 times faster than the rates measured for theparent materials at cyclic frequencies of 0.5 and 0.05Hz (Fig.23). With the

introduction of a 20 minute hold time at peak load, there is a more notable effecton da/dN-TOTAL. relative to that for the parent steels. The fracture paths followedin the continuous cycle tests were just contained within the 'visible' HAZ, in therelatively soft ICHAZ (Fig.24). In contrast, the fracture paths observed in holdtime tests tended to follow the coarse grain HAZ immediately adjacent to thefusion boundary. This is consistent with the observation that the creep crackgrowth resistance of CrMoV HAZ microstructures (and particularly the coarse grainHAZ) is notably less than that of the parent steel [41] . The creep ductility ofcoarse grain CrMoV weld heat affected zones is notoriously low, and the higherHSF-creep crack growth rates are therefore probably due to the effect of low creepductility being responsible for increases in both da/dN<=vct.ic and da/dNet»*?(Eqns.9-11).

At cyclic frequencies of 0.5 and 0.05Hz, fatigue crack growth rates in both theG-X22CrMoV 12 1 parent and HAZ structures are similar (Fig.25). However, incontrast to the low alloy creep resistant steel, there is no reduction inHSF-creep crack growth rates measured in the HAZ of the cast 12CrMoV steel, evenwith the introduction of a 20 minute hold time at peak load in the cycle. Thefracture path adopted in the 12CrMoV specimens is outside the 'visible' HAZ andfollows the soft SCHAZ, irrespective of hold time (Fig.26).

GENERAL OBSERVATIONS

The main collaborative effort of the COST 505 High Temperature Crack. GrowthWorking Group was based around a Creep Crack Growth Round Robin [29] . In additionto this activity, six members of the Group (Table 1) were also concerned with the

investigation of certain HTFCG properties for a number of high temperature turbinesteels. The high temperature fatigue crack growth studies covered a variety ofaspects which were generally unrelated, and the present Review has had to draw onfindings from outside COST 505 in order to complete the story. Nevertheless, theCOST 505 HTFCG results have added to the turbine materials database and to thegeneral understanding of the subject. They have also highlighted those areaswhere our knowledge is incomplete. These include:

(a) a quantitative understanding of the effect of long hold timesrepresentative of service applications on da/dNcvcuc and da/dNcM » forrelatively small cracks in critical turbine materials,

(b) an understanding of the effect of combined primary and secondary stress

140



creep loading transients on high temperature fatigue crack growth rates,(c) economic methods of accurately calculating A K E O , & J and C* for relatively

small defects in large complex turbine components.

Overall, the COST 505 High Temperature Crack Growth activity was extremelyproductive and a useful vehicle for exchanging ideas, and establishing workingcontacts throughout Europe.

CONCLUSIONS

The high temperature fatigue crack growth properties of a number of steam turbinematerials have been reviewed. Many of the results referred to in the text weregathered by members of the COST 505 High Temperature Crack Growth Working Groupwho were active during the period 1985 to 1988, but additional information hasalso been collated from the published literature. The following conclusionssummarise the current understanding of high temperature fatigue crack growth insteam turbine materials.

1. It is convenient to consider both low and high strain high temperature fatigue

crack growth rates in terms of two components, one due to cyclic loading andthe second due to creep, ie.

da/dNTOTAt, = da/dNcYci.ic + da/dNcREsp

The da/dNcYct.ic term i9 a power law function of ¿K=o and may be influenced bycrack tip damage due to oxidation and prior creep loading. da/dNcniEp isexpressed in terms of the C* parameter.

2 . At 530/550°C and for a range of steam turbine forging, casting and pipesteels, there is no apparent effect of creep on LSF crack growth rates forf>lHz (ie. the magnitude of da/dNciusEP is negligible).

3. HTFCG threshold AKo levels increase with decreasing frequency due to oxideblocking and crack closure whereas at somewhat higherAKs, LSF crack growthrates increase with decreasing frequency due to enhanced oxide growth.

4. For frequencies less than 1Hz, da/dNTOT»L is still mainly influenced by oxideassisted growth effects on da/dNcvcL.xc at relatively low AKs. With increasingA,K, the role of oxidation diminishes and creep becomes more important throughits effect on da/dNcvcuc and its contribution in terms of da/dNcnKEP.

5. In a simple engineering model da/dNcvcuc, for a given AK» o, increases to amaximum as the crack tip damage zone develops to the critical conditionnecessary for the onset of creep crack growth. The prior hold time necessaryto achieve this peak acceleration increases with increasing creep ductility.

6. At frequencies less than 10~ 3Hz when the cycle is in load control, fatiguecrack growth rates are determined by the rate of creep crack growth (ie. themagnitude of da/dNcrci.xc is small relative to that of da/dNcMutp). This isnot always the case when HTFCG is due to a strain controlled cycle. In thesecircumstances, stress relaxation occurs and da/dNcnEEP is not the dominantcrack growth component until lower frequencies.

7. The resistance of a material to creep-fatigue crack growth is stronglyinfluenced by creep ductility. The magnitudes of both cyclic and creep crackgrowth rate components are lower for steels with high creep ductility.

- 141



8. Prior long term thermal exposure (without load) has no influence on hightemperature fatigue crack growth rates, at least while the ageing treatment isnot responsible for a significant reduction in creep ductility.

9. In load controlled tests, high temperature fatigue crack growth rates throughthe weld heat affected zone of a cast lCrHoV steel are faster than thosethrough the parent material at frequencies of around 0.1Hz. The difference incracking rates increases dramatically with the introduction of a hold time atpeak load, coinciding with a change in fracture path from the partiallytransformed ICHAZ to the fully transformed coarse grain HAZ immediatelyadjacent to the fusion boundary.

1 0 . High temperature fatigue crack growth rates through the weld HAZ and parentstructures of a cast 12CrMoV steel are similar in both low frequencycontinuous cycle and peak load hold time tests. For the test conditionsadopted, all fracture paths in weldment tests are contained within the softSCHAZ.

ACKNOWLEDGEMENTS

The study has been conducted with the financial assistance of the Commission ofthe European Communities, under Contract no. COST 0015 UK (CH).

The author wishes to acknowledge the many constructive discussions withMr D.V. Thornton (GEC Alsthom) and members of the COST 505 High TemperatureWorking Group, particularly Dr J. Ewald (Siemens KWU), Dr T. Hollstein (FhG-IWM),Dr H. Kanbach ( A E G ) , and Dr. G.A. Webster (Imperial College).

142



REFERENCES

1. J. Ewald, C. Berger & H. Brachvogel; "Investigation on crack initiation andpropagation under static, cyclic and combined loading conditions of lCrMoNiVsteels at 530°C\ COST 505 D20/D21 Final Report, Siemens Report No. TW1187/89,1 9 8 9 , June.

2 . H. Kanbach; "Crack growth in welded turbine materials at elevatedtemperatures", COST 505 D35 Final Report, AEG, 1989, April.3. I. Ragazzoni; COST 505 13 Final Report, ENEL, to be issued.4. S.R. Holdsworth; "High temperature crack growth in turbine steels", COST 505

UK5 Final Report, GEC Alsthom Report No. RM872/89, 1989, November.5. G.A. Webster & F. Djavanroodi; "Elevated temperature crack growth in steam

turbine materials", COST 505 UK1S Final Report, Imperial College(Mech.Eng.Dept.), 1989, January.

6. G.A. Webster & F. Djavanroodi; "Determination of the crack growth behaviourand failure mode of pre-exposed material", COST 505 UK26 Final Report,Imperial College (Mech. Eng. D e p t . ) , 1989, January.

7 . J.F. Knott; "High temperature crack growth in steam turbine materials",COST 505 UK25 1st Interim Report, Cambridge university (Met. & Mat. Sci.

D e p t . ) , 1986, June.8. T.C. Lindley, C E . Richards & R.O. Ritchie; "The mechanics and mechanisms offatigue crack growth in metals", Conf.Proc. The Mechanics and Physics ofFracture, Churchill College, Cambridge, 1975, January.

9. P.C. Paris & F. Erdogan; "A critical examination of crack propagation laws",J.Basic Eng., 1963, 85, 528.

1 0 . M.O. Speidel; "Fatigue crack growth at high temperatures", Proc. Symp. HighTemperature Materials in Gas Turbines, Brown Boveri, Baden, 1973, March.

1 1 . S.J. Garwood; "Fatigue crack growth threshold determination", Welding Inst.R e s . Bull., 1979, 20, 262.

1 2 . R.T. Davenport & R. Brook; "The threshold stress intensity in fatigue", Fat.Eng. Mat. Struc, 1979, 1, 151.

1 3 . G.K. Haritos, T. Nicholas & G.O. Painter; "Evaluation of crack growth models

for elevated temperature fatigue", ASTM STP 945, 1988, 206.1 4 . K.J. Nix, N. Knee, T.C. Lindley & G.G. Cheli; "An investigation of fatiguecrack growth in ductile materials at high growth rates", CEGB (CERL) ReportN o . TPRD/L/3168/R87, 1987, October.

1 5 . Stress Intensity Factors Handbook, Ed. Y. Murakami et al, 1987, Pergamon.1 6 . R.P. Skelton b J.R. Haigh; "Fatigue crack growth rates and thresholds in

steels under oxidising conditions", Mat. Sei. Eng., 1978, 36, 17.1 7 . G.J. Lloyd; "High temperature fatigue and creep-fatigue crack propagation:

Mechanics, mechanisms and observed behaviour in structural materials", Fatigueat High Temperatures, Ed. R.P. Skelton, App. Sci. Pubi., 1983, 187.

1 8 . J.R. Rice; "Crack tip placticity and fracture initiation criteria", Proc. 3rdInt. Conf. Fracture, Munich, Part 2, 1-441.

1 9 . W. Elber; "The significance of crack closure", ASTM STP 486, 1971, 230.

2 0 . S.R. Holdsworth; "Remaining life assessment of high temperature turbinecastings", Inst. Metals Conf. Proc. Metals Development in Turbo-MachineryDesign, Churchill College, Cambridge, 1988, September, 136.

2 1 . J. Schivje; "Some formulae for the crack opening stress level", Eng. Frac.Mech., 1981, 14, 461.

2 2 . R.P. Skelton; "Cyclic crack growth and closure effects in low alloy ferriticsteels during creep-fatigue at 550°C", High Temp. Tech., 1989, 7, 3, 115.

2 3 . N.E. Dowling; "Crack growth during low cycle fatigue", ASTM STP 637, 1977, 97.2 4 . V. Kumar, M.D. German & C F . Shih; "An engineering approach for

elastic-plastic fracture analysis", EPRI Report No. NP-1931, 1981, July.

- 143



25. R.P. Skelton; "Review - The application of small specimen crack growth data toengineering components at high temperature", ASTM Conf. Proc. Low CycleFatigue - Directions for the Future, Lake George, 1985, September.

26. R.P. Skelton, S.M. Beech, S.R. Holdsworth, G.J. Neate, D.A. Miller & R.H.Priest; "Results of a Round Robin test on creep-fatigue crack growth in aferritic steel at 550°C", NPTEC Report to be issued.

2 7 . H. Riedel; Fracture at High Temperatures, MRE, Springer-Verlag, 1987.2 8 . R.A. Ainsworth, G.G. Cheli, M.C. Coleman, I.W. Goodall, D.J. Gooch, J.R.

Haigh, S.T. Kimmins & G.J. Neate; "Assessment procedures for defects in plantoperating in the creep range", Fatigue Fract. Engng Mater. Struct., 1986, 10,115.

29. T. Hollstein, F. Djavanroodi, G.A. Webster & S.R. Holdsworth; "Hightemperature crack growth in Alloy 800 and a lCrMoV steel. The results of anEGF Round Robin", Conf. Proc. ECF7, Failure Analysis - Theory and Practice,Budapest, 1988.

3 0 . J.R. Haigh; "The growth of fatigue cracks in turbine casing steels at hightemperatures under predominantly elastic loading", CEGB (CERL) Report No.RD/L/N9/74, 1974, January.

3 1 . J.R. Haigh, R.P. Skelton & C E . Richards; "Oxidation-assisted crack growthduring high cycle fatigue of a lCrMoV steel at 550°C", Mat. Sei. Eng., 1976,26, 167.

3 2 . G.J. Neate; "Crack growth in bainitic 0.5CrMoV steel at elevated temperatureunder cyclic loading conditions", ASME Conf. Proc. Advances in Life PredictionMethods, 1983.

33. D.N. Gladwin, D.A. Miller, G.J. Neate & R.H. Priest; "Creep fatigue andcreep-fatigue crack growth rates in parent and simulated HAZ type 321stainless steel". Fat. Fract. Engng Mater. Struct., 1988, 11, 5, 355.

3 4 . G.J. Neate: "Creep fatigue crack growth in 0.5CrMoV steel", Mat. Sci. & Tech.,1988, 4, June, 524.

35. R.H. Priest, D.A. Miller, D.N. Gladwin & J. Maguire; "The creep fatigue crackgrowth behaviour of a lCrMoV rotor steel". Proc. ASM Intern. Conf. FossilPower Plant Rehabilitation, Cincinnati, 1989, March.

36. D.N. Gladwin, D.A. Miller & R.H. Priest; "Examination of fatigue andcreep-fatigue crack growth behaviour of aged type 347 stainless steel weldmetal at 650°C", Mat. Sci. E> Tech., 1989, 5, 40.

3 7 . K.M. Nikbin, D.J. Smith St G.A. Webster; "An engineering approach to theprediction of creep crack growth", J. Eng. Mat. Tech., Trans ASME, 1986, 108,186.

3 8 . G. Thomas & R.A.T. Dawson; "The effect of dwell period and cycle type on thehigh strain fatigue properties of a lCrMoV rotor forging steel at 500-550°C",1980, Inst. Mech. Engrs. Conf. Proc. Engineering Aspects of Creep, Sheffield,Paper C335/80.

39. D. Armstrong Si G.J. Neate; "Creack growth in bainitic 0.5CrMoV steel undercreep-fatigue conditions", Mat. Sci. & Tech., 1985, 1, January, 19.

4 0 . P.J. Alberry S¡ W.K.C. Jones; "Structure and hardness of 0.5CrMoV and 2.25CrMosimulated heat affected zones", Metals Technology, 1977, December, 557.

4 1 . S.R. Holdsworth Si D.V. Thornton; "The effect of stress relief heat treatmenton the resistance to creep crack growth of the weld and heat affected zones ofCrMoV type joints", WI Conf. Proc. Residual Stresses In Welded Constructionand their Effects, 1977, November, 133.

- 144



Table 1 COST 505 High Temperature Crack Growth Working Group

Notes:

Organisation

Siemens-KWU (Mulheim-Ruhr)FhG-IWM (Frieberg)AEG (Frankfurt)ENEL (Milano)VTT (Espoo)GEC-Alsthora (Rugby)Imperial College (London)Cambridge University

Project

D20/D211

D22D351

I31

SF2UK51

UK18/UK261

UK251-2

project directly involved with high temperature fatigue crack growth testingonly active for one year

145



Table 2 Chemical Compositions and Mechanical Properties of Steels in COST 505 Programme

C

Si

Mn

P

S

Cr

Mo

Ni

V

Al

Cu

As

Sn

Sb

Ti

Zr

20C

RpO.2 MPaRm MPa

Elong %

RofA %

Cv J

530/550CRp0.2 MPaRm MPa

Elong %

RofA %

Project

lCrMoV

0.26

0.17

0.45

0.016

0.010

1.10

0.90

0.58

0.27

0.005

0.090

0.031

0.026

0.003

621

748

18

70

22/56

419

481

17

81

D20/D21

rotor steels

*

0.22

0.24

0.64

0.009

0.003

1.29

0.66

0.66

0.28

0.014

0.120

0.005

0.009

0.002

594

705

22

72

84/95

372

400

24

89

UK18

0.33

0.22

0.009

0.006

1.25

1.18

0.06

0.27

0.050

0.010

0.005

644

792

15

43

20/30

484

440

24

73

13

Turbine casting steels

lCrMoV

0.19

0.47

0.80

0.014

0.009

1.32

1.03

0.13

0.21

0.040

0.140

500

640

25

64

52/75

363

429

23

62

UK5

0.15

0.38

0.67

0.016

1.330

0.94

0.07

0.31

0.05

0.130

554

675

20

59

30/36

428

479

18

71

D35

2.25CrMo

0.12

0.34

0.59

0.015

0.005

2.40

1.00

0.19

0.020

0.070

0.010

0.034

0.011

310

521

26

72

50/53

208309

29

80

D35

12CrMoV

0.24

0.37

0.69

0.014

0.009

11.300.84

1.07

0.20

599

763

20

39

41/44

D35

2.25CrMo pipe

new

0.14

0.25

0.56

0.010

0.012

2.25

1.00

0.10

0.05

243

555

28

61

219469

26

68

UK26

se**

0.13

0.23

0.31

0.010

0.020

2.30

1.14

0.14

179

469

32

70

156249

40

68

UK26

* - CCG Round Robin steel** - service exposed

costrev2.wkl/srh

146 -



EXAMPLES OF PRIMARY AND SECONDARY STRESS CYCLES IN

TURBINE COMPONENTS

ROTORSteg. bore)

c rrotationalstress

\thermalstress

CASINGS

ROTORS (eg. rim)

*— ' ^ i

rotationalK stress

-v- Y-.' v i thermal

stress

PIPEWORK

0"

N thermal

' ^ „ stress

\ ' pressure\ i stress\ ,

pressure

stress" - - 1

v

\ ¡Jt iermalN g stress

^ .

Examples of primary and secondary stress cycles in steamturbine components

Fili Rff: Org. No. Fig. Sht. No . 1

147 -



LABORATORY CYCLE TYPES

( a ) ÇT E

Siemens

ENELA " " Ä ~ Ä " ~ "A A A /

t

(b) CT

Imp. Col.»h

t

A A AA A A /E

/ V ^ t

(c) (T E

AEG

1

V " " T T

/ V \/ r\ r\\J vt

(d) O" E

GEC A ^ ^ -

vAA V V V(e) CT E

GEC ^ A ^ AV \ / ^

'^ A V A T

I d e a l i s e d c y c l e t y p e s u s e d I n COST 505 HTFCG a c t i v i t y( a ) - ( c ) : l oa d c o n t ro l c y c l e s , ( d ) - ( e ) : s t r a i n c o n t ro l c y c l e s

Da u Fil i Rtf: Org. No. Flg. SM. No. 2

148



1.0E-01da/dN - mm/cycle

1 . 0 E - 0 2 b

1 . 0 E - 0 3 k

1.0E-04

1 . 0 E - 0 5 b

1.0E-06

LOW &K REGIME MID &K REGIME HIGH AK REGIME

Kc

HSFREGIME

1 . 0 E - 0 7 1 ÛKo

da /dN • C (ÙK) '

Log AK

LSFREGIME

Fat igua crack growth reg imes due to cycl ic loading

Filt Raf: Drg. No. Fig. Sht . No.3

149



Pmax

Po

Pmin

APeff=q0AP

Determination of crack opening load and effective load range

Fila R * f : D r g . N o . Fig, S h i N o , I*

- 150



Ko - MPa/m

0.1 0.2 0.3 0.4 0.S 0.8 0.7 O.S 0.9

R ■ Kmln/Kmax

Effect of mean stress temperature and frequency on K o for lCrMoV rotor steel [1 16]

i l t R i f Drg. No. Fig. SM. No. 5

151 -



da/dN mm/cycle

1.0E-02 e

1.0E-03

1.0E-04

1.0E-05

1.0E-07

d e c r e a s i n g f r e q u e n c y ( 5 5 0 C )

¿Ko

a l l f r eq u en c i e s ( 26C )

d e c r e a a i n g f r e q u e n c y ( 6 6 0 C )

Log&K

Schematic representation of effect of frequency on LSF crack

growth rates [16

Oats File Rrf: Org. No . Fig. Sit. No. 6

152



Di t t

da/dN - mm/cycle

10°

- 1

10

1Õ2

«ä

IO

155

1Õ̂

1 1 1 1 1 1 1 1 KrMoV l28CrMoNiV<.9) • Siemens [D20/21I Load control ■ Imp Col IUK18I

R=0-1

530/550°C

~ Frequencies>1Hz ~~

da = 7x 1 8 Û K2 5 n,m;CyCle

dN \ s

J^ Ï • 0 • — •

• 1

• 1 1 1

IO 20 30 M) SO 60 70 BO 90 100 AK - MPa/m

LSF crack growth r a t e s i n lCrMoV r o t o r s t e e l a t 530/550°C [ 1 . 5 ]

F1l«(taf: Drg.No. Fig. SM . No.7

153



Dits

10°

1 0 - 1 -

'm

I , io -2-u

■s. E £

•o O

I O ,

IO « T

j

I O 5 -

ï rMoV rotor steel ƒ I m p c r i a l C o U l g e

R=»1 j IUK18I

/ Veld* Hz

/

ff l x«3

Hz j S x I p H z

• r xf *

'* -<dr ^ ¿* ***í 2 | t e

i? f£*^£/^*<3f j jo 0"0"

^^r"^^

/ffr u Hz

10 100

ÍK ( Mp g V m )

T h e e f f e c t o f f r e q u e n c y on f a t i g u e c r a c k g r o w t h r a t e s f o rl C rM o V r o t o r s t e e l a t 5 5 0 ° C [ 5 ]

FlliRtf: Drg. No. Flg. Sht. No.ßa

154



10 ° T

K T 1

10 2 - ¡

10-3-J

l o - S

« ■ S

10_

As received 2-25CrMo pipe steel 550°r R=0

10

1xlõ2 Hz

Imperial College IIIK26)

AK (MPiVm) 10 '

The effect of frequency on fatigue crack growth rates for 2.25CrMo pipe steel at 550°C [6]

Fila R i f : D r g N o Fi g S h t N o 8 b

155



I

1

S

T E S T E G IN V I U » « i . 1 0 » M r . SO O C .

• B.OI.MHi.

• R.M.IOHi.

• Oi I Hl.

■ R-M.OlHz.

2.2SCrMo (UK25)

P A R E N T P U T E . C O A R S E ¡ R A I N E D M A R T E N S I T E I

= 0 - 5 , M H z , p p .

■ R . 0 5 . 1 H i . V E I U - M E T A L

A S - R E C E I V E O I

10Hz, pp.

I I , , , , I , , I - ■■ - I

2 0 3 «1 SO 6 7 1APPUEO ALTERNATING STRESS INTENSITY. AK.HPiÆ

The effect of frequency on fatigua crack growth rates for 2 25CrMo weldment microstructure at 5 °C in vacuus [7]

Dm RI« Rtf: O r g No. Fig S h t N o 8 c

156



da/dN- mm /cycle

10°

15'

IO*

« i

M *

« S

10°

1 1 1 1 1 1 1 1

GS-17CrMoV 5l1 AEG ID35)

Load control

R=0-1

S30-C

~ o -

* r » o 530-C

- œ$m# l C r M o V ^ / g

f o r g i n g /lines /

(f»IHzl

D-0-5Hz, o - 0 0 5 H z .

open points: parenl steel,

solid points: HAZ

I 1

0 20 30 M> 50 60 70 60 90 1

ÄK - MPa/ø

)0

n

The e f f e c t of f requency on fa t i gu e crack growth ra te s fo r

lCrMoV tu rbi ne ca st in g st e el a t 530°C [2]

O n * Fita Raf: Drg. No. Fig. S h t . N o . 8 d

157



da/dN - mm/cyc le

IO»,

Iff*

»*

IO?

rG-X22CrMoV121

Load control

R=0-1

530»C

i — i — i — i - rAEG 1035]

D - 0 5Hz; o - 0-OSHz,

open points: parent steel ,

solid points: HAZ

I I I I I I20 30 40 SO 60 70 00 90 HO

¿K • HPa/m

The effect of frequency on fatigue crack growth rates for12CrMoV turbine casting steel at 530°C [2]

Filt Ref: Drg. No. Fig. S h t. N o . 8 e

158



tia/dN - mm/cycle

MP l i l i

4 K r 3 5 M P a / m

Loid Control IR=01 ]

, 530 / 550 'C

^¿^C *x>> -^ v X ^ > * > ~ ~

i i i i

1

i

1

lCrMoV forcings

• Siemens (020/21)

■ Imp. Col. lUKlBI

i ENEL 1131

-

S ' \ '

i

152

» 3 -

10<-

«5 lõ > 103 152 1Í' 10' 10 ' 102

Frequency - d N / d t - H z

Fatigue crack growth rate versus frequency for lCrMoV rotor steel at 530/550°C and ^K=35MPa/õT [1,3,5]

FII« Rtf: Org. No . Flg. Sht.No9

159



Im

•̂ E E

5 ^ e

Comparison s t a t i c load 550°C [5]

10°

I O ' -

i o 2 :

i o - a -

' '

'

I O ' 4 -

St.fic and cyclic IUK,8

>

■a B* A

c

» Ê « * Ä *

/..: Fr

i ■ M OD

♦ B « • ♦ M X _

i - . . -■ M g B B

« B »

a

■fc •

■ BE M S T A T I C emisa

m BDI STATIC CT2VU• BDll STATIC CTIMO

a BDll STATIC CTKRO

• BDÍ STATIC CT3S/J0 ABM M M kiCTU/SOA.QI

* BD4 uoooi luCnVMR.OI

• BE4 r-0001 hi CTÎVJ0 K.O.t ♦ BE4 M 001 hi CTU/U JU0.1

K BEÎIf-O.OOOI hiCTKWOIUO.i

■ BE» r-aooi hi CTicno R-O.I

10° 101 10

2 10

3 IO

4 IO

5

C* (J /m 2h r)

□f low f requency f a t i g u e crack growth r a t e s and creep crack growth r a t e s for lCrMoV r o t o r s t e e l at

Oatf Fila Rif: Org. No. Flg. Sht. No. 10

- 160



lOOjjm

Crack tip damage in lCrMoV specimen creep pre-cracked for2,000 hours at 530°C [1]

D r g . N o . Fig. Sht . N o . 1 1

- 161



1.0E-01da / dNcyc l i c - mm/ cyc l e

1.0E-04

1.0E-OS

th(0)<th(1)<th(2)<th(3)<th(4)

¿K

Crack tip creep damage associated with a high strain fatiguecrack in a 0.5CrMoV turbine casting steel [20]

Dit« F il e R a f : D r g . N o . F ig . S h t . N o . 1 2

162



HOLDTIME

th(0)

th(1)

th (2)

th (3)

th (4)

p re crack b d f

cy c l i c c r a c k g rowth —c r e e p c r a c k g ro wth - v ^

fâ©

ß \ ¡Ã \ f

'a Ve Ve

c r e e pd a m a g e

zone

I n f l u e n c e o f p r i o r h o l d t i m e o n c y c l i c c r a c k gr o w th r a t e

Date File Ref: Org. No . Fig. Sht. No . 13

163 -



Dite

1 0 - 2 - ,

d

d

m

m

c

e

3

O

O

Ól

¿

¿

■ ■ ■

Symbol X A

D B D

AS-RECEIVED 1000h/530*C „ „ p . „ „ soooh/S3o-c P R E - A C E D

1000h/530*C-da/dl 3000h/530*C-da/dt fRE-CREEP

4000h/530 ,C-da/dt C R A C K E D

SIEMENS ID20/211

o

Q ■ D

% B

□ 4

i>*

JET

x íf < a D

JCX ■ V * X D O

x # i r fi T = 5 3 0 ° c

x # o « R = 0.1

F = 1 0 H z x S CT25S/CT25

10

The i n f l u e n c e o r a t e s i n lCrMoV

Fila Ref:

2 0 / IK ( M P a v m ) 5 0 1 0 D

C p r i o r c r e e p damage o n c y c l i c c r a c k g r o w t h r o t o r s t e e l a t 530 °C [ 1 ]

Drg. No. Fig. SM. No. K

164



s >Awls

100^jm

Crack tip creep damage associated with a high strain fatiguecrack in a 0.5CrMoV turbine casting steel [20]

D r g . N o . Fig. S h t. N o . 15

- 165



da/dN-mm/cycle

ir>0

1Õ2

1Õ3

1Õ'

1

AK=35MPa/m

Load Control ( R= 0 I )

SSO C \ s ~

VS \

-

1

1 1 1

\ %^

X \ ^

\ \ Q v

s ■ \ ^ \ a s ^ - ^ \

1 1 1

1 1

2-25CrMo pipe s tee l

■ new material

o ex-service material

Imperial College (UK26)

0

1 1

-

ro3 m »'

Frequency-dN/dt -Hz

The influence of prior exposure on creep-fatigue crack growth rates in 2.25CrMo pipe steel (ie. after 120,000h @ 565°C) [6]

File Ref: Org. No. F i g S h t N o 6

166



(a) Type I cyc le

A

(b) Type n cycle

t r

he-

Type I and Type II service cycles

Oat» Fila Raf: Drg. No. Fig. SM. No. 17

- 167 -



TYPE 1 CYCLE TYPE [I CYCLE

E

f .10 c

i0

5 i i

f 0 o

t«* O 3-VO

- . 1 0

— *o-5

VLmaxI 1 .

1 1 5,000 10,000

-05

L̂min

1— -V0

- v Lmax

1 1 i l

5.000 10.000

^Lmin

_» BOO

WOO

♦ XXI

z

i 0 ■o m o _ j

-100

-200

-300

— -300

\ . .200 \ \ P m a x , Pds

We

1 n 1 °

5.000 n.ooo

- - W )

— -200

- * 3̂00

— ^ * ^ P m a x

_Pds | |

P d T ^ = 1 5,000 n.ooo

- ^—^°~

Pmin

^ — P m i n

s M r M i— E 30

J 20

1 » o

s

- X » / / - K 1 1 1 1

0 5,000 10.000 0 5.000 » .000 Cycles Cycles

Typ i ca l t e s t records fo r l a rge 2.25CrMo SENB specimens sub ject t o Type I and Type I I HSF cyc les at 540°C [4 ]

Da ta F i l a Re f : D r g . N o . F i g . S h t . N o . 1 8

168



d a / d N Cy C |¡ c - m m /cyc l e

1 0 '

1Õ2

1 Õ 3 -

1 Õ < »

225C r Mo D i sp l a c em e n t co n t r o l5 W C SEN B 75 /100I cp mAVj_= 102 mm

25 50 75 100

C r ack l en g t h - m m

Variation of HSF crack growth rate with crack length indisplacement control bend test for 2.25CrMo at 540°C [4]

Data Fila Rff: Drg. No. Fig. Sht. No . 19

- 169



da /dN | .0 f a | -mm/c y c l e

101

10 2

10-3

10*

«S

I I I I I I I M 2-25CrMo Cast Steel 5 W C . 1cpm

Typenţ A ue t ì o f )

V b 0-56 J Type l i * 0-W tf>

- rnrr • 038 ^ J \j V r 0-33 >*■ f r V Ï 0-24 > • y /

S 0-19 > / ~ / j ó

- p^ -

/ dwell — /

(hours)

/ open points 0 / part solid points 0-5

/ solid points t

/100 Hz Data Line

1 1 1 1 1 1 1 1 1 0 20 30 In 50 60 70 80 90 BO 201 0

A K g q - M f W m

The i n f l u ence of cyc l e type and hold t ime on HSF crack growth r a t e s f o r cast 2.5CrMo a t 540°C [ 4 ]

D i u File R«f : Org. No . Fig. Sht. No. 2 0

170



da / dN-mm/ cyc le

id

1Õ2

I03

mi

1 1

¿K=3SMPa/m

2 ZSCrKo

55 0 C \ \ ¿s

K

X VS X. ^\ X. S

_ Cast 2.2SCrMo / ¿ * ~ ^ - ^ ^ ^

Type I s t r a i n con t r o l cycle ^ — . ^ .

CEC (UKSI

, ,

1 1

s

S. v

s ■ ^ ^ \ ■ \ ^ - ^ x ^

_\^J^V i i

i i

2.2SOMO pipe steel lud control cycle IR:0-11

y imperial College IUK26)

1 i

-

•

IO1 IO2

F requency -dN / d r -Hz

The influence of frequency and load/strain control mode on high temperature fatigue crack growth rates for 2.25CrMo at 540/550°C K̂=35MPa/K) [4,6]

Data F i l a Ra f : Drg. N o . Fig. Sht. No . 21

171



CT 25/50 fes p i e c e s

Specimen location in cast lCrMoV and 12CrMoV welaments [2]

Oat« FII« Rtf: Drg. No. Fig. S h t . N o . 2 2

- 172 -



da/dN-mm/cycle

10°

10

10-2

M í

io4

10-5

|Õ 6

I 1 1 1 1 1 1 1

GS-17CrMoV511 AE G I D35 I

Lead c o n t r o l

R=01 * J *

530»C é i

i

h ¡i è&

- ^ßi1 0 M o V ¿ / ¡ f forging / lines / ( M Hz )

□ - 05Hz . o - 0 0 5 H z ,

licked symbols: 20min hold,

open points: parent steel,

solid points: HAZ

1 1 l i l i l í

0 20 30 IO 50 60 70 SO 90 1

ÛK - HPa/f

)0

n

H i g h t e m p e r a t u r e f a t i g u e c r a c k g r o w t h r a t e s i n c a s t l C r M o V s t e e l ( G S - 1 7 C r M o V 5 1 1 ) w e i d m a n t s [ 2 ]

Data F i l « R« f : O r g . N o . F i g . S M . N o . 2 3

173



base material HAZ f i l ler metal T ?

2 5° -

> X a

Hardnes

CD

1. Hardness p ro f i l e , specimen B1.3.2

fractur*

hit» a i u r l l l ^

3 e \ ^ / \

» 1 ■ l ' 1 ' 1 ' l

- 2 - 1 D 1 2

Dis t ance in mm.

Fracture path in cast lCrMoV steel (GS-17CrMoV511) weldraent HTFCG testpiece (0.05<f<0.5Hz, t»=0) [2]

Drg. No. Fig. S h t . N o . 2U

174 -



d a / dN - m m / c y c l e

10 ° i

1 5 2 -

1 0 - 3 -

G - X 2 2 C r M o V I 2 1

L o ad co n t ro l

B=0-1

530" C

" i—i—r~rAEG (035)

D - 0-5HZ; o - 00 5Hz ; « - CCG

o p e n p o i n t s : p a r e n t s t e e l ,

so l id po in ts: HAZ

t i cked s ymb o l s : 20mi n ho l d

I I I I I—L20 to 50 60 70 90 90 100

AK - M P a / m

High temperature fatigue crack growth rates in cast 12CrMoVsteel (G-X22CrMoV121) weldments [2]

File Rlf: Org. N o . Fig. S M . N o . 2 5

175



base metal HAZ filler metal

3 0 0 -

O

> X

m in 2 5 0 -

Hardn

2 0 0 -

2. H a r d n e s s p r o f i l e , s p e c i m e n C 1.1

•

• — - e e ■ «<

I l - 3 - 2 - 1

i i

1 y^ | \ ^

1 f 1

1 ■ rJ—- r - ^ r -3 1 2 3

D i s t a n c e (mm)

u,

Fracture path in cast 12CrMoV steel (G-X22CrMoV121) weldment HTFCG testpiece (0.05<f<0.5Hz, t»»=0) [2]

Drg. No. Fig. S h t . N o . 2 6

176



The Communities res earch an d developme ntinformation se rvice

C O R D I S

A vital p art of your p rogramme 'sdissemination strategy

CORDIS is the information service set up under the VALUE programme to give quick and easy accessto information on Eunpean Community research programmes. It is available froo of-charge online viathe European Commission host organization (ECHO), and now also on a newly released CD-ROM.

CORDIS offers the European R tD community:

— a comprehensive up-to-date view of EC R&TD activities, through a set of databases and relatedservices,

— quick and easy access to information on EC research programmes and results,— a continuously evolving C ommission service tailored to the needs of the research com munity and

industry,— full user support, including doc umentation, training and the CORDIS help desk.

Th e CORDIS Databasas are:

R&TD-programme s - R&TD-projects - R&TD-pa rtne rs - R&TD-resultsR&TD-publications - R&TD-comdocuments - R&TD-acrony ms - R&TD-news

Make sure your programme gains the maximum benefit from CORDIS

— Inform the CORDIS unit of your programme initiatives,— contribu te information regularly to CORDIS databases such as R&TD-news. R&TD-puohcations and

R&TD-programmes,— use CORDIS databases, such as R&TD-partners, in the implementation of your programme.— consult CORDIS for up-to-date information on other programmes relevant to your activities.

— inform your programme participants about CORDIS and the importance of their contnoution to theservice as well as the benefits w hich they will derive from it,

— contribute to the evolution of CORDIS by sending your comments on the service to ;ne CORDISUnit.

For more in formation about con tributing to CORDIS,con tact the DQ XIII CORDIS Unit

Brussels LuxembourgMs I. Vounakis M. B. Niessen

Tel. +(32) 2 299 0464 Tel. +(352) 4301 33638Fax +(32) 2 299 0467 Fax +(352) 4301 34989

To register for online access to CORDIS, contact:

ECHO Customer ServiceBP 2373

L-1023 LuxembourgTel. +(352) 3498 1240Fax +(352) 3498 1248

If you are already an ECHO user, please mention your customer number.













European Commission

EUR 14678 — High-temperature crack growth in steam turbine materials

J. Ewald T. Hollstein, G. A. Webster F. Djavanroodi S. R. Holdsworth

dited by J. B. Marriott

Luxembourg: Office for Official Publications of the European Communities

1994 — VIII, 176 pp. , num. tab., fig. — 16.2 x 22.9 cm

Physical sciences series

ISBN 92-826-7536-X

Price (excluding VAT) in Luxembourg: ECU 20

Modern steam turbines must retain a very high reliability throughout theirservice life of typically 2000 00 hours, wh ich in practice extends over more

than 25 years. One of the features that must be considered at the designand manufacturing stages and during the assessment of 'fitness' carriedout periodically during the service life is the growth of the manufacturing-type defects at temperatures up to about 550 °C.

Within the concerted action research programme COST 505 on materialsfor steam turbines, one coordination group studied this problem with aview to enabling a more accurate evaluation of defect acceptability on thebasis of data gathered using laboratory test-pieces. The work of the groupwas structured under three headings which form the three parts of thismonograph:

Part I: Creep crack initiation and growth in term s of KPart II: Creep crack growth in 1 %C rMoV steel and Alloy 800H — an evalu

ation of the results of the CO ST 505 and an EGF round robinPart III: High tem perature fatigue crack grow th in steam turbine materials

In each part, solutions are given according to the current state of the art.None can be regarded, however, as giving a well established methodology for practical application. Further data will have to be determinedtogether with results from complex, simulative 'benchmark' tests thatremain to be conducted, before there will be sufficient critical evidence

upon which to base general rules for practical applications.









Venta y suscripciones • Salg og abonnement • Verkauf und Abonnement • nuAr¡oci<; Kai ouvCpoucc, Sales and subscriptions • Vente et abonnements • Vendita e abbonamenti

Verkoop en abonnementen • Venda e assinaturas

B E L G IQU E /B E L G IË

Ru « d» Louvam 42 / L e u v a n u w M 4 2

1000 Bruxelles

/ 1 0 0 0

Brussel

T « {021 51 2 » 2 6 Fa « ( 02 )511 01 8 4

Avenue du R a 20 2 / Koningslaen 2 02 1060 Bruxelles / 1 0 6 0 Brussei Te l (02) 53 8 51 69 Télex 63220 UNBOOK B Fa x (02) 53 8 OS 41

Autres rjrstributours/ Ovenge verkooppunten:

Ubcelr ie eu ropéenne ) E u r o p e t e boeknende l Ru e de la Loi 244/WetstreM 24 4 1040 Bruxe l les /1040 Brussel T « (021231 04 35 Fa x ( 0 2 , 7 3 5 08 6 0

Documen t delivery:

Breite Straße 78 -60 Postfach 1 0 0 5 3 4 50445 Köln Tel. (0 2 2 1 ) 2 0 20 -0 Telex ANZEIGER BONN 6 6 Fa x 2 02 92 7 8

GREECE /EAAAoA

G .C . De f tha roudek ia SA

trsernevonal Bookstore Nkrs Strest 4 10563 Aliisns Te l ( 0 1 ) 3 2 2 8 3 2 3 Telex 21B410ELEF Fa x 3 2 3 9 6 2 1

ESPAÑA

Trafalgar. 27 -29 28071 Madrid Tel. ( » 1 ) 5 3 6 22 9 5 Fa x (91) 53 6 23 4 9

reuma -P rense L lo ros , SA

C e s i e » . 37 28001 Madrid Te l (91)431 33 99 (Ubros)

43 1 32 2 2 (Susa ¿clones) 43 5 36 37 (a reccMin )

Telex 4 9 3 7 0 M P U - E Fex (91) 575 39 96 Sucursal:

Conssfo de Ciento. 39 1 06009 Barcelona Te l (93) 46 6 34 9 2 Fa x ( 9 3 ) 4 6 7 76 5 9

d a C e W u n y e

Rambla dsfs Eshjdrs, 11 8 (P i 08002 Barcelona Tel (93)302 6635 Ta l 9 3 ) 3 0 2 64 6 2 Fa x (93) 3 02 12 9 9

FRANCE

Service de s pubNcattono

26 . ru e D e s t o 75727 Pans Cedex ÍS Tel. ( 1 ) 4 0 56 77 01131 Fa x ( 1 ) 4 0 56 7 7 0 0

IRELAND

Go v e r n m e n l SuppSse Agency

4- 5 Harcourt Roed Durjun2

Ta l ( 1 ) 6 6 1 3 11 1 Fa x ( 1 ) 4 7 60 64 5

ITALIA

Licosa Sp A

vi a Duca rj C a l e b ™ 1/1

50125 Firenze Te l ( 0 5 5 ) 6 4 5 4 1 5 Fa x 64 12 57 Telex 570466 L ICOSA I

GRAND -DUCHÉ D E L U X E M B OU R G

5. ru e Raiftetsen 2411 üjxernbourg Te l 4 0 10 2 0 Fa x 4 9 06 6 1

Ru e de la Montagna 3 4 / rlergatraal 34 B t e 1 1 / B u s 1 1 1000 Bruxelles / 1 0 0 0 Brussel

Te l

( 0 2 ) 5 1 1 6 9 41

Fa x (02) 51 3 31 9 5

DANMARK ~

J. H. Schüret IrytonTajIion AVS

Herstedvang 10-12 2620 Albertslund TU 4363 23 00 Fa x (Sales) 43 63 19 6 9 Fa x (Managarnsnl) 4 3 S 3 1 9 4 9

DEUTSCHLAND

N E D E R L A N D

S O U C

Externe Fondsen Postbus 20014 25 (»EA 's -Grave r ihaga Ta l . (070) 3 7 69 88 0 Fa x 0 7 0 3 7 89 7 83

P OR T U GA L

k n p r a n u Newton i i

Casa da Moeda . E P Ru a D Francisco Manual de Malo. 5 1092 UsboB Codex Te l . ( 0 1 ) 6 9 34 14 Fa x (Ol) 69 3 1 66

D t s W b u k k x a d e U v r o e Be r t r a nd , L d . '

Gr u p o rsertrend, S A

Ru a da s Terras do s Valea, 4- A Apartado 37 2700 Amadora Codex Te l . ( 0 1 ) 4 9 5 9 0 50 Telex 15798 BERD IS Fax 49 60 255

UN ITED K IN GD OM

H U S O B o o k e (Agency section)

H M S O Publications Centre 51 rene Elms Lene London S W 8 5D R Tel. 1071)873 9090 Fax 873 8463 Telex 2 9 71 13 8

ÖS T E R R E IC H

nz sche Verìoge-d Urttvera ltå tabucftrrendhe

Kohlmarld16 1014 Wien Te l . ( 1 ) 5 3 1 6 1 0 T e l e x 1 1 2 50 0 B O X A Fa x (1 ) 53 1 61-161

SUOMI /F INLAND

eUuMsMftsiflsMI KeôeKrMjppia

K4*kuska tu1

P O Box 21 8 00381 HaêsinM T U . (0 ) 12 1 41 Fa x ( 0 ) 1 2 1 44 4 1

N O R G E .

NsHWsMn In to Csn l tc Bertrand Narveaons ve l 2 P O Box 6125 Enerstad 0602 Celo 6 Te l . (22) 57 33 0 0 Telex 79666 N IC N Fa x (22) 68 19 01

S V E R IGE

B T J A B

T raMMvgen 13 22100 Lund Te l (046) 18 M 0 0 F a s ( 0 4 6 ) 1 6 0 1 2 S

3 0 7 9 4 7

Skòlavordustlg. 2

10 1 Reykiavik T e l . 1 1 56 50

S C H W E IZ / SU ISSE / SV IZZERA

StamrjrenbectiatraBe 65 8035 Zurich Te l . ( 0 1 ) 3 6 5 5 4 4 9 Fa x ( 0 1 ) 3 6 5 5 4 11

Europ rees Kleeetce BK U d

66 . bdVitosha 1463 Sofia T e l / F a x 2 52 74 75

CESKÁ REPUBLIKA

H a v e k o v a 2 2 13 0 00 Praha 3 Te l . (2 ) 24 22 94 33 Fa x (2 ) 24 22 14 84

M A GY A R OR S Z A G

IS. Krucza 38 /42 00 -512 Warszawa Te l . (2 ) 62 1 99 93 , 6 2 6 - 2 8 4 2 International Fax&Phone (0-39) 12-00-77

ROMAN IA

65 . Strada Dionisio Lupu 70184 Bucuresti T e U F s x 0 1 2 9 6 4 6

9.604etiya Oktyabrya Avenue 117312MOSOOW Tel/Fax (095) 1355227

SLOVAK IA

S lovak Techn ica l U b r a r y

N m slobody 19 61 2 23 Bratislava 1 Te l . ( 7 ) 2 2 0 4 5 2

Fa x : (7 ) 29 5 78 5

C Y P R U S

Cyp ruaC I ndue t ry

Chamber Buerjng 38 Gnvas Drageres A 3 Deligkxgis Street P O Box 1455

I. (2) 449500/462312 X (2) 456630

LO A OS M e t a Te l . 6 6 44 88 Fa x 6 7 6 7 9 9

TÜRKJYE

Pree Gaze te Krtep Dergl Paze r ieme DeøtOTi T ica re t v e A Ş

r fe iaoaehe Sokak N 15 btanbutCagaloglu Tel. (I) 52 0 92 96 - 52 8 55 66 Fa x 5 20 64 57 Telex 23622 DSVO-TR

ISRAEL

R O Y hitorrratunal

P O Bo x 13056 41 Mishmar Hayardan Street Te l Aviv 61130 Tel. 3 496106 F a x 3 6 4 8 6 0 3 9

4 1 S h e r i l S l Cairo Tel/Fax 39 39 73 2

UN ITED S T A T E S O F AMER ICA / CANADA

Lanham . M D 20706 -4391 Te l . Toll Free (800) 27 4 4688 Fa x (301) 4 59 0056

Subscript ions only Un iquemen t a tex inemen ts

R e n o u l Pub l ish ing C o. Lt d

1294 Algoma Road Ottawa. Ontario K1 B 3 W 8 Te l . (613) 74 1 4 3 33 Fa x (613) 74 1 54 3 9 Telex 0534783

AUSTRALIA Hun te r ftablteeoone 56 A Glpps Street CcVtingwocd

Victorie 3066 Te l . ( 3 ) 4 1 7 5 3 6 1 Fa x ( 3 ) 4 1 9 7 1 5 4

JAPAN

IGnokiinrya Company Lt d

17-7 S»n luku 3-Chome ShinfukLhku Tokyo 160-91 Te l . (03) 34394 )121

tJountM DopArtUoSfit P O Box 55 Ctirlose Tokyo 15 6 Te l . (03) 3439 -0124

S OU T H -E A S T ASIA

Lega l U b r e r y Serv ices Lt d

Orchard P O Bo x 0523 Singapore 9123

Te l 7 3 0 4 24 1 F a i 24 3 2 4 7 9

S OU T H AFR ICA

Serlo 5t h Floor. Expon House O r Maude 6 Wea l Streets Sandten 2146 T e l . ( 0 1 1 ) 6 6 3 4 7 3 7 Fa x ( 011 )863 -6569

A U T R E S PAYS OT H E R C OU N T R IE S A N D E R E LANDER

Of f ice d ee p u M c a U o t r a órnetenos

2. ru e Merrier 2965 Luxembourg T e l . 4 9 9 2 6 - 1 T e k s x P U B O F L U 1 3 2 4 b F ax 4 6 85 73 /48 66 17

I f f i U I W ^ ^



NOTICE TO THE READER

All scientific and technical reports published by the EuropeanCommission are announced in the monthly periodical 'euroabstracts'. For subscription (1 year: ECU 60) please write to theaddress below.

o9z>

Date post:	08-Aug-2018
Category:	Documents
Upload:	imebla
View:	225 times
Download:	0 times

Caso Rotura

Documents