ISSN 1018-5593

European Commission

physical sciences
High-temperature crack growth in steam turbine materials

Report EUR 14678 EN

'^sj;'*-

.i?'*

European Commission

physical sciences
High-temperature crack growth in steam turbine materials
J. Ewald
Siemens Power Generation Group KWU Mlheim-Ruhr Germany

T. Hollstein
Fraunhofer Institut fr Werkstoffmechanik Freiburg Germany

G. A. Webster and F. Djavanroodi

Imperial College of Science and Technology London United Kingdom

S. R. Holdsworth
GEC-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 Commission
through Contract Nos COST 0032-D...1 COST 0014-D...2.3 COST 0015-UK...4 / July 1992

A hot,

AltZbr

D rectorate-G e n IfrTTT"" Science, Research and u]e>ftMteUR0P. Biblotfl.

1994

(Sj

EUR 14678 E r v

cl

Published by the EUROPEAN COMMISSION Directorate-General XIII Telecommunications, Information Market and Exploitation of Research L-2920 Luxembourg

LEGAL NOTICE Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of the following 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 life of typically 200 000 hours, which in practice extends over more than 25 years. One of the features which must be considered at the design and manufacturing stages and during the assessment of "fitness" carried out periodically during the service life is the growth of the manufacturing type defects at temperatures up to about 550C. Within the concerted action research programme COST 505, "Materials for Steam Turbines" one coordination group studied this problem with a view to enabling a more accurate evaluation of defect acceptability on the basis of data gathered using laboratory test-pieces. The work of the group was structured under three headings which form the three Parts of this Monograph: * * * Part I Part n Part HI Creep Crack Initiation and Growth in terms of K Creep Crack Growth in 1%CTMOV steel and Alloy 800H an evaluation of the results of the COST 505 and an EGF Round Robin High Temperature Fatigue Crack Growth in Steam Turbine Materials.

In each part solutions are given according to the current state-of-art. None can be regarded, however, as giving a well established methodology for practical application. Further data will have to be determined together with results from complex, simulative "benchmark" tests which remain to be conducted, before there 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 concerted action 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 been "Materials for Steam Turbines; COST 505". The countries represented in this programme were Austria, Belgium, Denmark, Finland, Germany, Italy, Sweden, Switzerland and United Kingdom together with the Joint Research Centre of the Commission of the European Communities. Jointly, organisations from these countries tackled a range of problems concerned with the improvement and reliability of steam turbines. One of the coordination groups containing the 7 organisations Usted in Table A on page 7 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 plant components on the basis of data gathered using laboratory test-pieces. The defects were understood to be those associated with the manufacturing process rather than defects initiated by creep exhaustion. The work of this group was structured under three headings: * creep crack initiation and (early) growth * creep crack growth * high temperature fatigue crack growth These subjects formed the basis of reports in which the work was reviewed and evaluated in the light of present day literature and experience in studies which were conducted with some financial assistance from the Commission of the European Communities. The round robin work in the second heading was also integrated into a task being undertaken by the European Group for Fracture (now ESIS) and reported by them in 1990. The three constituent parts which make up this Monograph are independent in terms of chapters, literature references, table and figure numbering. They are, however, preceded by a common Executive Summary and an integrated list of the symbols and abbreviations used 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. The assistance of Dr. S.R. Holdsworth with the preparation of the list of symbols is also gratefully acknowledged. J.B. Marriott Secretariat for COST Projects 501/505 EUCO/MST/01/92 JBM/tp/1/1397 Petten, July 1992 V-

Contents

Executive summary List of symbols used throughout

1 11

Part I Creep Crack Initiation and Growth in Terms of K 1. Basic considerations upon the use of K 2. Crack tip/farfield models 3. Test results on creep crack growth A. Discussion of results 5. Conclusions References Table Figures

21 23 27 33 38 43 44 46 47

Part II

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

69 71 73 76 78 79 83

1. Introduction 2. Material 3. Specimens and test procedure 4. Parameter determination 5. Numerical investigations 6. Results

VII

7. Discussion 8. Conclusions References Tables Figures

84 91 92 96 104

Part HI High Temperature Crack Growth in Steam Turbine Materials 1. Introduction 2. Fatigue crack growth regimes 3. High temperature fatigue crack growth correlation parameters 4. Low strain fatigue 5. High strain fatigue 6. High temperature crack growth in weldment microstructures 7. General observations 8. Conclusions References Tables Figures

129 131 131 132 134 138 139 140 141 143 145 147

VIII

COST 505 HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS Executive Summary J. Ewald. Coordinator Introduction Within COST 505 "Materials for Steam Turbines", the High Temperature Crack Growth Working Group contained 7 research groups, Table A. They investigated 1% CrMoV, 2 1/4 Cr 1 Mo and 1 Cr 0,5 Mo steel forgings, castings, and pipes with respect to static creep crack initiation and growth and to cyclic crack growth, between 530 and 550C, Table B. In addition to the mechanical testing, 2 groups (D22 and UK18) performed finite element analyses to control the validity range of C* application and to investigate whether cracked specimens creep under plain strain or plain stress conditions. Principal Findings The main conclusions which can be drawn from the results and evaluations of the work of 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, which means that the load line displacement rate of the specimens - which is decisive for the amount of C* - is dominated by plain stress behaviour. The behaviour of the specimen within the range of the tails and during the early creep crack stage may be described in terms of K with additional geometry factors such as a crack tip/farfield ratio. The real creep crack behaviour with higher creep crack rates can best be described by means of C*2 which is a version of C* including the load line displacement due to creep. However, it is probable that these displacement rates can rarely be measured in service for low stress, long time loaded specimens/components. Cyclic crack growth is accelerated by hold time. The hold time cyclic crack growth rates lie between the cyclic crack growth and the static crack growth curves. Numerical accumulation rates da/dNtotal = d a / d N ^ + da/dN,creep based on an equivalent AK, considering the cyclic crack closure effects and C* for hold time effects were investigated and used to describe the cyclic crack growth behaviour.
- 1 -

The solutions given in of this Monograph have been obtained according to the current state of the art, but none can be regarded as giving a well established methodology for practical application. In the area of static high temperature crack growth further data will have to be determined for specimens with different shapes and sizes under loading conditions which lead to long crack initiation times and low crack growth rates for more or less creep ductile materials. A broad data base will have to exist before the final decision can be made about the most economic and appropriate evaluation method. With respect to high temperature fatigue crack growth the situation is similar. There are many data, a considerable level of mechanistic understanding and various proposals for evaluation methods, but even here there remains a lack of critical evidence upon which to base general rules for practical applications. Furthermore, it is necessary to have a good data base for creep crack growth because according to the proposed accumulation rules for load controlled cyclic crack growth with hold times both "static" and "dynamic" types of data are required. However, there still remains the complex loading condition of strain controlled cycles arising from thermal stresses with superimposed stress relaxation. This type of behaviour is frequently met in service and so far it has not been possible to model the behaviour using relatively simple laboratory tests. At present the only way to obtain information is by performing expensive tests which closely follow the actual load cycles. These conclusions have been established on the basis of three studies, which were conducted with partial financial assistance from CEC, DG XH-Gl, to review the work of the Group and to examine separate aspects of the field. The summaries of these reports follow in the order: Part I Part n "Creep crack initiation and growth in term of K" by J. Ewald; "Creep crack growth in 1 % CrMoV steel and Alloy 800 H - an evaluation of the results of the OCST 505 and EGF round robin" by T. Hollstein, G.A Webster & F. Djavanroodi; "High temperature fatigue crack growth in steam turbine materials" by S. Holdsworth

Part HI -

Parti Creep Crack Initiation and Growth in Terms of K J. Ewald Siemens, Power Generation Group KWU, Mlheim-Ruhr, Germany It is clear 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 is reasonable from the physical point of view. On the other hand there are the people who try to apply creep crack initiation and growth data in terms of K for the description of components which have thick walls, long loading times, and low loading stresses. With respect to crack initiation it can be concluded: Ky,, (the fictitious or ideal elastic stress intensity factor) seems to be basically usable to describe the crack tip situation for crack initiation. The time for crack initiation increases with decreasing crack tip driving force (Klid), Fig. A. Specimen size and shape determine the damage mode and the related specimen behaviour. Therefore, it is necessary to use Kjd 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 "two criteria diagram for creep crack initiation" which covers the range of the tails from = f (KIid) plots and which is able to demonstrate the influence of different crack tip/farfield ratios KEd/Gn. Creep ductility dramatically influences the creep crack initiation and growth behaviour. Notch weakening materials should be avoided, since they tolerate only extremely short defects. The related data base for these materials is not yet available. The creep crack growth behaviour when there is only a limited crack increment may be described by plots of K,d over t with parameter lines for a, similar to Fig. A. However, with such diagrams, which are only valid for CTl-specimens, it is hardly possible to estimate the influence of specimen shape and thickness. Thus, crack growth with small increments should be rated with diagrams like Fig. B which shows the influence of specimen shape and specimen thickness, e.g. lateral constraint If it is necessary to describe 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 is considered in greater detail in Part II of this Monograph.

- 3-

The most important point for the future is to generate further data on specimens with different 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 appropriate evaluation method. Partii 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*, GA. Webster**, F. Djavanroodi** * ** Fraunhofer Institut fr Werkstoffmechanik, Freiburg, Germany Imperial College of Science and Technology, London, UK

Experiments and analyses have been performed on a 1% CrMoV steel and on Alloy 800H which have shown that creep crack growth in these materials is described most satisfactorily by the 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, Fig. C. 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 ft 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, Figs. C and D. Increased scatter, due to 'tails' in the early stages of cracking, Fig. D, 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 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 Materials S.R. Holdsworth GEC Alsthom Turbine Generators Ltd., Rugby, UK

The high temperature fatigue crack growth (HTFCG) properties of a number of steam turbine materials have been reviewed. Many of the results were gathered by members of 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 from the published literature to complete the overview. High temperature low and high strain fatigue crack growth rates may be considered in terms of two components, one due to cyclic loading and the second due to creep, ie da/dNtotal = d a / d N ^ + d a / d N ^ The da/dNjydj,. term is a power law function of AK^ (the equivalent cyclic stress intensity) and may be influenced by crack tip damage due to oxidation and prior creep loading. a./dNmep is expressed in terms of the C* parameter. At 530/550C 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 AK0 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 da/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 the onset of creep crack growth. The prior hold time necessary to achieve this peak acceleration increases with increasing creep ductility. At frequencies below 10"3Hz 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 a/Smcp). 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 not 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 the ageing treatment is not 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 lCrM oV 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 fully transformed coarse grain HAZ immediately adjacent to the fusion boundary. High temperature fatigue crack growth rates through the weld HAZ and parent structures of a cast 12CrMoV steel are similar in both low frequency continuous cycle and peak load hold time tests. For the test conditions investigated, all fracture paths in weldment tests are contained within the soft sub-critical HAZ.

Table A Research Projects

No. D 20/21 D 22 D 35 13 SF 2 UK 5 UK 18/22

Organisation Siemens IWM Frbg. AEG ENEL V i l Espoo GEC-A Imp. Coll.

Project Responsible Berger/Ewald Hollstein Kanbach Ragazzoni Rintamaa Holdsworth Webster/ Djavanroodi

- 7

Table B Materials/Tests Project Number 1% CrMoV 550C Round Robin

X X

Additional Materials

crack growth static cyclic

D 20/21 D 22 D 35 13 SF 2

1% CrMoV; 530C 1% CrMoV; 550C 1% CrMoV cast; 530C 1% CrMoV (used rotor; 540C) 1% CrMo V (used rotor; 550C) pipe mat; 550C 2 1/4 G 1 Mo cast; 538C 1% CrMoV; 550C

x (+ analytical work)
X X X X

X X

UK5 UK 18/22

X X

X X

X X

(analytical work)

8 -

scatterband, makroscop. Creep Crack Initiation Siemens 1989

100

1000 loading time ( h )

10000

100000

Fig. A

Creep Crack Initiation, 1 % CrMoNiV 530C, CT25S

Specim.Thickn. (mm)
, a/W = X CT 50/1 O O S , a/W = + CT100/200 , a/W = , a/W = V 0 9 , a/W = Y D60 , a/W = D D60 , a/W = A D60

D CT 25/50S

0.55 0.55 0.55 0.40 0.40 0.20 0.10

O
o

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 0 A 6 $100

XQ\
Hollstein ' 2 5 / C T 1 2 . S - 50 D ^ - ^ J f o m a, [ 3 ]

10

100 loading time

1000 (h)

10000

Maile/Tscheuschner |Siemens|Ragazzoni| Hollstein |Holdsworth|Rintamaa c * + A YO A B a O CT25 -,CT50 X D - O a o 0.004W(CT),0.01W(OENT) 0,5mm 0,3mm '0,2mm 0,5mm - 0,5mm spark eroded fatigue precracked

Fig. B

Creep Crack Initiation for different specimen thickness and size

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

tH R

8-

pi strain ( 10) pi. stress (=0) / pi. strain(sO) 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 t . sec

6.00 CE+06)

7.00

B.00

9.00

10.00

11.00

Fig. C

Experimental and numerical evaluation of C* for CT 12.5/40 specimen A B 2 B , w i t h and without crack g r o w t h

Q Q S O + M D X

BD3,BD,BDS.BW1,BW2,BW3,AHJ.AH*,AOS,AQ5,BCl,BC2,AN,AEl,AE2 AF.AB3.ABRAMI,AHI,All,AI2,ABS,AB,AES,AC1-6 AN1.AN3.AN,A03 AQ1.AQ2 (CT20/40) o . A (CTI5/S0)

RR5 (SENB 9.5/191 BB1 ( C T 6 . S/137) CCP5.CCP A1G.A17 (CN2S/S0)

(CNI2.S/S0) (CP.S/a) (CT50/100)

*t/

ABZA.AB1B API.AP2.AP3 AM1.AM2

(CZ15/0)

(SENT12/20)

RR7 ( C T 13/16)

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

- <

CTj/fcn 2 ]

Fig. D

Crack g r o w t h rate in 1 % CrMoV steel at 550C as a f u n c t i o n of C * .

10 -

List of symbols and abbreviations used throughout the Monograph symbol or abbreviation a a0 Aa
Aa

meaning crack length initial crack depth crack extension crack initiation criterion crack growth rate per unit time initial creep crack growth rate

Part where used

i;n;m
I;H

n
i I;II; da/dt used in III

. A A, (bk b(t, er) B B. B. C C(f, t, er)

steady state creep crack growth rate . uniaxial elongation at fracture impact energy remaining ligament (w-a) constant in creep crack growth Equation. function of hold time and ductility in creep crack growth expression specimen thickness effective B = B - (B-BJ2/B net section thickness constant in Paris FCG law function of frequency, hold time and creep ductility in da/dN,^ law constant in Norton minimum creep rate law

n
II; El also used byn

n
n

ml\

m
I;II;m

n
i;n m
ra

C'

n;m

- 11

symbol or abbreviation C*

meaning parameter characterising stress and strain rate fields at crack tip in creeping solid C* determined using Kumar-Shin based solution experimentally determined C* centre notched specimen

Part where used I;II;m

C*!

I I; included in C*by

c*2
CCP . CN \CNT J CT CTOD ACTOD ACTOD0 da/dN da/dN^ da/dN^ da/dN,,^ da/dt dr ds DENT

i;n

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 and cyclic loading crack growth rate per unit time crack tip ligament arc length on r double edge notched tension specimen

i;n;m

m m m m m m m
lu; used in i;n

n
r,n

12 -

symbol or abbreviation E E'

meaning elastic modulus elastic modulus including lateral contraction (i.e. including pi. e) uniaxial elongation at fracture frequency; dN/dt load fatigue crack growth elastic strain energy release rate geometry function in Kumar-Shih J solution geometry function in Kumar-Shih based C* (or C*i) solution constant in cyclic hardening law heat affected zone high strain fatigue high temperature fatigue crack growth non dimensional function of n intercritical HAZ

Parts where used

m
i

El f F FCG G gi(7w,) gi(7W,n)

II; A also used by i;n

m
I;H; P used in m

m n m
I;ffl

H HAZ HSF HTFCG In ICHAZ

m m m m n m

13 -

symbol or abbreviation J

meaning path independent line integral characterising stress and strain fields at crack tip in plastically deforming solid cyclic J theoretical elastic stress concentration factor elastic stress intensity factor cyclic K apparent elastic K

Parts where used i;m

AJ
k,

m
I I;II;m

K AK

m
IH; referred to as K M ini

Kc K* AK^
Kt

critical value of K leading to instability effective K cyclic K^ initial K fictitious (ideal) K K M at a0 creep crack initiation K maximum K
minimum K

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 in

KM
K

Wo

Kno
Knm
Kniin
A K

m m m

equivalent A K total AK

AK

14

symbol or abbreviation AK LSF M

meaning fatigue crack growth threshold A K low strain fatigue exponent in Paris law exponent in da/dN cyclic law stress exponent in Nortons (minimum creep rate) law number of cycles load load at end of dwell period load at start of dwell period maximum load minimum load crack opening load exponent in creep crack growth rate law effective load range creep damage zone size load ratio (R=KmJKaJ

Parts where used

m m m m
I;II;m

WJt O
n

N P
*DE

ni HI; F used in I;II

m m m m m m m n
HI; also used for radial distance from centre of bar in H I

*DS

^max

min

Po

q q
r

R'

geometrical variable related to crack tip farfield 2 condition ie.R' =

(K /0

- 15 -

symbol or abbreviation

meaning activation energy term uniaxial reduction of area at fracture 0.2% proof strength 1% proof strength tensile strength stress ratio (o^JR^

Part where used

R" RA

n
H; Z also used by II for same purpose

Rpo.2

i;n n n
i i

Rpi R Rff RK
R, SENB

stress intensity ratio (Kftjo/Km)) creep rupture strength single edge notched bend specimen single edge notched tension specimen subcriticai HAZ time time at which crack tip creep damage first forms time to failure hold time time at which crack tip conditions are responsible for creep crack initiation time to rupture transition time

I; referred to as a 0 in II

I;II;m

SENT

n m
I;;in

SCHAZ

t
d

m
II; referred to as tr in I

tf
h

m
I;I

ti

t.
'i

I; referred to as tf in II

i;n

- 16

symbol or abbreviation T
T

meaning temperature traction vector (T = a^np displacement rate vector energy term energy dissipation rate load line displacement crack opening displacement load line displacement cyclic VL crack tip farfield ratio
^o
= ff

part where used

n n n n n
I; H; referred to as VL in

U* 5U*/a V

v vL
AVL

m
Iu; referred to as V in I

m
I

v. v. w w*
X

^ D d o / i i o aebel

load line displacement rate due to creep specimen width stress work rate distance compliance/geometry function in K solution uniaxial reduction of area at fracture exponent in cyclic hardening law strain

i;n i;n;ra

n n m
H; RA also used by H

Y Z e

m
i;n;m

17

symbol or abbreviation
e

meaning creep strain strain parameter in primary term of creep strain expression uniaxial creep rupture ductility multiaxial creep ductility ductility exhausted in the ligament prior to the arrival of the crack cyclic plastic strain nominal total outer fibre strain range creep strain rate creep strain rate strain rate tensor minimum creep rate creep strain rate at reference stress in uniaxial specimen creep strain rate factor from limit load analysis factor depending on testpiece geometry stress

Part where used

cr

n n
n;m

*f

*.

n n m m
I;II; referred to as in

A e

Ae

T(0P)

n and de/dt in m

cr

II;see comments above

n n
I

'min f

de/dt

III; referred to as in

i;n

D
Oc a

n
i;n

i;n;m

18

symbol or abbreviation La
ff

meaning cyclic stress nominal stress as defined by ASTM nominal stress for components, net section stress for specimens net section stress nominal stress according to Siebel a ax for initial crack depth a plain specimen rupture strength reference stress yield strength integration path around crack tip connecting lower & upper crack face in a counter clockwise direction

Part where used

m
I I; for tensile specimens, alternatively referred to as ona in I HI; see comment above I I U; referred to as R,,,, in I I

ASTM

*
a

ncbcl

undici

0
ra aY

m n

- 19

HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS Parti Creep Crack Initiation and Growth in Terms of K

J. Ewald Siemens, Power Generation Group KWU, Mlheim-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 steadystate 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: 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
fc C

*' = t e

1=

(n+1) C*

1"

(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 the Norton 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 by the 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 displacement rates involved. It therefore makes sense to try to use the fictitious stress intensity factor KT. . to describe the
lid

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 employed to 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/farfield ratio, K .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 Kparameter. 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 *2version 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 Vnn+\

,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 reasonsoutlined 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 [1520] 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 ., 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 . , 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 61682, 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 NT), ^ASTM = F / B ( W a ) For specimens with bending fractions, (e.g. C T), ^ASTM
= F/(B (W_a)) (4)

( 1+ 3 (W + a)/(Wa)).

(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 bending3 fractions are referred to as O . , ,, r ^ 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 r lid n Siebel farfield parameters: (1) With increasing specimen size, the K T . ,/G~ . , , ratio ^ ^ lid n Siebel increases for the same K I . , (similar relationships arise r lid for the C*/S" . ratio) Fig. 1. This means that large 3 n Siebel ^ CTspecimens 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, Aa, 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 smaller specimens beyond this value. This is logical, as a crack growth increment of 1 mm in the ligament of the large specimen 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 the shape 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 " ) 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 pre-stressed c fracture. - 28 relatively early with subsequent creep crack growth into a low farfield. This high K T . ,/ situation would lead 3 lid n to a failure mode of crack tip damage, promoting leak before

If the same component had a relatively small flaw (low K T . ,) (Fig. 7b), creep crack initiation with subsequent crack growth into a highly prestressed farfield would only occur at a very late stage in life. This would lead to a ligament damage mode with low K . ,/" 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 / R m 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 Kvalue at r lido the crack tip within the component, Fig. 9, or a specimen. K 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 K T . , / C , such as is found in CT 1specimens. r ^ lido no 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 ratiolines 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 of
lid

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 the lid 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 weakening materials 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 that KTTr. , .... < K T , _ , ... for a certain time). IID creep brittle IID creep ductile Another 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 initiation time of specimens with equal K x . , and K T . ./G" -values. ^ ^ lid lid n Thus, the effect of the constraint should be taken into account by differing materials KT---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" with c specimen size. = c u i < increasing lid n Siebel 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 = 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 ., 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 twocriteria 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 cracklike 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"" ^ lido no ratio wich =gives rise to 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. 1315. 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 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 crack growth behaviour of laboratory specimens of various sizes and, ultimately, also of components on the basis of the ratio V = Iido/'''o 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 ., " . , , ? fti 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.
K

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 / ; 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 change the V = K T . , / 5 ~ ^ o o lido no is based on initial crack crack increment in order not to . . , , value too much, since this Siebel Sieb size

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 relation to 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.TI 530C. 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 530C 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 from COST 505 partners and also the results of a further German research group [21-23] for 1 % CrMoV at 550C. 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. 21, which had spark eroded crack 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, 50100 nun, Fig. 22), create the upper boundary. The small scale specimens (V = 1,73) 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, 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 .i, / ( T I d n ratios. Of course, there is some scatter, but this 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: K T j seems to be basically usable to describe the crack tip J r lid situation for crack initiation, Fig. 19, the time for crack initiation increases with decreasing crack tip driving force (KJJJ) Figs. 1822, 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 [5,1619] 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 using the 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 550C [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 the
lid

specimen types the curve has a crack initiation tail. Fig. 24 shows all Siemens/KWU data on CTI-specimens at 530 C together with some D35 data [16] from cast 1 % CrMoV and from the heat affected zone of a weldment made with the same steel grade*. 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 affected zone 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 (560C/10 h which is equivalent on the basis of Larson-Miller to 530C and 10 h) and with fatigue precracks which were initiated at 530C. Both pretreatments did not influence the crack growth rate. This means that creep crack growth or fatigue crack initiation 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 depthsa = 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 (KT. J-curves which represent the stress redistribution 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 initiation and 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 550C is faster than at 530C, 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 o = KT . , /CT , ,. Therefore, the models for use of the 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 fictitious o 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-specimens the size effect due to the shape/geometry factor is negligible 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 use both 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 above lid 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 displacement rate 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 s t r e s s , G T . , the local events (crack tip damage) are proposed to be assessed by C*, which is expressed in the following form: <

c* = 5~ , 2
ref

.
ref

R'

(6)

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 particular for 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 ' S obtained by using the stress intensity factor K, so R is defined as
R

'=

K2/5

"ref 2

(7)

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 .

which tends in the direction of the square value of V

(R~V

It should be noticed that reference [7] gives the same reasons for not using the C*.*jn - 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. , or of C* when there is only a slight increase of 4 a. Only the C* 0 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 . , 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 values, K _, for creep as materials characters lateral constraint into initiation. shown in Fig. 20 supply also the necessary crack initiation, which may be introduced for specimens with different size and the two criteria diagram for creep crack

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.) C onclusion

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 . , for the description of components which have thick walls, long loading times, and low loading stresses. For such applications, the use of K ., together with the parameter C to describe the farfield is recommended. C rack 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 . ,) plots. The creep crack growth behaviour with parameter lines for A a, limited

crack increment may be described by plots of K J over t with similar to Fig. 19. With such diagrams, which are only valid for C TIspecimens, however, it is hardly possible to estimate the influence of specimen shape and thick nessB. 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. 1315, or some c 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. 112130, 1980. [2] K.M. Nikbin, D.J. Smith and G.A. Webster, Proc. ASME Int. Conf. on Advances in Life Prediction Methods, Albany, pp. 249258, 1983. [3] R. Hollstein and R. Kienzler, Numerical simulation of creep crack growth experiments, COST 505D22, Institut fr Werk stoffmechanik, Freiburg, FRG, January 1987. [4] A. Saxena, EPRI Meeting, Sess. 6, Raleigh, North Carolina, Sept. 1214, 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. 389398, 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. Kumaul, J. Granacher, K. Maile, R. Tscheuschner und W. Eckert, Kriechrieinleitungs und Kriechriwachstums verhalten unter Bercksichtigung des Greneinflusses, Ab schlubericht AIFVorhaben Nr. 6038, 1988. [9] J. Ewald, K.H. Keienburg und K. Kumaul, Hinweise auf Mechanismen und Einflugren zur Beurteilung des Bauteil verhaltens im Kriechbereich anhand von Kleinproben, VDI Bericht Nr. 354, pp. 3957, 1979. [10] J. Ewald and K.H. Keienburg, A two criteria diagram for creep crack initiation, Int. Conf. on Creep, Tokyo, pp. 173178, 1418 April 1986. [11] J. Tscheuschner, Anri und Rifortschrittsverhalten 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 Kconcept, 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. 185195, 1989. [14] K.H. Kloos, J. Granacher und R. Tscheuschner, Kriechrifort schrittsverhalten des Stahles 28 CrMoNiV 49, Z. Werkstoff techn. 18, pp. 390398, 1987. 44

[15] J. Ewald, C. Berger and H. Brachvogel, Investigation on crack initiation and propagation under static, cyclic and combined 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 at elevated 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 turbine steels, COST 505 UK5 Final Report, CEG Alsthom Report No. RM 872/89, November, 1989. [19] G.A. Webster and F. Djavanroodi, Elevated temperature crack growth 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 von Bruchmechanik-Parametern zur Beschreibung des Kriechriwachstums, Arbeitsgemeinschaft fr Warmfeste Sthle, 11. Vortragsveranstaltung "Langzeitverhalten warmfester Sthle", 25.11.88, Herausgeber: VDEh. [22] J. Granacher, R. Tscheuschner, K. Maile, W. Eckert und J. Ewald, Ermittlung und Beschreibung des Riausbreitungsund Riwachstumsverhaltens bei hohen Temperaturen, DVM, Vortrge 21. Arbeitskreis Bruchvorgnge, Bad Nauheim, 1989. [23] R. Tscheuschner und W. Eckert, Kriechriverhalten von Kleinund Groproben bei betriebsnaher Langzeitbeanspruchung, DVM, Vortrge 21. Arbeitskreis Bruchvorgnge, Bad Nauheim, 1989. [24] S.M. Beech, J.W. Selway and A.D. Batte, Factors influencing crack development in 1 % CrMoV steam turbine rotor forging steels, 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", Materialprfung 4, pp. 41-43, 1962. [27] Htte I, Ingenieur-Taschenbuch, 28. Auflage 6, Abschnitt: Festigkeitslehre, pp. 847-851.

45

Table 1 Author Holdsworth Slemens/KWU ^ >

i

Literature

Material Condition partly creep ductile creep ductile partly creep ductile ductile partly creep ductile creep brittle creep brittle partly creep ductile

Temp. C

Type of Specimen CCPl fatigue CT ) precrack CT) fatigue CT) precrack CT eroded crack tip CT eroded crack tip

t l /t r

aj* mm ~ ^ 0.2 0.2

18 15 22 11

550
550 550

0.5 0.3 0.4-0.75 0.3-0.4

'v 0.5 'v 0.5 0.5 0.1 0.1

Malle Tscheuschner Tscheuschner

550 550

0.3
0.11-0.18 0.2-0.5 0.15-0.5 0.22-0.48

Beech

24

CCP short cracks a Q = 1-3 mm

0.1

v 0

*ia A : criteria for crack Initiation

200

400

600

800

1 000

nominal (net-section) stress [N/mm 2 ]

Fig.

Stress Intensity and Nominal Stress as Function of Specimens Size

47

21CrMoNiV57. T=550C CT12.5/40 spec men AB2B m i n =l3-IO- 2 0 <J 6 5 [h- , l, O inMPa

-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+06

Fig. 2a.

Experimental and numerical evaluation of C for CT 12.5/40 specimen AB2B, with and without crack growth, creep law [3]. 21CrMoNiV57, 550C CT 12.5/40, AB2B

oc'

: Experiment"

: Eq (15)

f-*tt-V
3*"
ori oo

'

3000

Fig. 2b.

Experimental and numerical values of crack growth [3] - 48 -

JfnK;
ASTI

ah
0\ f n Siebel

ab

TTTI

at

Q
Il

Q
u

F F' ,. . ., w + a ,
) w a (7ASTM ~ r; -(1+3 B (w - a , a W

F F
(7nSiebel= r B ( w a)

w a

- _ _fc. c ASTM f Siebel

for 0.5 ^

'1.43

Fig. 3.

Nominal Stresses in the Farfield of Specimens with Bending Portions.

On =

^ =

^ASTIVI

completely redistributed

"general yield condition"

Fig. 4.

Nominal Stress in symmetrically loaded Specimens.

49

400

'^"a/Wl
1

CT 25/

f CT 50

( C T 100 ai

i/

a. a c
10'

I/W.157

_,

>icf

Trtie t /

i>\d

Fig. 5.

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

F m"' r "-1 \

3 0

.V

... -V - ...

- v -

8.
r.

4)

1 j 7 i
c/w = 357
KJ -

o 8v

irf1-

1 r i(

A CT KW. ep.

3 . TO".10'

f B

KJ1 Stress Intensity Foctor K^ / MPoVT

F i g . 6.

Theoretical creep growth characteristic of different specimens with identical crack increase.

- 50

K.gf On high ratio predominant crack tip damage

^ S

Ig loading time t -

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

Ig loading time t -

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

51

crack tip damage ^> witho t h o u t j ; with acr. creep crack 1.0
_ Kj,de K
" UD

(leakage)

T-

0.5

1.5

2.4 3.9

Fig. 8.

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

Fig. 9.

Explanation of the Normalized Values R for Stress and Stress Intensity.

- 52

with

without SENT, ductile brittle (Beech. Selway. Bane) small manufacturing defects in smooth spec (Taylor, Battel castings with manufacturing defects (ductile) DENT'. 3 0 x 112. a / w - 0.45 OENT *. 4 0 x 65. a/w - 0.1 and 0.4 *) with spark-eroded crack-tip >12 O | (leakage)

micro crack init.

macro crack init.

o I
eng. crack nit.

A
V

Ki*/oM0 I crack tip damage ~ > i

io ~i\jfZT
without J5 with acr. creep crack

0 . 5

1 . 0
KiidoU)

1.5

2.4 3.9

Fig.

10.

Two-Criteria-D iagram for Creep Crack Creep D uctile l%CrMoV-steel, 530 C.

Initiation,

-5=.'

1 % C r M o V - S t e e l ; 530 CC

0,5 burst

mixed damage

v"
1 1 r-

crack tip damage w i t h creep crack

J>

0,5

K no for Creep Crack

Fig.

11.

D raft

for

Two-Criteria-D iagram for Notch Weakening

Initiation

Materials.

- 53

250 Fig. 12.

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 specimen shapes and specimen sizes (IfCrMoV steel, CT compact tension specimen, K notched round bar specimen) [14].

AGA AGB

28CrMoNIV49, 600 C
600 "C

Ligament Damage 1000

Crack-Tip Damg.

a 5 1 0 s mm/h i - 1 10'mm/h

soo
10 1 0 1 10 s 1 0 ' (h) K8 1 Notched Round Bar (T.-rVAit

200

K17 j

CT with Side Grooves

((Tnosj#|>#| _ ,,^.43,

Creck.Tip/Farf.eld Ratio V0 - Kuo/o,** (mm"1) Fig. ,,J for constant K,d 13. K i n i t i a l crack t i p for creep d u c t i l e

crack growth rate, , versus the a, vi far f i e l d r a t i o , Vg = Kj;j/a nO' l % C r M o N i V material, 600 *C

- 54

1 \ CrMoNiV. 5 5 0 C. creep ductile Lig. D. 1500 b E 2 I- M i x e d Mode Damage

0 5 9 1 Crac k-Tip D mage

I O 3 mm/h

AGA

1 0 ' m m / h (Tscheuschner) / 8 / 5 5 0 "C

1 IO"* m m / h 1

1 1 1 1 1 1
1

K, -A lid

120- V

,X*
<*s^^

^
<^ - 1 10

1000

jr

500 1 10 m r n/h 200-

1 1 y^. 1
1

/^

* ^

X
3

>

^^*

IO'
J

|h|

10 /9/

mm/h

A M A (Maile) CT0.5 ( W - 4 0 )

H
CTI CT2

A 6 IO'5 mm/h A-CTI CT4 -CT0.5(W-40mm) D22 (Hollstein/Kiaruier) / 10 /

M -4- ^ + n
i KB

CTO. 5

r-L-T6 7 8 9

-4-

Crack-Tip /Farfield Ratio

V 0 KuoiOnObMi ( m m , / 2 )

Fig. 14.

K ^ d for constant crack growth rate, , versus the initial crack tip/farfield ratio, V Q K lid /cr nO' for l%CrMoNiV, 550 C.

- 55 -

2000

E ^ ^ z s s? 1500

1000

500

T 1 2 3 4 5 6 7 8 9 C'ock-Tip/Farlield Ratio Vo~ K<jo/<7>io sii (mm" 2 )

Fig. 15.

K^id f r c o n s t a n t growth rate versus the initial crack tip far field ratio, V Q = K l i d / a n 0 , for 12% Cr steel, 550 C, creep ductile.

56

CCP

35

E E

30

O) c
0)

r"*=

(D

20

4 CT25/50S [15] CT25/50S [18] XA CCP25 [18] ROUND ROBIN CT25/50S [ 5 ] 1%CrMo(Ni)V , T = 550C

15'

100

1i

1000 loading time ( h )

10000

100000

Fig. 16.

COST 505 - Results

Summarized Data
D Seg.3 as received condition A Seg.4 as received condition Y Seg.6 aged 560C, 10000h X Seg.9 as received condition precracked at 530C " + Seg.10 as received condition Seg.10 as received condition from pot, drop measm.

35

Ol 00

1000

loading time ( h )

10000

Fig. 17.

Creep Crack Growth, 28CrMoNiV49 (KWU 1508), 530C, CT25S

KA - ^Pa/m 50 10M(iV \ S.

1
N

>. "s s
'S. i

Rotor steel

N .

20 v

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

\^Aa=0-5mm (Ref.lSl.
o

^^^rcT25,530 C-: : 10
a/w=0.36

'^

^DENT-10:

-TCT251 - V_ ifSfo-cra =02mm

i"
5

7 *"" ^

JM L AasO-OlmniilRefZylb]:---.-^ "~- . _ : _ ; ; _ : ^ _ . L =..~~ . j u-y. ::==.:;

:

: - ?~' fes-' . . i =1-5-.

-, -:. = i r = '

"-= -

:1

-i

_."=7
10*
i

::

)\i '

-rv 10

"1 103

"'

"~~

-~:

105 10^ Time fo initiation-hours

Example Depending 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 time may v a r y between 20h (Aai=0.01mm) and 20,000h (&ai=2ram) a t K. = 20MPa/m".

Fig.

18.

Creep

crack

initiation

times

in

lCrM oV r o t o r

steels

[18]

59 -

60

Aa D 0.5 mm A 1.0 mm X 2.0 mm spark eroded crack tip cast mat. [16] 1 mm forged [6] 0.5 mm

50
co

Q.

1
O

40

scatterband, makroscop. Creep Crack Initiation Siemens 1989

30 -l
20

10

1iii -

TTII-

100

1000 loading time ( h )

10000

100000

Fig. 19.

Creep Crack Initiation, 1% CrMoNiV 530C, CT25S

60-

50
ra
Q.

_ 40-

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 1%CrMo(Ni)V,T = 550C

30

20

Hollstein 'CT12.5-50 Jrom ,[ t ]

10 10 loading time
Creep Crack Initiation Aa Criteria

10000 M a i I e/Ts c h euschner Siemens Ragazzoni Hollstein Holdsworth Rintamaa D. x + A Y A H a < CT25 -,CT50 X D o u e
-0,5mm - 0,5mm 0,3mm

0.004W(CT),0.01W(DENT)

Fig. 20

spark eroded Creep Crack Initiation with Parameter K

0,2mm

fatigue precracked Uri I 3

0,5mm

K.

iH/mrrO
tf30 CrMoNiV 4 11, AMA, 550 C
A - Cs25. q/t=0.55 7-09. q/W =0,4 a = CT100.q/W =0.55 O - Cs50. q/ty=0.55 O - D60. (v^W =0.* B-060. q^W =02 = D60.(VVi=O.1

3*10*

10'
Fig. 2 1 . Stress Intensity versus crack initiator time for a R = 0,01 W, l%CrMoNiV, 555C [21]

62 -

60

5U
ra o. Z. 40
TO

a/W = 0.55 a/W = 0.55 a/W = 0.55 D g a/W = 0.40 D60 a/W = 0.40 D60 a/W = 0.20 D60 a/W = 0.10 a/W = 0.50 a/W = 0.50 a/W = 0.20 11 %CrMo(Ni) V,T = SSO'C

CT 25/50S CT 50/1 OOs CT100/200

u>

30

20-

/Hollstein CT12.5-50 J r o m ,[ 3 ]

10 10
loading time Maile/Tscheuschner x + Y D A Siemens Ragazzoni Hollstein Holdsworth Rintamaa CT25 -,CT50 X B y e a O D -o
~ 0,5mm 0,3mm 0,2mm ~ 0,5mm

10000

Creep Crack 0.004W(CT),0.01W(DENT) ~ 0,5mm Initiation Aa > spark eroded Criteria

Fig. 22

fatigue precracked Creep Crack Initiation with Parameter Spec. Thickness

10 .

X CT 12 D CT 25
A CT 50 Y CCP25 + SENT 12 1%CrMoV,T = 550<'

Y B
T

B B

Sua
Y B B

10"

'

CTU
m B B

B
:

AV

T3

re
B
B

jyk

10"

Qi-iW
ET
B

^s&P
a
+

^BB 3 aa B
B

10"
10

B B 0

20 K|id

(MPaYm)

50

100

Fig. 23.

COST 505-Group 6, Summarized Data

10" D A + X Y D 20, T = 530 C (forged new material) D 35, T = 530 C (casting) 1%CrMoV 10 -2

welded,HAZ

E E 10"
O) UI

*
ra
D ^ 4.

10 -4

+
ti
Y Y

10' 10 20 K, i d (MPa\/m) 50 100

Fig. 24.

COST 505-Group 6, Summarized Data, CT25

10"

Summarized Data
D Seg.3 as received condition A Seg.4 as received condition Y Seg.6 aged 560C, 10000h X Seg.9 as received condition fatigue precracked at 530C + Seg.10 as received condition

10"

E
O) O)

10*

# 2/ I lv I

9 X

10"
/

*
^

lvYl

I I
10"
10
20 K||d (MPatfm) 50 100

Fig. 25.

Creep Crack Growth, 28CrMoNiV49 (KWU 1508), 530 C, CT25S

10

1 % CrMoNiV-Steel 530 C, CT 25s

a* ot 2.0 m m 1.0 m m 0.5 m m rameter lines for accumulated crack d e p t h fatot

cr>

RI o

crack growth curves

20

Kild (MPa i/rfT)

50

100

Fig. 26.

Creep Crack Initiation and Growth

10"

10 -2

Aaqes . 2,0 mm 1,0 mm D 0,5 mm

je

10"
O) 00

7o
T3

SP
10-

c*/\

10' 10

20 K||d

(MPa^m)

50

100

Fig. 27.

Creep Crack Initiation, 28CrMoNiV49 KWU (1508), 530C, CT25S

HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS Partii 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 fr 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 800C and 1% CrMoV steel at 550C, 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 different laboratories 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 parameters 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) [28]. Collaboration was achieved through the Versailles Agreement on Advanced Materials and Standards (VAMAS) [29]. 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 286 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 Rpo . .2 _ and ultimate tensile stress R m, conventional tensile tests were conducted at room temperature and 550C 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 the pO. 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 crack growth 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 -

25C 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 550C to a good approximation by a Norton law m i. = 1.3 I O 2 0 <r6'5 L [h1] , < r in MPa. (1) n 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. 12): I . = 7.06 1 0 9 min e0369 [h1]f
v

in MPa.

(2)

When the creep strain and strain rate are represented by a law according to Garofalo, to include primary creep then,

74

RI"
G

and (3)

cr = e o t 1 " 6
A

I + min* + l min . [h _ 1 ],

icr V e " r r = eo

where from Fig. 13 and 14 e o = 2.16 o ic 10 m e_ o and

4

and R' can be described by

0.0088 <r Q e ( r in MPa) .

R" = 3.08 I O 3 o 0 ' 5183 [h 1 ]

Alloy 800 H The test material was a 32% Ni 20% C r 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 strain rate . for the Alloy min 800 H at 800C may be described by Norton's law with stress in MPa: ? . = 5.7 I O 1 6 min o6'5 [h _1 ]
L J

uniaxial

creep

(4)

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 PROCEDURE 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 550C. Loads were chosen mainly to give approximately 4 mm of crack extension in approximately 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 displacement methods, all and tests, crack it was required using that load line of

length,

electrical

potential

should

be recorded

continuously.

Other methods

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 538C and 594C mainly on CT specimens with 25% side grooves.

- 77

PARAMETER DETERMINATION

Crack Growth Rate The creep crack growth rate was determined from the crack length versus time records. In general, these were constructed 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 concerning the 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 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 points or 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 v * = *c F I T ' <5> n where F is load, v_ is the load line displacement rate due to
C

- 78

creep alone, Bn is the net section thickness and b is the 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 common c 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 = TnW

(6)

where T J is obtained from limit-load analysis. The values used in the unified approach for T J are [34,36,37]: for CT specimens: for SENB specimens: for SENT, CN specimens: T I = 2+0.52b/W T J= 2 i j = 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 T J have been obtained using the expressions listed above in Fig. 19 with those calculated from (see Refs. [38-42]), including those

compared

alternative derivations given by Eq.

based on numerical simulations incorporating the material law (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
9u

C* = [ W* dy - T gir

ds

(?)

The equivalence between (7) and (8) is only valid for a viscous material under steady-state creep conditions, when elastic strains are negligible. e mn W* is the stress work rate W* = T <r. .d. . 1 oJ 3 ^ and U* is defined as U* = o
Vn

F d v " ,

where the following notations are used: r ds o-ij Ti = o-ij nj eij ui integration path around the crack tip connecting the lower and the upper crack face in a counter-clockwise direction, arc length on r, components of the stress tensor, components of the traction vector, c o m p o n e n t s o f t h e strain rate t e n s o r , components of t h e displacement rate vector.

The s u m m a t i o n c o n v e n t i o n i s implied f o r repeated i n d i c e s . S u b r o u t i n e s t o c a l c u l a t e J a n d C * integrals h a v e been implemented into t h e finite-element p r o g r a m m e A D I N A [ 4 4 ] . S i n c e a l l q u a n t i t i e s entering Eq. (7) a r e k n o w n i n t h e c o u r s e of a n F E M c a l c u l a t i o n , t h e line integral c a n readily b e evaluated. More effective a n d easily extendable to t h r e e - d i m e n s i o n a l p r o b l e m s i s t h e c a l c u l a t i o n o f r e l e a s e rates by t h e v i r t u a l c r a c k e x t e n s i o n m e t h o d [ 4 5 ] . B o t h m e t h o d s h a v e b e e n e m p l o y e d a n d lead t o t h e same r e s u l t s w i t h i n t h e machine-dependent accuracy, s e e [ 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 redistribution time 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. 20, 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, i j is determined numerically by comparing Eqs. (5) with (7) or (8), then T J values of 0.81 (0.78) and 0.71 (0.64) are evaluated for the creep laws (1) and (3), respectively. The T J values without brackets hold for plane stress conditions and those in brackets for plane strain conditions. These values are in fairly good agreement with T I = n/(n+l) = 0.87 determined from Eq. (6) and which were used in 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 T I values mentioned above. In Fig. 21, 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 effective thickness B = B-(B-B )2/B [46]. Calculations, with and

81

(1). It was assumed, as determined in [3], that after an incubation period of 1 hour = 7.41 10~3 c*0'72 with in mm/s and C* in N/(mm>s). (9)

These finite-element calculations gave i j values of 1.86 and c 1.82 for plane stress and plane strain, respectively, which are in good agreement with T J = 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-stress calculations 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], i j = 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 adoptei of i i adopted in the unified approach as reasonable approximations.

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 given for K and C* in Figs. 25 and 26, respectively. It is apparent that the unified approach has only little influence on the extent of the 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 800 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 29, respectively. Using a uni fed approach had no influence on the extent of the 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 and C* is about the same as for

determined by

[38] for the same batch of material and

different geometries and sizes of specimens That is shown by the power law correlation = 0.0034 c * 0 , 7 3 3

(see Fig. 30).

(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 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 550C and Alloy 800 H at 800C.

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, 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. I n 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 I I 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. I t can occupy up to about 3 0% 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 = 3 C*'85/e (11)

with 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 difference method, 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 creep deformation, 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 = TSyc*

<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 follows.

of this behaviour. This can be

obtained

as

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 is given by

= 11) [ ^ ] n / ( n + 1 ) (c'rc)V(n+D

(13 )

where In is a nondimensional function of n and state of 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
n/(n+1) /V _ . C*l ,'_!/< + l) o=b[?]" "'"^dr)V(n

of
(14)

which is similar in form to the steadystate relation Eq. (13) v '. Because the dr and rc are raised to a small fractional power in Eqs. (13) and (14),

k s

o WIT

s 'a

<15>

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
L n J

(16)

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 damage occupies 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. The results 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 correlate with 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 steel, 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 by the 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, Springer Verlag, Berlin, Heidelberg, New York, 1987. [2] Ewald, J., Berger, C , Brachvogel, H., Investigations on Crack Initiation and Propagation under Static, Cyclic and Combined Loading Conditions of 1% CrMoV Steels at 530'C; COST 505-D20/D21, Annual Progress Reports and Final Report, Siemens-KWU, Mlheim, June 1989. [3] Hollstein, T., Kienzier, R., Numerical Simulation of Creep Crack Growth Experiments; COST 505-D22, Annual Progress Reports and Final Report, IWM Freiburg, December 1988. [4] Kanbach, H., Crack Growth in Welded Turbine Materials and Elevated Temperatures; COST 505-D35, Annual Progress Reports and Final Report, AEG Frankfurt, March 1989. [5] Raggazzoni, S., High Temperature Crack Growth in Steam Turbine 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 in Turbine Steels; COST 505-UK5, Annual Progress Reports and Final Report, GEC Turbine Generators Ltd., Rugby, 1989. [7] Webster, G.A., Djavanroodi, F., Elevated Temperature Crack Growth in Steam Turbine Materials, COST 505-UK18, Annual Progress Reports and Final Report, Imperial College, London, January, 1989. [8] Rintamaa, R., Salonen, J., Auerkari, P., Residual Life and 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, priv. comm. P., Hritier, J, UNIREC, Firminy,

[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 of Strength, Kiev, priv. comm. [18] Maile, K., MPA, Stuttgart, priv. comm. [19] Mandorini, V., IRB, Milano, priv. comm. [20] Nazmy, M., ABB, Baden, priv. comm. [21] Piques, R., ENSMP, Evry, priv. comm. [22] Rantala, J., Imatran Voima, Vantaa, priv. comm. [23] Remke, M., RWTV, Essen, priv. comm. [24] Rdig, M., IRW-KFA, Jlich, priv. comm. [25] Saxena, A., Han, J., Georgia Tech., Atlanta, priv. comm. [26] Tscheuschner, R., Granacher, J., IfW, Darmstadt, priv. comm. [27] American Society for Testing and Materials, ASTM Committee E 24.04: Cooperating Program on Creep Crack Growth, Chairman: A. Saxena. [28] Japanese 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 Report, Issue 1, February 1989, Ed. T.B. Gibbons, NPL, Teddington.ton. [30] Tscheuschner, R., IfW, TH Darmstadt, priv. comm. [31] Kempf, B., Bothe, K., Gerold, V., Study of Fatigue and Creep-Fatigue Interaction in a High-Temperature Alloy, COST 501-D11, Final Report, Stuttgart, 1987. [32] Drossier, E., Danzer, R., Aigner, H., Mitter, W., Lebensdauer von Alloy 800H unter kriechnahen Bedingungen, COST 501-Al, Endbericht, Leoben, 1987. [33] ASTM-E 399-78, Standard Method of Test for Plane Strain Fracture Toughness of Metallic Materials, Annual Book of ASTM Standards, Section 3, Vol. 03.01. [34] ASTM-E 813-81, Standard Test for Jic, A Measure of Fracture Toughness, Annual Book of ASTM Standards, Section 3, Vol. 03.01.

- 93

[35] Saxena A., Han, J., Evaluation of Crack Tip Parameters for Characterising Crack Growth Behaviour in Creeping Materials, ASTM Task Group Report, Joint Task Group: E24.08.07/E24.04.08, 1987. [36] Webster, G.A., Crack Growth at High Temperatures, in Engineering Approaches to High Temperature Design, Eds. B. Hilshire and D.R.J. Owen, Pineridge Press, 1983. [37] Hollstein, T., Djavanroodl, F., Webster, Holdsworth, S.R., High Temperature Crack Growth in 800 H and 1% CrMoV Steel - The Results of an EGF Robin; in: Failure Analysis - Theory and Practice, Ed. E. Czoboly, EMAS (1988) Vol. II, 656-668.668. G.A., Alloy Round ECF7,

[38] Hollstein, T. and Kienzier, R., Fracture Mechanics Characterization of Crack Growth Under Creep and Fatigue Conditions, IWM Report W 2/87, Freiburg, February 1987. [39] Kienzier, R. and Hollstein, T., Experimental and Numerical Investigations of Creep Crack Growth, Proc. 3rd Intern. Conf. Creep and Fracture of Engineering Materials and Structures, Swansea, The Institute of Metals, 1987, 563-576. [40] Hollstein, T. and Kienzier, R., "Numerical Simulation of Creep Crack Growth Experiments", IWM Report Z13/87, Freiburg, December 1987. [41] Koterazawa, R. and Mori, T., Applicability of Fracture Mechanics 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 Engineering Approach for Elastic-Plastic Fracture Mechanics Analysis, Topical Report No. EPRI NP-1931, Research Project 1237-1, General Electric Co., Schenectady, July 1981. [43] Landes, J.D., Begley, J.A., A Fracture Mechanics Approach to Creep Crack Growth, ASTM STP 590, American Society of Testing and Materials (1976) 128-148. [44] Bathe, K,.-J., ADINA, A Finite-Element Program for Automatic Dynamic Incremental Nonlinear Analysis. Report AE 84-1, Massachusetts Institute of Technolgy, Cambridge, Mass., USA (1984). [45] Parks, D.M., The Virtual Crack Extension Method for Nonlinear Material Behavior. Comp. Methods Appi. Mech. Eng. 12 (1977) 353-364.

94

[46] Shih, C F . , DeLorenzi, H.G., Andrews, W.R., Elastic Compliance and Stress Intensity Factors for Side-Grooved Compact Specimens. Int. Journ. of Fracture 13 (1977) 544-548. [47] Siegele D., Ockewitz, A., Hollstein T., Berechnung des ij-Faktors fr eine 3-Punkt-Biegeprobe, IWM-Bericht V 25/87, Freiburg, July 1987. [48] Nikbin, K.M., Smith, D.J. and Webster, G.A., An Engineering Approach to the Prediction of Creep Crack Growth, J. Eng. Mat. and Tech., ASME, 108, 1986, 186-161. [49] Tscheuschner, R., Maile, K., Stichprobenartige Untersuchung des Kriechriverhaltens von Proben mit angeschwungener und von Proben mit erodierter Ristartfront, 26. Oktober 1989 [50] S. Kubo, K. Ohji, and K. Ogura, An Analysis of Creep Crack Propagation on the Basis of the Plastic Singular Stress Field, Engineering Fracture Mechanics 11, (1979) 315-329. [51] H. Riedel, The Extension of a Macroscopic Crack at Elevated Temperature by the Growth and Coalescence of Microvoids, in Creep in Structures, (A.R.S. Ponter and D.R. Hayhurst, eds.) Springer-Verlag, Berlin Heidelberg (1981) 504-519. [52] J.L. Bassani, Creep Crack Extension by Grain Boundary Cavitation, in Creep and Fracture of Engineering Materials and Structures (B. Wilshire and D.R.J. Owen, eds.), Pineridge Press, Swansea (1981) 329-344. [53] Saxena, A., Creep Crack Growth under Non-Steady-State Conditions, Fracture Mechanics, Vol. 17, ASTM STP 905, 1986, 185-201. [54] Rdig, M., Kienzier, R., Nickel, H., Schubert, F., Ermdungsund Kriechriwachstum in Rohren eines Rhrenspaltofens 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 0.22

Si 0.24

M n 0.64

P 0.009

S 0.003

Cr 1.29

Mo 0.66

Ni 0.66

V 0.28

Al 0.014

Cu 0.12

Sn 0.009

Heat Treatment : 6 h 930 C, oil 10 h 690C, air Room temperature mechanical properties:

Longitud., removal position: near surface

R p o z , MPa 613 Rm, MPa 727 A, 7. 19.8 Z,l 72 Av,J 139, 158, 163 near KA1, l e f t side (see f i g . 1)

Rp0.2, MPa 623

Rm, MPa 744

A, 7. 19

Z.7. 69

Av,J 92, 8 4 , 85

near DX, right side (see fig. 1)

Table 1: Details of t h e 1% CrMoV test material given by the steelmaker

96 -

Specimen

Position

R po.2 [MPa]

R m [MPa]

Z [%]

Ac 5 [%] 21.0 22.0 20.5

A, , n L=60 [%] 19.0 19.0 18.0 19.0 19.0 16.5

AB 7 AA8A AA63
1 - AA65
<U

center 0.23 R

483 536 594 608 561 600 615

635 676 705 712 683 714 721

70 72 68 71 72 72

R/2
surface center

Ol E 1

a. CG61
O

/
21.5

CG63 CG65

R/2
surface

AB8
AA8B AA62 AA64
0

center 0.23 R

316 355 372 388 380 375 397

345 389 400 410 398 400 416

86 89 89 88 89 89

27.5 24.5 23.5

22.5 17.0 16.5 20.5 20.5 20.5

R/2
surface center

o CG60 LT)
LD I I

/
25.5 22.0

CG62

R/2
surface

1 CG64

Table 2: R e s u l t s of the t e n s i l e t e s t s on 1 % CrMoV s t e e l (R radius of bar), from [3]

- 97

Bezeichnung der Probe Kohlen&toll *;. Silizium Mangan Phosphat Schwein Stichlolt Aluminium losl Aluminium Chrom Molybdn Vanadium Nichel Kobalt Wollram Tilan Niob Eisen Kupier Zink Zinn Blei Antimon Arsen 'I. *'. *! /. 'f. '

X/K

=0.1
.21 .25 .61

X/H .20 .25 .62

0.25

X/R . .22 .25

.6M

0.5

X/R

.0.9
.22 .25 .63

.011 .002

.011 .002

.012 .003

.012 .003

.009

.009

.008

.008

''. *' 'f. *' V. *' *' *'. ' "' V. *'. ' "' */.

1.29
.6H

1.29

1.32
.67 .27 .66

1.31

.65
.27 .6s

.65
.27

.26

.61

.66

.11

.11

.11

.11

.008

.00?

.008

.009

.00? .006

.002 .006

.002 .005

.002 .007

"/.
Round Robin M a t e r i a l X/R-Distance from c e n t e r 21CrMoNiV57 P o s i t i o n ; AA

Table 3: Chemical composition over c r o s s s e c t i o n , from [2]

- 98

Axial Position

Stress (MPa)

Time to X Strair (h)

Rupture

0.1

0.2

0.5

1.0

2.0 118 185 770

2,250 7,900

5.0 184 333

1,383 4,000

time(h) EL(X)

RA(X) 78.4 75.8 39.9 12.9

BA

(O CO

263 247 216 185 154 139 216 216

2 2 7 17 42 105 5 4

8 9 29 75 290 565 24 17

34 36 142 428
1,600 3,089

69 92 366
1,102 4,170 7,700

224 429
1,595 4,495 12,486 17,908 1,577 1,362

22.3 18.7

9.0 6.8 5.7 3.9

14.3 13.2

5.0 5.7
43.9 40.8

DA AA

117 100

320 302

690 635

1,275 1,150

Table

4: Creep rupture test results for 1% CrMoV steel at 550"C, from [ 6 ] ; see Fig. 1 for key to axial position

Spec.-No.

Stress [MPa]

Time to % str in [h]

Rupture

0.2 285 260 235 190 155 135 105 80 3.5 4 8 100 530
1425 3000 12000

0.5
22 22 80 460
2000 3900 13500

1.0
36 65 200
1010 4400 9500

time[h] El[%] RA[%]

AK 6 AK 7 AK 4 AK 5 AK 2 AK 3 AK 8 AK 1

88 237 900 4193 10661

22.5 25.7 19.3 10.8 5.9

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 and VOD Process +) Hot forming to 250 mm Hot forming to a 90 mm Rolling to a 19,5 mm Hot forming to 28 x 103 mm Production of seamless tubes by extruding Cold tube rolling

Solution annealing 1130* 30'/ Water quenching

Solution annealing 1125 - 1130* 30'/ Water quenching .+) VOD ... vacuum oxygen decaburisation

Solution annealing 1130" 30'/ Water quenching

Chemical composition of the Incoloy 800 H investigated: Si Weight 0.07 0.46 Mn 0.68 0.020 0.004

Cr
20.26

Ni 31.11

Al 0.34

Ti 0.31

Fe bal

( % )

Plate dimensions : 28 x 103 mm Heat treatment: 1130' + 30'/Water Mechanical data: Testing conditions: Round bar, d or = 5 MPa/s

= 8.0 mm, L = 40 mm

Temperature [C] 25 800

[N/mm2

[i
R

%
573 577 282 270

P1

P0.2

Z 74 74 74 74

289 293 190 192

250 254 155 161

46 46 52 52

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. Specimen Specimen Sideno. type groove [*] 1 AF 2 20 CT 25/50 2 3 4 S 6 7 S 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 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 CT 20/50 CT 20/50 CT 20/50 CT 25/50 CT 25/50 SENT 12/20 SENT 12/20 CT 20/40 CT 25/50 CT 25/50 CT 25/50 CT 25/50 CT 63.5/127 CT 12.5/40 CT 12.5/40 CT 25/50 CT 25/50 CT 25/50 CT 50/100 CT 50/100 CN 12.5/50 CN 12.5/50 SENB 9.5/19 CT 13/26 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. Specimen Specimen Sideno. type groove [*] BF1/CCP5 CN 26/51 20 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 BF2/CCP6 CN 26/51 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 AD CT 20/40 CT 20/40 CT 20/40 SENB 6/6 20 20 20 20 0 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20

F fkN] 84.7 107.6 12.5 12.5 10.0 0.88 11.5 18.5 26.7 11.1 19.1

a/W

0.509 0.504 0.548 0.563 0.550 0.528 0.566 0.555 0.535 0.628 0.545

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

10.5 17.9 16.1 16.0 10.8 10.8 26.7 10.9 14.1 20.0 20.0

0.557 0.561 0.536 0.536 0.515 0.506 0.491 0.550 0.564 0.470 0.375

1-6/ CT 25/50 1-6 CT 25/50 CT 25/50 CT 12.5/40

BD 3 BD 5 AK 4

Table 7: Data of fracture mechanics specimens; 21 CrMoNiV 5 7 T = 550C; constant force

- 102

Laboray No Specimen Ne Specimen type Side IMP 1 A76 CT 25/50 IMP 2 A77 CT 25/50 RWTUV 3 A89 CT 20/40 RWTUV 4 A90 CT 25/50 A69 1 ATOM 5 CT 12.5/25 1 ATOM 6 A70 CT 12.5/25 KFA A64 7 CT 25/50 KFA 8 A66 CT 25/50 UNIREC g A93 CT 20/40 UNIREC 10 A94 CT 20/40 UNIREC 11 A95 CT 20/40 FHIWM 12 A13 CT 25/80 FHIWM 13 A604 CT 25/50 ENEL 14 A91 CT 25/50 ENEL 15 CT 25/50 A92 Pelten 16 A72 CN 4.8/12 Petten 17 A73 CN 4.8/12

qroove % F rkNl 11.5 20 6.36 20 4.3 20 7 20 20 1.7 20 1.6 20 6 20 5.5 20 20 20 20 9 20 4 20 5.21 20 4.38

a/W Ki [MPaVml 11.5 0.6 0.6 15 0.6 15.6 20.2 10.8 0.6 9.7 0.6 0.5 12.32 10.88 0.5 11 14 8 14.4 9.9 11.16 9.35

ai imm/hrl 3.89E-07 6.67E-06 3.46E-05 7.61 E-05 2.05E-06 5.28E-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 = 800C; constant force

103

to

E E A .J_ AA -2*JS A

LA

D "X

1000 ^charpy specimens

2000

I 3000

mm

^charpy specimens

Fig. 1: Specinen layout for the tension tests

Rptt2 IMPQ]

21CrMoNiV5 7

70025 C

600-

25C

500

400-

550C

550C

300

~i 0

0.5 distance from centre x/R

Fig. 2: 0.2% proof stress RPo.2 and ultimate tensile strength Rn for different radial and axial positions (see Fig. 1, from [3])

104

T=20"C D 150 - 3 2UrMoNiV57

D AA. tang /rad O DA, tang/rad

'

0,9

0,8

0,7

0,6

0,5

0,4

0.3

0,2

0,1

Distance fram center x/R

Fig. 3: Charpy energy variation with radius, from [2]

O
D

O a
a

a ISO

a _o

eg o
CD

21CrMoNiV57 D AA, x/R= 0 55, tang/rad, FATT=2SC O DA. x/R= 0.55. tong/rad, FATT=25C

tb

5 0

8
o

.4
G -160 -120 -80 -tO
*0 60 120 160 Temperatin C

Fig. 4: Charpy energy versus temperature, R/2 - position, from [2]

105 -

E 150 O

100 - '

21 CrMoNV57 D AA,x/R=0.rang/rod,FATT= 55 "C O DA.x/R=0, tang /rad, FATT= 35 'C

50 -

oo B

OD D -160 -120 -80 -40 to 80 120 160 Temperature n "C

Fig. 5.:Charpy energy versus temperature, centre position, from [2]

21CrMoNiV57 Av in J 20C 80C AA, ran/rod DA. " "

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

* -(>

* *

^ ^^^afc^'i, ^ ^ '::A

J?*
y BOP* *v.

9827

0 . 1 mm
k..

9831
1% C rMoV steel, longitudinal

Fig. 7: Microstructure of sections

Creep strain 7

6 f-550C 5

216MPa

AA/ / / -0A,

U 3 2
1
.4?^

/ /
/ ' / ' ' / ' '' /

R A

/''"/ ''''/
.si''Cs^ 9 *? ,3P ^^ AA BA
2600 mm

,
500

T I I !
AK

D A I B
CG

1000

1500 Time - hours

Fig. 8: The effect of axial position on the creep properties of 1% CrMoV at 550"C, from [6]
e [%]

100O

2000

3000

4000

500

t [h]

Fig. 9: Creep properties at 550'C, ( : Region BA [6], Region AK [26]; for axial position see Fig. 1) 108

500

200

100

A section AA 1 section B A > Holdswofth (1986) [ G ] f section OA J O section A K . Granacher. Tscheuschner 11388} 1261

20

10

W
20 0 10

-&

102

W3 Time to Rupture [h)

10'

Ws

Fig. 10: The creep rupture properties of 1% CrMoV steel at 550'C

21 CrMoNiVS7, 550C

100

200 a [MPQ]

Fig. 11: The e f f e c t of s t r e s s on minimum creep r a t e a t 550*C

- 109

Erran [ h"1

21CrMoN57. 550C

10"

10-5
m n =7,06-10- 9 e W 3 6 9 d

10

-6 _

10" _L 100
j L

150

200

250 tf[MPa]

F i g . 12: S t r e s s dependence on t h e minimum creep r a t e

'0

100

200

300 dWPa]

F i g . 13: Determination of the parameter co, creep law (3)

110

log R

21 CrMoNiV57, 550 C

-1,5

R=0,00308-cr0-5183
-2

80 1 , 9

100 , 1 2

150

200 250 300 2 , 3

d [MPa] l o g d (3)

2 , 5

Fig. 14: Determination of the parameter R " creep law

YfXVW/

< J

Fig. 15: Microstructure of Alloy 800 H

- 111

l-

10" 10"
' :

10 10

i
^

^d^ &

xx V
L^

s ^

S"

*** in

- 1 5 . 2 4 + 6,4 Log v [h~ , i

10 10
[Mpa]

Fig. 16a: he stress dependence of m i n (full symbols) and of e at 1% plastic strain (open symbols); round bar results (19.5 mm ), see Table 6, from [31]

[Mpa]

Fig. 16b: Relationship between the life time tt and the applied stress o , from [31]

112

1 4 - 14

55

27 Ti .27 .27 27. 2i .27.

,

27

2: ,27 ,27 27, 27 .53 , 53 22 55

A A

A A A A A A A A A A A A A B C D E F G H I K L M N 0

A P

A Q

t\ -?

A S

800

(mm)

Schnitt A-A

m
-io i

(mm)

(mm)

27

A I B

27

ATE

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

500
A70

471 108
A 61
A62

475 [ A63

1 "
AGS

[11
A66 A67

o|o
A69

A64

o|o
AGS

o|o o|o o|o o|o o|o o|o o|o

o|o

O'

O'

ojo ojo

ojo ojo o o o|o o|o ojo ojo o|o

AB7 A89

o|o o|o
A8G ABS

A 90

A91

A92

A93

A9

A95

olo o|o o|o o|o o|o o|o o|o o|o

F i g . 18: A l l o y 800 H cut-up plan

A A

2.0

-ffrlV-f).
^ ( 2 + 0 . 5 2 b / W ) , Ref.(3M

Ref. (35)

15

(2) (38-41) plane stress A O plane strain V A X Specimen CN CT CN CT

FE-Results: Reference

2n-1

1.0 -

0.5 n Fig. 19: Values of uc, a comparison of different approaches; CT: a/W 0.54, CN: a/W 0.43, 7 and are functions of a/W, from [37] - 114 -

21CrMoNiV57, 550 C CN 12.5/50, AQ1

R, -Rt E cr =e 0 {1-e} + 2O

e m i n -t, Eq.(3) a 65 [h- , ] l d i n MPa

m i n = 1.3-10"
^J

plane stress

-i
0.00 1.00

1
2.00

1
3.00

1
4.00

1
5.00

1
6.00

1
7.00

1
8.00

1
9.00

1
10.00

r
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 CT 12.5/40 specimen AB2B, with and without crack growth, creep law (1), from [3] 115

Fig. 22: Experimental and numerical values-3of crack growth, crack growth law: = 7.41 IO C*"'72 (a in mm/s, C* in N/(urn-s)), from [3]

" " .
in -

B B

^
U

V
B

0 Z 1

"T MS s E

'B

:
D D

X*

ue dk * #
'

T u k H

o
B H 1

O 1 1 I K

^
K
[

m G )
m

U D DO

RB

:
10 KS in D 0

'fei sfB'
Wfifl
l/ir

F
'l
1

ko* J
..
1

4
o

o #

- i? i \

: I 4 K IMPa]

1 1
9

Fig. 23: Crack growth rate in 1% CrMoV steel at 550*C as a function of stress intensity factor K participants' evaluation 116

C" IkJ/fmelili

5 C

J0

-A

[N/inm/sccl

Fig. 24 Crack growth rate in 1% CrMoV steel at 550'C as a function of C* integral - participants' evaluation

U3.B[HBD3.,BWI.BW2IBW3.AII3.AH4.A03.AB5.AB6.BCI,BC2.AN6.AEI.AE2 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) BE3I.BE32(CTIO20) BBI1CT63J/I37) A16, AP (CT 13/30) RR7 (CT 1 3/2) CCP3.CCPICN257S R U (SENB I2J/50)

a * X . A a g

K IMPaVm)

Fig. 25: Crack growth rate in 1% CrMoV steel at 550'C as a function of K, unified evaluations 117 -

Q
+

*qoMH COILHDI A*JAU|CTIU-^

tlCTOi'1)

C* ( J / m 2 h r )

Fig. 26: Crack growth rate in 1% CrMoV at 550'C as a function of C*, unified evaluations

1 R I t o y 6BBH , T-B00*C

10

1000

a Ml , U l ; CT H / W Ul ; CT M/40 O ut , ui-, cr H/sa , Ul;C D 4.8II ; CT /SO l; CT IS/SO A M9,, 0 ; C T II. S/IS M . AI. U / l . U/2; VM/* U4,, MS, M t ; C T H/SO 0 U ) ,, A M . U S ; CT n / 4 0

^ r ^
o a

.f*

H'

Fig. 27: Crack growth rate in Alloy 800 H at 800*C as a function of C* integral - participants' evaluation

118

ATM77.A0JWI.AW>64,A66.AMCT')

A.A3.AW.A(CT2(Wtl)

10'
M AW.A70 (CT UJ/25) M AIUCTH/W)

IO"'

IO"4'

10

101
K (MPAVW)

10*

Fig. 28: Crack growth rate in Alloy 800 H at 800"C as a function of stress intensity factor K, participants' analysis.
iu o
X
A76. A77,A64.A66.A90.A<)1,A9 , A 6 W | C T : 5 - S 0 ) Afc9.A93.A94.A95 (CT 20/40)

tp

IO' :
o

Af.<>.A70(CT 12.5/25) A72.A73(CI*HI/|2)

o D

A13(CT2S/>0)

io -

ooo o o /

lo'DO
t

i /t
' = 0003-C* 0 - 7 3 3 . r muli a [ J

f/9

io -

'A? /i A'
i

c* M M
L

ImmsJ

lO'-

io J IO 2

IO 3
7

IO4

C* (J/m hr)

Fig. 29: Dependence of creep crack growth rate on experimental C* for Alloy 800 H at 800'C (unified evaluation)

119

SL
= 0.0034 C
0.733

Incoloy 800H, T 800t;

Specimen (LT) Looding

PTC

CN
X

CT +

CT

CT
o D F = const V = const = const

125/50

25/BO

12.5/50

25/50

12 5 0

B/W

207* side grooves

4 3

C* , N/mm/sec

Fig. 30: Creep crack growth rate as a function of C * for Incoloy 800 H (Z= constant rate of cross head displacement). From [38].
o
a

CT63J/I37 CT5CH00 CT25I50 CT2IV40 cr IISMO CT 13/36 CT 13/30

+
a

m
4

/y

v. i:

(Titrai

>mP

'S

*?
C (J/m'hr)

Eq(11)

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 21CrMoNiV57, T = 550 "C CT 50/100, 20V. SG F = 32.1 k N , ao/W = 054

a [mm]

0.4

- 2

02

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 .2G .258 .25G

J .254 .252 0. \ .25

u E .248

CT50/100-AP3. 20V. SG

a.
" .244 .242 .24 -j 1 1 1 1 1 1 1 1 1 1 1 i i i i

200

400

G 00 time [h]

B 00

1000

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 50/100 specimen AP2

'

Fig. 35: Crack opening displacement at the load line and final creep crack growth values as a function of time; CT 25/50 specimens

*V

T^^W^
. T

-li
li
X

,A Creep crack H *-f > ;,/, ^ ^ !^ V : ? & 3

Spark eroded crack ,* ^_.\ \ eroded cracks V w^^kjtr*'~ 305*^,

Fig. 36: Creep crack initiation in the middle of the crack front, 524 h after loading, from [49]

- ^ ^ V ^ U l l r l i . wr--*-""

^fesf Fatigue crack "j" Creep crack

^Taw'SuKrtSB v ? > <"V

T2
^

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 at crack tip

t' (J/m'hr)

Fig. 39: Prediction of transient crack growth at 550'C in 1% CrMoV steel for some compact tension specimens

124

CT6-VH7

( CT 1 0 0
a

CTJS50
CT 20140

/
' ,.*

a:
F-

CT 12 J/40 CT1VI6 CT 15/30

/ ?

CTiaao

:*** //;:vy
v*

Eq.111)

C* (J/m*hrl

Fig. 40: Effect of specimen size on creep crack growth rate in CT specimens with 'tails' omitted

CT
CN

A*
SENT SEND

* # *
E E IO2

?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 550C EGF550*C JSPS S38C JSPS S38-C ASTM538C ASTMS38C

C (J/m 2 hr)

EGF 550*0 ~ I"" !

1-

EGFS50"C JSPS 594-C

10" 2 1

JSPS 594C ASTM594C ASTM 5940C

10'

10

10"

C* (J/m2hr) Fig. 42: Comparison between EGF data at 550*C with ASTM and JSPS [28] results at a) 538'C and b) 594*C [27]

126

21 Cr Mo Ni V 57 T = 550 C

X
jf

jS

Q X

x "X x
1.0E- 06X X X X
.

x // /
X

*X x
X X

jy^

*
X D

i = 6t4-10"13-K*'
Ref

*x

13]

o
07x

s*

CN 26/51 CT 25/50 ; CT 20/50 ; CT 20/40 CN 12.5/50 CT 12.5/40 : CT 13/26 ; CT 15/30 SENB 6/6 ; SENB 9.5/19 CT 50/100 : CT 63.S/127 SENT 12/20 60 70 80

30

x n a o 40

1.0E- 08-

K [MPa*fm

Fig. 43: Initial crack growth rate vs. initial stress intensity factor

127 -

HIGH TEMPERATURE CRACK GROWTH IN STEAM TURBINE MATERIALS 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 plant industry to assist in the specification of acceptable defect sizes, in remaining life assessment and in failure diagnosis. It is also used as an aid to material selection. Fracture mechanics parameters have been used to model crack propagation at elevated temperatures for over 20 years. However, for much of this period, a major difficulty has been the inherent size and geometry dependence of the available crack growth correlating parameters when creep processes have dominated the fracture process. In practical terms, this has meant that it has not always been possible to accurately assess defect tolerance in large power plant components on the basis of results from relatively small laboratory testpieces. The main aim of the COST 505 High Temperature Crack Growth Working Group (Table 1) was therefore to pool the resources of a number of European industrial organisations and academic institutions in a joint consideration of the problem. The following Review examines the results of the Group investigations relating 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 cracking mechanisms resulting from a wide spectrum of loading cycle types. Stress-strain cycles may be linear elastic or elastic-plastic and have been applied in load or strain control. HTFCG may be due to one or a combination of pure cyclic, primary stress creep or secondary stress creep loading, and is dependent on such factors as material properties, temperature, strain rate, cycle type and hold time. The service cycles experienced by high temperature steam turbine components such as rotors, valve chests and inner casings are complex and comprise a variety of loading transients conceivably spanning the entire range of behaviour. In laboratory tests mounted to generate HTFCG data for the above components, it is often convenient to adopt an idealised isothermal cycle which closely models the most damaging service loading transient at the specific location being assessed. For example, the operational cycle experienced by a rotor is composed of a combination of rotational and thermal transients (Fig.l). The particular combination varies with position in the rotor, and may be dominated by either the primary stress or secondary stress cycle. Hence load control cycle tests may be adopted to model the rotational stress cycle at the rotor bore, whereas strain control cycle tests may be selected to model the thermal stress transient at a rim position. A similar situation exists for valve chests, inner casings and main steam pipework, which experience operational cycles comprising superimposed pressure and thermal transients. Thermal stress variations tend to provide the most likely driving force for cracking at critical locations in valve chests and casings whereas pressure stress cycles are likely to be more influential in the case of pipework. The load and strain control cycles selected by the COST 505 HTFCG investigators are typical of idealised isothermal cycles adopted to model service loading transients in such components (Fig.2). FATIGUE CRACK GROWTH REGIMES Fatigue crack growth behaviour is conveniently considered in terms of three regimes (Fig.3) [8]. These are a lowAK regime close to the fatigue crack threshold, A,K. a mid AK regime in which propagation rates are modelled by a power law [9] (Eqn.l), and a high AK regime in which K M approaches K c At low AK levels close to AK, the magnitude of da/dN is very sensitive to small increases 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 at elevated temperatures. In the mid AK regime, crack growth rates are effectively modelled by an expression of 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 less sensitive to microstructure and mean stress effects. There have been a number of modifications to the Paris Law which, for example, minimise its dependency on material properties and temperature [10] or extend its range of applicability into the low and high AK regimes [12,13]. In the highAK regime, da/dN becomes increasingly sensitive to the level of AK, and particularly to KMJUC as critical Kc or plastic collapse is approached. Depending on the deformation and fracture characteristics of the material, crack growth rates can be strongly influenced by size and geometry. In these circumstances, AK is not the most effective correlating parameter, and alternative cyclic load functions have to be employed to minimise any dependence on size and geometry (see below). In addition to the factors already listed, da/dN in this regime is strongly dependent on microstructure, temperature, environment and frequency (ie. strain rate). Tearing fatigue [14] does not fall within the scope of this Review. In the following text, the term low strain fatigue (LSF) is used to refer to the load/strain transients resulting in linear elastic loading cycles and crack growth in the low and mid AK regimes (Fig.3). Load/strain transients responsible for cyclic plastic loading involving some degree of general yield in tension and/or compression are referred to as high strain fatigue (HSF) cycles. HSF crack growth rates are due to higher AKs and may be influenced by the load/displacement control mode, particularly when there is superimposed creep loading. Since HSF cycles can involve yield in compression but not tension, there are circumstances when the effective AK responsible for crack opening may be linear elastic. Consequently, it is not possible to rigidly fix the lower bound of the HSF crack growth rate regime in Fig.3, and caution with the terminology used in this area is necessary. HTFCG CORRELATION PARAMETERS Low strain fatigue crack growth rates expressed as a function of AK are independent of size and geometry for a wide spectrum of engineering materials at temperatures below the creep range. This means that crack growth rates determined using laboratory specimens may be reliably applied to large components. AK is also an effective correlating parameter in the LSF regime at elevated temperatures when frequencies are relatively high (ie. f>lHz). Cyclic stress intensity factor is a function of cyclic stress, A0\ and crack size, 'a' (Eqn.2), and K solutions are now available for a wide range of laboratory testpiece and component geometries [15]. AK = YA<ra- s (2)

132

where Y is a compliance/geometry function. It is sometimes convenient to consider the mechanics of fatigue crack growth in terms of cyclic CTOD [16,17]. For example, da/dN = CTOD/2 is regarded as a useful 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 E OV (3a)

Assuming that local crack tip flow stresses are increased by hardening to 2V, an estimate of maximum cyclic CTOD is given by:
CTOD = 0.22 (AKTQT)3 (3b)

(1 - R)
where R = KMIN/KMAX.

E C-

In practice, fatigue crack growth is the result of cyclic crack tip opening and does not occur when the crack is closed [19]. Eqn.3b is a reasonable approximation of the situation when the R ratio is positive (ie. when KMIN and KMAX are both due to tensile loading) and when there is no premature crack closure, for example due to crack face oxidation. However, this is not necessarily the case when the crack tip is loaded in compression for part of the cycle (ie. when R is negative). It is then more appropriate to think in terms of crack propagation being due to the effective AK responsible for crack opening, ie. ACTOD = 0.22 AKg.^)* E V 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)

(3c)

where Po is the crack opening load determined experimentally from P-Vc hysteresis loops (Fig.4) [4,20]. This approach is not a simple matter in practical situations, but there are now a number of empirical formulations to estimate qo [21,22], eg. qo = (1 - R/2)/(l - R) (4b)

AKeff is then simply defined by Eqn.5, and the well documented K solutions for a wide range of laboratory testpiece and component geometries are still applicable [eg.15].
AKarr = qo AKTOT (5)

As the magnitude of A K increases, the crack tip plastic zone increases to a size which is no longer small relative to other significant dimensions. Fatigue crack growth rates expressed in terms of A K = r r a r e no longer size and geometry independent when general yield occurs in tension, and alternative parameters are required to describe HSF crack growth rates. Cyclic J integral has been shown to be independent of size and geometry in both low and high strain fatigue regimes [23], and there are now 3 solutions available for a large number of standard geometries [23,24]. For example, for a simple

133

power law hardening material ( 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 line integral for the geometry of interest. The physical basis for using &3 to correlate 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 it then becomes possible to adopt a single parameter throughout the whole range of AK (Fig.3). Equivalent A K is simply AK=pr corrected for plasticity, and the correction may be applied either via a J or an equivalent energy calculation [20,25]. AK*o = A K ^ r Tl fl AfeP Afep EE"1"1 (7)

where Afep is cyclic plastic strain. HSF crack growth rates expressed in terms of A K E O are independent of size and geometry [26], At low load/strain cycling rates and/or with the introduction of hold times, creep processes can influence the rate of crack propagation at high temperatures. Crack extension due to creep is not effectively correlated in terms of \K (or \K=o) because of crack tip stress relaxation effects. In these circumstances, creep crack growth rates are described using the C* energy rate line integral which has been shown to characterise stress and strain rates in the vicinity of a crack tip subject to steady state creep conditions [27], The practical application of this parameter is still relatively limited. Nevertheless, C* solutions are becoming available for an increasing number of specimen and component geometries [27-29]. For example, for a material obeying the secondary creep rate law de/dt = C'o"": C* =C'gi(a/W,n) (5x'"*l> a (8)

where gi(a/W,n) is a function representing the crack tip line integral for the geometry of interest [24], Creep crack growth rates for the COST 505 lCrMoV rotor type steel are independent of specimen thickness and geometry when expressed in terms of C* [29]. LOW STRAIN FATIGUE LSF without Creep The term low strain fatigue was introduced above to refer to the load/strain transients resulting in linear elastic AK*rr and crack growth in the low to mid AK regimes (Fig.3). LSF crack growth rates due to cyclic stresses applied in either load 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 [1,16]. Since at 550C the higher frequency 10Hz/air AKo values are similar to those determined at a much lower frequency in vacuum, the increase in threshold is at least partly due to the reductions in elastic modulus and yield strength incurred at the higher temperature. This trend is anticipated by reference to Eqn.3c (ie. ACTODo is inversely proportional to E and V). In air, AKo increases further with decreasing frequency as a direct consequence of premature crack closure due to crack face oxidation [16].

- 134

LSF crack growth rates increase with increasing temperature. In vacuum, crack propagation rates are inversely proportional to elastic modulus and yield strength (ie. da/dN o . ACTOD I*(AK) 2 /EOY) , both E and OV decreasing with increasing temperature. Modulus normalised crack propagation laws have been proposed to account for the effect of temperature [10], but their application should really be limited 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 becomes increasingly influential in controlling fatigue crack growth rates [16,30,31]. Even at relatively high frequencies, LSF crack propagation rates are generally not accurately predicted on the basis of room temperature behaviour and a modulus normalised growth law. Crack propagation rates are notably accelerated by oxide assisted growth at low AK levels in excess of AKo, and the effect is enhanced with decreasing frequency (Fig.6). High frequency LSF crack growth rates for lCrMoV rotor steel at 530/550C (f>lHz) are shown in Fig.7 [1,3,5]. This data was collected as part of the COST 505 HTFCG programme, 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 turbine applications is limited since defects are generally not tolerated in components subject to this type of loading, eg. blading. Those parts in which certain defects may be acceptable tend to operate at cyclic frequencies much lower than 1Hz and in these circumstances, oxidation and creep become increasingly important considerations. The bores of rotors and main steam pipe welaments are examples of high temperature component locations for which low frequency LSF crack growth data determined in load control may be applicable (Fig.l). LSF-Creep As frequencies are reduced by either lowering the cyclic load/strain rate and/or by extending hold time durations, crack face oxidation and creep strain accumulation at the crack tip increasingly influence LSF crack propagation rates. In addition, as time dependent processes become more influential, the cycle control mode also becomes important. For example, the load is sustained during the hold time in a load controlled cycle whereas it relaxes when the cycle is controlled within strain limits (Fig.2). Consequently, lower frequency LSF crack growth rates are usually faster when the cycle is applied in load control. The effect is not great at low AK but becomes pronounced in the HSF regime (see Fig.21). Providing that the magnitude of the peak tensile stress is sufficiently low to avoid general yield in tension or creep deformation, high temperature fatigue crack growth rates may still be described in terms of AK=-F, using a power law of the type given by Eqn.l, but with C being very dependent on frequency. It is also noticeable that the m exponent tends towards 2 with increasing temperature. The validity limit to using linear elastic fracture mechanics is provided by a critical &CTOD, above which creep deformation causes cracking [30]. At very low AK, crack face oxidation is primarily responsible for retarding fatigue crack growth rates due to crack closure, as seen by the effect on AKo in Figs.5 and 6 [16]. There is a marked beneficial effect of reducing frequency on the apparent AKo for lCrMoV rotor steel at 550C in air.

135

At somewhat higher &K, the combined effects of enhanced oxide growth and creep on crack extension more than outweigh the influence of crack closure (Fig.6). Initially, strain assisted oxidation is the primary driving force for the observed increases in crack growth rate but with increasing &K (ie. &K>20MPa/l for lCrMoV at 550C), the relative effect of environment diminishes and creep increasingly becomes the controlling factor [7,31]. Hence for K>&Ko, the main influence of reducing frequency at elevated temperatures is to increase LSF crack growth rates. The effect is shown for lCrMoV rotor steel (Fig.Ba) [5], 2.25CrMo pipe steel (Fig.8b) [6], 2.25CrMo weldment 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]. The data in Fig.8a also indicate the tendency for Kc to reduce with decreasing frequency. The effect of frequency on LSF crack growth rates for the COST 505 lCrMoV rotor steel cycled in load control at 530/550C is conveniently summarised for &K=35MPa/JT in Fig.9. For f>lHz, da/dNxoxxi. is fatigue dominated and frequency independent (ie. da/dNror*!. = da/dNcvcuc). Static load creep crack growth rates are also plotted in Fig.9 and it is clear that da/dN is creep dominated for f<10 _3 Hz (ie. da/dNroTAi. = da/DNcRKp). The point is also demonstrated in Fig.10 [5], in which low frequency LSF crack growth rates per unit time (ie. da/dt = da/dN x dN/dt) are consistent with static load creep crack growth rates for CT10/20 and CTas/so specimens. Crack growth rates are plotted here in terms of C* determined by substituting PMAX into ON*T in Eqn.8, in addition to using the appropriate value for gi(a/w,n) [24]. A similar conclusion has been drawn with load control test data for a O.SCrMoV pipe steel [32]. At intermediate frequencies (ie. 10*3<f<lHz, Fig.9), fatigue crack growth rates comprise 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 and crack tip creep processes. For example, da/dNcvctic may be accelerated by the presence 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 time and creep ductility, ie.
da/dNcvcL.ic = C(f,th,6!.) (AKj)-"- ,:h -*' , > (10)

where en is uniaxial creep rupture ductility. For a given AKio, da/dNcv.ic typically increases to a plateau level with increasing hold time (Fig.12) [33,34,35]. The behaviour is associated with an increase in the magnitude of C(f,t,&it) and some reduction in m(f,tt,ei) due to the 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 damage is first formed), any changes to the values of C(f,th,6i) and m(f,tn,fen) are mainly due to oxidation. The damage intensity and zone size increase with increasing hold time to ti (ie. the time at which crack tip creep damage achieves a critical condition resulting in the onset of creep crack growth due to local ductility 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 brittle steels [36]. Both the extent to which da/dNc-rcnc is accelerated and the hold time dependence of C(f,t>,eR) & m(f,tt,en) are determined by 6R, since creep ductility influences ta, ti and the intensity and size of the damage zone. For example, ta, ti and the critical crack tip damage density for creep crack initiation increase with increasing creep ductility. Creep crack growth rates are conveniently expressed in the form of Eqn.lla [37], ie. da/dtcREEP = b_ (C*)"
6R

(11a)

and alternatively as
th

da/dNcREEP = f b(th,6R) (C*)(C*) dt

(lib)

where b and q are constants dependent on material and temperature, which may be influenced by prior cyclic damage. The exponent q is typically in the range 0.8 to 1. The term b(t,eR) is a function of tn and 6R, since creep crack growth does not occur until t>ti. In the intermediate frequency regime of Fig.9, creep-fatigue crack growth rates are the summation of Eqn.10 and Eqn.llb (ie. as given by Eqn.9). The magnitudes of both crack growth rate components decrease with increasing 6R, and hence turbine steels with high creep ductilities are more resistant to high temperature fatigue 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 process when cyclic primary loads are linear elastic, when loading rates are slow and/or when hold times (steady running periods) are long. One COST 505 project investigated the situation for a lCrMoV (28CrMoNiV 4 9) rotor steel when periods of creep crack growth (ie. 1,000 to 4,000h @ 530C/Kx-30MPa/m) are followed by a period of low AK cyclic loading at 530C [1]. In these circumstances, cyclic crack growth rates are initially retarded until the crack has extended beyond the influence of the prior creep damage (Fig.14). The crack tip damage in the creep pre-cracked CTis/so specimens was typically discontinuous grain boundary microcracking (Fig.15) [1]. The condition is not dissimilar 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 latter examples 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 Ko levels increase due to oxide blocking and crack closure while da/dNxoTAi. increases for somewhat higher &Ks when oxide assisted crack growth becomes more influential [16,31] (Fig.6). The behaviour displayed in Fig.14 is consistent with cyclic crack growth rates being retarded by crack closure at relatively low initial levels of AK. Prior ageing for up to 5,000 hours at 530C (with no applied load) has no influence 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 are apparently unaffected by 120,000 hours service at 565C (Fig.16) [6]. It is noted that reductions in creep ductility were not reported for either the lCrMoV rotor steel after 5,000 hours at 530C or the 2.25CrMo pipe steel after 120,000 hours at 565C. The possibility of a reduction in HTFCG resistance should not be discounted 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 cyclic primary stress loading in regions of stress concentration, but is more likely to be due to thermal loading. The thermal transients experienced at, for example, steam inlet locations in valve chests and inner casings include on-load periods at relatively constant temperature when creep deformation is most likely to be due to secondary stresses (Fig.l). Consequently, with reference to steam turbine applications, most effort has been directed to the measurement of HSF crack growth rates 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 real components are invariably complex, and that engineering assessments usually assume idealised cycle types. For example, at turbine start-up, through section thermal gradients can be responsible for the generation of large compressive strains in excess of the cyclic yield strain (OA, Fig. 17). During steady running, temperature gradients are low and thermally induced strain levels are effectively zero. Nevertheless residual tensile stresses can be initially high due to reversed plasticity following compressive yield during start-up (B, Fig.17). Creep strain accumulation occurs as these stresses relax while the turbine is on load (BC, Fig.17). On shut-down, components may cool down slowly or rapidly depending on their location in the unit. Thermally induced strain transients are minor on slow cooling (CO, Fig.17a), whereas more rapid cooling is responsible for a tensile peak strain (CDO, Fig.17b). These two service cycle types are referred to as Type I and Type II respectively, and form the basis of the UK CEGB/Turbinemakers' HSF endurance database [38]. HSF crack growth rates have been measured for a 2.25CrMo turbine casting steel at 540C using the two cycle types described above [4]. The approach adopted in this COST 505 study was novel in that large SENB-rs/ioo specimens with short pre-cracks were used to model the constraint existing in the wall section of a turbine casting. Typical test records are shown in Fig.18, and these show how peak and effective load ranges and the loads at the start and end of the dwell period reduce with cycle number in this type of test. The result is that crack growth rates 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 qualitatively similar to that displayed by thermal fatigue cracks in service. This contrasts with crack growth rate records determined from fatigue tests conducted in load control which continuously increase with increasing crack length. The effect of dwell period on HSF crack growth rates is shown in Fig.20. For a Type I cycle, da/dNTorju. increases significantly with increasing hold time. Crack growth rates resulting from a cycle with t=4h are 4 to 5 times faster than those for tn=0. In contrast, there is a relatively small influence of increasing hold time for a Type II cycle (Fig.20). The higher residual stress level at the start of 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 during the dwell period for tn>t<=, although the evidence indicates that crack extension by this mechanism was limited in the highly ductile cast 2.25CrMo steel, even for a Type I cycle with a 4 hour hold time. The observed differences in growth rate due to cycle type and hold time were mainly related to the extent of the crack tip creep 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 [34,36]. Furthermore, the effect on da/dNcycLic and the contribution of da/dNcREEp becomes increasingly more notable in turbine steels at longer hold times (ie. 16<t<200h) [33-36]. There is a marked effect of load/strain control mode on HSF crack growth rates in 2.25CrMo at 540/550C, particularly at low frequency (Fig.21). With decreasing frequency, propagation rates due to a Type I cycle shape applied in strain control are increasingly lower than the da/dNTOTAL. values resulting from the same cycle shape and K E O applied in load control. This is because the magnitude of da/dNcREEP reduces during the hold time of a strain control cycle due to stress relaxation and a reducing C*. In contrast, da/dNcREEP increases during the hold time 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 in practice, parts such as steam pipework, valve chests and turbine casings are joined by welded connections and it is therefore important to know the properties of weldment microstructures. The incidence of defects in weld metal and heat affected zone regions is not uncommon either during manufacture or in service, and it may be necessary to assess their acceptability using fracture mechanics methodology. In many cases, the properties of the weld metal and the HAZ can be inferior to those of the parent steel and therefore need to be taken into account in 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 is complex [40]. For example, each weld bead may comprise a region of as-welded columnar 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 the fusion boundary on one side and regions of fine grain HAZ elsewhere. Modern welding procedures aim to limit the extent of coarse grain HAZ regions such that they are not continuous and can, in the limit and with extreme care, be wholly refined by re-austenisation during a subsequent weld pass. Typically, steam pipe weld heat affected zones extend no more than 2 to 3 mms from the fusion line and this band contains not only the fully transformed HAZ microstuctures referred to above, but also a transition region comprising a similarly complex mixture of partially transformed intercritical tempered ICHAZ and untransformed subcriticai tempered SCHAZ microstructures. There are two schools of thought as to how the properties of weldment microstructures should be established. The first is to identify the metallurgical region responsible for minimum properties and then to simulate the appropriate microstructure as a homogeneous matrix in a block of material of sufficient size to yield the requisite number of laboratory specimens. This approach is adopted to minimise data scatter and to ensure that the properties determined have not been influenced by surrounding microstructures. The alternative opinion acknowledges that it is very difficult to precisely simulate individual HAZ

139

microstructures and that in practice the performance of these regions is almost certainly influenced by the properties of the surrounding matrix. The protagonists of this view support the use of specimens machined from real welds, even with the associated difficulties in producing a weldment which is truly representative 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 real weldments 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/o and C29/SO specimens cycled in load control. The two testpiece types were manufactured with pre-cracks running adjacent and parallel to weldment fusion boundaries (Fig.22). During testing, propagating cracks followed the HAZ microstructure offering minimum resistance to cracking. The fatigue crack growth rates determined for the HAZ structures of two GS-17CrMoV 5 11 steels are 2 to 3 times faster than the rates measured for the parent 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 effect on da/dN-TOTAL. relative to that for the parent steels. The fracture paths followed in the continuous cycle tests were just contained within the 'visible' HAZ, in the relatively soft ICHAZ (Fig.24). In contrast, the fracture paths observed in hold time tests tended to follow the coarse grain HAZ immediately adjacent to the fusion boundary. This is consistent with the observation that the creep crack growth resistance of CrMoV HAZ microstructures (and particularly the coarse grain HAZ) is notably less than that of the parent steel [41]. The creep ductility of coarse grain CrMoV weld heat affected zones is notoriously low, and the higher HSF-creep crack growth rates are therefore probably due to the effect of low creep ductility 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 the G-X22CrMoV 12 1 parent and HAZ structures are similar (Fig.25). However, in contrast to the low alloy creep resistant steel, there is no reduction in HSF-creep crack growth rates measured in the HAZ of the cast 12CrMoV steel, even with the introduction of a 20 minute hold time at peak load in the cycle. The fracture path adopted in the 12CrMoV specimens is outside the 'visible' HAZ and follows the soft SCHAZ, irrespective of hold time (Fig.26). GENERAL OBSERVATIONS The main collaborative effort of the COST 505 High Temperature Crack. Growth Working Group was based around a Creep Crack Growth Round Robin [29]. In addition to 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 turbine steels. The high temperature fatigue crack growth studies covered a variety of aspects which were generally unrelated, and the present Review has had to draw on findings from outside COST 505 in order to complete the story. Nevertheless, the COST 505 HTFCG results have added to the turbine materials database and to the general understanding of the subject. They have also highlighted those areas where our knowledge is incomplete. These include: (a) a quantitative understanding of the effect of long hold times representative of service applications on da/dNcvcuc and d a / d N c M for relatively 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 extremely productive and a useful vehicle for exchanging ideas, and establishing working contacts throughout Europe. CONCLUSIONS The high temperature fatigue crack growth properties of a number of steam turbine materials have been reviewed. Many of the results referred to in the text were gathered by members of the COST 505 High Temperature Crack Growth Working Group who were active during the period 1985 to 1988, but additional information has also been collated from the published literature. The following conclusions summarise the current understanding of high temperature fatigue crack growth in steam 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 and the 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 by crack tip damage due to oxidation and prior creep loading. da/dNcniEp is expressed in terms of the C* parameter. 2. At 530/550C and for a range of steam turbine forging, casting and pipe steels, there is no apparent effect of creep on LSF crack growth rates for f>lHz (ie. the magnitude of da/dNciusEP is negligible). 3. HTFCG threshold AKo levels increase with decreasing frequency due to oxide blocking and crack closure whereas at somewhat higherAKs, LSF crack growth rates increase with decreasing frequency due to enhanced oxide growth. 4. For frequencies less than 1Hz, da/dNTOTL is still mainly influenced by oxide assisted growth effects on da/dNcvcL.xc at relatively low AKs. With increasing A,K, the role of oxidation diminishes and creep becomes more important through its effect on da/dNcvcuc and its contribution in terms of da/dNcnKEP. 5. In a simple engineering model da/dNcvcuc, for a given AKo, increases to a maximum as the crack tip damage zone develops to the critical condition necessary for the onset of creep crack growth. The prior hold time necessary to achieve this peak acceleration increases with increasing creep ductility. 6. At frequencies less than 10~3Hz 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/dNcrci.xc is small relative to that of da/dNcMutp). This is not always the case when HTFCG is due to a strain controlled cycle. In these circumstances, stress relaxation occurs and da/dNcnEEP is not the dominant crack growth component until lower frequencies. 7. The resistance 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.

- 141

8. Prior long term thermal exposure (without load) has no influence on high temperature fatigue crack growth rates, at least while the ageing treatment is not responsible for a significant reduction in creep ductility. 9. In load controlled tests, high temperature fatigue crack growth rates through the weld heat affected zone of a cast lCrHoV 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 a change in fracture path from the partially transformed ICHAZ to the fully transformed coarse grain HAZ immediately adjacent to the fusion boundary. 10. High temperature fatigue crack growth rates through the weld HAZ and parent structures of a cast 12CrMoV steel are similar in both low frequency continuous cycle and peak load hold time tests. For the test conditions adopted, all fracture paths in weldment tests are contained within the soft SCHAZ.

ACKNOWLEDGEMENTS The study has been conducted with the financial assistance of the Commission of the European Communities, under Contract no. COST 0015 UK (CH). The author wishes to acknowledge the many constructive discussions with Mr D.V. Thornton (GEC Alsthom) and members of the COST 505 High Temperature Working Group, particularly Dr J. Ewald (Siemens KWU), Dr T. Hollstein (FhG-IWM), Dr H. Kanbach (AEG), and Dr. G.A. Webster (Imperial College).

142

REFERENCES 1. J. Ewald, C. Berger & H. Brachvogel; "Investigation on crack initiation and propagation under static, cyclic and combined loading conditions of lCrMoNiV steels at 530C\ COST 505 D20/D21 Final Report, Siemens Report No. TW1187/89, 1989, June. 2. H. Kanbach; "Crack growth in welded turbine materials at elevated temperatures", 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 behaviour and failure mode of pre-exposed material", COST 505 UK26 Final Report, Imperial College (Mech. Eng. Dept.), 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. Dept.), 1986, June. 8. T.C. Lindley, C E . Richards & R.O. Ritchie; "The mechanics and mechanisms of fatigue crack growth in metals", Conf.Proc. The Mechanics and Physics of Fracture, Churchill College, Cambridge, 1975, January. 9. P.C. Paris & F. Erdogan; "A critical examination of crack propagation laws", J.Basic Eng., 1963, 85, 528. 10. M.O. Speidel; "Fatigue crack growth at high temperatures", Proc. Symp. High Temperature Materials in Gas Turbines, Brown Boveri, Baden, 1973, March. 11. S.J. Garwood; "Fatigue crack growth threshold determination", Welding Inst. Res. Bull., 1979, 20, 262. 12. R.T. Davenport & R. Brook; "The threshold stress intensity in fatigue", Fat. Eng. Mat. Struc, 1979, 1, 151. 13. G.K. Haritos, T. Nicholas & G.O. Painter; "Evaluation of crack growth models for elevated temperature fatigue", ASTM STP 945, 1988, 206. 14. K.J. Nix, N. Knee, T.C. Lindley & G.G. Cheli; "An investigation of fatigue crack growth in ductile materials at high growth rates", CEGB (CERL) Report No. TPRD/L/3168/R87, 1987, October. 15. Stress Intensity Factors Handbook, Ed. Y. Murakami et al, 1987, Pergamon. 16. R.P. Skelton b J.R. Haigh; "Fatigue crack growth rates and thresholds in steels under oxidising conditions", Mat. Sei. Eng., 1978, 36, 17. 17. G.J. Lloyd; "High temperature fatigue and creep-fatigue crack propagation: Mechanics, mechanisms and observed behaviour in structural materials", Fatigue at High Temperatures, Ed. R.P. Skelton, App. Sci. Pubi., 1983, 187. 18. J.R. Rice; "Crack tip placticity and fracture initiation criteria", Proc. 3rd Int. Conf. Fracture, Munich, Part 2, 1-441. 19. W. Elber; "The significance of crack closure", ASTM STP 486, 1971, 230. 20. S.R. Holdsworth; "Remaining life assessment of high temperature turbine castings", Inst. Metals Conf. Proc. Metals Development in Turbo-Machinery Design, Churchill College, Cambridge, 1988, September, 136. 21. J. Schivje; "Some formulae for the crack opening stress level", Eng. Frac. Mech., 1981, 14, 461. 22. R.P. Skelton; "Cyclic crack growth and closure effects in low alloy ferritic steels during creep-fatigue at 550C", High Temp. Tech., 1989, 7, 3, 115. 23. N.E. Dowling; "Crack growth during low cycle fatigue", ASTM STP 637, 1977, 97. 24. 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 to engineering components at high temperature", ASTM Conf. Proc. Low Cycle Fatigue - 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 a ferritic steel at 550C", NPTEC Report to be issued. 27. H. Riedel; Fracture at High Temperatures, MRE, Springer-Verlag, 1987. 28. 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 plant operating in the creep range", Fatigue Fract. Engng Mater. Struct., 1986, 10, 115. 29. T. Hollstein, F. Djavanroodi, G.A. Webster & S.R. Holdsworth; "High temperature crack growth in Alloy 800 and a lCrMoV steel. The results of an EGF Round Robin", Conf. Proc. ECF7, Failure Analysis - Theory and Practice, Budapest, 1988. 30. J.R. Haigh; "The growth of fatigue cracks in turbine casing steels at high temperatures under predominantly elastic loading", CEGB (CERL) Report No. RD/L/N9/74, 1974, January. 31. J.R. Haigh, R.P. Skelton & C E . Richards; "Oxidation-assisted crack growth during high cycle fatigue of a lCrMoV steel at 550C", Mat. Sei. Eng., 1976, 26, 167. 32. G.J. Neate; "Crack growth in bainitic 0.5CrMoV steel at elevated temperature under cyclic loading conditions", ASME Conf. Proc. Advances in Life Prediction Methods, 1983. 33. D.N. Gladwin, D.A. Miller, G.J. Neate & R.H. Priest; "Creep fatigue and creep-fatigue crack growth rates in parent and simulated HAZ type 321 stainless steel". Fat. Fract. Engng Mater. Struct., 1988, 11, 5, 355. 34. 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 crack growth behaviour of a lCrMoV rotor steel". Proc. ASM Intern. Conf. Fossil Power Plant Rehabilitation, Cincinnati, 1989, March. 36. D.N. Gladwin, D.A. Miller & R.H. Priest; "Examination of fatigue and creep-fatigue crack growth behaviour of aged type 347 stainless steel weld metal at 650C", Mat. Sci. E > Tech., 1989, 5, 40. 37. K.M. Nikbin, D.J. Smith S t G.A. Webster; "An engineering approach to the prediction of creep crack growth", J. Eng. Mat. Tech., Trans ASME, 1986, 108, 186. 38. G. Thomas & R.A.T. Dawson; "The effect of dwell period and cycle type on the high strain fatigue properties of a lCrMoV rotor forging steel at 500-550C", 1980, Inst. Mech. Engrs. Conf. Proc. Engineering Aspects of Creep, Sheffield, Paper C335/80. 39. D. Armstrong S i G.J. Neate; "Creack growth in bainitic 0.5CrMoV steel under creep-fatigue conditions", Mat. Sci. & Tech., 1985, 1, January, 19. 40. P.J. Alberry S W.K.C. Jones; "Structure and hardness of 0.5CrMoV and 2.25CrMo simulated heat affected zones", Metals Technology, 1977, December, 557. 41. S.R. Holdsworth S i D.V. Thornton; "The effect of stress relief heat treatment on the resistance to creep crack growth of the weld and heat affected zones of CrMoV type joints", WI Conf. Proc. Residual Stresses In Welded Construction and their Effects, 1977, November, 133.

- 144

Table 1 COST 505 High Temperature Crack Growth Working Group Organisation Siemens-KWU (Mulheim-Ruhr) FhG-IWM (Frieberg) AEG (Frankfurt) ENEL (Milano) VTT (Espoo) GEC-Alsthora (Rugby) Imperial College (London) Cambridge University Notes: project directly involved with high temperature fatigue crack growth testing only active for one year Project D20/D211

D22 1
D35

UK5 UK18/UK261 UK25 1 - 2

I31 SF2 1

145

Table 2

Chemical Compositions and Mechanical Properties of Steels in COST 505 Programme lCrMoV rotor steels * Turbine casting steels lCrMoV 0.33 0.22 0.009 0.006 1.25 1.18 0.06 0.27 0.050 0.010 0.005 0.19 0.47 0.80 0.014 0.009 1.32 1.03 0.13 0.21 0.040 0.140 0.15 0.38 0.67 0.016 1.330 0.94 0.07 0.31 0.05 0.130 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 0.011 594 705 22 72 84/95 372 400 24 89 UK18 644 792 15 43 20/30 484 440 24 73 13 500 640 25 64 52/75 363 429 23 62 UK5 554 675 20 59 30/36 428 479 18 71 D35 310 521 26 72 50/53 208 309 29 80 D35 D35 599 763 20 39 41/44 243 555 28 61 179 469 32 70 12CrMoV 0.24 0.37 0.69 0.014 0.009 11.30 0.84 1.07 0.20 2.25CrMo pipe new 0.14 0.25 0.56 0.010 0.012 2.25 1.00 0.10 0.05 se** 0.13 0.23 0.31 0.010 0.020 2.30 1.14 0.14

C Si Mn P S Cr Mo Ni V Al Cu As Sn Sb Ti Zr 20C RpO.2 MPa Rm MPa Elong % RofA % Cv J 530/550C Rp0.2 MPa Rm MPa Elong % RofA % Project

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

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

621 748 18 70 22/56 419 481 17 81 D20/D21

219 469 26 68 UK26

156 249 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)

ROTORS (eg. rim) rotational stress -v'vi thermal

cr

rotational stress

*'

Y-.
pressure stress

\ thermal stress CASINGS 0" ' ^ \ \ \ ' i ,

N

stress

PIPEWORK thermal stress

"--1

pressure stress

Jtiermal N g stress

Examples of primary and secondary stress cycles in steam turbine components

Fili Rff:

Org. No.

Fig. Sht. No. 1

147 -

LABORATORY CYCLE TYPES

(a)
Siemens ENEL

AAA/
C T

A""~"~"
t

AAA/
E
E

A A A

(b) Imp. Col.

t
(c) (T

AEG
1

/ V \/
A^ ^-

V " "T

/V^ r\ \J r\ v 1
t t

(d)

O"

G E C

vAA

VV
E

(e)

C T

GEC

^A ^A V \/ ^

'

^AVAT

Idealised cycle types used I n COST 505 HTFCG a c t i v i t y ( a ) - ( c ) : load control cycles, ( d ) - ( e ) : s t r a i n control cycles

Dau

Fili Rtf:

Org. No.

Flg. SM. No. 2

148

1.0E-01

da/dN - mm/cycle
LOW &K REGIME MID&K REGIME HIGH AK REGIME Kc

1.0E-02b

1.0E-03k 1.0E-04

HSF REGIME

1.0E-05b da/dN C (K) ' 1.0E-06 1.0E-071 Ko LSF REGIME

Log AK

Fatigua crack growth regimes due to c y c l i c loading

Filt Raf:

Drg. No.

Fig. Sht. No.3

149

Pmax

APeff=q 0 AP

Po

Pmin

Determination of crack opening load and effective load range

Fila R*f:

Drg. No.

Fig, Shi No, 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/Kma x

Effect of mea n stress, tempera ture a nd frequency on &Ko for lCrMoV rotor steel [1,16]

Filt Rif:

Drg. No.

Fig. SM. No. 5

151 -

da/dN - mm/cycle 1.0E-02 e

1.0E-03 decreasing frequency (550C) 1.0E-04 all frequencies (26C) 1.0E-05

decreaaing frequency (660C) Ko 1.0E-07 Log&K

Schematic representation of effect of frequency on LSF crack growth rates [16)

Oats

File Rrf:

Org. No.

Fig. Sit. No. 6

152

da/dN - mm/cycle 10 1 KrMoV l28CrMoNiV<.9) Load control R=0-1 530/550C 10' "~ F requencies>1Hz
-1

1 1 1 1 1 Siemens [D20/21I Imp Col IUK18I

~ ~

12 da = 7x 10"8 K2 5 n,m;CyCle

dN

s '

IO'

J^

155

1^ IO

1 20

1 30

M)

S O

1 60 70 B O 90 100 AK - MPa/m

LSF crack growth rates i n lCrMoV rotor s t e e l at 530/550C [1.5]

Ditt

F1l(taf:

Drg.No.

Fig. SM. No.7

153

10 rMoV rotor steel R=1 j / Veld* Hz 10-1

Impc!rial CoUlge

IUK18I

'm
s.

I, u
E o
O

io- -

f
f r xf

lx 3 Hz *

jSxIpHz

IO"",

i? f*^ /^*<3f'' jjo " "

0 0

'* -<dr

^**** 2|te

IO" T

^^r"^^ /ff
r u Hz
100 K (MpgVm)

j
IO"510

The 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 lCrMoV r o t o r s t e e l a t 550C [ 5 ]

for

Dits

FlliRtf:

Drg. No.

Flg. Sht. No.a

154

10
KT1!

As received 2-25CrMo pipe steel 550r R=0

Imperial College IIIK26)

10" 2 - 1xl2 Hz 10-3-J

lo-S
S
10_ 10 AK (MPiVm) 10'

The effect of frequency on fatigue crack growth rates for 2.25CrMo pipe steel at 550C [6]

Fila Rif:

Drg. No.

Fig. Sht.No.8b

155

TESTEG IN V I U i.10''Mr. SOO C. B.OI.MHi. R.M.IOHi. K.Oi. I Hl. R-M.OlHz. R . 0 5 . 1 Hi. (VEIU-METAL (AS-RECEIVEOI

2.2SCrMo (UK25)

PARENT P U T E . (COARSE (RAINED MARTENSITEI

=0-5, (MHz, pp.

& I

1
S

10Hz, pp.

I 20 30 APPUEO ALTERNATING

I - - I I, , , , I , , 1 SO 60 70 10 STRESS INTENSITY. AK.HPi

The effect of frequency on fatigua crack growth rates for 2.25CrMo weldment microstructure at 500C in vacuus [7]

Dm

RI Rtf:

Org. No.

Fig. Sht.No.8c

156

da/dN- mm/cycle 10 GS-17CrMoV5l1 Load control R=0-1 S30-C

1 1 1 1 AEG ID35)

15'

IO* ~

o * r o 530-C

$m# lCrMoV^/ forging/ lines / (fIHzl g D-0-5Hz, o -005Hz. open points: parenl steel, solid points: HAZ

M*

10 0

I 20

1 30

M >

50

60 70 60 90 1 )0 K - MPa/n

The 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 lCrMoV t u r b i n e c a s t i n g s t e e l a t 5 3 0 C [ 2 ]

for

On*

Fita Raf:

Drg. No.

Fig. Sht.No.8d

157

da/dN - mm/cycle IO, G-X22CrMoV121 Load control R=0-1 530C

iiii-r
AEG 1035]

Iff*

IO?

D - 0 5Hz; o - 0-OSHz, open points: parent steel, solid points: HAZ

20

30

I 40

I S O

I 60

I I I 70 00 90 HO K HPa/m

The effect of frequency on fatigue crack growth rates for 12CrMoV turbine casting steel at 530C [2]

Filt Ref:

Drg. No.

Fig. Sht. No. 8 e

158

tia/dN - mm/cycle MP 4Kr35MPa/m Loid Control I R = 0 1 ] , 530/550'C l i l i

1
i

1 lCrMoV forcings Siemens (020/21) Imp. Col. lUKlBI ENEL 1131

152

3-

^^C *x>>-^

vX^>*>~~
i i i 10'

'

\ '
i 10' 102 Frequency - d N / d t - H z

10<-

i l">

103

152

1'

Fatigue crack growth rate versus frequency for lCrMoV rotor steel at 530/550C and ^K=35MPa/T [1,3,5]

FII Rtf:

Org. No.

Flg. Sht.No9

159

10
S t . f i c and cyclic
IUK,8

>

a
B* A

IO"'-

*' *

/..:!Fr
^
S
Im

E E
e

io"2:

5 ^
i M * OD B M X _

i - . . -
g B B* " f c
B

io-a-

*a

1 ' '

'
IO'4 10 10 1 102

BEM S T A T I C emisa m BDI STATIC CT2VU BDll STATIC CTIMO a B D l l STATIC CTKRO BD STATIC CT3S/J0 A B M M M kiCTU/SOA.QI * B D 4 uoooi luCnVMR.OI BE4 r-0001 hi CTVJ0 K.O.t BE4 M 001 hi C T U / U JU0.1 K BEIf-O.OOOI hiCTKWOIUO.i BE r-aooi hi CTicno R-O.I

103

IO4

IO5

C* (J/m 2 hr)

Comparison f low frequency fatigue crack growth rates and s t a t i c load creep crack growth rates f o r lCrMoV rotor steel a t 550C [ 5 ]

Oatf

Fila R i f :

Org. N o .

Flg. Sht. N o . 10

- 160

lOOjjm

Crack tip damage in lCrMoV specimen creep pre-cracked for 2,000 hours at 530C [1]

Drg. No.

Fig. Sht. No. 11

- 161

1.0E-01

da/dNcyclic - mm/cycle

1.0E-04

th(0)<th(1)<th(2)<th(3)<th(4)
1.0E-OS

Crack tip creep damage associated with a high strain fatigue crack in a 0.5CrMoV turbine casting steel [20]

Dit

File Raf:

Drg. No.

Fig. Sht.No.12

162

HOLD TIME th(0)

precrack b d

cyclic crack growth creep crack growth

f

-v^

\ Ve

f Ve

th(1)

'a

th(2)

creep damage zone

th(3)

th(4)

I n f l u e n c e of p r i o r hold time on c y c l i c crack growth r a t e

Date

File Ref:

Org. No.

Fig. Sht. No. 13

163 -

10-2-,
Symbol

X A
D B D

SIEMENS ID20/211 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 f RE - CREEP 4000h/530,C-da/dt C R A C K E D

o
Q

% B

da/dN - mm/cycle

O I

i>* JET*"

V* X
x
'- - --

JCX

x f

"

<aD
fi T=530c R = 0.1 F = 10Hz CT25S/CT25

D O

# # x

ir o S

3 l

10

'

20

/IK

(MPavm)50

10

The i n f l u e n c e o C p r i o r c r e e p damage on c y c l i c c r a c k growth r a t e s i n lCrMoV r o t o r s t e e l a t 530C [ 1 ]

Dite

Fila Ref:

Drg. No.

Fig. SM. No. K

164

s >Awls

100^jm

Crack tip creep damage associated with a high strain fatigue crack in a 0.5CrMoV turbine casting steel [20]

Drg. No.

Fig. Sht. No. 15

- 165

da/dN-mm/cycle ir>0 AK=35MPa/m Load Control (R=0I) SSO" C \ s ~ 1

1 2-25CrMo pipe steel new material ex-service material

\ %^
VS

Imperial College (UK26)

X$\ ^
\
\ Q

1 2

v
0

1 3

ss \ ^^ - \ a ^\

1'

ro3

Frequency-dN/dt-Hz

'

The influence of prior exposure on creep-fatigue crack growth rates in 2.25CrMo pipe steel (ie. after 120,000h @ 565C) [6]

File Ref:

Org. No.

Fig. Sht. No. 16

166

(a) Type I cycle

(b) Type n cycle

tr

A
he-

Type I and Type II service cycles

Oat

Fila Raf:

Drg. No.

Fig. SM. No. 17

- 167 -

TYPE 1 CYCLE E

TYPE [ I CYCLE .10 *o-5 v

f .10
c

VLmaxI 5,000 ^Lmin 1 10,000

i* '
ii

05

Lmax

f 0
o

"O 3-VO 1

t*

1
-05 -V0 -300

1
i 5.000 ^Lmin

1
l 10.000

_ BOO

W O O \ . \ \ P m a x , Pds XXI We z i o m o
_j

.200

^ * ^ P m a x

1 1
5.000 -

1 n 1
n.ooo -W)

_Pds

PdT^
5,000

n.ooo

-100

- ^^~
Pmin

-200

-200

-300

* ^ P m i n
s

^300
M i

s E

M r

k
"

30

J" 2 0

- /X
0

1 1
10.000 Cycles K 0

1 /
o

1
5.000

1
.000 Cycles

5,000

Typical t e s t records f o r large 2.25CrMo SENB specimens subject to Type I and Type I I HSF cycles a t 540C [ 4 ]

Data

Fila Ref:

Drg. No.

Fig. S h t . N o . 1 8

168

da/dN C y C | c - mm/cycle

10'
225CrMo 5WC Icpm AVj_= 102mm Displacement control SENB 75/100

1 2

13

1<

25

50

75

100

Crack length-mm

Variation of HSF crack growth rate with crack length in displacement control bend test for 2.25CrMo at 540C [4]

Data

Fila Rff:

Drg. No.

Fig. Sht. No. 19

- 169

da/dN|. 0 f a |-mm/cycle

101

I
Typen V
Type l i

I I I I M

2-25CrMo Cast Steel 5 W C . 1cpm

A
*

ue tof) b 0-56 J
tf> 0-W

10"2

- "rnrr
\j
r

038
> >/

^
>*

J
f

V
V

r 0 3 3
0-24 S 0-19

y / ~/j

10-3

10*

- p^ f
/

open points part solid points solid points

dwell (hours) 0 0-5 t

/100 Hz Data Line

S
0

1
20

1
30

1 1 1 1 11 1
In 50 60 70 80 90 BO

0 201
AKgq-MfWm

The influence of cycle type and hold time on HSF crack growth rates f o r cast 2.5CrMo a t 540C [ 4 ]

Diu

File Rf :

Org. No.

Fig. Sht. No. 2 0

170

da/dN-mm/cycle 1 K=3SMPa/m 2 ZSCrKo

550 C

"

\ \

id'

K
X" X.VS ^'
\ s

1 2

X. S % S.

I03

Cast 2.2SCrMo / *~^-^^^ Type I strain control cycle ^ . ^ . CEC (UKSI

_\^J^V
i i
1

s ^ ^ \ \ ^ - ^ x ^

2.2SOMO pipe steel l u d control cycle IR:0-11 y imperial College IUK26)

i IO1 IO2 Frequency-dN/dr-Hz

mi

The influence of frequency and load/strain control mode on high temperature fatigue crack growth rates for 2.25CrMo at 540/550C (^K=35MPa/K) [4,6]

Data

Fila Raf :

Drg. No.

Fig. Sht. No. 2 1

171

CT 25/50 fes pieces

Specimen location in cast lCrMoV and 12CrMoV welaments [2]

Oat

FII Rtf:

Drg. No.

Fig. Sht.No.22

- 172 -

da/dN-mm/cycle 10
GS-17CrMoV511 Lead c o n t r o l

AEG ID35I

R=01 530C 10'

*J* i

10-2

i &

- &^ i& 10MoV/ forging / lines / (MHz) f -05Hz. o-005Hz, licked symbols: 20min hold, open points: parent steel, solid points: HAZ 1 0 20 1 30 IO l i 50 l 60 i l

io4

10-5

| 6

70 S O 90 1 )0 K - HPa/f n

High temperature f a t i g u e crack growth s t e e l (GS-17CrMoV511) weidmants [2]

rates

in

cast

lCrMoV

Data

Fil Rf:

Org. N o . F ig. S M . N o . 2 3

173

base material

HAZ

filler metal

1. Hardness profile, specimen B1.3.2

fractur* hit a i u r l l l ^

5
>
X a

25

Hardness

3
CD

\^

1
-2

l
-1

'

1
D

'

1
1

'

l
2

Distance in mm.

Fracture path in cast lCrMoV steel (GS17C rMoV511) weldraent HTFCG testpiece (0.05<f<0.5Hz, t=0) [2]

Drg. No.

Fig. Sht. No. 2U

174

da/dN -mm/cycle 10 i G-X22CrMoVI2 1 Load control B=0-1 530" C

"iir~r
AEG (035)

152-

10-3-

D - 0-5HZ; o - 005Hz; - CCG open points: parent steel, solid points: HAZ ticked symbols: 20min hold

I
20 to

I
50

I
60

IL

70 90 90 100 AK - MPa/m

High temperature fatigue crack growth rates in cast 12CrMoV steel (G-X22CrMoV121) weldments [2]

File Rlf:

Org. No.

Fig. SM. No. 25

175

base metal

HAZ

filler metal

2. H a r d n e s s p r o f i l e , specimen C 1.1

""""
300-

u,

i
O

i y^ | \ ^

>
X m in
250-

1f

Hardn

- e

<

200-

I -3

'

l -2

-1

1
3

rJ1 2

r-^
3

r-

D i s t a n c e (mm)

Fracture path in cast 12CrMoV steel (GX22C rMoV121) weldment HTFCG testpiece (0.05<f<0.5Hz, t=0) [2]

Drg. No.

Fig. Sht. No. 2 6

176

The Communities research and development information service

CORDIS
A vital part of your programme's dissemination strategy
CORDIS is the information service set up under the VALUE programme to give quick and easy access to information on Eunpean Community research programmes. It is available froo of-charge online via the European Commission host organization (ECHO), and now also on a newly released CD-ROM. CORDIS offers the European RtD community: a comprehensive up-to-date view of EC R&TD activities, through a set of databases and related services, quick and easy access to information on EC research programmes and results, a continuously evolving Commission service tailored to the needs of the research community and industry, full user support, including documentation, training and the CORDIS help desk. The CORDIS Databasas are: R&TD-programmes - R&TD-projects - R&TD-partners - R&TD-results R&TD-publications - R&TD-comdocuments - R&TD-acronyms - R&TD-news Make sure your programme gains the maximum benefit from CORDIS Inform the CORDIS unit of your programme initiatives, contribute 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 the service as well as the benefits which they will derive from it, contribute to the evolution of CORDIS by sending your comments on the service to ;ne CORDIS Unit. For more information about contributing to CORDIS, contact the DQ XIII CORDIS Unit Brussels Ms I. Vounakis Tel. +(32) 2 299 0464 Fax +(32) 2 299 0467 Luxembourg M. B. Niessen Tel. +(352) 4301 33638 Fax +(352) 4301 34989

To register for online access to CORDIS, contact: ECHO Customer Service BP 2373 L-1023 Luxembourg Tel. +(352) 3498 1240 Fax +(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 Edited 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 their service life of typically 200000 hours, which in practice extends over more than 25 years. One of the features that must be considered at the design and manufacturing stages and during the assessment of 'fitness' carried out periodically during the service life is the growth of the manufacturingtype defects at temperatures up to about 550 C. Within the concerted action research programme COST 505 on materials for steam turbines, one coordination group studied this problem with a view to enabling a more accurate evaluation of defect acceptability on the basis of data gathered using laboratory test-pieces. The work of the group was structured under three headings which form the three parts of this monograph: Part I: Creep crack initiation and growth in terms of K Part II: Creep crack growth in 1 %CrMoV steel and Alloy 800H an evaluation of the results of the COST 505 and an EGF round robin Part III: High temperature fatigue crack growth 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 determined together with results from complex, simulative 'benchmark' tests that remain 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 nuAroci<; Kai ouvCpoucc, Sales and subscriptions Vente et abonnements Vendita e abbonamenti Verkoop en abonnementen Venda e assinaturas
BELGIQUE/BELGI IRELAND Governmenl SuppSse Agency Ru d Louvam 42 / L e u v a n u w M 42 1000 Bruxelles / 1 0 0 0 Brussel T {021 512 2 6 Fa (02)511 01 84 4-5 Harcourt Roed Durjun2 Tal ( 1 ) 6 6 1 3 111 Fax ( 1 ) 4 7 60 645 ITALIA Avenue du R a 202 / Koningslaen 202 1060 Bruxelles / 1 0 6 0 Brussei Tel (02) 538 51 69 Licosa SpA via Duca rj Caleb 1/1 50125 Firenze Tel (055)64 5 4 1 5 Fax 64 12 57 Telex 570466 LICOSA I GRAND-DUCH DE LUXEMBOURG Europrees Kleeetce BK Ud 66. bdVitosha 1463 Sofia Tel/Fax 2 52 74 75 CESK REPUBLIKA UNITED STATES O F AMERICA / CANADA SCHWEIZ / SUISSE / SVIZZERA ISRAEL StamrjrenbectiatraBe 65 8035 Zurich Tel. (01)365 54 49 Fax (01)365 54 11 ROY hitorrratunal PO Box 13056 41 Mishmar Hayardan Street Tel Aviv 61130 Sklavordustlg. 2 101 Reykiavik Tel.11 56 50 TRKJYE Pree Gazete Krtep Dergl Pazerieme DetOTi Ticaret ve A rfeiaoaehe Sokak N 15 btanbutCagaloglu

Tel. (I) 520 92 96 - 528 55 66

Fax 520 64 57 Telex 23622 DSVO-TR

Tlex 63220 UNBOOK B

Fax (02) 538 OS 41 Autres rjrstributours/ Ovenge verkooppunten: Ubcelrie europenne) Europete boeknendel Rue de la Loi 244/WetstreM 244 1040 Bruxelles/1040 Brussel T (021231 04 35 Fax ( 0 2 , 7 3 5 08 60 Document delivery:

Tel. 3 496106
F a x 3 6 4 8 60 39

5. rue Raiftetsen 2411 jxernbourg Tel 40 10 20 Fax 49 06 61 NEDERLAND SOU C Externe Fondsen Postbus 20014 25(EA's-Graverihaga Tal. (070) 37 69 880 Fax 070 37 89 783 PORTUGAL k n p r a n u Newtonii Casa da Moeda. E P Rua D Francisco Manual de Malo. 5

41SherilSl

Rue de la Montagna 34 / rlergatraal 34 Bte11/Bus11 1000 Bruxelles / 1 0 0 0 Brussel Tel ( 0 2 ) 5 1 1 6 9 41 Fax (02) 513 31 95

Cairo Tel/Fax 39 39 732

DANMARK
J . H. Schret IrytonTajIion AVS Herstedvang 10-12 2620 Albertslund

Havekova22 130 00 Praha 3 Tel. (2) 24 22 94 33 Fax (2) 24 22 14 84 MAGYARORSZAG

Lanham. M D 20706-4391 Tel. Toll Free (800) 274 4688 Fax (301) 459 0056

TU 4363 23 00
Fax (Sales) 43 63 19 69 Fax (Managarnsnl) 43 S 3 1 9 49 DEUTSCHLAND

1092 UsboB Codex

Tel. ( 0 1 ) 6 9 34 14 Fax (Ol) 69 31 66 DtsWbukkxa de Uvroe Bertrand, L d . ' Grupo rsertrend, SA Rua das Terras dos Valea, 4-A Apartado 37 2700 Amadora Codex Tel. ( 0 1 ) 4 9 59 050 Telex 15798 BERDIS IS. Krucza 38/42 00-512 Warszawa Tel. (2) 621 99 93, 6 2 6 - 2 8 4 2 International Fax&Phone

Subscriptions only Uniquement atexinements Renoul Publishing Co. Ltd 1294 Algoma Road Ottawa. Ontario K1B 3 W 8 Tel. (613) 741 43 33 Fax (613) 741 54 39 Telex 0534783 AUSTRALIA Hunter ftablteeoone 56A Glpps Street CcVtingwocd Victorie 3066 Tel. ( 3 ) 4 1 7 5 3 6 1 Fax ( 3 ) 4 1 9 7 1 5 4 JAPAN IGnokiinrya Company Ltd 17-7 Snluku 3-Chome ShinfukLhku Tokyo 160-91 Tel. (03) 34394)121 tJountM DopArtUoSfit P O Box 55 Ctirlose Tokyo 156 Tel. (03) 3439-0124 SOUTH-EAST ASIA Legal Ubrery Services Ltd Orchard P O Box 0523 Singapore 9123 Tel 7 3 0 4 24 1 F a i 24 32 4 7 9 SOUTH AFRICA

Breite Strae 78-60 Postfach 1 0 0 5 3 4 50445 Kln Tel. (02 2 1 ) 2 0 20-0

Telex ANZEIGER BONN 6 6

Fax 202 92 78 GREECE/EAAAoA G.C. Deftharoudekia SA trsernevonal Bookstore Nkrs Strest 4 10563 Aliisns Tel (01)322 8 3 2 3 Telex 21B410ELEF Fax 3 2 3 9 6 2 1 ESPAA

Fax 49 60 255
UNITED KINGDOM H U S O Booke (Agency section) HMSO Publications Centre 51 rene Elms Lene London SW8 5DR

(0-39) 12-00-77
ROMANIA

Tel. 1071)873 9090 Fax 873 8463

Telex 29 71 138 STERREICH

65. Strada Dionisio Lupu 70184 Bucuresti T e U F s x 0 1 2 96 46

Trafalgar. 27-29 28071 Madrid Tel. (1)536 22 95 Fax (91) 536 23 49 reuma-Prense Lloros, SA C e s i e . 37 28001 Madrid

nz'sche Veroged Urttveralttabucftrrendhe Kohlmarld16 1014 Wien Tel. ( 1 ) 5 3 1 6 1 0 Telex112 500 B O X A Fax (1) 531 61-161 SUOMI/FINLAND eUuMsMftsiflsMI Ke^oeKrMjppia K4*kuskatu1 PO Box 218 00381 HasinM T U . (0) 121 41 Fax ( 0 ) 1 2 1 44 41 NORGE .

9.604etiya Oktyabrya Avenue 117312MOSOOW

Tel/Fax (095) 1355227

SLOVAKIA Slovak Technical Ubrary N m slobody 19 612 23 Bratislava 1 Tel. ( 7 ) 2 2 0 4 5 2 Fax : (7) 295 785 CYPRUS CypruaC Induetry Chamber Buerjng 38 Gnvas Drageres A 3 Deligkxgis Street PO Box 1455

Tel (91)431 33 99 (Ubros)

431 32 22 (Susa clones) 435 36 37(areccMin) Telex 4 9 3 7 0 M P U - E

Fex (91) 575 39 96 Sucursal:

Conssfo de Ciento. 391 06009 Barcelona Tel (93) 466 34 92 Fax (93)467 76 59 daCeWunye Rambla dsfs Eshjdrs, 118 (Pi 08002 Barcelona

Serlo
5th Floor. Expon House O r Maude 6 Weal Streets Sandten 2146 Tel.(011)6634737 Fax (011)863-6569 AUTRES PAYS OTHER COUNTRIES ANDERE LANDER Office dee puMcaUotra rnetenos

Tel (93)302 6635

Tal 9 3 ) 3 0 2 64 62 Fax (93) 302 12 99 FRANCE

NsHWsMn Into Csnltc Bertrand Narveaons vel 2 PO Box 6125 Enerstad 0602 Celo 6 Tel. (22) 57 33 00 Telex 79666 NIC N Fax (22) 68 19 01 SVERIGE

I. (2) 449500/462312 X (2) 456630

Service des pubNcattono 26. rue D e s t o 75727 Pans Cedex S Tel. ( 1 ) 4 0 56 77 01131 Fax ( 1 ) 4 0 56 77 00

BTJAB TraMMvgen 13 22100 Lund Tel (046) 18 M 00 Fas(046)16012S 307947

LOA OS Meta Tel. 66 44 88 Fax 67 67 99

2. rue Merrier 2965 Luxembourg Tel.49926-1 TeksxPUBOFLU1324b Fax 46 85 73/48 66 17

IffiUIW^^

NOTICE TO THE READER

All scientific and technical reports published by the European Commission are announced in the monthly periodical 'euro abstracts'. For subscription (1 year: ECU 60) please write to the address below.
o 9 z

>

Price (excluding VAT) in Luxembourg: ECU 20

ISBN ^-flSb-TSab-X

OFFICE FOR OFFICIAL PUBLICATIONS OF THE EUROPEAN COMMUNITIES

***

L-2985 Luxembourg