Martin-Luther-Universitt Halle-Wittenberg Holm Altenbach, Konstantin Naumenko and Yevgen Gorash Zentrum fr Ingenierwissenschaften Non-isothermal creep-damage model Lehrstuhl

fr Technische Mechanik for heat resistant steels D-06099 Halle

in application loading range

National Technical University Kharkiv Polytechnic Institute Chair Dynamics & Strength of Machines Ukraine 61002 Kharkiv

1. Basic Assumption of the Constitutive Creep Modeling

1.1 Basic assumptions and facts
Introduction. A new appropriate approach to the phenomenological modeling of creep and damage behavior based on the well-known concepts of continuum damage mechanics and creep mechanics is developed. In the frames of this work a comprehensive non-isothermal creep-damage model for a wide stress range is proposed. It is based on the available creep and rupture experimental studies and microstructural observations for advanced heat resistant steels. The proposed approach takes into account the following features, which are important for the comprehensive creep and damage modeling for structural analysis of industrial components: The uniaxial form of the proposed non-isothermal creep-damage model describing primary, secondary and tertiary creep stages is formulated (see Fig. 3.3). The model consists of the constitutive equation for the creep strain rate governing steady-state creep behavior and two damage evolution equations for accumulation rates of brittle and ductile damage parameters governing tertiary creep behavior and rupture. To take into account the primary creep behavior and stress relaxation effects a strain hardening function (see Fig. 2.2) is utilized in the constitutive equation. The creep constitutive equation (see Fig. 2.1) shows the stress range dependent behavior presenting the power-law to linear creep mechanism transition with a decreasing stress. To take into account the primary creep behavior a strain hardening function is utilized. The constitutive equation in the form of serious connection of linear and power-law components extended with strain hardening function well describes the creep strain rates for a wide stress range and stress relaxation process under the loading values relevant to in-service conditions of industrial applications. To characterize creep-rupture behavior the constitutive equation is generalized by introduction of two damage internal state variables and appropriate evolution equations (see Fig. 3.3). The description of long-term strength behavior is based on the assumption of ductile to brittle damage character transition with a decrease of stress (see Fig. 3.1). Two damage parameters show different ductile and brittle damage accumulation characters based on Kachanov-Rabotnov concept, but the similar time-torupture dependence (see Fig. 3.1). Since creep and damage effect are heat-activated processes, the creep constitutive and damage evolution equations are extended with the temperature dependence using the Arrhenius-type functions (see Fig. 2.1 and Fig. 3.1). Such a temperature dependence is found applicable only for a quite narrow range of high temperatures. The set of creep material parameters for the 9Cr-1Mo-V-Nb (ASTM P91) heat-resistant steel valid in the stress (0 300 MPa) and temperature (550 650C) ranges relevant to in-service loading conditions is identified. The identification procedure is based on manual fitting of the available experimental data presented by the min. creep strain rate (Fig. 2.1), time-torupture (Fig. 3.1) and rupture strain vs. stress dependencies (Fig. 3.2). The unified multi-axial form of creep-damage model is presented (see Fig. 4.1). To analyze the failure mechanisms under multi-axial stress states the isochronous rupture loci and time-to-rupture surface are presented (see Fig. 3.5). They illustrate that the proposed failure criterion based on the dependence of time-to-rupture on stress include both the maximum tensile stress and the von Mises effective stress. But the measures of influence of the both stress parameters are dependent on the level of stress with ductile to brittle failure transition (see Fig. 4.4 - 4.7).

1.2 Phenomenological modeling

= const T = const = * Fracture

1.3 Stress Dependence of the Minimum Creep Rate

Schematic illustration of the power-law dependence Diffusional flow or Harper-Dorn creep Minimum Creep Strain Rate (cr) n1 < n2 < n3 n1 ~ 1 Low-alloy steels: 2 n2 8 n3 > 8 High-alloy steels: 8 n2 12 n3 > 12 experimental creep tests extrapolation Power-law creep or viscous glide

III
& = g w (s ,H, w ,T ) w

Creep Strain (cr)

n3

2 > 1

n2

I
& = g (s ,H, w ,T ) H H

& cr = g e (s ,H, w ,T ) e

Time (t)

n1

e & cr = g e (s ,H, w ,T ), etcr Creep Constitutive Equation =0 = 0 & H = g H (s ,H, w ,T ), H t = 0 = 0 Evolution Equation (Hardening/Recovery) w & = g w (s ,H, w ,T ), wt = 0 = 0 Evolution Equation (Softening/Damage)

Sigma () After: Yavari & Langdon (1982), Ashby & Jones (1996), Dimmler et al. (2002, 2008)

2. Formulation of Creep Constitutive Law and Primary Creep Modeling

2.1 Minimum creep strain rate vs. stress at various temperatures
Application Range Creep Strain (cr) Low Moderate High

2.2 Primary creep stage fitting of the test creep curves at 600C
0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0 2500 5000

2.3 Primary creep strain vs. stress for the steel P91 at 600C
Low Stress Moderate Stress High Stress

Creep Strain (cr)

10 1
Minimum Creep Strain Rate (cr) Experimental data at 600C, after [1, 2, 3] Experimental data at 625C, after [1, 4] Experimental data at 650C, after [1, 5] Transition from viscous creep to power-law creep mechanism 600C n -1 s cr & = A(T ) s 1 + 625C Model: e s ( T ) 0 650C
-Q A(T ) = Ac exp c RT and -Q s0 (T ) = As exp s RT

Creep constitutive equation: s H n -1 cr & = A s H 1 + e s 0

0.07 0.06 0.05 0.04 0.03 0.02

cr Primary Creep Strain (epr )

cr epr

Strain hardening function: cr H (ecr ) = 1 + a e - b e

0.007
cr epr

0.008
cr -3 epr max = 7.76 10

10 10 10 10 10 10 10 10 10

-1

-2

12

0.007
= 125 MPa
0 2000 4000 6000 8000 10000

= 120 MPa
7500 10000 12500 15000 17500

0.01 0

0.006 0.005 0.004

cr epr = 4.56 10-3

Primary creep strain values: cr cr cr epr = epr + sec (s ) - esec (s )

-3

1
0.12 0.1 0.08 0.06 0.04 0.02 0 0 200 400

Time (t), h
0.12

Time (t), h
Primary creep constants of the steel P91 at 600C: a = 0.5, b = 300 e
cr pr

-4

Creep Strain (cr)

Creep Strain (cr)

-5

-6

Ac = 2300 MPa

-1

h ,
-1

-7

Qc = 200000 J mol , As = 0.658 [MPa ],

Secondary creep constants of the steel P91 at 600C: A = 2.5 10-9 MPa -1 / h, n = 12, s0 = 100 MPa
cr epr

0.1 0.08 0.06 0.04

Primary creep strain function (Boltzmann Function produces a sigmoidal curve): , s -s 1 + exp C where the transition constant C = 9.5 and the transition stress s = 87 MPa corresponding to mean primary cr creep strain epr = 4.56 10-3 + e (s ) = e
cr pr cr pr max cr cr epr min - epr max

0.003 0.002

-8

1 1 1 10
Stress (), MPa

-9

Qs = 36364 J mol , n = 12

-1

= 150 MPa
600 800 1000 1200

0.02 0 0 5 10 15 20 25

= 200 MPa
30 35 40 45

0.001 0

cr -3 epr min = 1.35 10

100

1000

Time (t), h Model without strain hardening function H( ) Model with strain hardening function H(cr)
cr

Time (t), h Test creep curves, after Sklenika et al. (2002) Test creep curves, after Orlov et al. (1998)

Reference: [1] Kloc & Sklenika (1997); [2] Kloc & Sklenika (2004); [3] Sklenika et al. (2005); [4] Gaffard et al. (2005); [5] Kloc & Fiala (2005).

50 100 s = 87 MPa

150 Stress (), MPa

200

technical operating region

II

250

experimental creep tests

Power-law breakdown

300

3. Tertiary Creep Modeling and Failure Criterion Formulation

3.1 Non-isothermal long-term strength curves for the steel P91
1000
Experimental data, after Kimura (2006), Bendick et al. (1993), Kloc et al. (1998), Sklenika et al. (2005), Cerjak and Letofsky (1996), Gaffard et al. (2005), NRIM Creep Data Sheet No.43 (1996), ECCC Data Sheets (2005), etc.:
0.4

3.2 Creep-Rupture Strain vs. Stress at 600C

Application Range Low Moderate Creep Strain (cr) High

3.4 Creep curves of the steel P91 for low and moderate stresses at 600C
0.1 0.09 0.08

3.5 Time to Rupture Surface at 600C Time to Rupture t*, h

190 000 - 200 000 180 000 - 190 000 170 000 - 180 000 160 000 - 170 000 150 000 - 160 000 140 000 - 150 000 130 000 - 140 000 120 000 - 130 000 110 000 - 120 000 100 000 - 110 000 90 000 - 100 000 80 000 - 90 000 70 000 - 80 000 60 000 - 70 000 50 000 - 60 000 40 000 - 50 000 30 000 - 40 000 20 000 - 30 000 10 000 - 20 000 0 - 10 000

t * (s ) =

B
1- k n-k % n -1 + svM smax t s0

High

High stress: Mod. stress:

550C

600C

625C

650C

0.1

Creep Strain (cr)

0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0

11.5 1

0.5 1

100 MPa 90 MPa 80 MPa 70 MPa 60 MPa 50 MPa

Time to Rupture t*, h

2.0E+05 1.8E+05 1.6E+05 1.4E+05 1.2E+05 1.0E+05 8.0E+04 6.0E+04 4.0E+04 2.0E+04 0 -1.5
28

% 0 = 100 MPa, s k = 0.5

n = 12

%0 s

0.01

Moderate

100
Model: 550C 600C 625C 650C

0.005 10 100 1000

Stress (), MPa Application Range Experimental creep-rupture data, after Polcik et al. (1999), Sklenika et al. (2002), Wu et al. (2004), Kimura (2006) Primary creep strain e (s ) in the form of Eq. (2.3)
cr pr

-1.5 -1.0 -0.5 0 0.5 -1.0

(n-k )

Stress (), MPa

Locus of Elongation at Failure

210
4

% 0 (T ) Ductile to brittle transition stress s B(T ) Time to rupture : t * = n -1 % 0 (T ) ] + s n - k s1- k [s

% Q Q % 0 (T ) = Bs exp s , B(T ) = Bf exp f and s RT RT -14 (n-k ) -1 Bf = 2.0 10 h MPa , Qf = 698000 J mol , 1 % = 44940 J mol , k = 0.5 Bs = 0.205 [MPa ], Q s

410

610

810

110

1.210 1.410

1.610

1.810

-0.5

1.0 0 0.5 1.0 1.5 1.5

sI %0 s

Time (t), h

B = 1.25 10 h MPa

%0 sII s

Low

1 0.5
4

Creep strain accumulated during secondary and tertiary stages before the rupture t* : l + 1 s n -1 l + 1 C = 0.009 2 %s 1 AB + k = 0.5 l1 s0 l2 + 1 - n cr k es/ t (s ) = = C s with n -1 %0 s1- k s + s n-k l1 = 0.532 l2 = 17.383 Rupture creep strain: e* (s ) = ecr (s ) + ecr (s )
pr s/ t

Creep constitutive equation :

3.3 Non-isothermal uni-axial creep-damage model

n

Damage evolution equations : and &d = w 1 , t * (l2 + 1) (1 - wd )l2 % Q % 0 (T ) = Bs exp s s RT

10 10 10

sH A(T ) s H & cr = A(T ) e + , 1 - wb [s0 (T ) ]n -1 1 - wd -Q A(T ) = Ac exp c RT and

H (e ) = 1 + a e
cr

- b ecr

&b = w t* = B(T ) % 0 (T ) ] s1- k [s

n -1

1 t * (l1 + 1) (1 - wb )l1

10

Time to 10 Rupture (t*), h

10

10

-Q s0 (T ) = As exp s RT

+ s n-k

Q B(T ) = Bf exp f RT

and

4. Example of Creep-Damage Modeling for Structural Analysis

4.1 Isothermal multi-axial creep-damage model
Creep constitutive equation: Creep constants of the steel P91 at 600C:
A = 2.5 10-9 MPa -1 / h, n = 12, s0 = 100 MPa
a = 0.5, b = 300
n s H 3 s s H A cr vM vM & = A e + n -1 1 w s 1 w b 0 d 2 s vM - b ecr cr H (eeq ) = 1 + a e eq with strain-hardening function:

4.3 ABAQUS model: geometry, loadings and FE-mesh

Temperature: 600C

4.4 Redistribution of the maximum tensile stress max t in the location of brittle rupture
max t, MPa
78.08 71.68 65.28 58.88 52.48 46.08 39.68 33.28 26.88 20.48 14.08 7.68 1.28

4.5 Redistribution of the von Mises effective stress vM in the location of ductile rupture
vM, MPa
90.28 83.13 75.99 68.85 61.71 54.56 47.42 40.28 33.13 25.99 18.85 11.70 4.56

max t, MPa
63.80 58.59 53.39 48.18 42.97 37.76 32.56 27.35 22.14 16.94 11.73 6.52 1.32

vM, MPa
82.36 75.82 69.28 62.74 56.20 49.66 43.11 36.57 30.03 23.49 16.95 10.41 3.87

Damage evolution equations: 1 1 &b = * &d = * w and w l1 t (l1 + 1) (1 - wb ) t (l2 + 1) (1 - wd )l2 with time-to-rupture function:
t * (s ) = B
1- k n-k % n -1 + svM smax t s0

Balance loading: 8.67 MPa

l1 = 0.532,
k = 0.5,

l2 = 17.383

% 0 = 100 MPa, s

B = 1.25 1028 h MPa ( n - k )

Stress, MPa

50 40 30 20 10 0 0 20000 40000 60000 80000 100000 120000 140000 160000 180000

Stress, MPa

4.2 Typical installation of the steam turbine quick-stop valve (VQS) in a power station
Control Components Inc.

Material: 9Cr-1Mo-V-Nb (ASTM P91) heat-resistant steel

Boundary conditions: symmetry

70 60

b a

70 60 50 40 30 20 10 0 0

Heat Recovery Steam Generator

Elastic properties: E = 1.12 105 MPa = 0.3 = 1.26 105 K1

Internal pressure: 20 MPa

FE-Model: C3D8R (8 Nodes) 7740 Elements 10085 Nodes

Maximum Tensile Stress (max t) Von Mises Effective Stress (vM)

Von Mises Effective Stress (vM) Maximum Tensile Stress (max t)

c
20000 40000 60000 80000 100000 120000 140000 160000 180000

4.6 Evolution of damage parameters b and d

1

Time, h
Brittle damage parameter d

Damage Parameter

Steam Turbine

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Brittle damage parameter (b) Ductile damage parameter (d) max t, MPa
69.64 63.64 57.65 51.65 45.65 39.66 33.66 27.66 21.66 15.67 9.67 3.67 0.00

Brittle rupture location vM, MPa

61.55 56.44 51.33 46.22 41.11 36.00 30.89 25.78 20.67 15.56 10.45 5.34 0.00

Time, h

Ductile rupture location

Ductile damage parameter d
0.282 0.259 0.236 0.214 0.191 0.168 0.146 0.123 0.101 0.078 0.055 0.033 0.01

VQS VPC

Condenser

0.996 0.922 0.849 0.775 0.701 0.628 0.554 0.480 0.407 0.333 0.260 0.186 0.112

Condensate Pump

20000

40000

60000

80000

100000

120000

140000

160000

180000

Time, h

1st Principal Total Strain (tot)

4.7 Accumulation of creep strain

0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0 20000 40000 60000 80000 100000 120000 140000 160000 180000

4.8 Summary of FEA results

Problem definition. The isothermal form of the creep constitutive model, presented on Fig. 4.1, is applied to simulate the transient creep behavior and to predict the damage evolution in the typical power plant component. As the component to be analyzed the casing of a steam turbine quick-stop valve is chosen. Such a component has to control and/or to stop the steam flow from a steam generator into a high pressure turbine, see Fig. 4.2. The proposed creep-damage model with appropriate creep constants for the steel P91 at 600C summarized on Fig. 4.1 are incorporated in the FE-code of CAE-software ABAQUS by the means of user-defined creep subroutine.

Location of ductile rupture (Fig. 4.5) Location of brittle rupture (Fig. 4.4)

The 3D-solid geometry of the valve casing designed in the CAD-software SolidWorks is transferred to the FEMbased CAE-software ABAQUS, meshed and provided with elastic material properties and initial-boundary conditions, as illustrated on Fig. 4.3. The FE-simulation was performed in two analysis steps (linear static and non-linear transient) with the following FEA results: The creep behavior simulation has predicted the failure of the component in t* = 164000 hours = 18.7 years. The obtained FEA results show the critical damage accumulation in 2 locations in the final moment of time t*. The first location is situated on the outer surface of the valve casing, as illustrated on Fig. 4.4, and is caused

Time, h

by the brittle damage parameter b critical concentration. In this possible place of brittle rupture initiation the damage accumulation is dominantly governed by the maximum tensile stress max t, as shown on Fig. 4.4. The second location is situated on the inner surface of the valve casing as illustrated on Fig. 4.5 and is caused by the ductile damage parameter d critical concentration. In this possible place of ductile rupture initiation the damage accumulation is dominantly governed by the von Mises effective stress vM, as shown on Fig. 4.5. Since the character of the both damage parameters ( b and d) evolution is found out to be equal in the both locations (see Fig. 4.6), the decision about the type of

rupture is done basing on the redistribution character of the stress parameters ( max t and vM) . Thus, the dominant stress parameter ( max t or vM) defines the type of rupture (brittle or ductile, respectively). Additionally, this assumption is proved by the different evolution character of the first principal total strain tot in different rupture locations, illustrated on Fig. 4.7. The comparison of the creep curves shows, that the ductile rupture location has accumulated almost two times more creep strain than the brittle rupture location. Due to Fig. 4.7, the ductile rupture location has more prevalent tertiary creep stage of the creep curve, but in the both locations the rupture occurs in the same time t*.