9 Oct 2013 ... DIN EN13480-3. FDBR ... The allowable stress in AD2000 is equal to DIN EN
13480-3 and FDBR. ..... 13480-3:2012, Beuth Verlag, Berlin. 2.
Numerical simulation of similar and dissimilar welds of martensitic steels and Nickel alloys D. Hüggenberg, P. Buhl, T. Klein, A. Klenk Materialprüfungsanstalt Universität Stuttgart (MPA), Pfaffenwaldring 32, 70569 Stuttgart, Germany 39th MPA-Seminar October 8th and 9th, 2013 in Stuttgart
Abstract In components of power or process plants operating at high temperatures weldments often become the weakest links due to the specific characteristics of the weld or heat affected zones. In design and life assessment this is considered by weld strength factors reducing the allowable stresses. For common materials material and temperature specific factors are available. These factors are usually obtained by evaluation of crossweld creep tests. Thus, they are applicable for creep loading and describe the weld strength of a fully loaded and at least in the long term range approximately homogeneous stress distribution. However, this often does not reflect the real situation in components. Numerical simulations considering the specific properties of heat affected zones in ferritic and martensitic steels are able to predict the stress and strain situation and the stress redistribution more accurate. In this paper assessments of martensitic welds in piping systems using numerical simulation are described. The demand for efficiency increase in fossil fired plants which lead to the development of 700°C-technology, causes the necessity of the application of Nickel alloys in at least parts of boiler and turbine components. This implies also the necessity of dissimilar welds in such components. In the paper similar assessments for dissimilar welds are shown allowing conclusions regarding stress redistributions and explanation of failure mechanisms.
1
Introduction / Motivation
Welded joints in piping systems are subjected to complex loading conditions. On the one hand the development of the damage depends on the primary stresses, which act permanent like the internal pressure and weight of the system. On the other hand there are secondary stresses, which influence the redistribution of the stresses. These stresses occur due to restraint of the thermal expansion in components with inhomogeneous temperature distributions. A welded joint consists of the heat affected zone, the base and weld material. Due to the different material properties within the weld and its specific loading conditions it is very difficult to evaluate the most critical location and to describe stress redistribution during operation. The design of components in the current regulations DIN EN 12952-3 and DIN EN 13445-3 is carried out with equations for linear elastic material behavior. For components, where irreversible deformations occur due to plastic deformation or creep strain accumulation the use of these regulations lead to conservative results because the redistribution of stresses and the reduction of local stress peaks are usually not considered. Even the treatment of additional forces, bending and torsion moments of the piping system which influence the size and direction of maximum principal stresses is not clearly handled by these regulations. Therefore nonlinear viscoplastic simulations can be carried out to calculate the redistribution of the stresses and the accumulation of inelastic strains in components with welded joints with regard to its different material properties.
2
Comparison of technical rules for lifetime assessement of components for elevated temperatures
Table 1 shows an overview of the procedures of the standards and regulations which are used for design or lifetime evaluation: DIN EN 13480-3 [1], FDBR [2], AD2000 [3] and TRD [4], with regard to the calculation of welded joints in pressurized components. Regulation
DIN EN13480-3
FDBR
σeff
AD2000
(
(
TRD
)
)
(
(
)
)
σallow
Additional reduction of σallow
-20%
-20%
-
-
Table 1: Overview of the procedures of different regulations
The effective and the allowable stress are treated identical in the regulations DIN EN 13480-3 and FDBR. The allowable stress in AD2000 is equal to DIN EN 13480-3 and FDBR. TRD includes differences in the determination of the allowable stresses for 200.000 h of creep. There is used a denominator of 1 and in the other regulations it has a value of 1,25. The calculation of the effective stress is different in the TRD and AD2000. All these procedures make use of equivalent stress formulations but do not consider effects of multiaxial stress states on the material or component behavior. The influence of primary and secondary stresses in welded joints is not handled in a consistent manner. Due to this reason a comparative assessment using these procedures and a detailed inelastic calculation is of interest.
3 3.1
Numerical Investigations for welds in pipework Welded joint with a high bending moment
For comparion with the results obtained using standard regulations, an inelastic finite element simulation of a pipe section with a detailed modeled welded joint was performed. The pipe section has the material properties of the martensitic steel P92 and is subjected to internal pressure and bending moment. The level of the bending moment was defined according to the FDBR regulation with the aim to get an ineligible load factor in the welded joint. The critical value of the bending moment amounts to MB=6,1*108 Nmm. Furthermore the internal pressure and the temperature were assumed to be 85 bar respectively 618°C for the whole simulation for a service period of 200.000 h. At the determined temperature the time dependent material behavior of the P92 base material was modeled by a modification of the incremental Graham Walles creep law which is given by equation (1) and (2).
d 10 A1 m1 10 A2 dt 1 D 1 D n1
3 dD AD1 10 dt q
nD1
n2
m2
3 mD1 AD 2 10 q
(1) nD2
mD 2
(2)
Equation (1) is the modified Graham-Walles creep law whereby it is possible to describe the primary, secondary and with the damage parameter D the tertiary creep effects. The damage parameter D is calculated with equation (2) and depends on the multiaxiality factor q. The parameter AD1, nD1, mD1, AD2, nD2, mD2, A1, n1, m1, A2, n2 and m2 are material dependent parameter that are fitted with uniaxial creep tests. For modeling purposes the welded joint was divided into four areas representing the intercritical, the fine and the coarse grained zone and the weld metal as shown in figure 1. The different creep behavior of these areas was determined with uniaxial creep tests of specimens with simulated microstructure (weld simulation using Gleeble test). The resulting creep curves were the basis of the approximation of the Graham-Walles parameter for each area. The same material properties as the base material were assigned to the welde metal. The creep curves of the different areas are shown in figure 2. Using an user defined material (UMAT) the implementation of the time dependent material behavior of material P92 in a finite element calculation was performed.
Figure 1: finite element model of a pipe section with a detailed modeled heat affected zones (HAZ)
Figure 2: Creep curves of the three areas of the HAZ and WM at a stress of 80 MPa
For geometrical modeling two reference points in cylindrical coordinates were defined. The first reference point was used for the application of the bending moment and the second to fix the model. Both reference points were coupled with „kinematic couplings“ and one degree of freedom in radial direction was coupled to the free surfaces of the pipe section. By choosing these boundary conditions expansion of the pipe as a result of the internal pressure is possible. With the described model a stress and a strain controlled simulation were performed. In the stress controlled simulation all loads were treated like primary loads and in the strain controlled simulation all loads with the exception of the internal pressure were treated like secondary loads. In figure 3 the stress and creep strain distribution in the weld region is shown in detail. The top of the contour plot represents the inner side of the pipe wall, the bottom of the picture shows the outer side of the pipe wall. In the upper row of the figure the equivalent Mises stress after loading and after 200.000 h of creep is depicted, in the lower row the equivalent creep strain after 200.000 h of creep is shown. For the 200.000 h state a distinction has been made between the stress controlled and the strain controlled case. After the first loading step the equivalent Mises stress on the inner side of the pipe is slightly higher than on the outer side. Since both weld metal and heat affected zones have been modeled using identical elastic material parameters, they cannot yet be distinguished in the contour plot in this state. After 200.000 h the influence of the weld on the stress distribution becomes apparent in both stress and strain controlled case, the maximum equivalent stresses are encountered in HAZ 1, the minimal equivalent stresses can be found in HAZ 3. Again in both cases (stress and strain controlled) the highest creep strains appear in HAZ3 on the inner side of wall.
Figure 3: Contourplots of the equivalent stress and creep strain in the weld for the stress and strain controlled loadcase
Figure 4 shows evaluations of the equivalent Mises stress and creep strain along a path from inner to outer pipe surface in HAZ3 for stress and strain controlled loading case. Stress controlled :
Strain controlled:
Figure 4: Path evaluation for equivalent Mises stress and creep strain from inner to outer surface along the HAZ 3
The finite-element-simulations have been evaluated with respect to time and the equivalent stresses and creep strains are given in figure 5.
Figure 5: Temporal courses of the equivalent Mises stresses and creep strains evaluated at the respective maximum for the stress and strain controlled simulation
The development over time clearly shows that the stress controlled case yields significantly higher stresses and strains than the strain controlled case. In addition, the Mises equivalent stress obtained by analytical formula valid for linear elastic thickwalled pipes is plotted in the diagram. The path evaluation of the HAZ3 shows a maximum accumulated creep strain close to inner surface. In order to assess the results of the FE-simulations, values taken from different locations (base metal: BM, heat affected zone: HAZ1 and HAZ3) at several time points are compared to the 100,000-creep strength of P92 base metal at 618 °C. Normalized
0h
5.000 h
200.000 h
stress
HAZ1
HAZ3
BM
HAZ1
HAZ3
BM
HAZ1
HAZ3
BM
Mises
0.93
0.93
0.93
0.70
0.47
0.58
0.70
0.45
0.58
Axial
0.23
0.23
0.23
0.30
0.29
0.30
0.30
0.31
0.31
Table 2: Comparison of the finite element results with 100,000 h-creep strength acc. to VdTÜV und ECCC
In the regulations AD2000 and TRD the bending moment is not considered in the effective stress calculation. The comparison of the evaluation of local values of effective and axial (perpendicular to the weld) stress shows that after short relaxation periods the stresses in the critical zones especially in the intercritical heat affected zone are small. It is also visible that the axial stress which is the critical one since it is perpendicular to the weld is significantly smaller than the effective stress. This is in coincidence with the estimated strains (see Fig. 4).
Dissimilar welded joints
3.2
The 700°C power plants currently under development will utilize Ni-base alloys such as Alloy 617. Due to technical and economic reasons CrMoV-steels will be used for components or parts of components which are subjected to temperatures < 650 °C. Therefore dissimilar joining of Ni-base alloys and Cr-steels is necessary in these plants. The following investigation focuses on welds between Alloy 617 and a 2Cr-steel. Figure 6 shows the creep-rupture strength at 550 °C. In creep tests two competing failure mechanisms are observed:
Failure in the fusion line (FL) between ferritic/martensitic steel and nickel based weld metal. The crack usually occurs on the outer surface. Failure in the heat affected zone (HAZ) of ferritic/martensitic steel. The crack usually occurs in the specimen center.
Figure 6: Creep-rupture strength of 2Cr/A617-welds at 550°C [5]
Several research projects, [6]-[10], have identified the intercritical heat affected zone (HAZ3) as the “weakest link” in welded components under service conditions. The heat input during the welding procedure affects the microstructure of this zone [6]. Inelastic analysis is able to obtain detailed information on the deformation and failure behavior of welds. Figure 6 shows contour plots of an axisymmetric finite-element analysis at different time points. The increase in stress in the Ni-base weld metal near the fusion line and (due to the lower creep resistance) the concentration of creep strain in the intercritical HAZ3 can be seen. In Figure 7 the development of stress and creep strain over time is shown for two different loads. The time to rupture in the experiment is marked respectively. In case of the lower load (92 MPa) the creep strain rate in the HAZ is so small that the axial stress in the weld metal becomes predominant and as a consequence crack growth completely takes place in the fusion line. The higher load (125 MPa) leads to higher stresses in the weld metal and higher creep rates in the heat affected zone. Thus, the two competing damage mechanisms in weld
metal and HAZ become nearly equally important. On the one hand the higher creep rate promotes failure in the HAZ. On the other hand the maximum principal stress in the fusion line increases, even compared to the equivalent stress, which still leads to cracks in the fusion line.
Figure 7: Regions of highest stress and strain in a creep-rupture specimen (FEM-simulation)
250
200
Mises stress / MPa
Mises stress / MPa
200
150 Mises (below surface) Mises (center) axial stress (below surface) axial stress (center)
100
50
150
100
Mises (below surface) Mises (center) axial stress (below surface) axial stress (center)
50
SV1Z19, 550°C, 125 MPa
SV1Z20, 550°C, 92 MPa 0 0
2000
4000
0
6000
0
time / h
4000
6000
8000
10000
12000
2,0
equivalent creep strain / %
equivalent creep strain / %
2000
time / h
2,0
1,5
1,0
0,5
HAZ3: intercritical zone
1,5
1,0
0,5
HAZ3: intercritical zone
SV1Z19, 550°C, 125 MPa
SV1Z20, 550°C, 92 MPa
0,0
0,0 0
2000 time / h
4000
6000
0
2000
4000
6000
8000
10000
12000
time / h
Figure 8: Development of stresses in fusion line and strain in HAZ over time of 2Cr/A617-welds at 550°C (FEM-simulation)
The different stress distributions in weld metal and HAZ are illustrated in figure 8. The highest values of maximum principal stress in the weld metal are found at the outer surface, while the mostly stressed region in the HAZ is near the center, which corresponds to the crack positions found in the specimens.
weld metal_Mises weld metal_SMax HAZ3_Mises HAZ3_SMax
SV1Z20, 550°C, 92 MPa
stress / MPa
stress / MPa
300 280 260 240 220 200 180 160 140 120 100 80 60 40 20
0
2
300 280 260 240 220 200 180 160 140 120 100 80 60 40 20
SV1Z19, 550°C, 125 MPa
4
radius [mm]
0
2
weld metal_Mises weld metal_SMax HAZ3_Mises HAZ3_SMax
4
radius [mm]
Figure 9: Stress distributions in weld metal and HAZ at time to rupture (FEM-simulation)
To summarise, the following findings have been presented: • • •
Creep tests of dissimilar welds have shown significant creep strength reduction Two competing damage mechanisms have been identified Numerical simulation of crossweld specimens gave evidence to the development of different damage mechanisms by correlation to the development of stresses and strains
Since the numerical calculations can explain the damage occurring in specimens similar calculations for components can be used to make predictions on the failure behavior of components. LITERATURE 1. 2. 3. 4. 5.
6. 7.
8.
9. 10.
Metallic industrial piping – Part 3: Design and calculation; German version EN 13480-3:2012, Beuth Verlag, Berlin FDBR-Taschenbuch Rohrleitungstechnik 1: Planung und Berechnung, Günter Wossog, Vulkan, 2005 AD2000 Regelwerk Taschenbuch – Ausgabe 2009, Verband der TÜV e.V., Berlin TRD - Technische Regeln für Dampfkessel mit den Vereinbarungen der Verbände: Taschenbuch-Ausgabe 1998/2, Heymann T. Klein, C. Feuillette and A. Klenk: „Verhalten von optimierten Werkstoffen und Werkstoffverbunden für 700-720°C-Dampfturbinenbauteile“, Teilvorhaben MPA Stuttgart zu Verbundvorhaben COORETEC DT3, 2010 K. Maile et al.: „Zeitstandverhalten von P91-Schweißverbindungen“, 19. Vortragsveranstaltung AGW/AGHT, Düsseldorf, 1996 R. U. Husemann, et al.: „Langzeiteigenschaften von Schweißverbindungen moderner Stähle für Dampferzeuger“, 23. Vortragsveranstaltung AGW/AGHT, Düsseldorf, 2000 K. Maile et al.: “Behavior of similar welds in T24 Tubes and E911 pipes”, 3rd HIDA and Integrity Conference, Integrity of High Temperature Repair Welds, OeirasLisabon, Portugal, 16-18 September, 2002, pp. 69/80. M. Bauer et al.: “Evaluation of cross weld creep strength of 9-11% Cr steels”, Proceedings Int. Conf. on Plant Life Extension, Cambridge, 16-18 April 2004 M. Bauer, A. Klenk, K. Maile and E. Roos: “Numerical Investigations On Optimisation Of Weld Creep Performance In Martensitic Steel”, 8 th Seminar of Mathematical Modelling of Weld Phenomena, Seggau, 2008
