proceeding of fifteenth abaqus user group2000p16_W

1、R5 CREEP LIFE ASSESSMENT OF A WELDEDTRUNNION COMPONENTMr R D Patel, MEng, CEng, MIMechEBritish Energy Generation Ltd, Barnett Way, Barnwood, Gloucester, GL4 3RS, U.K.ABSTRACTThree-dimensional ABAQUS finite element (FE) analyses are described as part of a creep life assessment of an ex-service welded

2、 1/2CMV trunnion component. One quarter of the component was modelled, with appropriate boundary conditions applied on planes of symmetry. Elastic and elastic-perfectly plastic ABAQUS FE analyses were carried out. In particular, the RIKS algorithm within ABAQUS was used to determine the limit load a

3、nd hence the reference stress of the component. Rupture data for the constituent materials of the weldment were then used to estimate the time to rupture of the vessel, and this predicted time to failure compared to the actual vessel test time. The effects of varying the means of applying the system

4、 loading to the trunnion arms were investigated in order to evaluate a moment limit load. Using mean materials data, the total creep damage at the end of the test, including prior service loading, was estimated using the British Energy R5 high temperature assessment procedure to be less than 0.5. Th

5、is estimate was less than 0.8 using lower bound data which suggests that total failure of the trunnion would be predicted not to have occurred. Post-test examination showed that there were no signs of the vessel being close to failure, consistent with the analysis predictions.1 INTRODUCTIONThe R5 pr

6、ocedure 1 is routinely used within British Energy for life assessments of high temperature plant and has been extensively validated using full-scale vessel tests. This paper describes the assessment of an ex-service welded ferritic 1/2CMV steel component tested under steady loading. As is normal in

7、1/2CMV components, the weld metal used is 21/4Cr1Mo. The assessment uses R5 in conjunction with ABAQUS 2 FE analyses to determine the reference stress and hence determine a predicted time to failure of the test vessel. Similar assessments of other welded ferritic pressure vessel tests have been repo

8、rted by Budden 3. Full details of the R5 assessment of the trunnion vessel are given in Patel et al. 4. This paper concentrates on the FE analysis.The component assessed is a welded trunnion removed from service following detection of cracking in a weld on one arm of the trunnion. Examination of the

9、 second trunnion arm revealed the presence of extensive creep cavitation at the main pipe side of the weld. The main pipe section containing the two trunnion arms was therefore removed from plant and replaced with a new section of pipe. The removed pipe section was fabricated into a test vessel, but

10、 including only the cavitated trunnion arm (Fig. 1), and was tested under near-service loading at accelerated temperature (Table 1).2001 ABAQUS UK Users Group Conference9Service ConditionsTest ConditionsTemperature527C585CTrunnion Hanger Load46.03kN50.20kNInternal Pipe Pressure3.60MPa3.50MPaService

11、and Test Times91,658 hours15,431 hours1. Service and test conditions of 1/2CMV trunnion vessel.Main pipeTrunnion armThe main pipe was blanked off following removal of the crackedarm Ram1.The trunnion vessel mounted in the test facility.(The ram loading the remaining trunnion arm can be clearly seen)

12、2 REFERENCE STRESS APPROACH TO RUPTURE LIFE CALCULATIONThe limit load reference stress, sref, as defined in R5, for a homogeneous component is given byo= PW y(1)PrefLwhere PL is the plastic limit load corresponding to yield stress sy and PW is the working load. The ratio sy/PL is independent of yiel

13、d stress for a homogeneous component. For combined loads, the limit load is evaluated assuming proportional loading. Then the rupture reference stress for ductile materials is calculated in R5 fromsRref= 1+ 0.13c -1sref(2)where sref follows from equation (1) and c , the stress concentration factor,

14、is given byc = s E,maxs ref(3)In equation (3), sE,max is the maximum elastically-calculated value of equivalent stress for the same set of loadings as are used to obtain sref, so that c is independent of load level. Equation (2) applies to rupture of initially uncracked bodies and is limited to mode

15、rate stress concentrationswith c4. Modifications to equation (1) to account for the effect of welds are discussed in Sections3 and 6. The effects of crack growth are not addressed within this paper.3 RUPTURE DATArefCreep rupture data are used in conjunction with the R5 procedure to estimate the time

16、 to failure for various metallurgical zones of the weldment. For the purpose of assessment, 1/2CMV weldments made using 21/4Cr1Mo filler metal are characterised as consisting of four distinct metallurgical regions: (i) parent 1/2CMV steel; (ii) 21/4Cr1Mo weld metal; (iii) intercritical (Type IV) hea

17、t affected zone material; and (iv) coarse grained high temperature heat affected zone (HAZ) material. The high-temperature HAZ is characterised as fully coarse grained for conservatism. For each region, the predicted failure life is obtained based on a calculated sRfor that zone and thecorresponding

18、 rupture data. The overall failure time of the weldment is then the minimum calculated value for the various weld zones. Full details of the rupture equations employed, for each of the metallurgical regions, can be found in Patel et al. 4.4 FINITE ELEMENT MODELTable 1 lists the service and test cond

19、itions of the 1/2CMV trunnion vessel, and Figure 2 shows the vessel dimensions used in the FE analyses. The mean radius to thickness ratio for the main pipe of the trunnion vessel is Rm/T=9.4. The end restraints on the main pipe section in the test configuration (Figure 1) are assumed to impose symm

20、etry about the plane in the pipe normal to the axis of the trunnion arm. Therefore, only one quarter of the vessel is analysed because of symmetry, with the appropriate boundary conditions applied on planes of symmetry, see Figure 3. The FE mesh consists of 1365 three-dimensional ABAQUS 2 C3D20 or C

21、3D20R elements and 7385 nodes. ABAQUS C3D20R elements are used for the elastic-plastic analyses and C3D20 elements for the elastic analysis.PipeL = 1500mmTrunnionH = 500mmRi = 231.5mm Ro = 257.5mmArmri= 148.5mm ro = 165mm1. Trunnion vessel dimensions used in analyses.SurfaceTest loading conditions1

22、Pipe internal pressure loading = 3.5MPa2 Pipe pressure end loading = -14.7MPa3 Trunnion loading = -4.842MPa4 No rigid body motionUx = 05 Symmetry Uy = 06 Symmetry Uz = 0 7, 8, 9 & 10 UnloadedThe trunnion load is applied at 2 elements as shown.2. Trunnion vessel FE mesh and test loading conditions.Th

23、e level of mesh refinement was important, but there was not a requirement for an extremely refined mesh since the reference stress being evaluated was a global parameter. However, close to the intersection of the main pipe and the branch, the level of mesh refinement was greater than in the rest of

24、the vessel in order to capture any stress concentrations that were expected to occur in this region.The elastic material properties used in the FE analyses are obtained from BS806 5 and are listed in Table 2. For the elastic-perfectly plastic analyses, the material is arbitrarily assumed to have ayi

25、eld stress sy=100MPa. The pressure internal to the main pipe was applied as a pressure load to the element faces, and the system loading as pressure load to two element faces of the trunnion arm to avoid excessive local deformation in the mesh. The trunnion arm itself does not contain any pressure l

26、oading.Material PropertiesYoungs modulus 527C172,300MPaYoungs modulus 585C166,500MPaPoissons ratio0.292. Material properties of 1/2CMV trunnion vessel.5 FINITE ELEMENT ANALYSES5.1 Calculation of maximum equivalent stressAn elastic FE small displacement analysis of the trunnion, for both test and ser

27、vice conditions, was performed to calculate the maximum elastic value of equivalent stress in equation (3). This was then used in the calculation of the rupture reference stress from equation (2). The FE analysis under test conditions took 2.11 minutes (total CPU time) to run on a Silicon Graphics O

28、rigin 2000 computer.5.2 Calculation of limit loadThe limit load was calculated by using the RIKS algorithm within ABAQUS 2 assuming small displacement. Here elastic-perfectly plastic FE limit load analyses, for both test and service conditions, were performed where the internal pressure and system l

29、oads were increased in proportion. The RIKS algorithm within ABAQUS is a method of calculating the limit load and is generally used to predict such unstable geometrically non-linear collapse of a structure. In order to run this algorithm one has to use the *STATIC, RIKS command in the ABAQUS input d

30、eck. In particular, the load proportional factor (LPF) value at the end of the step, obtained from the .sta file, is used in equation (6), shown below, to determine the limit load, PL.PL = LPF PW(6)Using equation (6) to determine PL, one can then use equation (1) to calculate sref. Hence the rupture

31、 reference stress can be obtained from equation (2). The RIKS FE analysis took approximately 43 hours (total CPU time) to run for the test condition case on a Silicon Graphics Origin 2000 computer.5.3 Moment limit load validation of FE modelHere, a comparison of the moment limit loads obtained from

32、the FE analyses and from the Miller solution for a cylinder under bending 6 is made. The FE moment limit load is calculated byapplying only the trunnion end load (system loading). Three analyses were performed, each using the same mesh but with different levels of elastic reinforcement to support th

33、e system loading. One can then see how the moment limit loads vary according to how the system load is applied to the trunnion arm. Each analysis is described below:(i) Analysis (a):FE mesh with no elastic reinforcement(ii) Analysis (b):FE mesh with some elastic reinforcement(iii) Analysis (c):FE me

34、sh with ring elastic reinforcementtrunnion loading alone was applied to the vessel. One can see that local deformation no longerAnalysis (a) is simply the FE mesh with no elastic reinforcement. Figure 4 shows the local deformation which occurred when the trunnion loading alone was applied. One can s

35、ee that without any elastic reinforcement there is a large amount of local deformation. Analysis (b) is where some elastic reinforcement was applied near the point of application of the trunnion load. This was done to try to prevent local deformation from occurring when the trunnion loading alone wa

36、s applied to the vessel. However, Figure 5 shows that local deformation did still occur, in this analysis, when the trunnion loading alone was applied. Analysis (c) is where the trunnion is reinforced by a ring of elastic elements, see Figure 6. Figure 7 shows the deformation when theoccurs. This is

37、 considered to be a more realistic model of the actual test loading.3. Analysis (a): no elastic reinforcement. Application of trunnion load alone. Displaced shape at final step. Note displacement factor is 1.4. Analysis (b): some elastic reinforcement. Application of trunnion load alone. Displaced s

38、hape at final step. Note displacement factor is 2.RING OF ELASTIC ELEMENTS5. Analysis (c): ring elastic reinforcement. FE mesh showing trunnion reinforced by a ring of elastic elements.6. Analysis (c): ring elastic reinforcement. Application of trunnion load alone. Displaced shape at final step. Not

39、e displacement factor is 5.Table 3 shows a comparison of the moment limit loads. One can see that when no elastic reinforcement is used the FE moment limit load is approximately 66% less than that obtained from Miller. Collapse is dominated by the local deformation in this case. With the use of some

40、 elastic reinforcement on the trunnion, this difference is reduced to approximately 32%. When elastic ring elements are used there is good agreement between the Miller and FE moment limit loads, with the FE moment limit load being within 6% of the Miller solution. Collapse is then occurring local to

41、 the intersection.Miller (MNm)FE (MNm)Difference (%)Moment Limit Load (with no elastic reinforcement)0.1620.05665.69Moment Limit Load (with some elastic reinforcement)0.1620.11131.64Moment Limit Load (with elastic ring reinforcement)0.1620.1535.5023. Comparison of Miller and FE moment limit loads.20

42、01 ABAQUS UK Users Group Conference136 RESULTS AND DISCUSSION6.1 Stress analysisUnder the test condition, a maximum von Mises stress of 69.2MPa, at the internal surface of the pipe, was obtained from the elastic analysis. The reference stress results for the test conditions are given in Table 4, fol

43、lowing Patel et al. 4. The working load, PW, plastic collapse limit load, PL, reference stress, sref, equivalent elastic stress, sE,max , stress concentration factor, c, and rupturereference stress,Rsref, are tabulated. Note that, as a sensitivity study on the effect of meshrefinement on the compute

44、d peak elastic stresses, three maximum equivalent stresses are considered in Table 4, namely 75MPa, 85MPa and 95MPa. This covers uncertainty in the details of the weld profile, noting that the elastic stresses are, strictly, unbounded at sharp corners such as at the intersection of the trunnion arm

45、and main pipe. For the service condition, a maximum von Mises stress of 70.4MPa was obtained from the elastic analysis, again at the internal surface of the pipe. However, as in the test case condition, elastic equivalent stresses of 75MPa, 85MPa and 95MPa were used as a sensitivity analysis. The re

46、ference stress results for service conditions are also given in Table 4. Figure 8 shows the displaced shape at the end of the RIKS FE analysis under test loading conditions. From this figure it can be seen that there was no excessive local deformation on the trunnion arm where the trunnion loading w

47、as applied despite the lack of a full ring of elastic reinforcement. The reason for this was because the pressure and trunnion loading were applied proportionally in the RIKS analysis. Therefore, this analysis was considered to provide a realistic value of the limit load. Note that in Section 5.3 ex

48、cessive local deformation could occur when the trunnion load alone was applied.PW (MPa)PL (MPa)sref(MPa)o E,max(MPa)cRref(MPa)TrunnionTest conditionsCASE i3.5011.93529.33752.5635.3CASE ii3.5011.93529.33852.9036.6CASE iii3.5011.93529.33953.2437.9Service conditionsCASE i3.6012.129.85752.5135.7CASE ii3

49、.6012.129.85852.8537.0CASE iii3.6012.129.85953.1838.34. Rupture reference stress results 4.7. Test loading conditions. RIKS analysis: Increase pressure and trunnion loading proportionally. Displaced shape at final step. Note displacement factor is 3.refResults for the creep damage due to the test an

50、d service conditions are given in Table 5 for the limiting Type IV zone, following Patel et al. 4. The lives quoted in Table 5, corresponding to the Type IV material zone, give the lowest rupture time across the weldment, and thus limit the predicted life of the vessel. Damage is calculated, using R

51、obinsons rule, by linearly summing the ratios t/tf( sR ) from the service and test conditions. Here, t is the corresponding time at load andtf is the creep rupture time at the associated calculated value of rupture reference stress,Ro.refFollowing R5, the homogeneous rupture reference stress is mult

52、iplied by a factor, k, which quantifies stress redistribution within a weldment. For stress states where the maximum principal stress is essentially parallel to the fusion line and overall stress redistribution occurs due to the differing creep deformation response of the weldment zones, k may diffe

53、r from unity and varies between zones 1. This is judged from the FE analyses to be the case for the trunnion vessel, where the stresses near the weld are controlled by the pressure load. The weld stress redistribution factor, following the advice in R5 1, is taken as k = 0.7 for the weld and k = 1.4

54、 for the coarse- grained HAZ. For Type IV material, k=1 is assumed for conservatism. For all of the cases examined, the total damage does not exceed 0.8, so that the trunnion is predicted to have not failed at the end of the test.Test time (h)Estimated test failure time (h)Test damageService time (h

55、)Estimated service failure time (h)Service damageTotal damagemean dataCASE i15,431623630.247916585000000.1830.431CASE ii15,431582140.265916585000000.1830.448CASE iii15,431543610.284916585000000.1830.467lower bound dataCASE i15,431431120.358916583000000.3060.663CASE ii15,431397890.388916583000000.306

56、0.693CASE iii15,431367420.420916583000000.3060.7265.Trunnion damage calculation results (Type IV zone) for various load cases 4.6.2 Experimental results and comparison with R5 estimatesMicro-crack initiation occurred on the outer surface of the component, at the trunnion/main pipe intersection, at about 12576h into the test and extended marginally in the circumferential direction during the subsequent 2855h of testing. There was no evidence of through-thickness crack growth from any of the monitoring techniques applied to the vessel throughout the test. After a total of 15431h on test t