Comparison of computed and measured residual stresses in a mock
Transcription
Comparison of computed and measured residual stresses in a mock
VTT TECHNICAL RESEARCH CENTRE OF FINLAND LTD Fatigue affected by residual stresses, environment and thermal fluctuations (FRESH) Heikki Keinänen Safir 2014 final seminar, March 2015 FRESH 2014 Research topics Task 1.0 Thermal fatigue simulation, subtasks 1 & 2 - computational models and material parameters for thermal fat. simulation Task 2.0 Weld residual stresses, subtasks 1, 2 & 3 - more realistic estimates for welding residual stresses and their magnitude under operational conditions to be utilised in structural integrity assessments Task 3.0 Categorization of stresses - to assess the level of conservatism in the existing plastic strain correction factor definitions Task 4.0 Fatigue caused by turbulent mixing - computational methods to assess fatigue caused by turbulent mixing 22/04/2015 2 Trueflaw thermal fatigue data - use in FRESH project (task 1 – thermal fatigue simulation) Trueflaw manufactures cracks for NDE training and qualification using in-situ thermal fatigue loading Each produced crack can be viewed as a thermal fatigue test Thousands of cracks produced, mostly nuclear cases In FRESH, parametric models were built to simulate these tests using elastic-plastic FE analyses Aim of the work was to determine the strain amplitude data for each test and combine the numerical and experimental results into a strain-life fatigue curve Known temperature load and crack nucleation data (input) 22/04/2015 Simulated stress/strain distribution (output) Measured failure cycles and simulated stress/strain distribution and amplitudes (result) 3 Simulation models Calculation results for each case Example loading 22/04/2015 Calculated strain cycles for all cases 4 Result data as a strain-life curve 22/04/2015 5 Comparison of computed and measured residual stresses in a mock-up pipe (task 2 – weld residual stresses) General details of the computational analyses Summary of the analysis performed for the mock-up pipe and comparison of the computational results to the measured ones 22/04/2015 6 Analysis procedure • Abaqus general purpose FE code was utilised in the computation • Sequential thermal and mechanical analyses, small strains and displacements assumed • Temperature dependent material properties • Mixed hardening material model of Abaqus including both isotropic and kinematic hardening • Anneal temperature of 1100-1400 °C 22/04/2015 7 Analysis procedure Welding speed, voltage, current 22/04/2015 Properties for isotropic/kinematic hardening model 8 Comparison of computed and measured residual stresses in a mock-up pipe - TVO has manufactured mock-up plates and pipes containing multiple butt-welds. - In the DEMAPP MACY - project residual stresses have been measured with multiple methods. - Computational simulation of welding of the pipe was performed 2 years ago, measurement results are available now: [1] Miikka Aalto, Residual stress relaxation due to thermal loads in boiling water reactor nuclear power plant pipe welds, Licentiate’s thesis, 15.12.2014. 22/04/2015 9 - 22/04/2015 Pipe material: SA 376 TP 304 Weld material: AISI 318L Pipe outer diameter 323.85 mm Wall thickness was 17.45 mm Length of mock-up 400 mm 10 Run 1 2 3 4 5 6 7 Voltage Current Torch speed Interpass temperature [V] [A] [mm/s] [°C] 13.5 13.5 22 22 22 22 22 84 90 77 100 100 100 100 0.3289 0.3929 1.0221 1.2386 1.1923 1.4570 1.6201 25 56 43 21 70 48 88 - ¼ of the circumference - Abaqus FE code - Pass by pass modelling - Temperature dependent mat. props. 22/04/2015 11 The properties of Eshete 1250, which is a fully austenitic chromium-nickel steel having a rather similar chemical composition as AISI 304 steel, were utilised. The reason for using the Eshete 1250 properties is the availability of material properties for the isotropic/kinematic hardening material model of Abaqus. For parent metal the utilised stress-strain properties are considerable higher than presented in [1]. 22/04/2015 12 Computed and measured [1] temperatures at weld root during welding of passes 6 - 9. Time scales has been adjusted. 22/04/2015 13 22/04/2015 14 22/04/2015 15 Computed and measured [1] axial residual stresses at weld root. Measurements are done by hole drilling method. Computational results at the middle of the circumference. 22/04/2015 16 Axial residual stresses as a function of depth in the roots of three pipe butt welds before thermal cycling. Pipes W5 and W6 were exposed to pressure testing. Weld W6-7 is measured with ring-core method from the depth of 1 to 6.5 mm. All other measurements are done by hole drilling method [1]. 22/04/2015 17 Computed and measured [1] hoop residual stresses at weld root. Measurements are done by hole drilling method. Computational results at the middle of the circumference. 22/04/2015 18 Hoop residual stresses as a function of depth in the roots of three pipe butt welds before thermal cycling. Pipes W5 and W6 were exposed to pressure testing. Weld W6-7 is measured with ring-core method from the depth of 1 to 6.5 mm. All other measurements are done by hole drilling method [1]. 22/04/2015 19 t = 17.45 mm, Q/t = 78…102 MJ/m2 Computed axial and circumferential (hoop) residual stress (MPa) after welding at the middle of the weld. Three dimensional and axisymmetric results are shown. Inner surface corresponds to coordinate value of 0 mm. 22/04/2015 20 t = 17.45 mm, Q/t = 78…102 MJ/m2 Computed axial and circumferential (hoop) residual stress (MPa) after welding at the middle of the weld. Three dimensional and axisymmetric results are shown. Inner surface corresponds to coordinate value of 0 mm. 22/04/2015 21 Conclusions • The computed axial and hoop as-welded stresses are tensile at the inside surface in the weld root area • The computed hoop stresses are tensile through the wall at the middle of the weld • The computed axial stresses are tensile on the inner surface and compressive on the outer surface at the middle of the weld 22/04/2015 22 Conclusions • There is large variation in the measured stresses. Some discrepancies exist, e.g. negative hoop stresses at the weld root? • Additional measurement points or even other methods, e.g. iDHD would be useful/necessary? • The computations could be re-run with lower stress-strain curve for the parent material and maybe with a 180deg model (at least). Optimal situation would be that the real cyclic stress strain curves are measured 22/04/2015 23 Conclusions • In case of welding residual stresses there is uncertainty in computations and measurements. Both methods should be utilised to support each other to get more reliable results. 22/04/2015 24 Reference: • Computation of Welding Residual Stresses in a Multi-Pass Welded Mockup Pipe, Research Report VTT-R-08364-12 Other performed analyses: • • • • • • • Comparison of computed residual stress states by 3D and 2D models and evolution of residual stresses due to operational loading (in feedwater nozzle), Research Report VTT-R04886-14. Computation of (RPV) cladding welding stresses, Research Report VTT-R-06020-14. Simulation of weld overlay, Research Report VTT-R-00485-14. Computation of Welding Residual Stresses in a Multi-Pass Welded Mock-up Pipe. Keinanen, H. 22nd International Conference on Structural Mechanics in Reactor Technology 18-23 August, San Francisco, California, USA. Keinänen, H. Weld repair simulation for the Mock-up 2 of EU FP7 STYLE Project. Baltica IX. International Conference on Life Management and Maintenance for PowerPlants. Pertti Auerkari & Juha Veivo (eds.). VTT Technology 107. VTT. Espoo (2013), 24. Keinänen, H., Alhainen, J., Karppi, R. & Verho, M. 2009. Control and Exploitation of Thermal Distortions in Welded T-joints. 20th International Conference on Structural Mechanics in Reactor Technology (SMiRT 20), Espoo, Finland, August 9. Computation of welding residual stresses in a mock-up nozzle, Research Report VTT-R02265-11. 22/04/2015 25 TECHNOLOGY FOR BUSINESS