Abstract Scope |
Introduction:
The focus of this study is temper bead welding (TBW), which is a reliable method of repairing pressure vessels in the nuclear industry. These pressure vessels are subjected to PWHT during fabrication to lower as-welded hardness and increase the component’s fracture toughness. The components are then put in service, but at this point it becomes infeasible to perform repair welds without forming brittle martensite. To remedy this the temper bead welding technique is used to provide tempering to the brittle HAZ, by using the heat generated by adjacent and subsequent weld passes. The multiple reheats generated by surrounding welds in TBW raise the temperature of the HAZ between the minimum tempering temperature (MTT) and the austenite transformation start temperature (Ac1) to provide a tempering effect, similar to conventional PWHT.
Technical Approach:
The overarching goal of this study is to validate a TBW model-based design of experiment (DoE) computational module, which can be used to perform efficient weld simulations for optimization of TBW procedures. Experimental and modeling trials were performed to validate previously developed tempering response predictive equations for the HAZ of Grade 22 steel and the accuracy of SysWeld predicted HAZ thermal histories. The later are used in combination of the tempering response equations to predict the HAZ hardness in temperbead welding.
Three two-layer weld overlays of Alloy 625 on Grade 22 steel plates were produced using GTAW cold wire procedures, while varying the heat input. Thermocouples were used to record the thermal history of each weld pass, and the thermal histories were analyzed to extract the peak temperature and Grange Baughman Parameter (GBP) values. The peak temperatures and GBP values were then input into the predictive equations to determine predicted hardness values for each thermocouple location. These thermocouple locations were sectioned and extracted from the weld overlay, then hardness tested at the thermocouple tip area. Hardness values from these tests indicate the accuracy of the HAZ predictive hardness equations.
In parallel, the experimental weld overlays were modeled in the SysWeld FEA software, to replicate the experimental welding. These models were developed using the actual bead geometry, obtained from profilometer data which was recorded during welding. After the models were made, the welding process was simulated. This involved calibrating the models through adjusting the welding heat source geometry, heat source location, and arc efficiency. The calibration procedure started by validating a single bead, followed by two-beads, after which the full model was ran with all weld beads included.
Results/Discussion
Hardness maps at the thermocouple locations indicated that the predicted hardness equations are accurate. The hardness values around the thermocouple tip were averaged to determine a single hardness value for each thermocouple. A total of 17 thermocouple locations were extracted from the three sets of overlays to be hardness tested. It was found that 14 locations had hardness values that agreed well with the predicted hardness. Only 3 samples had deviations from the predicted hardness greater than 30 HV.
The FEA model validation involved extracting thermal histories from the models at the same HAZ locations as where the thermocouples were located in the experimental welds. The thermocouple locations, in terms of distance from the fusion boundary and distance from the weld start point, were used to select the appropriate nodes for comparison of thermal histories. Generally, the SysWeld predicted thermal histories matched well the thermocouple recorded thermal histories after the calibration process, and minute adjustments to the LOAD definition were made to further improve the model accuracy.
Conclusion
There are two types of validation being discussed in this study. The first is the validation of the predictive HAZ hardness equations which were developed earlier in the project. The second validation is the FEA models predictive accurately of experimental HAZ thermal histories. Both validation trials will be used for calibration of the TBW model-based DoE module, which is needed for its successful application in optimization of TBW procedures. |