Abstract Scope |
Intro
Formation of martensite and high-hardness zones during field and repair welding is common but usually undesired. Temper bead welding is a technique that deposits weld beads over previous weld layers to temper the martensite in the heat affected zone (HAZ), when post weld heat treatment (PWHT) is inapplicable. However, the efficiency of temper bead welding is affected by many factors (number of layers, number of beads in each layer, heat input, etc.). This project aims to develop a methodology for accurate prediction of the outcome of temper bead welding.
Technical Approach
A computational model is developed to estimate the effect of temper bead welding. The model considers following welding parameters as input: substrate material, welding filler metal, shape of substrate, number of overlay layers, number of beads at each layer, heat input for each layer, cooling time. The above parameters are incorporated in a FEA model that simulates the temper bead welding using the SysWeld software. The simulation generates a set of thermal histories for all nodes in the HAZ. The thermal histories are analyzed to predict the microstructure and hardness value. With this computational model, a design of experiment (DoE) software module is developed to perform parallel simulation of welding cases and enable comparison of different welding parameters. A current trial version of the DoE module utilizes the heat input for each layer and layer thickness as main variables, and has been used to optimize these variables at different welding scenarios. While the DoE module relies on specific software environment on host machine, a website is developed to provide rapid access to the module from any devices with internet connection.
Result/Discussion
A DoE with 27 cases is run using the model. The main variables in these 27 cases are the heat input, the power ratio, and the layer thickness. The goal of this DoE is to find an optimal combination of heat input, power ratio, and layer thickness to reduce the hardness and the content of fresh martensite in the HAZ. The DoE module generates a single case report for each case which includes predictions of hardness, microstructure, number of effective tempering reheats, Grange-Baughman parameter, etc. A summary report is also generated that compares the results of all DoE cases. The summary report of the DoE performed in this study indicated that lower heat input tends to produce lower hardness and less fresh martensite in HAZ across all DoE cases.
Conclusion
A design of experiment module for computational optimization of temper bead welding, which is based on finite element analysis, has been developed and demonstrated. This approach is applicable for prediction of the hardness and microstructure in the heat affected zone of weld overlays and for comparative evaluation of the tempering efficiency and optimization of welding procedures. |