An Improved Plasticity-Based Distortion Analysis Method for Large Welded Structures

Predicting welding-induced distortion in large and complex structures is a computationally intensive task.  Welding simulation on welded structures such as mining truck and wheel loader frames can take anywhere between 1 and 3 days of wall-clock computing time, depending on the number of welds modeled, to predict distortion induced by a single welding sequence. This makes it impractical to perform welding sequence optimization, which typically requires simulations of several hundred test welding sequences, using full thermo-elastic-plastic analyses on these structures. 

A plasticity-based distortion prediction method was developed to address the computationally intensive nature of welding simulations. Plastic strains, which are typically first computed using either two-dimensional (2D) or three-dimensional (3D) thermo-elastic-plastic analysis on finite element models of simple weld geometry, are mapped to the full structure finite element model to predict distortion by conducting a linear elastic analysis.  This method was improved in this work to consider the effect of weld interactions on plastic strains for optimizing welding sequences.

This improved method was tested on two large-scale welded structures – a light fabrication and a heavy fabrication, by comparing against full-blown distortion predictions using thermo-elastic-plastic analyses. 3D solid and shell models were used for the heavy and light fabrications respectively to compute plastic strains due to each weld. Quantitative comparisons between this method and thermo-elastic-plastic analysis indicate that this method can predict distortions fairly accurately – even for different welding sequences, and is roughly 1-2 orders of magnitude faster. We conclude from these findings that, with further technical development, this method can be an ideal solver for optimizing welding sequences.