Prediction Method of Strength Robustness Affected by Arc Welding Sectional Dimensions

2024-32-0038

04/18/2025

Features
Event
2024 Small Powertrains and Energy Systems Technology Conference
Authors Abstract
Content
The arc welding process is essential for motorcycle frames, which are difficult to form in one piece because of their complex shapes, because a single frame has dozens of joints. Many of the damaged parts of the frames under development are from welds. Predicting the strength of welds with high reliability is important to ensure that development proceeds without any rework. In developing frames, CAE is utilized to build up strength before prototyping. Detailed weld shapes are not applicable to FE models of frames because weld shapes vary widely depending on welding conditions. Even if CAE is performed on such an FE model and the evaluation criteria are satisfied, the model may fail in the actual vehicle, possibly due to the difference between CAE and actual weld bead geometry. Therefore, we decided to study the extent to which the stresses in the joint vary with the variation of the weld bead geometry. Morphing, a FE modeling method and design of experiment method, was utilized to derive the distribution of stress variation in the joint for each dimension of the weld bead shape (throat thickness, leg length, flank angle, etc.). The number of calculations would be enormous if every combination of weld bead dimension values were considered exhaustively. Therefore, the “Latin Hypercube Sampling (LHS)” [1] method of design of experiments can be used to reduce the number of combinations while distributing the probability of occurrence of parameters in a full distribution. Using these methods, the influence of weld quality on frame strength was clarified within a feasible computation time.
Meta TagsDetails
DOI
https://doi.org/10.4271/2024-32-0038
Pages
7
Citation
Hada, Y., and Sugita, H., "Prediction Method of Strength Robustness Affected by Arc Welding Sectional Dimensions," SAE Technical Paper 2024-32-0038, 2025, https://doi.org/10.4271/2024-32-0038.
Additional Details
Publisher
Published
Yesterday
Product Code
2024-32-0038
Content Type
Technical Paper
Language
English