StressDiffVis: Visual Analytics for Multi-Model Stress Comparison

Structural analysis is essential in modern industrial design, where engineers iteratively refine geometry models based on stress simulations to achieve optimized designs. However, comparing stress distributions across multiple model variants remains challenging due to the complexity of stress fields, which are high-dimensional, unevenly distributed, and dependent on intricate geometric structures. Existing tools primarily support single-model analysis and lack dedicated functionalities for multi-model comparison. As a result, engineers must rely on manual, cognitively demanding visual inspections, making it difficult to systematically identify and interpret stress variations across design iterations. To address these limitations, we propose StressDiffVis, a visual analytics approach that facilitates stress field comparison across multiple structural models. StressDiffVis employs a volumetric representation to encode stress distributions while minimizing occlusion, enabling voxel-wise difference analysis for model comparison. To support localized analysis, we introduce model segmentation, grouping voxels with similar stress patterns across models. StressDiffVis integrates these techniques into an interactive interface with a tree view, organizing models by the iterative design process, and a comparison view, using a matrix layout for detailed comparisons. We demonstrate the effectiveness of StressDiffVis through two case studies illustrating its utility in comparative stress analysis. In addition, expert interviews confirm its potential to enhance engineering workflows.