Processed result data for graph neural network-based surrogate modelling of finite element stress fields in stiffened cylindrical shells

This dataset contains processed result data supporting the manuscript entitled “Graph Neural Network-Based Surrogate Modelling of Finite Element Stress Fields in Stiffened Cylindrical Shells and Out-of-Distribution Generalisation”. The original data.zip contains the processed data supporting the baseline results reported in Tables 1–8 and Figures 4–13, including aggregate model-performance metrics, sample-wise error metrics, node-level prediction data for representative stress-field visualisation, parameter-range information and dataset-partition information. This new version additionally includes revision_additional_data.zip which contains the processed CSV data underlying the additional validation, sensitivity and computational analyses introduced during peer review. These data support Tables 9–15 and Figures 14–20, including peak-stress and high-stress-region error analysis, representative field and spatial-error data, mesh-sensitivity analysis, training-set-size sensitivity, training and inference computational profiling, finite-element data-generation timing, and finite-element benchmark validation. Large raw finite-element intermediate files, full training tensors, model checkpoints, source code, temporary files and software-specific cache files are not included because of file size and software-dependency constraints.