Evaluating numerical simulation accuracy for full-scale high-strength steel ship structures: Insights from the ISSC 2025 Ultimate Strength Committee benchmark on transversely stiffened panels

Journal article, 2026

The demand for sustainable ship design has driven the use of high-strength steel to reduce structural weight, although this introduces buckling challenges due to unchanged elastic properties. Supported by the ISSC 2025 Ultimate Strength Committee, this study evaluated the ability of numerical simulations to predict the nonlinear response and ultimate strength of stiffened panels subjected to transverse compression. The benchmark consisted of full-scale blinded experimental tests that were conducted in parallel using a deck-like structure with thin plating prone to elastic buckling. The finite element models produced by participating researchers were compared, focusing on the complete end-shortening curve rather than just ultimate strength. Despite identical input geometry and minimal modeling guidance, results varied widely, revealing the significant influence of user-defined assumptions. The inclusion of additional data on material properties in the second study phase led to greater result dispersion due to the different strategies adopted for the hardening model. Key variability sources included the modeling of initial imperfections, material constitutive laws, and residual stresses from welding. The study highlights the need for consistent modeling and improved experimental data collection, particularly regarding boundary conditions and residual stress effects. While including welding stresses improved stiffness predictions, uncertainty in boundary behavior limited the assessment of ultimate strength impacts. The study also evaluated compliance with classification society rules (e.g., CSR, DNV, UR-S35), offering insights into how nonlinear numerical analyses complement or challenge regulatory frameworks based on closed-form expressions. Recommendations are made for improving simulation reliability and result validation.