Journal of Welding Technology
Vol 8, No 1 (2026): June

Comparative finite element analysis of IIW linear stress extrapolation and haibach methods for structural hot-spot stress prediction in welded plate joints

Regan Rahadian Pambudi (PEM Akamigas)
Richard Alexander (Universitas Diponegoro)



Article Info

Publish Date
01 Jul 2026

Abstract

Accurate prediction of structural hot-spot stress (SHSS) is essential for evaluating welded joints. This study compares the International Institute of Welding (IIW) Linear Stress Extrapolation (LSE) and Haibach methods using finite element analysis performed with the open-source software CalculiX. Three weld geometry representations (no weld, chamfer, and fillet) and two mesh densities (fine and coarse) were investigated, resulting in six finite element models. Both SHSS methods were applied to evaluate stress prediction, mesh sensitivity, weld geometry effects, and computational performance. The IIW LSE method consistently predicted higher SHSS than the Haibach method, while mesh refinement produced variations below 4%. In contrast, weld geometry had a greater influence on SHSS prediction, with the chamfer model producing the highest stress. Fine meshes required approximately five to six times longer computational time than coarse meshes while providing only marginal improvements in SHSS prediction. Weld geometry representation was therefore found to have a greater influence on SHSS prediction than mesh density. The IIW LSE method is recommended for conservative structural assessment, whereas the Haibach method is suitable for preliminary engineering analyses requiring simpler post-processing.

Copyrights © 2026






Journal Info

Abbrev

Welding_Technology

Publisher

Subject

Industrial & Manufacturing Engineering Materials Science & Nanotechnology Mechanical Engineering

Description

The main scope of the journal is to publish original research articles in the area of Welding Technology The main focus of the journal is on experimental research. The scope of the journal ...