PSI - Issue 81

Dhanies Wahyu Ardyrizky et al. / Procedia Structural Integrity 81 (2026) 458–464

463

Based on these results, it is confirmed that there is a strong correlation between the material strength and its mechanical response to temperature variation. This phenomenon involves the degradation of the elastic modulus above 200°C, the accumulation of irreversible strains, stress relaxation due to viscoplastic effects, and the formation of residual stresses during cooling. Such responses are crucial for understanding the structural integrity of ships following fire accidents. Understanding this hysteresis phenomenon can provide valuable insight into predicting reductions in structural stiffness and the onset of material failure under thermal loading. 4.2. Mesh convergence study As the mesh size becomes finer, as indicated by the increasing number of elements, the resulting axial force tends to approach a specific, stable value. The relationship between these parameters is shown in Fig. 8. The corresponding numerical results obtained from the mesh-convergence study are summarized in Table 2. This pattern indicates that the numerical results have reached convergence, meaning that the computational outcome is no longer significantly dependent on mesh density within a specific range of element sizes. The convergence line in Fig. 8 represents the constant limiting value of axial force obtained from a series of mesh variations. At mesh sizes of 50 mm or larger, fluctuations in the results persist. This occurs due to the low resolution in capturing the stress and temperature gradients within the structure. However, for mesh sizes of 45 mm or smaller, the results begin to stabilize and approach the convergence line. With approximately 40000 elements at a mesh size of 45 mm, an adequate balance between computational accuracy and processing time is achieved. The difference in maximum axial force obtained is only about 1% from the convergence line. Additionally, this mesh configuration offers a significant computational advantage, approximately seven times faster than the smallest mesh observed. The mesh-convergence study included all simulations run with the same solver settings and hardware, ensuring a fair and consistent comparison of computational cost.

Table 2 Mesh convergence results and representative computational runtimes Mesh size (mm) Number of elements Max axial force (kN)

Runtime (s)

30 35 40 45 50 55 60 65 70 75

88060 61056 45094 39729 30858 24846 21297 18360 16195 14652

2332.08 2370.15 2388.58 2401.97 2448.61 2411.64 2467.05 2494.33

25515

9892 5133 3619 2756 1985 1287

896 614 420

2546

2514.13

Fig. 8. Mesh convergence-based axial force.