PSI - Issue 79

Ibrahim T. Teke et al. / Procedia Structural Integrity 79 (2026) 17–25

23

Table 4. DIN 1623 steel similar MTS specimen life prediction capability with prestress. Load (kN) Experimental Prediction Without Prestress Prediction Study 3.7 9970 8975 10,196 Ertas & Akbulut (2021) 3 39000 27716 29,349 Ertas & Akbulut (2021) 2.7 52470 50325 51,572 Ertas & Akbulut (2021) 2.4 93550 97954 100,890 Ertas & Akbulut (2021) 2.2 163546 164200 169,340 Ertas & Akbulut (2021) 2.1 433452 218420 225,420 Ertas & Akbulut (2021)

3.5. General Insights Across all specimen types, omitting thermal prestress led to consistent overestimation of fatigue life, especially in dissimilar joints where residual stress fields are steep and localized. The Morrow model combined with thermal strain induced prestress re duced life prediction errors dramatically, aligning within ±2× of experimental results in over 90% of cases. This confirms the model’s capability to serve as a robust, general -purpose tool for fatigue life assessment of spot-welded structures. Fig. 2 presents a log-log comparison between the predicted and experimental fatigue life values across all investigated specimen types and literature sources. The dashed diagonal line indicates perfect agreement, while the dotted lines represent the ±2× sc atter band, a common engineering benchmark for fatigue prediction models.

Fig. 2. Log- log plot comparing predicted and experimental fatigue life using the Morrow model with thermal prestress. The ±2× error band (dotted lines) shows strong agreement across TS, SLJ, and MTS joints for both similar and dissimilar materials.

The majority of the data points fall within this acceptable error range, demonstrating the reliability of the proposed Morrow-based fatigue model combined with thermal prestress. Notably, predictions for dissimilar joints (Mg – Al ) showed substantial improvement when prestress effects were included, whereas similar material joints (DIN 1623 steel) exhibited minor differences between the prestressed and non-prestressed cases. This consistency highlights the model’s robustness and general applicability across varying joint geometries and loading conditions.