Thermo‑mechanical fatigue life and probabilistic fracture assessment of a large-bore diesel engine block using integrated S-N, SMART technology, and non‑linear regression models

• Integrated S-N and LEFM crack‑growth modelling via SMART tool. • Combustion pressure identified as primary fracture‑risk driver. • Combustion temperature rise 400-800 °C reduces fatigue SF by 39 %. • Cylinder diameter increase 30-198 mm lowers fatigue SF from 3.5 to 0.3. • Monte-Carlo analysis shows cracked-life P f >0.13, dominating failure risk. Large-bore, high-cylinder-count diesel engine blocks are subject to severe cyclic thermos-mechanical loading; however, fracture-based fatigue assessments of such complex structures are scarce. In this study, an integrated deterministic–probabilistic framework is developed to evaluate the thermo-mechanical fatigue life of a 20V MTU engine block, combining classical stress-life modelling with crack-growth analysis. This is achieved using the Separating Morphing and Adaptive Remeshing Technology (SMART) framework within a sequentially coupled finite-element framework. This hybrid strategy enables direct simulation of crack initiation and propagation under realistic combustion-induced pressure and temperature gradients, which eliminates the need for manual remeshing. Non-linear regression and sensitivity analyses quantify the influence of geometric, thermal, and loading parameters on static and fatigue safety factors. Statistical analysis based on Spearman’s rank correlation coefficient indicates that combustion pressure is the most influential variable, followed by stress ratio and flaw size. A Monte-Carlo-based structural reliability analysis is also incorporated to capture variability across static, fatigue, and cracked-life limit states. The results highlight that cracked-life is the dominant contributor to total failure probability, and that a combustion temperature rise from 400°C to 800°C reduces the fatigue safety factor by 39%. Enlarging the cylinder diameter from 30 mm to 198 mm decreases this quantity from 3.5 to 0.3. The proposed probabilistic methodology offers a novel approach for predicting fracture-critical durability in large engine blocks under coupled thermo-mechanical service conditions