This simulation investigates the development and application of machine learning (ML)-assisted multi-hazard fragility surfaces for earthquake-tsunami resilience analysis. The primary goal is to generate 3D fragility surfaces that integrate probabilistic damage states under sequential earthquake and tsunami loading, providing a computationally efficient alternative to physics-based simulations. Methodology & Testing: Data-Driven Fragility Surface Generation: - The ML model was trained using randomly selected 2D fragility curves for earthquakes and tsunamis. - A Random Forest (RF) regression algorithm was used to estimate joint failure probabilities across varying hazard intensity measures. - The model was validated against physics-based fragility surfaces from finite element simulations. Monte Carlo-Based Risk Assessment: - Fragility surface predictions were integrated into a Monte Carlo simulation framework to compute structural damage probabilities. - Multiple building archetypes, including reinforced concrete (RC) and unreinforced masonry (URM) structures, were tested under regular and long-duration earthquake motions. Comparison with FEMA Combinational Rule: - The ML-generated fragility surfaces were compared to FEMA’s superposition-based approach for estimating multi-hazard damage states. - Results demonstrated that the ML-assisted fragility model provides more accurate estimates at higher hazard intensities, where FEMA’s approach tends to underestimate cascading damage. Outcome & Key Findings: - The ML-based approach successfully synthesized continuous earthquake-tsunami fragility surfaces from discrete 2D fragility curves, reducing computational costs by over 90% compared to traditional physics-based methods. - The generated fragility surfaces improved predictions of community-level structural damage and economic loss assessments. - The methodology supports structural retrofitting analysis, allowing for horizontally shifting fragility curves to represent mitigation scenarios. - The results highlight the importance of accounting for hazard interactions in multi-hazard resilience planning. Reusability & Applications: - The dataset can be adapted to different structural types and hazard environments for multi-hazard fragility modeling. - Researchers can use this dataset to benchmark ML-assisted fragility functions against traditional fragility methods. - The ML-based fragility surfaces can be integrated into decision-support tools, such as IN-CORE, for urban disaster resilience planning. - The approach can be extended to other multi-hazard scenarios, including hurricane-induced storm surges, landslides, and fire-following-earthquakes. This simulation dataset provides a valuable resource for engineers, researchers, and policymakers interested in multi-hazard fragility analysis, structural risk assessment, and disaster resilience planning.