SAFEMAP

Structural Assessment and Optimization using Fast and Efficacious Multi-data AI Approaches with Hybrid Multi-Physics under Natural Hazards

Principal Investigators 

Prof. Roland Wüchner,
Chair of Structural Analysis, TUM School of Engineering and Design

Prof. Felix Dietrich,
Professorship for Physics-Enhanced Machine Learning, TUM School of Computation, Information and Technology

Abstract SAFEMAP will focus on developing a fast prediction method that produces results comparable to traditional methods using multi-source data, such as real-world measurements, scaled-down lab tests, and physics-based simulation data. We will use artificial intelligence (AI) integrated with hybrid multi-physics to model and assess structural responses to natural hazards, with a particular focus on wind effects. Wind is a critical challenge due to its dynamic interaction with structures, potentially leading to significant damage and safety concerns. The main challenge is the solution of a coupled problem involving fluid-structure interaction (FSI), which is computationally expensive due to the turbulent 3D wind flow in large domains. Moreover, shape optimization in an early design scenario to effectively reduce the wind effects on a building is computationally extremely expensive with fully coupled FSI in numerical wind tunnels, and prohibitive in physical wind-tunnels. The proposed fast and efficacious wind load prediction approach will enable the development of an AI-accelerated FSI and shape optimization of the building hull under wind loading. We will utilize the fast multi-X approach to search the design space while employing traditional analysis for safety-critical decision-making, improving the design phase of structures under natural hazards, such as wind.