PI: James Ricles
Co-PI(s): Maryam Rahnemoonfar, Muhannad Suleiman
University: Lehigh University
Industry partners: Eastern Exterior Wall Systems, Hindle Power
This project will develop an artificial intelligence (AI)-enabled, multi-physics real-time cyber-physical simulation (RTCPS) framework to assess the resilience and performance of coastal infrastructure subjected to storm-induced inundation, including storm surge, wave loading, and associated fluid-soil-foundation-structure interaction (FSFSI) effects. Recent extreme storm events have highlighted the vulnerability of coastal infrastructure systems and the need for advanced simulation tools capable of capturing coupled hydrodynamic, geotechnical, and structural response under realistic loading conditions.
The proposed research integrates expertise from the Civil and Environmental Engineering and Computer Science and Engineering departments to extend RTCPS to complex coastal multi-physics problems. In the proposed framework, the soil-foundation domain and selected structural or energy-dissipating components are modeled physically, while the two-phase (air-water) fluid domain and superstructure are represented numerically using AI-based surrogate models. This partitioning is essential because soil-foundation behavior under extreme loading is highly nonlinear, path-dependent, and difficult to model reliably using constitutive soil models, whereas high-fidelity numerical simulation of two-phase fluid flow is computationally prohibitive for real-time execution. RTCPS enables these challenges to be addressed simultaneously by physically capturing the most uncertain geotechnical behavior while numerically modeling the surrounding fluid and structural response in real time.
Neural network models will be trained to reproduce nonlinear, history-dependent hydrodynamic forces induced by storm surge and wave action, including two-phase air-water effects, while significantly reducing the computational demands associated with traditional high-fidelity computational fluid dynamics approaches. High-performance computing (HPC) resources will be used to generate large-scale training datasets and to develop advanced recurrent neural network architectures capable of capturing transient two-phase fluid-structure interaction (FSI) behavior suitable for conducting real-time cyber-physical simulations.
The trained AI models will be integrated into the NHERI Lehigh real-time cyber-physical simulation framework to perform RTCPS of coastal infrastructure under inundation scenarios. The project will leverage the unique experimental, computational, and control capabilities of the NHERI Lehigh Real-Time Multi-Directional Experimental Facility along with the Soil-Foundation-Structure Interaction Facility that are both located within the ATLSS Engineering Research Center.
The outcomes of this project will provide validated tools for performance-based assessment and mitigation of coastal infrastructure subjected to extreme storm conditions, advance AI-enabled real-time cyber-physical simulation methodologies, and support the training of graduate students in multi-physics simulation of coastal hazards. Overall, this research supports White House priorities in climate resilience, infrastructure modernization, trustworthy artificial intelligence, disaster risk reduction, and workforce development by advancing AI-enabled cyber-physical tools for assessing and mitigating storm-induced impacts on critical coastal infrastructure.