This proposal covers the computational needs of two research projects in structural engineering. Both aim to improve the prediction of structural behaviour, one through physics-based simulation and the other through data-driven methods.
Project 1 – Fluid–structure interaction simulations. This project studies how structures respond to fluid loading using transient two-way fluid–structure interaction (FSI) simulations. The simulations use Ansys Fluent coupled with Ansys Mechanical through System Coupling, and capture the interaction between the flow field and the structural response: aerodynamic or hydrodynamic forces, vibration and stress histories. Simulations cover a range of flow conditions and structural configurations. With meshes of several million cells and dynamic mesh motion, they require parallel CPU resources and substantial storage for transient results.
Project 2 – Deep learning on structural health monitoring data. This project develops deep learning models for predicting structural behaviour from large-scale, long-term structural health monitoring (SHM) data, such as accelerometer, strain gauge and environmental sensor measurements. The models are used to predict structural response under varying environmental and operational conditions, detect anomalies, and identify early signs of damage, as well as forecast future values of sensor data. Training on years of high-frequency sensor data requires GPU resources and large project storage.
Both projects contribute to safer and more cost-effective management of infrastructure. The proposal requests CPU resources for Project 1, GPU resources for Project 2, and shared project storage for simulation output and SHM datasets.