INTEGRATING COMPUTATIONAL FLUID MECHANICS AND ARTIFICIAL NEURAL NETWORKS FOR PREDICTING FLUID–STRUCTURE INTERACTIONS IN ADVANCED MECHANICAL MATERIALS

Abstract

Fluid–structure interaction (FSI) plays a critical role in advanced mechanical materials used in aerospace, biomedical devices, energy systems, and high-performance engineering applications. Conventional computational fluid dynamics (CFD) and structural mechanics approaches provide accurate predictions of coupled multiphysics behavior; however, they are computationally expensive for nonlinear and transient problems. Recent advances in machine learning and neural surrogate modeling offer promising alternatives for improving computational efficiency while maintaining predictive accuracy.

This study proposes a hybrid computational framework that integrates computational fluid mechanics with artificial neural networks (ANNs) for efficient prediction of FSI responses. The framework utilizes CFD-based simulations for data generation and ANN-based surrogate models to learn complex nonlinear relationships between input parameters and FSI outputs, including displacement, stress distribution, and pressure interactions.

The results show that the ANN model achieves high prediction accuracy with a coefficient of determination (R²) of approximately 0.97 and low error metrics, while maintaining strong agreement with simulation results. Furthermore, the proposed approach reduces computational time from several hours required by conventional CFD–FSI simulations to near-instant predictions, enabling efficient large-scale analysis.

The integration of computational fluid mechanics and artificial intelligence provides a reliable, scalable, and computationally efficient solution for modeling complex FSI systems in advanced mechanical materials.