Thermodynamic Consistent Neural Networks for Learning Material Interfacial Mechanics
This work provides a more robust data-driven modeling approach for material interfacial mechanics, which is crucial for engineers and material scientists predicting and understanding interfacial failures in multilayer materials.
This paper addresses the challenge of modeling traction-separation relations (TSR) for material interfaces using sparse experimental data. The authors propose a Thermodynamic Consistent Neural Network (TCNN) that incorporates physical principles into its loss function, leading to accurate predictions of the TSR surface while significantly reducing violations of physical laws compared to traditional neural networks.
For multilayer materials in thin substrate systems, interfacial failure is one of the most challenges. The traction-separation relations (TSR) quantitatively describe the mechanical behavior of a material interface undergoing openings, which is critical to understand and predict interfacial failures under complex loadings. However, existing theoretical models have limitations on enough complexity and flexibility to well learn the real-world TSR from experimental observations. A neural network can fit well along with the loading paths but often fails to obey the laws of physics, due to a lack of experimental data and understanding of the hidden physical mechanism. In this paper, we propose a thermodynamic consistent neural network (TCNN) approach to build a data-driven model of the TSR with sparse experimental data. The TCNN leverages recent advances in physics-informed neural networks (PINN) that encode prior physical information into the loss function and efficiently train the neural networks using automatic differentiation. We investigate three thermodynamic consistent principles, i.e., positive energy dissipation, steepest energy dissipation gradient, and energy conservative loading path. All of them are mathematically formulated and embedded into a neural network model with a novel defined loss function. A real-world experiment demonstrates the superior performance of TCNN, and we find that TCNN provides an accurate prediction of the whole TSR surface and significantly reduces the violated prediction against the laws of physics.