Saturday 01 March 2025
Researchers have been working on a way to generate complex mathematical models using artificial intelligence, and their latest study has made significant progress in this area. The team used a large language model called ChatGPT to create finite element code for geotechnical engineering applications.
Finite element method is a powerful tool used to solve complex problems in various fields, including civil engineering, physics, and chemistry. It involves dividing a system into smaller parts, called elements, and then using mathematical equations to calculate the behavior of each element. This process allows researchers to simulate real-world phenomena, such as the movement of soil or water.
Geotechnical engineers use finite element method to study the behavior of unsaturated soils, which are common in many natural and man-made structures. These soils can exhibit complex behavior due to the interactions between water, air, and solid particles. By using ChatGPT to generate the code, researchers were able to create a more accurate model of this behavior.
The team tested the model with three different scenarios: the dissipation of excess pore water pressure in one-dimensional space, time-dependent differential settlement of a strip footing, and gravity-driven seepage. In each case, they provided ChatGPT with necessary information, such as balance and constitutive equations, problem geometry, initial and boundary conditions, material properties, and spatiotemporal discretization and solution strategies.
The results showed that the model required minimal code revisions when using the FEniCS finite element library. This is because FEniCS has high-level interfaces that enable efficient programming. However, the MATLAB code generated by ChatGPT needed extensive prompt augmentations and direct human intervention. This highlights the importance of understanding the mathematical formulation and numerical techniques in generating accurate models.
The study demonstrates that large language models like ChatGPT can assist humans in implementing numerical models, but they are not yet ready to replace human programmers entirely. The researchers hope that their work will pave the way for future studies on using AI in geotechnical engineering applications.
The implications of this research are significant, as it has the potential to revolutionize the field of geotechnical engineering. By providing a more accurate and efficient model of unsaturated soil behavior, researchers can better understand and predict the performance of structures and infrastructure under various environmental conditions. This knowledge will be crucial in designing and building safer, more sustainable structures that can withstand natural disasters and other hazards.
Cite this article: “AI-Assisted Modeling of Unsaturated Soil Behavior”, The Science Archive, 2025.
Geotechnical Engineering, Finite Element Method, Artificial Intelligence, Chatgpt, Unsaturated Soils, Numerical Models, Programming, Fenics, Matlab, Code Generation
