Grad Research Experience: Honghui Du, Structures

Honghui Du, advised by Qizhi He, is studying Structural Engineering.
He is supported in his studies through a Sommerfeld Fellowship.

"I particularly appreciate the comprehensive support from my supervisor, who is always available to discuss any concerns I have. This collaborative atmosphere makes our research group a comforting and stimulating place to work. Moreover, my project is at the cutting edge of technology, integrating knowledge from various interdisciplinary fields, which expands my expertise and perspective."

My research aims to develop advanced computational methods to enhance modeling and simulation capabilities for multiphysics systems. These methods are enabled by state-of-the-art techniques in artificial intelligence (AI) and machine learning, and they can be applied to multiphysics processes in fields from geophysics to bioengineering. Better models could enhance our understanding of how complex systems, as well as materials—from underground porous medium to human tissues—behave in response to external loading.

A prime example of our progress is a solver we call NIM. This is a hybrid solver that creates customized approximation functions by integrating AI with traditional numerical methods. This approach addresses the challenges of high dimensional parameter spaces in neural networks while avoiding the tedious theoretical differentiation of constitutive models. Thus, it helps us achieve more efficient prediction and analysis of material
properties, as well as rapid and accurate assessments of material behaviors.

Another example of progress is the methodology we detailed in a paper
published in Computers and Geotechnics (2023) that enables the accurate modeling and simulation of density-driven flow in porous media. This research enhances our understanding of the long-term behavior of CO2 in aquifers. This understanding is vital for geological carbon sequestration, which is being pursued to help to combat climate change.