Coupled Multiphysics Modeling with Machine Learning Presented By: Mr. Kevin Wandke, PhD Candidate, University of Illinois at Urbana-Champaign 

Abstract: Improvements in computer hardware have enabled engineers to simulate real world  systems with increasing precision and speed. However, as the rate of improvements in  computing hardware has slowed, researchers have begun exploring alternative ways to improve the capabilities of computer modeling technologies. In this research talk, I will examine one such  approach that leverages recent advances in machine learning and parallel computing to simulate  coupled multiphysics systems. I will begin with a brief overview of existing approaches to  simulating these systems, as well as the drawbacks of these approaches. Next, I will discuss how  machine learning can replace these traditional approaches. After that, I will examine a workflow  to utilize machine learning to replace conventional iterative solvers as we solve a real-world  simulation problem. To conclude, I will discuss some additional research directions I hope to  explore, as well as ways I would be excited to involve students in this project. 

Biography: Kevin Wandke received his B.S. in Mechanical Science and Engineering in 2019, and his M.S. in Electrical Engineering in 2022, both from the University of Illinois at  Urbana-Champaign. He is currently pursuing his Ph.D. in Electrical and Computer Engineering at University of Illinois at Urbana-Champaign. He previously worked at Argonne National laboratory as a member of the SULI program, and at  General Electric's Global research center as an intern in the  
Edison Engineering Program. Additionally, he is the recipient of the Chancellor's Scholarship, Olsen award for Excellence in Undergraduate Teaching, and a Mavis Future Faculty Fellow in the Grainger College of  Engineering. Kevin’s research focuses on two major themes, computational modeling, and  pedagogy. Within the theme of computational modeling, he has focused on ways to use machine  learning to accelerate the modeling of complex multiphysics processes. Within the area of  pedagogy, Kevin has focused on ways to provide adaptive instruction, as well as ways to deliver  content through various modalities to improve the learning experiences of students.