Part-Time Research Assistant

Penn State University•University Park, IL

About The Position

The College of Earth and Mineral Sciences, John and Willie Leone Family Department of Energy and Mineral Engineering is seeking applicants for a part-time Research Assistant. The project involves developing a data-driven, physics-informed modeling framework to enhance the predictive capabilities of population balance models (PBMs) for stirred media milling. This research aims to address limitations in conventional PBMs by integrating Discrete Element Method (DEM) and Computational Fluid Dynamics (CFD) simulation data to understand microscale mechanisms governing particle breakage. A physics-informed graph neural network (PI-GNN) framework will be created, representing the mill as a spatial graph. The PI-GNN will learn spatially distributed PBM closure terms using local physical descriptors from DEM-CFD simulations. These learned closures will be integrated into a PBM solver to predict particle size distribution evolution and identify high-energy breakage regions. The ultimate goal is to establish a multiscale modeling framework linking particle-scale physics with macroscale milling performance for more accurate predictions of grinding efficiency, serving as a proof-of-concept for machine learning-based closures in comminution models.

Requirements

Part-time position
Must be a current Penn State student (not employed previously at the university) or an external applicant.
Successful completion of background check(s) in accordance with University policies.

Responsibilities

Develop a data-driven, physics-informed modeling framework to improve the predictive capability of population balance models (PBMs) for stirred media milling.
Integrate DEM-CFD simulation data to resolve the underlying microscale mechanisms governing particle breakage.
Construct a physics-informed graph neural network (PI-GNN) framework where the mill domain is represented as a spatial graph.
Train the PI-GNN to learn spatially distributed PBM closure terms using local physical descriptors obtained from DEM-CFD simulations.
Embed learned closures into a PBM solver to predict particle size distribution evolution and identify high-energy breakage regions within the mill.
Establish a multiscale modeling framework that links particle-scale physics with macroscale milling performance, enabling more accurate and physically grounded predictions of grinding efficiency.
Serve as a proof-of-concept for incorporating machine learning–based closures into mechanistic comminution models.