Authors:

Jalen Magee

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Fluid mechanics use various methods to solve different types of practical problems. Conventional methods suffer from long computation times and need hands-on manual intervention, prompting us to seek a better method. In our study, we investigate the potential of physics-informed machine learning (PIML) and neural networks in the field of fluid mechanics, specifically through the Nonlinear-Elasto-Visco-Plastic (NEVP) model. Our research is driven by combining machine learning techniques using Python with physics principles, thereby shedding new light on fluid mechanics. The proposed methods entail collecting experimental data using a rheometer to measure and analyze the flow properties of the fluids with the data including the shear stress, the elastic modulus, etc. Next, in Python, we will be using PIML, more specifically, the DeepXDE library that specializes in physics-informed neural networks (PINN) to train the NEVP model using supervised learning, allowing the network to incorporate the experimental data. Adding on, our results showcase that the NEVP model features enhanced accuracy and efficiency in capturing the elastic behavior and the stress buildup of certain complex fluids. Through this study, we present an innovative approach for fluid mechanics research to progress by harnessing the power of PIML and PINNs to learn the NEVP model and thus the behavior of complex fluids (such as thermal greases) and how they flow.

Source:

Purdue University / 2023

Topics:

No topics listed

Co-authors:

Jalen Magee