New Jersey Institute of Technology
Department of Mathematical Sciences

Capstone Laboratory

From Structure to Performance of Porous Media using Image Processing, Topological Data Analysis and Machine Learning


Instructor: Lou Kondic

Lab Assistant: Zhaoshu Cao



Fluid flow (flux) through porous media is crucial in various industrial processes and natural phenomena, including oil recovery and groundwater filtration. The internal structure of these materials plays a crucial role in determining how fluid flows through them, making their characterization essential for optimization processes. To help achieve this characterization, we use synthetic data and work towards connecting topological properties of these data with flow properties. We utilize topological data analysis, a method that helps us quantify material structure, for this analysis. For more details, see the poster, and the group reports below.





Group 1: Producing the data. Students' report, presentation

Group 2: Simplifying Porous Media to Networks. Students' report, presentation

Group 3: Machine Learning Applied to Both the Original Data and the Networks. Students' report, presentation