team photo


Team 16

Team Members

Faculty Advisor

Jason Dumont
Daniel Kloyzner
Dylan Carzello

Dr. Matthew Stuber

Sponsor

Other

sponsored by
sponsor logo

Self-Learning Flow Reactor

It can take many experiments to refine a chemical process when finding the most optimal way to run it. This can make optimization difficult to do by hand in a cost-effective way, especially with the possibility of human error which increases with the number of iterations done with a particular reaction. The goal of this project is to create a system that when given a set of parameters, and a particular measurement to optimize, will use optimization algorithms to adjust a reactor process in real time. This will both remove the element of human error from the process, as well as save time otherwise taken up by continuously stopping the process to adjust different variables before restarting. The project uses Matlab for both the optimization algorithm itself and for the use of a “digital twin” of the system. The digital twin is used to calibrate and test the algorithm to compare with results from using the system in a lab environment. The reactor, when in a physical setting, will give data of the outputs to the ReactIR software. This data is then passed to the algorithm in Matlab, which will then connect to the Labview program to in turn adjust the pumps and heat exchanger of the reactor setup. The changes made will then cause changes in the output data, which will continue to refine and optimize the process. This project will cause increases in lab productivity, while also decreasing costs. In the long term, this can impact the speed of development for life-saving drugs, and can speed the advancement of medical science, as well as development in several other fields of chemistry.