A Computer Vision\nApproach toward Verifying CFD Models\nof Stirred Tank Reactors

Mixing is one of the most important nonchemical considerations\nin the design of scalable processes. While noninvasive imaging approaches\nto deliver a quantifiable understanding of mixing dynamics are well-known,\nthe use of imaging to verify computational fluid dynamics (CFD) models\nremains in its infancy. Herein, we use colorimetric reactions and\nour kinetic imaging software, <i>Kineticolor</i>, to explore\n(i) the correlation of imaging kinetics with pH probe measurements,\n(ii) feed point sensitivity for Villermaux–Dushman-type competing\nparallel reactions, and (iii) the use of experimental imaging kinetic\ndata to qualitatively assess CFD models. We report further evidence\nthat the influences of the stirring rate, baffle presence, and feed\nposition on mixing in a tank reactor can be informatively captured\nwith a camcorder and help experimentally verify CFD models. Overall,\nthis work advances scarce little precedent in demonstrating the use\nof computer vision to verify CFD models of fluid flow in tank reactors.