Additive manufacturing allows for creating complex reactor geometries, expanding design spaces. This paper presents a machine learning-assisted approach for designing chemical reactors, using high-dimensional parameterizations, computational fluid dynamics (CFD), and multi-fidelity Bayesian optimization. The approach focuses on coiled reactors, identifying design features leading to ~60% improved plug flow performance compared to conventional designs. This demonstrates the potential of combining advanced manufacturing with augmented intelligence for enhanced reactor performance.

Tom Savage, Nausheen Basha, Jonathan McDonough, James Krassowski, Omar Matar, Ehecatl Antonio del Rio Chanona