Abstract

Optimisation based on surrogate models is becoming popular for engineering problems due to its reduced computational efforts. In this research, we aim to maximise the plug flow performance of coiled reactors operating under oscillating conditions for a fixed geometry. This is done through Bayesian optimisation that uses Gaussian processes as a surrogate model and is coupled with computational fluid dynamics (CFD) simulations in OpenFOAM through the PyFoam library. We run a transient analysis with ScalarTransportFoam solver where the tracer is injected into the water as a working fluid to obtain residence time distribution which is then fitted with the tank-in-series model to get the plug flow performance. We explore the parameter space for amplitude (1-8 mm) and frequency (2-8 Hz) for a fixed Reynolds number of 50. The optimal conditions for plug-flow performance correspond to the Strouhal number St > 1 and oscillatory Reynolds number Re0