LAUSR

Abstract In this study, a C5 + 0.5 BPD microchannel Fishcer-Tropsch process with a U-type cooling system was modeled using a cell coupling method, and multi-objective optimization was conducted using an artificial neural network as a surrogate model. Two objective functions (reactor core volume and maximum process temperature rise, ΔTmax) were to be minimized using seven design variables as optimization variables. Reactor core volume represents a reactor's compactness, which is essential for a micro-channel reactor, whereas ΔTmax is highly related to reactor stability. A Pareto optimal solution was obtained for a feasible ΔTmax range of 3.8–6.8 K. The optimal reactor core volume for ΔTmax of 3.8 K was 1.45 times larger than that for ΔTmax of 6.8 K. As ΔTmax increases, the total reactor length is shortened while the total width and height remain relatively constant. A sensitivity analysis of Pareto optimization was conducted for two types of parameters: 1) coolant flow rate, and 2) fixed design parameters. Coolant flowrates over 750 LPM were found to be inefficient for the given conditions. Fixed design parameters were closely related to the capabilities of the reactor fabricator. The present study suggested a priority order for modifying fixed design parameters to increase compactness. Suitable points can be selected based on the specific requirements of plant conditions.