The inverse optimization of exhaust hood by using intelligent algorithms and CFD simulation

Abstract

A method that couples artificial neural network (ANN), genetic algorithm (GA) and computational fluid mechanics (CFD) was proposed to inversely optimize the geometric configuration of exhaust hood and operation parameters. The optimal objectives of exhaust hood optimization are the limitation concentration of emission at the exit and the minimum deposition on the exhaust hood walls of cut tobacco. The design variables are air flow rate, valve open values and the pressures at the test points. Twelve different orthogonal cases based on the 3-D CFD model were chosen to train and validate the ANN in order to obtain the relationship between the optimal objectives and the design variables. The CFD cases and those of ANN were used by the genetic algorithm (GA) to find the optimal design variables satisfying the design objectives. Both CFD and experimental methods were used to verify the performance of the optimized exhaust hood. The results showed that the emission at the exit and the wall deposition of cut tobacco was significantly reduced by optimizing the pressure distribution in the exhaust hood. The optimal exhaust hood structure and operation conditions were obtained.