Exercising hybrid model to design an optimized electrospun polyamide-6 nanofibrous mat for air filtration applications

Abstract

There are several factors affecting the filtration efficiency and pressure drop in the nanofibrous filters including the basis weight of the mat, the thickness of the mat, and the nanofiber diameter. Besides, such parameters are dependent on manufacturing factors which should have been taken into account before manufacturing. In this study, first an experiment based on central composite design (CCD) to investigate the effect of electrospinning factors such as polyamide-6 concentration, applied voltage, nozzle-to-collector distance, and electrospinning time on main structural parameters consisting of basis weight, thickness, and nanofiber diameter is performed. Then, three models with similar inputs but different outputs to establish relationships between electrospinning factors and main parameters utilizing two procedures including response surface methodology (RSM) and artificial neural network (ANN) are developed. Next, the outputs of a procedure with best-performing models are used to compute two analytical models to estimate filtration efficiency (considering interception and diffusion effect) and pressure drop. Finally, a genetic algorithm (GA) is coupled to the analytical models to optimize electrospinning factors using a cost function constructed by filtration efficiency and normalized pressure drop. The results highlighted that ANN-based models have more average goodness value (≈2.00) than RSM ones (≈1.83). Also, the GA has the capability of resulting in an optimized sample with higher quality factor (0.015) than the optimum one (0.011).