Optimizing Melt-Blowing Nozzles for Small-Diameter Fibers: An Artificial Neural Network Framework

Abstract

In this study, two feedforward artificial neural networks (ANNs) were trained on experimental data to predict the melt-blowing (MB) fiber diameter of hot-melt adhesive and polypropylene fibers based on process operating conditions and nozzle geometry. These ANNs enabled a sensitivity analysis to investigate the effects of input parameters on the fiber drawing ratio. The results indicate that higher air–polymer flux ratios and extrusion temperatures, along with nozzles having an air impact point close to the nozzle exit and a low polymer-to-air area ratio, facilitate the production of small-diameter fibers. Furthermore, the ANNs incorporated a comprehensive set of input parameters characterizing the melt-blowing process and were trained using cross-validation and regularization techniques to enhance their generalization. This enabled the design of optimized nozzles for small-fiber production. Additionally, a nozzle design optimization framework based on ANNs is proposed to optimize new MB nozzles and enhance existing designs according to established industrial objectives and fiber compositions.