Artificial neural networks discover the effects of particle-size distribution on the appearance of powder coatings

Powder coating has been heavily applied in many fields due to its zero VOC emission and excellent protective performance. The appearance of powder coating such as glossiness and roughness is paramount for its practical utility. Among various factors influencing appearance, the particle-size distribution (PSD) stands out as a significant feature. Nevertheless, the impact of PSD on powder-coating appearance remains largely empirical and insufficiently quantified. This study undertakes an in-depth investigation into the intricate relationship between PSD and powder-coating appearance, utilizing advanced machine learning models. The PSDs of the powders were characterized using a classification into seventy-five particle-size bins ranging from 0.100 to 1000 μm, each associated with its respective volume percentage. In this work, gloss at 60° was employed to measure the coating glossiness, whereas the Ra value (Arithmetic Average Roughness) was utilized to assess the coating roughness. Artificial neural network (ANN) models identify the impactful particle-size bins with positive and negative effects on static and dynamic flowability.