Grey and black-box modelling based on neural networks and artificial immune systems applied to solid dissolution by rotating disc method

The dissolution rates of urea, sodium bicarbonate, and sodium carbonate in water and aqueous solutions were determined using the rotating disc technique. The experiments showed that the dissolution rate increases with increasing disc surface area, temperature, and rotating speed, while it decreases with the solute concentration increase in the dissolution medium. The comparison between experimental values for the dissolution rate and those calculated from Levich equation evidenced a satisfactory agreement in the case of the urea dissolution and poor compliance for the sodium bicarbonate and sodium carbonate dissolution. This poor results and the lack of a good model for making predictions in different situations determined the generation of empirical and semiempirical models (black and grey box approaches) which include neural networks developed with Clonal Selection algorithm (belonging to the Artificial Immune System class) and combination between neural network and phenomenological model. Satisfactory results were obtained with neural networks (black box models) and hybrid models (grey box models).