Artificial Neural Network–Based Models
The methodology for the development of models discussed in Chapters 4 through 7 of this book is based on a simplifying representation of the process of applying appropriate governing physical laws (e.g. laws of conservation of mass, momentum and heat and laws of kinetics and thermodynamics). The effect of randomness in the process on the process’ performance and product quality was considered using stochastic methods. Population balance models are a special type of stochastic model. However, in some situations, either there is no clear understanding of the process or there is not sufficient time to develop a model based on first principles that require a clear understanding of the process. For example, in cases of complex chemical reactions, the number of reactions is so large that it becomes almost impossible to obtain a unique global reaction rate equation. The change of rate-determining steps with operating conditions may not be clearly understood. In the absence of a chemical reaction, the performance of process and equipment may depend upon the flow regime; this may not be characterised easily, or there may not be sufficient time to develop a sound model for these aspects.