Self-optimising control a crystallisation process

Abstract

Consulta en la Biblioteca ETSI Industriales (Riunet)

[ES] With the passing of time, companies, aware of the necessity to improve their process efficiency and performance and also instigated by environmental regulations and deadlines, are paying a much closer attention to control and optimisation schemes when designing their processes. Crystallisation processes, very common in food, pharmaceutical and chemical industries, are no exception. An evaporative-type crystallisation process is studied in this work. It is modelled with population balance equations based on the experimental research carried out by Mesbah et al. (Mesbah et al., 2011). The concept of self-optimising control is relatively new. Firstly applied to continuous processes, and later extended to batch operations, self-optimising control basically tries to achieve optimal condition for the previously selected controlled variables by maintaining the manipulated variables at constant set points. Leaving the on-line control task to feedback controllers, self-optimising control ensures that, even though disturbances occur, the controlled variables will still be near-optimal, with an acceptable loss with respect to their truly optimal value. In contrast to self-optimising control, model predictive control depends strongly on the on-line task. Models of the process are used to predict the future, and then the optimal manipulated variable for this predicted future is calculated by minimising an objective function. Self-optimising control is implemented on the crystalliser, and the results obtained are compared to those obtained before by Salamanca Gonzalez (Salamanca Gonzalez, 2011) under the MPC approach. Different scenario simulations will be studied in order to test and compare the behaviour of both approaches under different situations. Conclusions are drawn from the analysis of these results.