The quality of ice-cream is at a large extent determined by the smooth texture and creamy mouth feel which are closely related to the crystal size and product viscosity. Model based computer aided process engineering may enable the design of optimal operation conditions so as to maximize ice-cream quality while minimizing, for example, energy consumption. In this regard, the development of predictive mathematical models describing the ice-cream crystallization process is object of intensive research. However, mathematical modelling is a time consuming task that involves several steps from the definition of the questions to be addressed to the development of a final model with satisfactory predictive capabilities. In this concern, model identification, plays a crucial role in model development since it involves structure characterisation, identifiability analysis, parameter estimation and experimental design. The objective of this work is to perform a complete study of the model identification for the ice-cream crystallization process. A simple yet accurate mathematical model relating the process inputs and outputs was derived, based on the method of moments. The AMIGO toolbox (Advanced Model Identification using global optimization) was used to iteratively identify model unknowns from experimental data obtained in the pilot plant located at CEMAGREF and to design new experiments to improve model predictive capabilities.
展开▼
