Model identification of batch cooling crystallizations through calorimetry and image analysis

For most batch crystallization processes it remains difficult to obtain relevant on-line information on both the dissolved solid concentration in the liquid phase and the crystal size distribution (CSD). Moreover, the reported experimental results are generally based upon simplified population balance models such as moment equations which are known to contain insufficient information on the CSD. In order to cope with these two difficulties, the aim of this article is to present a new approach for parameter estimation and modelling of batch cooling crystalizations. Measurements of supersaturation were made with a Mettler RC1 calorimeter coupled with an on-line in situ laser sensor. Estimates of the final CSD were also obtained through image analysis. In addition to such a measurement strategy, the population balance equations were solved to simulate the time variations of the CSD, and used to identify nucleation and growth kinetic parameters. In order to evaluate the method, the isothermal crystallization of adipic acid in water was experimentally investigated. Profiles of the heat release due to crystal growth were computed from the on-line energy balance of the crystallization, and final CSDs were obtained off-line after image analysis. On-line information about the nucleation mechanism was also provided by the laser sensor. Through the fitting of both calorimetric and granulometric experimental data, these operations allow one to select the appropriate crystallization kinetic models and to estimate the corresponding parameters. For the reported batch experiments, the comparison between experimental and simulated variables such as the supersaturation and the final CSD was satisfactory. (C) 1997 Elsevier Science Ltd.