Tomography-controlled Batch Crystallization
State of the art: Batch crystallization is a wide-spread industrial process to purify substances and transform dissolved compounds to solid products. Prominent examples are crystallization of sugar or salt, concentration of fruit or coffee extracts, purification of bulk or fine chemicals, production of pharmaceuticals, zeolites and more. Crystal product quality should be according to set specifications in order to avoid expensive reprocessing steps. Controllable process variables are dissolved compound concentration and fluid mixing. An effective but yet under-explored way of process control is via ultrasound field application. Ultrasound can be used to control crystal size, crystal habit, time of induction, agglomeration and aggregation properties. All these properties are directly controllable by ultrasound power, frequency and pulse duration, which influence the course of agglomeration and aggregation in all typical crystallization modes (precipitation, evaporation, cooling) depending on supersaturation and crystal size. As such growth dependent phenomena as well as surface fouling problems are common also in other industrial processes (e. g. aerobic or anaerobic fermentations, distillation of bio-raffinates), there is a grand leverage for process improvement. Process control has been in minor role in crystallization process design. In-line solution concentration measurements (FTIR, NIRS) and in-line particle size analysers are difficult to use in fast solidification and give only local data which are not representative throughout the reactor volume. Tomography-based sensors are a perfect solution and can be easily used in the most typical crystallization processes, providing their process interface is designed in a suitable way.
TOMOCON objectives: Within TOMOCON the efficient control of batch crystallization will be demonstrated. On a rather high TRL 7, Electrical Resistance Tomography (ERT) will be implemented to measure the spatial distribution of the solidification via changes in conductivity in the bulk. In parallel ultrasound tomography (UST) will be developed, which is yet TRL 2 for this application, but would have distinct advantages in costs, complexity and non-intrusiveness (clamp-on), when operated simultaneously with ultrasound actuation. The challenge here is the nonlinearity in ultrasound propagation in a field with strong phase change and local sound scattering, which is physically complex due to the fact that modes of sound propagation change from longitudinal to longitudinal and transversal for phase change to solid. The latter requires fundamental scientific analyses with experiments and numerical simulations. Both types of tomography will be used to measure suspension density during crystallization and, this way, derive the distributed solidification state. As control variable feed rate of precipitating solution and ultrasound power are considered. Control algorithms can be e. g. fuzzy, PID, neural network based control. For process modelling both general simplified reactor models (stirred tank plus mixing model) and advanced single phase CFD models up to the onset of crystallization will be developed and employed. Also population balance modelling (PBM) is studied using method of moments (QMOM) for reactor operation simulation. Besides control of ultrasound power the controller concept shall explore the potentials of further including precipitating feed rate and stirrer control to change the mixing state under tomographic observation. This results in a complex multi-actuator control. As in the micro-wave drying demonstration knowledge-based control shall be employed. Eventually this application shall be a demonstration case for a proper human-machine interface allowing an operator to interfere manually, e. g. to assist the design of the process by expert knowledge. Lab-scale demonstration will be carried out in the Lappeenranta University of Technology crystallization lab, with support from Sulzer and DuPont with respect to later upscaling.