PURPOSE: This paper presents a soft sensor that can be effectively used for in-line monitoring of the primary drying step of a pharmaceuticals freeze-drying process in vials. METHODS: Process modeling and product temperature measurements are used to estimate the residual amount of ice in the vial and the heat transfer coefficient from the shelf to the product in the vial. The resistance of the dried cake to vapor flow is determined through the heat balance equation at the interface of sublimation. Mathematical simulation and experimental tests have been carried out to validate the estimations provided by the soft sensor. RESULTS: Accurate estimations of the dynamics of the product until the end of primary drying are obtained, as well as of the heat and mass transfer coefficients, even in the case of a highly non-uniform batch. The reduction in the number of variables directly estimated by the soft sensor allows increasing the robustness of the tool with respect to other sensors presented in the literature. CONCLUSIONS: The proposed soft sensor is thus effective for process monitoring and it allows using model-based tools for cycle development in lab-scale units, where thermocouples are usually available, and for process monitoring in industrial-scale freeze-dryers, in case wireless sensors are used.
PURPOSE: This paper presents a soft sensor that can be effectively used for in-line monitoring of the primary drying step of a pharmaceuticals freeze-drying process in vials. METHODS: Process modeling and product temperature measurements are used to estimate the residual amount of ice in the vial and the heat transfer coefficient from the shelf to the product in the vial. The resistance of the dried cake to vapor flow is determined through the heat balance equation at the interface of sublimation. Mathematical simulation and experimental tests have been carried out to validate the estimations provided by the soft sensor. RESULTS: Accurate estimations of the dynamics of the product until the end of primary drying are obtained, as well as of the heat and mass transfer coefficients, even in the case of a highly non-uniform batch. The reduction in the number of variables directly estimated by the soft sensor allows increasing the robustness of the tool with respect to other sensors presented in the literature. CONCLUSIONS: The proposed soft sensor is thus effective for process monitoring and it allows using model-based tools for cycle development in lab-scale units, where thermocouples are usually available, and for process monitoring in industrial-scale freeze-dryers, in case wireless sensors are used.