| Model spray dried materials were used in order to determine if it is possible to develop a system capable of predicting spray drying formulations and processes for a variety of materials. These model materials were maltodextrin, colloidal silicon dioxide, dextrose, sorbitol, lactose, and povidone.; Extensive experimentation revealed that four of the six selected model materials were able to be spray dried. While some of the experimental results were quite predictable, other results did not follow expected trends and may have been influenced by material factors that are not addressed by theoretical relationships. Consequently, development of a predictive model for the spray drying process needed to address material factors in addition to process parameters.; Several critical physical characterization factors of each model raw material were selected for inclusion in the feasibility prediction process. Each factor had a direct effect on the ability of the material to be spray dried, which was expressed in a decision tree format. Additionally, a method for simulating the evaporation stage of the spray drying process was developed using DSC instrumentation.; Six material characteristics and six processing parameters were found to be predictive of the physical characteristics of the spray dried materials produced using both nozzle configurations. Artificial neural networks (ANN) were shown a suitable tool for the determination of these critical independent variables. Also, ANN were capable of building models which describe and predict the physical characteristics of spray dried powders. Specifically, a three-layer backpropagation network was shown to effectively model the process. Genetic algorithms used in conjunction with these ANN were capable of predicting the optimal processing parameters required to achieve specified powder characteristics.; It was determined that a predictive system composed of a database, decision trees, neural networks, and genetic algorithms can be constructed for evaluating the feasibility of a formulation to be spray dried and for calculating the optimal processing conditions required to produce a spray dried powder with user-specified physical characteristics. |
