In the Bayer process, used for the production of alumina from the bauxite ores, the caustic-to-alumina molar ratio in the digester slurry leaving the digesters train is a critical operating parameter. Overall process productivity depends on the ability to stabilize the digestion caustic-to-alumina molar ratio. It is desirable to operate the plant at a target molar ratio that is as low as possible to ensure that production and processing efficiences are not lost, yet high enough for complete alumina extraction from the bauxite and stable processing to precipitation. This paper describes the development and validation of a simple, but accurate, simulation model for the digestion process. The model developed is used to gain understanding of the dynamics of the digestion process, to determine an optimal value for the caustic-to-alumina molar ratio, and to test the performance of possible control schemes, thus minimizing the expenses involved in experimentation with poorly designed control schemes on the physical plant. The design, tuning, and testing of a feedforward-predictive-feedback control scheme are outlined. Sensitivity analysis has shown the dependency of the closed-loop performance on the model structure.