ABSTRACT
384Design studies are important in assessing the feasibility of supercritical extraction (SCE) processes since they properly weigh factors such as the scale of the process, the value of the product, the need for a nontoxic solvent, etc. Once the most promising design is located, the decision to invest in the SCE process can be made with confidence. In this chapter, both experimental and simulation approaches to design are discussed. The experimental studies aim to determine the effect of the key process parameters, and where possible, to collect fundamental data to aid in the development of more rigorous models for simulation and design optimization. Examples discussed include the decaffeination of coffee beans, the extraction of edible oils, and the dehydration of alcohols. Applications of simulation and optimization methods have been less common, due primarily to the difficulty of modeling the complex systems. Good success, however, has been obtained when these techniques have been applied to the dehydration of alcohols and ketones. Several examples are discussed that indicate that SCE can be competitive for these separations on the basis of energy consumption. A more complete study, which includes the cost of the equipment, demonstrates that SCE is not competitive since the products are low-value, high-volume chemicals. Other work is described that attempts to model more competitive SCE processes, such as the extraction of lecithin from soya oil and the isolation of (β-carotene from fermentation broths; however, the designs for the recovery of these high-value, low-volume chemicals are more uncertain since simplified models were used. Finally, the transient behavior and control of SCE processes is discussed. Although this challenging problem is only beginning to be addressed, initial studies indicate that maintenance of the proper phase distribution in the extractors and separators is difficult to achieve with conventional control schemes. As design models are extended to simulate the dynamics of SCE processes, new model predictive control algorithms can be expected to significantly improve the control of the phase distribution.
