Search

The residence-time distribution (RTD) studies are needed to characterize mixing conditions, flow patterns, and the extent of conversions and reactions of the biopolymers in any plasticating or cooking extruder. For the supercritical fluid extrusion process (SCFX), mixing of the fluid and starch-based biopolymers is an important step affecting cell nucleation and growth. RTD was studied in the mixing zone of the extruder, where CO 2 and gelatinized starch are brought together. Three levels of each of the process variables (screw speed and CO 2 injection pressure) were investigated, using specially designed apparatus for color-dye injection, and a die-restrictor to change the pressure profile and holdup volume in the extruder. Control RTD plots were for no CO 2 injection conditions at the same pressure, and the RTD of the full extruder section. Experimental RTDs were evaluated by: (1) the Wolf-White model, (2) a cascade of continuously stirred reactors (CSTR) model and (3) a model based on plug-flow in series with cascade of CSTRs. The last approach was most successful in capturing all aspects of the observed RTD curves including their tails. Injection of CO 2 at the highest pressure (110.3 bar) increased the RTD compared to no CO 2 -injection conditions. There were significant differences in the plug flow component of the RTD as well as differences in average residence time as a function of both screw speed and CO 2 injection pressure conditions.

Fractionation of anhydrous milk fat using supercritical CO 2 provides various fractions of different physicochemical and functional properties that should enhance its utilization. Based on a mass transfer study of a pilot plant setup of a continuous countercurrent fractionation of anhydrous milk fat by supercritical CO 2 , economic analysis of manufacturing plants was improved. Equipment was sized for six commercial-scale plant capacities, ranging from 800 to 10,000 tonnes/yr of anhydrous milk fat. Analysis considers capital investment, annual manufacturing expenses, revenues, sales, and payback time. The equipment scale-up index was .53, which is the usual literature value. For a plant processing 10,000 tonnes/yr of anhydrous milk fat, conversion cost is $.15/kg over a base price of $1.98/kg. Such a plant would have a capital investment of $4.4million and annual manufacturing costs of $21.1million. Example selling prices are calculated assuming a desired rate of return of 20% or a payback period of 5 yr, which further adds $.11/kg to the base price and conversion cost, yielding annual sales of $22.2million. The economics of the process are attractive, even though further optimization of operating conditions has not yet been performed, primarily because of the continuous nature of the process, the heat regeneration systems designed, and the large plant capacities considered. Thus, the notion that supercritical processing is expensive may be valid only for batch or semi-continuous systems rather than for continuous large-scale systems such as those studied herein.

Twin-screw extruders exhibit highly non-linear and interactive dunamic behavior which makes their modeling and control difficult. Transfer function models determined from experimentally observed behavior are not only adequate in capturing the dynamics of the extrusion process, but are also invaluable tools in the design of multivariable control strategies for extruders. Suitable forms of transfer function models are taken from the literature to characterize a typical process used in cornmeal extrusion. The parameters for the transfer function models are also adapted from the literature and are used as an example in process control analysis and design. The degree of process loop interactions expected in this extrusion process for various controlled variable/manipulated variable pairings is studied using the relative gain array (RGA) technique. For minimal interaction, the MIMO system with screw speed and barrel heating as manipulated variables and product temperature and motor torque as controlled variables was found best. This system, using feed rate and moisture content as input disturbances and with realistic constraints on the manipulated variables, was simulated using standard software for controlled system performance. PID controllers without additional decoupling gave adequate performance results for various controlled system requirements.