Variable retort temperature optimization using adaptive random search techniques

Global optimization algorithms and software based on adaptive random search techniques show considerable promise as more rapid and efficient approach to process optimization in the food industry. This paper describes use of the method in finding optimum variable retort temperature profiles that would maximize quality retention or minimize process time without compromising target lethality or minimum required quality retention in the case of thermal processing of canned foods. Results agreed well with those previously published by others who used more traditional approaches for similar optimization problems. Results also showed that the method lent itself well to the use of a cubic spline approximation for the dynamic temperature profiles, thereby reducing significantly the number of variables and dimensional space of the problem, in contrast to other methods, while producing superior results.