MODELING FOR OPTIMAL PRODUCTION DECISIONS AND PERFORMANCE CONTROL IN AQUACULTURE.

Abstract

One result of the search for inexpensive alternative sources of protein has been the rise in interest in aquaculture, the rearing of aquatic organisms under controlled conditions. In this dissertation we examine several management approaches to the efficient rearing of aquatic animals, using mathematical modeling to discover optimal production decisions. In addition we demonstrate the feasibility of simultaneous decision and performance control, providing empirical support for a theoretical extension of traditional variance analysis techniques. The results of three studies are included. In the first we model a situation in which the manager of an aquaculture system must decide when and how many animals to stock initially, how many animals to harvest each period, and when to restock an enclosure in order to maximize contribution. We consider both limited and unlimited growing seasons, solving mixed-integer and linear programs. We examine the effects of technological improvements on production strategies. Consistent improvement in contribution is noted, along with some variation in strategy. In the second study we introduce seasonal variation in revenues and lengthen the growing season. The resulting large-scale real-world mixed-integer problem necessitates the use of a heuristic and two strategies, selective expansion and sieve, in order to achieve a near-optimal solution within a reasonable length of time. In the third study we focus on the uncertainty inherent in the aquaculture environment. We provide empirical evidence of the feasibility of a performance evaluation system which gives explicit consideration to the effects of environmental uncertainty and incorporates intraperiod adaptive behavior on behalf of the individual responsible for implementation of model-specified activities. The system we describe may be used in the simultaneous evaluation of individual and model performances, thus clarifying responsibilities for variances and improving production control.