Optimization with Multiple Performance Criteria. Application to Minimization of Parameter Sensitivlties in a Refinery Model

In complex optimization problems there often exist several noncomparable performance criteria which a designer must consider. This situation is formulated as a vector maximum problem, for which the available theory on proper efficient solutions is outlined. A practical scheme is proposed for problems in which the criteria can be ranked in order of importance. A simplified refinery model is formulated in which it is desired to maximize total yearly profit as well as minimize the sensitivity of the profit to variations in refinery conditions. Detailed nonlinear programming results are presented.