Multiple Attribute Evaluation in Engineering Decision Support Using Limited Compensation and Reference Designs

This paper investigates multiple attribute design evaluation in computer-aided engineering decision support. A new method is developed, which is partly based on insights gained from existing methods. The new method can take into account limited compensation among multiple attributes, makes full use of raw evaluation data and provides a systematic computational procedure that is convenient to use in a computerised decision support system. This method is based on an extended concordance analysis and a modified discordance analysis for initial information processing. A preference matrix is thus constructed which forms a basis for subsequent ranking of alternative designs. The relative distance of a feasible design to given reference designs is employed as a measure for alternative ranking. Two engineering evaluation problems are examined to demonstrate the application and potential of the method.