A comparative multiobjective programming study (Q1108189)

Multiobjective programming seeks to aid decision making in complex problems where it is difficult to explicitly state decision-maker (DM) utility. A variety of techniques have been presented, seeming to trace a movement from a direct nonlinear programming approach to more indirect methods of searching for higher utility. Experiments have been conducted on small problems comparing a number of these approaches usually employing specialized computer codes. This study presents a larger decision problem involving an oil pipeline operations planning model involving three objectives. Basic concepts of multiple objective programming are applied, utilizing only a standard linear programming code. A nonlinear utility function is assumed, and a comparative study of how the major multiobjective programming techniques perform is reported. Evaluation factors include accuracy in reflecting the nonlinear utility function, enhancement of DM learning, and burden upon the DM.