Auction implementation problems using Lagrangian relaxation

Abstract

Lagrangian relaxation (LR) is used as an auction method for bidding in a deregulated environment. This paper focuses on the problems that occur while using LR as an auction method. Identical or similar units can prevent LR from finding the optimal solution. If the units are similar, LR may have trouble selecting some of them for the optimal solution. A feasible solution may thus not be found. Since a heuristic selection of which unit data to alter has to be applied, this is inequitable to the unit(s) not selected and may result in lawsuit(s). Because the dispatcher has to use heuristic selection, there is no "fair" solution to these problems. The authors suggest previously published methods to implement auctions. Sensitivity analysis has also been applied in this paper to show that the percentage difference of similarity between units may affect the solution found.