Payment minimization auction with demand bids and partial compensation of startup costs for deregulated electricity markets

Deregulated electricity markets use an auction mechanics to select supply offers and demand bids from market participants, and set market clearing prices (IMCPs) to charge and pay for energy and ancillary services. It has been pointed out that currently the cost minimized by ISO's is not the cost actually paid by consumers. A problem that directly minimizes consumers' payment cost under the simplifying assumptions of given system demand and full compensation of startup costs was solved by using a novel generalized surrogate optimization framework. Building on this paper presents a payment cost minimization problem with demand bids and offers compensated for startup costs by the excess of offer values over energy market values in a day. A new formulation is developed to properly consider the role of demand bids in market settlement, and model offers' compensations. Building on the augmented Lagrangian relaxation approach and the generalized surrogate optimization framework, a method is developed to solve this new problem by forming and solving subproblems for individual offers and demand bids by using the generalized surrogate optimization framework. Since different cost structures of the Lagrangian are caused not only by different combinations of selected offers, but also by different combinations of selected demand bids, there is complexity in solving subproblems. Our idea is to solve demand bids independent of offer subproblems for a given set of multipliers. With this, the complexity of individual subproblems is not expanded as compared to the payment cost minimization problem with fixed demand. Meanwhile, difficulty in solving offer subproblems caused by the non-time-additive compensations is overcome by redefining offers' compensations to be independent variables subject to linear inequality constraints, and relaxing these constraints by using multipliers. Numerical testing shows that this method is effective, and near optimal solution can be obtained for medium-sized problems.