## What Objective Function Should be Used in ISOs' Day-ahead Markets?

#### What Objective Function Should be Used in ISOs' Day-ahead Markets?

October 7, 2013 - 11:10 amWinston Chung Hall, 205/206

Abstract

In deregulated U.S. electricity markets (e.g., the day-ahead market), Independent System Operators (ISO’s) currently use an auction mechanism that minimizes total supply bid costs to select supply bids and their levels for energy and ancillary services. This “Bid Cost Minimization” problem is NP-hard due to its combinatorial nature, but because of its separability, it can be effectively solved by using the Lagrangian relaxation technique or other mixed-integer programming methods for near-optimal solutions. Furthermore with given demand, existing unit commitment and economic dispatch software can be readily adapted to solve the problem by replacing units with supply bids. After the auction problem is solved, markets are then settled where payments are calculated based on uniform Market Clearing Prices (MCPs) or congestion-dependent Locational Marginal Prices (LMPs). The above auction and settlement mechanisms are inconsistent and consumer payments could be significantly higher than the minimized total supply bid cost. This gives rise to “Payment Cost Minimization,” an alternative auction mechanism that minimizes consumer payments. Illustrative examples have shown that with the same set of supply bids, payment cost minimization leads to reduced consumer payments as compared to bid cost minimization. This observation has led to serious discussions among stakeholders of electricity markets as to which of the two auction mechanisms is more appropriate. However, while methods for bid cost minimization abound, very limited approaches have been reported for payment cost minimization. The difficulties of the latter result from its distinct feature that market-clearing prices explicitly appear in the objective function as decision variables in contrast to bid cost minimization where market prices are not involved but determined afterwards in an ex post manner. Also, market-clearing prices couple with bids in a very subtle manner, causing problem inseparability and thus the failure of traditional decomposition techniques including Lagrange relaxation.

In this presentation, a brief history about the debate on auction objective functions in ISO markets is first presented, and issues and concerns on the two auction objectives are highlighted. Then our recent progress on solving payment cost minimization problems under various market conditions are summarized. Our key idea is to use “surrogate optimization” to overcome the difficulties caused by the problem inseparability of payment cost minimization. With surrogate optimization, a relaxed problem does not need to be solved optimally as required by the traditional subgradient method. Rather, an approximate solution to the relaxed problem is sufficient if the “surrogate optimization condition” is satisfied. The relaxed problem can thus be optimized with respect to a particular bid one at a time until the condition is satisfied. This approach has been used to solve a payment cost minimization problem for an energy market with uniform MCP’s, and extended to solve problem under various market conditions (e.g., partial compensation of startup costs, demand bids, and transmission capacity constraints, etc). Also, strategic behaviors of market participants under bid cost minimization and payment cost minimization are investigated by using a game theoretic framework. Lastly, several questions and concerns regarding payment cost minimization are discussed.

Biography

Joseph Yan is the Principle Manager of Fundamental Modeling & Analysis at Southern California Edison (SCE). For the past decade, he has led the strategy development of SCE’s generation portfolios to optimize the value of these resources and reduce the cost of serving its customers. He has also actively engaged in California electricity market stakeholder processes representing SCE and made extraordinary contributions to the development of the ISO markets. His research interest includes operation research, optimization, unit commitment/scheduling and transaction evaluation, and optimal simultaneous auction in deregulated ISO/RTO markets. Joseph Yan holds a Ph.D. in Electrical and Systems Engineering of the University of Connecticut.