Nested Newsvendor Optimal Commitment Policies in Day-Ahead and Hour-Ahead Electric Capacity Forward Markets

Abstract

This thesis investigates load-purchasing policies from the day-ahead and hour-ahead electric capacity forward markets within the PJM Interconnection through the application of nested newsvendor logic. Newsvendor logic provides insight into an optimal order quantity that will minimize expected costs given a demand distribution and the unit costs of purchasing above or below realized demand. The application of nested newsvendor logic to electric forward markets considers the day-ahead and hourahead forward market load purchasing decisions sequentially, where each forward market is characterized by different unit costs of overage and underage which are determined by the respective market’s electric marginal generation costs and the amount of capacity purchased from the hour-ahead forward market is a function of the day-ahead load purchasing decision. The hour-ahead unit cost of underage is the spot market locational marginal price, which is characterized by unusually high volatility and instantaneous price spikes. The model first is run with aggregated 2010 PJM Interconnection data. Optimal day-ahead orders are shown to be less than average dayahead demand forecasts. This thesis will also formulate a second nested newsvendor model assuming a proportion of total electric load is satisfied through offshore wind generation. Wind power is an intermittent resource and hard to forecast, therefore the proportion of total load purchases satisfied through the day-ahead forward market is shown to increase under higher levels of wind penetration—suggesting integration of high levels of wind penetration may threaten PJM grid reliability.