 |
Center for Nonlinear Studies |
We consider the management of electric vehicle (EV) loads within a market-based Electric Power System Control Area. EV load management achieves cost savings in both (i) EV battery charging and (ii) the provision of additional regulation service required by wind farm expansion. More specifically we develop a decision support method for an EV Load Aggregator/Energy Service Company (ESCo) that controls EV battery charging. At the beginning of each period in a 24 hour cycle, the ESCo purchases firm energy from the real-time wholesale market and bids for a non-firm block of energy which the ESCo commits to allow the Independent System Operator (ISO) to schedule up or down for regulation service over 5 second intervals. The ESCO’s regulation service bid may or may not be accepted depending on the clearing of the real time regulation service market. The ESCo is also assumed to have access to information about local distribution network congestion constraints, namely the maximal additional load that may be applied along a specific low voltage distribution network feeder without stressing transformer and other distribution hardware tolerances. This retail-transactions-market information is employed together with wholesale market information on expected wind farm generation and clearing prices to make optimal feasible decisions regarding the quantity and bid prices for firm and non-firm energy nominations. Note that the quantity committed to regulation service must be backed by additional battery charging capacity allowing response to an up regulation service command, and, moreover, by setting aside sufficient unused feeder capacity so as to accommodate a commensurate increase in consumption that may be requested by the ISO, even momentarily, in conjunction with the regulation service bid. Wind farm generation forecasts affect the clearing price and the demand for regulation service. This is crucial to the ESCo’s decision on how much non-firm regulation service capacity and at what price it is profitable to bid for. A hierarchical decision making methodology is proposed for hedging in the day-ahead market and for playing the real-time market in a manner that yields regulation service revenues and allows for negotiated discounts on the use-of-distribution-network payments. The proposed methodology employs a rolling horizon look-ahead stochastic dynamic programming algorithm solved approximately by linear programming. Its implementation and the observed numerical/computational experience are also reported.
Host: Misha Chertokov, T-4