Operations Research Seminar: Timothy Duglas Mount

The Role of Distributed Storage in Electricity Markets in the USA: Research versus Practice

Timothy Douglas Mount, Cornell University

This presentation will describe analyses using a stochastic form of multi-period Security Constrained Optimal Power Flow (the MATPOWER Optimal Scheduling Tool (MOST)) that determines the optimum dispatch and reserves for the next 24 hours for power plants on a test network.  Given this stochastic framework, the model determines how distributed energy storage (e.g. deferrable demand) can be managed by aggregators to minimize the expected cost of their purchases from the grid, deliver the energy services needed to their customers, and still provide ramping services to the grid to mitigate the variability of wind generation.  The two additional requirements for making this type of two-sided market viable are to use 1) stochastic forecasts of the hourly price of energy to determine an optimal bidding strategy with threshold prices for charging and discharging the storage, and 2) a receding-horizon optimization that allows aggregators to modify their bids and ensure that the physical constraints on storage capacity are not violated. 

Nevertheless, there are still problems for managing distributed storage on the annual peak-load days because the typical operating criterion used by system operators minimizes the expected operating costs, and therefore, it ignores the potential savings in the capital costs of installed generating capacity associated with shifting load and reducing the system peak.   Consequently, some form of Critical-Peak Pricing (CPP) is needed for long-run economic efficiency to ensure that storage is used primarily to shift load away from the peak on peak-load days rather than to provide ramping services.

Given these conclusions about how to manage distributed storage efficiently, the limitations of four major policy practices in the USA will be discussed.  These are 1) net-metering, 2) using administered prices for distributed resources, 3) markets for ramping (flexibility), and 4) extending nodal pricing and trading onto distribution systems.  These practices are all substantially more complicated than our proposal for a two-sided market, with aggregators purchasing energy for their customers from the wholesale energy market, and in addition, requiring that these aggregators maintain a stable power factor by installing local equipment (e.g. smart invertors) to deal with local voltage problems.  This simple market for aggregators is consistent with the standard way that many wholesale customers currently operate.  The overall conclusion is that regulators should focus more on designing markets that encourage load to follow the real-time supply of renewable generation, and replace the long-established planning practice of treating load as an exogenous input and installing enough generating capacity to meet the annual peak system load.