Characterizing capacity of controllable DERs to provide energy storage service to the power grid

The aim of this project is to develop methods to estimate the capacity of distributed energy resources to provide demand side services to the power grid, in particular, Virtual Energy Storage (VES) from a collection of flexible loads. Virtual energy storage refers to small controlled changes in the electrical power consumption of loads, such as air conditioners, so that the resulting variation appears like the charging and discharging of a battery to the rest of the power grid.

Virtual energy storage is expected to be cheaper than real batteries since no new hardware is needed; the only change is software. A bottleneck in deploying this technology has been the lack of tools to answer questions about the equivalent battery capacity of large collections of loads, such as, "How many smart residential air conditioners do we need to provide the same service as a 5MW/3MWh Li-ion battery?".

The project will develop mathematical techniques to answer such questions and more, and thereby enable grid operation and planning with smart loads of the future. The project will thus contribute to increasing the reliability and resiliency of the nation’s electricity supply. An intellectual merit of the proposed methods is that they are "future proof" since they are independent of the algorithms used to coordination the loads.

Another intellectual merit is the data-driven methods proposed, which do not need simple reduced order models of DERs that are difficult to obtain. A broader impact of the project is to help grid operators adopt smart load technology by providing computational methods for operation and planning. Other broader impacts include enhancements to the curriculum in the PI's institution with research results and attracting underrepresented and minority undergraduates to STEM research.

Because each load must satisfy the consumer's local quality of service in providing grid support, the capacity of a large collection of loads to provide VES has a complex dependency on these local quality of service constraints. Apart from MW/MWh numbers, grid operators also need to determine what kind of demand deviation is feasible for a collection of loads so that no one has to violate its local constraints.

For instance, the indoor temperature should not become too hot or cold when an air conditioner is providing demand side service. These challenges are addressed in two distinct time scales: short term planning (in hours) and long-term planning (in months). For short term planning, a deterministic framework is proposed in which feasible demand trajectories are described via a set of low dimensional constraints.

A key innovation is to reduce high-dimensional integer constraints due to lock-out constraints at loads into low-dimensional convex constraints for the collection through ensemble averages. For long term planning, a statistical framework based on spectral densities is proposed, which enables characterizing all possible samples paths of feasible demand deviation signals.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.