(679a) The Impact of Microgrids on the Grid: Integration of Distributed Renewables Via Network-Constrained Affine Robust Unit Commitment

Distributed generation (DG) of electrical power is a promising alternative to the current paradigm of centralized generation followed by unidirectional distribution to the end user. It has the potential to reduce transmission congestion and losses while also providing resiliency to infrastructure disruptions. Additionally, local sources of renewable energy, such as wind and solar, can be incorporated to reduce emissions. Microgrids are an attractive approach to the technological realization of DG because they can use locally controllable resources such as energy storage systems (ESS) and fast-responding microturbines to mitigate the inherent intermittency and stochasticity of renewables. However, the current high cost of ESS technology means that standalone renewable DG is not yet economically viable. Thus, grid connection is required if widespread adoption of renewable DG is to be achieved and this results in increasing uncertainty and temporal variability of the residual load (demand minus local generation) which must be served by conventional generation.

In order to operate with grid-connected renewables in a way that is both reliable and economically optimal, the grid controlling entity (e.g. independent system operator) needs to determine optimal day-ahead unit commitments (UC) in the presence of uncertainty. A common approach is two- or multi-stage robust UC [1]. When complete recourse is allowed for wait-and-see decisions, network-constrained problems are computationally intensive without the use of approximations. However, accounting for these network constraints is important for capturing the coupled effects of renewable uncertainty and variability.

In view of this, we propose in this work a two-stage, network-constrained multi-objective unit commitment model based on affine robust optimization. The objectives are to minimize grid operating cost and emissions; the relative weighting of these is determined a-priori by a user-decided parameter. We consider uncertainty in residual load forecasts, that is, the difference between conventional load and local renewable generation. Restricting the recourse decisions to be affine with respect to this uncertainty allows for computationally efficient and heuristic-free solution of the model.

In addition to our UC model, we propose a grid configuration which corresponds to a truly distributed infrastructure. We include renewables-only microgrids at every bus in the transmission network, a departure from the common approach of considering larger capacity wind and/or solar farms at only one or a small number (compared to the total number of buses) of locations. The design of such a distributed system allows the controllable nature of microgrids to be fully exploited because local resources can be used to shape the residual load and dampen its uncertainty (see [2] for an example of a market structure which promotes this type of residual load regulation).

We solve the unit proposed commitment model to determine optimal operating cost and emissions for a grid represented by a version of the IEEE One Area Reliability Test System which has been modified to include (i) connection to an electricity market and (ii) microgrids at each bus which make up 15% to 25% of annual load through solar generation. We find that incorporating distributed generation which tracks solar availability results in higher cost and emissions than a grid operating without renewables, except at low penetrations. We also investigate limiting the temporal variability and uncertainty of the residual load in a manner inspired by the energy exchange regulations proposed in [2]. This type of regulation provides increasingly superior economic and environmental performance compared to unregulated distributed generation as penetration of renewables increases, as well as lower operating costs and similar emissions to a present-day grid at all levels of renewable incorporation.

[1]	M. Tahanan, W. V. Ackooij, A. Frangioni and F. Lacalandra, "Larges-scale unit commitment under uncertainty," 4OR, vol. 13, no. 2, pp. 115-171, 2015.
[2]	M. Zachar and P. Daoutidis, "Microgrid/macrogrid energy exchange: A novel market structure and stochastic scheduling," IEEE Transactions on Smart Grid, vol. 8, no. 1, pp. 178-189, 2017.