Leveraging the potential of load aggregators in residential clusters: A Model Predictive Control (MPC) framework for the Toronto metropolitan area

To fully achieve the potential of grid-interactive efficient buildings, virtual communities are essential to reduce the pressure on the grid and the carbon footprint of residential sector by enabling load aggregation. For instance, the coordinated control of a cluster of hundreds or thousands of residential buildings can provide the grid operator with additional energy flexibility. Yet, when multiple load aggregators are connected to the same substation, there is a risk that using rigid tariff structures for load management may result in a “stiff” peak shifting, thus defeating the ultimate purpose of flexibility enhancement. Therefore, with the increasing penetration of aggregators, determining an appropriate pricing structure is essential to steer the load management of substations managing residential customers. This paper proposes a methodology for deploying demand-side flexibility at the aggregated level, considering a variable number of households and aggregators in competition. The building models consist of reduced-order resistance-capacitance (RC) thermal networks. Smart thermostat data and ambient temperature and heating power measurements are used for model calibration purposes. Model Predictive Control is used to produce forecasting and optimal strategies up to one day ahead for different pricing structures: (a) static time of use; (b) dynamic time of use, and (c) dynamic pricing. As a proof of concept, the methodology is applied to thousands of homes in the metropolitan area of Toronto, Ontario. The load profile analysis at the aggregated level describes the grid stability to overperform by almost 100% of the reference performances by deploying medium-to-high participation in DSM. The optimal aggregation procedure suggests that the portfolio's symmetry brings the highest benefit and that a three-tier tariff impacts more than two-tiers in load flattening.