Optimal Operation of Thermostatically Controlled Loads in Residential Buildings under Uncertainty: From a Building to an Electricity System Perspective

Advanced control strategies, and model predictive control (MPC) in particular, are gaining widespread attention for building climate control, since they can systematically save energy and/or costs with simultaneous thermal comfort improvement, as well as adapt the energy demand in correspondence with the available (renewable/residual) supply via demand response. The performance of any MPC strategy is dependent on the accuracy of the mathematical model describing the thermal loads and on the quality of the forecasts of disturbances, such as weather and occupant behavior. Deviating model parameters and inaccurate disturbance predictions are shown to give rise to increased energy costs and comfort violations if not properly accounted for, and require real-time corrective actions, thereby jeopardizing the participation in possible demand response programs. In contrast to the additive forecast uncertainty, the uncertainty on the building model parameters is typically not explicitly taken into account in MPC applications. Nevertheless, the building controller model is not always capable of capturing the building dynamics in detail, due to the unavailability of sufficient data and/or experts' knowledge to set up the model, and/or due to the impossibility to correctly describe the complexity of the underlying physics. In that case, the parametric uncertainty of the controller model can become non-negligible, and hence, should be accounted for. Therefore, the main goal of this dissertation is to develop and assess a stochastic model predictive control (SMPC) strategy for building climate control and demand response under combined additive (disturbance forecast) and parametric (model) uncertainty (more specifically referred to as SMPC-ap). The presented approach is tailored to the class of linear time-invariant systems represented by a state space model. Analytically reformulated chance constraints are combined with affine disturbance feedback (ADF) to reduce conservatism. The SMPC development consists of two important parts. First, starting from the conventional deterministic optimal control problem, a convex stochastic problem formulation is derived, explicitly accounting for additive as well as parametric uncertainties. Second, an appropriate mathematical model of all relevant uncertainties is obtained, serving as an essential input to the stochastic optimal control problem. Here, an important additional contribution, in particular, is the derived probabilistic description of the parameters of a physics-based building controller model. The thermal characteristics of the building envelope, and the derived controller model parameters, are determined based on sparse, publicly available data via the probabilistic characterization method of De Jaeger et al., without intensive on-site data collection, which is fundamentally different from what is done in current research. To fully assess the potential added value of the proposed SMPC-ap strategy, its impact is examined at building level, as well as at electricity system level. At building level, the advantages of the SMPC-ap approach are investigated for two different application domains, being optimal control and integrated optimal control and design. Regarding optimal control, the main focus is on the attainable thermal comfort improvement by hedging against uncertainty, and on the associated energy costs. The enhanced uncertainty anticipation of the SMPC-ap approach is shown to induce improved thermal comfort in closed-loop simulations compared to the conventional deterministic MPC (DMPC) strategy and the state-of-the-art SMPC-a strategy only accounting for additive uncertainties. These gains are most prominent in buildings equipped with floor heating (representing large thermal inertia) and characterized by the combination of a large model uncertainty and a large nominal heat demand, and this at the expense of limited increases in energy use. For all considered cases, irrespective of the installed heat emission system, 90% of the thermal comfort improvement relative to the DMPC strategy can be realized with a relative increase of at most 9% in energy use. Regarding optimal design, the suitability of the SMPC-ap approach for obtaining a more appropriate, yet robust, heat supply system size is illustrated, by incorporating the SMPC-ap strategy in an integrated optimal control and design approach. Capacity reductions of 3-5 kW are shown to be achievable in an individual building for a heating system initially sized at 15 kW without increasing thermal discomfort compared to an analogous approach incorporating a DMPC strategy. At electricity system level, the focus is on the impact of the stochastic control strategy on the resulting demand, and on how this demand of a group of buildings can be coordinated via demand response to lower the system operating cost. Due to the incorporation of ADF in the open-loop control problem, the SMPC-ap strategy is able to simultaneously optimally schedule the demand for electrical energy, reserve capacity and real-time flexibility, required to guarantee thermal comfort. This discloses very valuable information for an aggregator or system operator, since the load uncertainty can be revealed and controlled ahead of real time. It is demonstrated that the day-ahead coordination of the demand for reserve capacity in addition to the energy demand is able to reduce the system operating cost, and hence, enables a more cost-efficient electrification of the residential heating sector. Cost reductions up to 10.7% are shown to be achievable for a demand side consisting of 900 000 flexible heat pumps combined with low-temperature radiators. These insights support the discussion on the need for flexibility markets for low-voltage/residential consumers, and demonstrate the added value of implementing the proposed SMPC-ap strategy for demand response under uncertainty in this context.

Jaar van publicatie:2022

Toegankelijkheid:Open