Demand response from thermostatically controlled loads: modelling, control and system-level value

Abstract

The research area of this thesis concerns the efficient and secure operation of the future low-carbon power system, where alternative sources of control and flexibility will progressively replace the traditional providers of ancillary services i.e. conventional generators. Various options are engaged in this challenge and suit the innovative concept of Smart Grid. Specifically, this thesis investigates the potential of demand side response support by means of thermostatically controlled loads (TCLs). This thesis aims to quantify the impact that a population of thermostatically controlled loads has on the commitment and dispatch of a future power system characterized by a large penetration of renewable energy sources (e.g. wind) that are variable and intermittent. Thanks to their relative insensitivity to temperature fluctuations, thermostatic loads would be able to provide frequency response services and other forms of system services, such as energy arbitrage and congestion relief. These actions in turn enhance the power system operation and support the strict compliance with system security standards. However, the achievement of this transition requires addressing two challenges. The first deals with the design of accurate device models. Significant differences affect the devices’ design included in the same class, leading to different system-level performances. In addition, the flexibility associated to TCLs would be handled more easily by means of models that describes the TCLs dynamics directly as a cluster rather than considering the appliances individually. Second, it is not straightforward achieving satisfactory controllability of a cluster of TCLs for the considered applications. The complexity lies in the typical operation of these devices that has only two power states (on and off) whereas the desired response is continuous. Moreover the control strategy has always to comply with strict device-level temperature constraints as the provision of ancillary services cannot affect the quality of the service of the primary function of TCLs. This thesis addresses the challenges exhibited. Detailed thermal dynamic models are derived for eight classes of domestic and commercial refrigeration units. In addition, a heterogeneous population of TCLs is modelled as a leaky storage unit; this unit describes the aggregate flexibility of a large population of TCLs as a single storage unit incorporating the devices’ physical thermal models and their operational temperature limits. The control problem is solved by means of an initial hybrid controller for frequency response purposes that is afterwards replaced by an advanced controller for various applications. Provided these two elements, a novel demand side response model is designed considering the simultaneous provision of a number of system services and taking into account the effect of the load energy recovery. The model, included in a stochastic scheduling routine, quantifies the system-level operational cost and wind curtailment savings enabled by the TCLs support.