Power systems are experiencing a period of rapid evolution. The previous status quo of large centralised generators operating is being replaced by a paradigm within which sustainability, flexibility and competition are key priorities.
Vertically integrated power utilities have been dismantled and competitive market places have been established to encourage the most effective use of generation and network resources. The push towards sustainability has resulted in the introduction of emission limits, carbon taxes, and most importantly going forward, ambitious renewable energy targets. Under current operating practices, large amounts of expensive and carbon intensive system operating reserves are often required to ensure the security of power supply and this is a particular issue on power systems with high penetrations of uncertain renewable generation.
A number of solutions have been proposed to remedy this situation. Flexible generation resources are typically employed to maintain the system balance, while interconnection between power systems and regions can increase geographical diversity and smooth ﬂuctuations in renewable power output. Electricity storage can also be used to balance periods of over- and under- supply of power. Demand response is a further option that is widely explored in the literature, but to date has had limited widespread usage. Demand response (DR) is regarded as an elegant solution to the issues of uncertain and ﬂuctuating power supply, as the potentially signiﬁcant latent ﬂexibility of electrical demand can be harnessed to provide the required power system services to support renewable power generation. It is important to note that the beneﬁts of demand response for renewable resources are neither the only, nor the primary, driver for demand response.
DR is not a new phenomenon and has been employed in various forms across the globe for decades. The most obvious form of demand response is systematic load shedding, a last resort to avoid system blackout; however more sophisticated approaches have been implemented in a number of power systems.
Time of use (TOU) rates where consumers are subject to expensive tariffs during ﬁxed peak hours, or cheaper rates during night hours, have traditionally been used to incentivise reduced peak consumption, . The objective of TOU rates is to reduce the difference between the peaks and troughs of the demand proﬁle, thereby reducing the need for generator cycling or part-load operation. This allows a more efﬁcient usage of generation, transmission and distribution resources.
Traditional approaches for demand response were adopted due to the predictable and cyclic nature of electricity demand and the dispatch-able nature of generating resources. While this is appropriate in power systems dominated by conventional generation, systems with large penetrations of renewable resources require demand, and the system as a whole, to behave in a ﬂexible manner on a continuous basis.
Crucially, this will allow the optimal usage of the renewable resource and ensure that the system balance is maintained. The concept of continuous demand response, and in particular the use of price signals to elicit this response, was proposed as far back as 1988 in the seminal work of Schweppe et al. on spot pricing of electricity.
Response and price signals
In this work it was proposed that price signals at a resolution of ﬁve minutes could be used to maximise the economic efﬁciency of the power system, revealing the true cost of electricity provision to consumers and thereby providing an economic signal to maintain the system balance. The use of price signals to this effect is termed indirect load control. At a time resolution exceeding ﬁve minutes, it was deemed that direct load control was required to ensure the stability of the system. This view is shared by Callaway and Hiskens, who prefer the use of direct control for all ancillary services as the system operator has greater certainty when demand is controlled directly rather than indirectly through a price signal where the price response must be predicted.
Under indirect control, the aggregator has limited information about the demand that is being controlled, and must estimate the price response of its demand portfolio. Prices are then issued to induce an expected response. Prices can be geographically varying, up to the resolution of information available to the aggregator, which may be at the level of several hundreds or thousands households. Direct control involves direct communication with individual appliances, and detailed information on their interactions with the surrounding environment.
This is more computationally and communicationally intensive, but allows a more precise response and individual control set-points can be sent to each appliance, facilitating control of DR at the highest possible geographic resolution.
Beneﬁts of demand response
The beneﬁts of demand response are widely lauded; advances in modelling and IT capabilities have made demand response an attractive option to increase power system ﬂexibility. This will consequently allow a more efﬁcient use of system assets and resources.
The ﬂexibility provided by demand response can be used to meet the ﬂuctuations of renewable generation and facilitate a higher penetration than could be achieved by relying on conventional generation alone. Although the energy cost of renewable resources, for example wind generation, is typically quite low, the associated system costs can be substantial
Operating costs are increased as both online (spinning) and quick start (standing) reserve generation is required to manage the frequent and often extreme ﬂuctuations in the wind power output. has been highlighted as a mechanism to facilitate higher penetrations of wind generation, while also reducing the system cost of its integration.
Traditionally variability and uncertainty from wind generation has been managed through a combination of ramping and part-load operation of conventional generating plant, interconnection to neighbouring regions, and storage.
While DR has the potential to bring about a great number of beneﬁts, there are a number of challenges that must be overcome before it can be considered as a valuable contribution to the power system.
The overriding issue is the lack of experience and understanding of the nature of demand response. Too much of the work in this ﬁeld is based upon simplistic models with superﬁcial results. At this crucial stage in the development of demand response it is imperative that a clear and concrete understanding of demand response is established, so that a realistic evaluation of its suitability for the provision of system services can be determined.
Demand is clearly a highly diverse and complex resource, varying according to a multitude of external factors. Despite the limited understanding of the nature of demand response, particularly at the system level where the response of demand from many different sectors and applications is aggregated, it is clear that the resource is highly diverse, so using a single model type to represent all demand is unrealistic.
However, the ability of demand response to support power system stability, to leverage the existing assets, to offset expensive network upgrades and to support the continued push to decarbonise generation is clear.
When evaluating demand response, it is imperative that it is considered in the context of the entire energy system. Demand response alone may offer certain beneﬁts, however when the interaction with other system components is considered demand response may become a very attractive option
Going forward the ability of network users – from the largest generators, to the smallest households – to provide flexibility to the network will be cruical