Pattern recognition residential demand response: An option for critical peak demand reduction

Influencing households to adopt sustainable energy consumption behaviour is important to the transition towards a sustainable energy future. However, if one aims at influencing the energy consumption habits of people, one should also be able to estimate the resulting effects on the entire energy system. Residential demand response to reduce load on the electricity network has largely been impeded by information barriers and a lack of proper understanding of consumers 19 behaviour. What are not well understood and are of great interest include load disaggregation, the behaviour of customers when responding to demand response request, load shifting models and their impact on the load curve of the utility. There is concern among demand response practitioners, for example, that demand response in the residential sector may simply move the peak problem with scale from one point in time to another. However, unavailability of appliance-level demand data makes it difficult to study this problem. In this paper, a generalized statistical model for generating load curves of the individual residential appliances is presented. These data allow one to identify the relative contribution of the different components of the residential load on a given residential feeder. This model has been combined with demand response survey in a neighbourhood with 400 households in Christchurch, New Zealand, to determine the effect of customers 19 behaviour in reducing the neighbourhood 19s winter peak demand. The results show that when customers 19 are given enhanced information, they would voluntarily act to reduce their peak demand by about 10% during the morning peak hours and 11% during the evening peak hours. The demand responsiveness of the individual appliances is also presented. The effectiveness of customer behaviour modification in maintaining system reliability is also presented