Residential Electricity Demand: Peaks, Sequences of Activities and Markov chains (REDPeAk)

点击翻译按钮获取中文摘要

Peak electricity demand is becoming an increasingly significant problem for UK networks as it causes imbalances between demand and supply with negative impacts on system costs and the environment. The residential sector is responsible for about one third of overall electricity demand (DECC, 2013). During peak demand, electricity prices in wholesale markets could fluctuate from less than 0.04 Euros/kWh to as much as 0.35 Euros/kWh (Torriti, 2015). In the future the peak problem is expected to worsen due to the integration of intermittent renewables in the supply mix as well as high penetration of electric vehicles and electric heat pumps. Understanding what constitutes peaks and identifying areas of effective load shifting intervention becomes vital to the balancing of demand and supply of electricity. Whilst there is information about the aggregate level of consumption of electricity, little is known about residential peak demand and what levels of flexibility might be available. REDPeak will fill this gap.The overall aim of REDPeak is to analyse the variation in sequences of activities taking place at times of peak electricity demand with a view to identify clusters of users which might provide flexibility for peak shifting intervention. The project will analyse 10-minute resolution time use activity data from the UK Office for National Statistics Time Use Survey with a view to derive information about occupancy and synchronisation of activities. Markov chains will be used to model load profiles in combination with appliance-specific parameter data. Since Markov chains have proven effective at generating electricity load profiles except for peak times, REDPeak will develop Hybrid Monte Carlo modelling to account for demand moving in larger steps during peak periods. Sequence analysis will be used to mine activities at periods of peak electricity demand. REDPeak will cluster respondents according to sequences of activities and analyse to what extent appliance-specific control variables explain activities at specific times of the day. Three datasets will be used for direct validation between metered data and time use data. Findings on sequence analysis will feed into algorithms for automated demand management or Demand Side Response.