Aggregate Load Forecast with Payback Model of the Electric Water Heaters for a Direct Load Control Program

Abstract

Domestic electric water heaters (DEWH) hold a large share of residential load in North America. The aggregated load profile of electric water heaters follows a similar pattern to the total household load profile, which means that changing the profile of DEWH load can significantly change the shape of the aggregated load profile. To change the load profile, the controller requires an estimation of future load profile and the payback effect of the control action on the forecasted load. This paper presents a load forecast module that uses a Kalman filtered neural network to forecast the aggregated controllable load combined with a statistical payback model to identify the impact of the control action on the load forecast. The proposed method was used by the University of New Brunswick as part of a pilot project named PowerShift Atlantic that aims to provide more than 11MW of ancillary services by controlling more than 1200 controllable loads. The experimental results on the real pilot project shows that the forecast method can be adapted with the dynamic behaviour of the customers. The payback model was also verified by applying various control signals on the pilot project.

Sustainability matchmaking: Linking renewable sources to electric water heating through machine learning

“…The model has been shown to have good estimates of daily hot water usage, but updates are required to provide additional variables relevant to user activity as inputs for the model. In [99], Shaad et al developed a prediction module that uses NN to forecast the aggregated EWHs energy usage in real-time. The mean absolute error rate was recorded in the range of 6.5 kW to 7.8 kW when predicting aggregated data from 95 water heaters.…”

Multi-Objective Demand Side Management at Distribution Network Level in Support of Transmission Network Operation

“…and Z loads (P Z . ), as shown by (3), while at the same time constraining these shares to ensure that demand composition is preserved. The constraints are given by (3a) which preserves the ratio of controllable loads, defined by ∆ and ∆ , before (left hand side of the equation) and after DSM (right hand side), and by (3b) which keeps the disconnected loads within corresponding flexibility limits.…”

“…National Grid (the UK transmission system operator) has set a goal of securing 30-50% of balancing capability from demand side sources, mainly provided by the aggregators, by 2020 [2]. Many pilot sites around the world are investigating the potential of aggregated DR, including a pilot site with aggregated DR from 200 water heaters in the US [3], where the aggregator has to report the load forecast of its controllable load and the payback effect of the control action to the virtual power plant, with the intention to provide more than 11 MW of ancillary services by controlling over 1200 loads.…”