Forecasting household consumer electricity load profiles with a combined physical and behavioral approach

In this paper, a simulation model that forecasts electricity load profiles for a population of Swedish households living in detached houses is presented. The model is constructed of three separate modules, namely appliance usage, Domestic Hot Water (DHW) consumption and space heating. The appliance and DHW modules are based on non-homogenous Markov chains, where household members move between different states with certain probabilities over the days. The behavior of individuals is linked to various energy demanding activities at home. The space heating module is built on thermodynamical aspects of the buildings, weather dynamics, and the heat loss output from the aforementioned modules. Subsequently, a use case for a neighborhood of detached houses in Sweden is simulated using a Monte Carlo approach. For the use case, a number of justified assumptions and parameter estimations are made. The simulations results for the Swedish use case show that the model can produce realistic power demand profiles. The simulated profile coincides especially well with the measured consumption during the summer time, which confirms that the appliance and DHW modules are reliable. The deviations increase for some periods in the winter period due to, e.g. unforeseen end-user behavior during occasions of extreme electricity prices.