Demand forecasting methodology to support flexibility estimation

Demand side flexibility will play a significant role in balancing the increasingly decarbonised electricity grid. Heat pumps have a great potential to provide flexibility by using stored heat to shift the demand to times with higher levels of renewable generation. In this paper, a demand forecasting methodology which can be used to support flexibility estimation is presented. Using Gaussian Mixture Models to characterise historical load profiles for the buildings that are being analysed, a small number of recurrent daily profiles can be inferred. The transition probabilities between these characteristic profiles are then used with a Markov Chain Model to predict the daily energy consumption of an air-source heat pump. The methodology can be applied to a single asset or multiple assets within an energy community.