Monday August 26, 2019 from 15:30 to 17:10
Application of demand response in cold room: evaluation of deep learning approach for modelling
Mahdjouba AKERMA1, Hong Minh Hoang2, Alan Le Montagner3, Denis Leducq4, Nedra MELLOULI-NAUWYNCK5, Anthony Delahaye 6, Ronia Ben Abdallah4.
1Irstea, Antony, France; 2Irstea, Antony, France; 3Irstea, Antony, France; 4Irstea, Antony, France; 5Université Paris 8, Saint-Denis, France; 6Irstea, Antony, France
The few technologies actually used to store electricity need to be improved while new solutions to overcome the intermittency of the renewable energy need to be developed. Demand response (DR) is one of the levers that can help to balance the electricity grid. Warehouses and cold rooms by their high thermal inertia are promising candidates for applying DR. However, the DR impacts (temperature rise and energy impact) can present a challenge to the DR deployment. These impacts were evaluated in the present work using two approaches for the modelling of a cold room and its refrigeration system:“white box” (based on energy balance) and “black box” (machine learning). An experimental pilot to measure DR effects in a loaded cold room was also developed. The comparison between these two approaches and the experimental results were performed to select a suitable method to be implemented. Several DR scenarios were tested in order to choose the right condition for DR application.