Towards Energy Efficiency in Schools : The Case of Qatar

The rapid growth of cities and municipal areas has played a major role in impacting the global environmental status and embarked on a significant increase in energy consumption. Buildings account for 40% of the total energy consumption globally. Qatar is a prime example of a country where extreme climate imposes unique challenges when managing energy consumption with temperatures in summer months can reach an average maximum of 40 C°. In fact, air conditioning, due to the region's extreme heat, represents 70% of the annual peak electricity consumption in the Gulf Cooperation Council (GCC) countries. Energy monitoring and the understanding of consumption patterns are crucial to managing energy resources and controlling the environmental impact. Hence, establishing adequate tools to generate practical recommendation reports is of immense value as it supports conscious and viable practices for energy management systems. Machine Learning algorithms and Neural Network models are utilized to perform cooling power disaggregation and power consumption forecasting for educational institutions is an example. According to the study findings based on Qatari schools’ datasets, we observe a significant impact of outdoor temperatures on school power consumption and a substantial reliance on cooling systems, which account for about 50% of total annual energy use.

Occupancy levels also affect energy usage, with slightly higher consumption during active school hours. The Random Forest model effectively disaggregated cooling power within schools, achieving an average R-squared value of 0.93. Transfer learning further demonstrated adaptability across multiple schools, with an average R-squared value of 0.79. Lastly, power forecasting using LSTM models showed promising accuracy, with an average R-squared value of 0.87, highlighting the potential of these technologies in enhancing energy management and sustainability in educational settings.