Data driven feedback for optimized and efficient usage of decentralized air conditioners

In the context of developing countries, buildings account for around one-third of aggregate energy consumption with decentralized air conditioners (AC) being the major contributor. The possibility of room-level control, together with buildings substandard thermal insulation make decentralized ACs, an attractive target for energy conservation. Our overall objective is to provide targeted feedback for optimized and efficient use of ACs by the occupant without affecting their thermal comfort. In our first work, we developed a system PACMAN for nonintrusive (using ambient temperature information) prediction (prior-usage) and estimation (post-usage) of AC energy consumption. We adapted a thermal model from existing literature and validated our approach using the data collected in an in-situ study conducted across seven homes in Delhi (India). The dataset contained around 2200 hours of usage data from the different ACs, room types, and thermostat temperatures. We achieved an average accuracy of 85% and 83% with the best accuracy of 97% and 93% for the estimation and prediction of AC energy consumption respectively, across all homes. During the study, we realized that easy-to-collect ambient information such as temperature (using various pervasive and ubiquitous devices such as smartphones) can generate actionable feedback for the occupants. We also observed that human activity, room structure, and several such factors impact the performance of PACMAN. Thus, we next performed controlled experiments to understand the effect of these factors on PACMAN and thus enhance our thermal model. One of the main conclusions was that sensor position plays a crucial role in the optimal control of AC. Also, windows exposed to the environment outside the room makes the most significant impact on AC energy consumption. In our ongoing work, we focus on finding the optimal position for the thermostat to reduce AC energy consumption without affecting occupant's thermal comfort. In future, we plan to propose a framework to detect anomalies in AC within time, as delay in that might lead to appliance failure or inefficient outcome.