Minimizing energy consumption in a real-life classroom assignment problem

This work addresses a classroom assignment problem (CAP) in the context of a large-scale Brazilian federal educational institution. In practice, such problem must be solved at the beginning of every term. Currently, the CAP arising in the referred institution is solved manually, which is not only an arduous task, but also very time-consuming, often leading to inefficient solutions. By analyzing the manual solution from an energetic perspective, one can verify that there are potential losses. For example, it is not desirable to assign classes with few students to rooms with large capacities, which in turn tend to have higher energy costs. The objective of this study is to minimize the energy consumption associated with the usage of the locations where lectures can take place, while meeting the requirements specified by the institution. To solve different versions of the problem, several scenarios were suggested and solved by a mathematical formulation of the problem based on integer linear programming. The model developed was tested on instances involving up to 3046 classes and 97 locations. All of the proposed scenarios were capable of achieving a significant reduction in energy consumption compared to the manual solution, with up to 26% of energy savings.