Multiobjective simulation and optimization of an occupied university building

In the past decade, building simulation and optimization techniques have gained momentum in the modeling of energy performance due to economic and environmental pressure to make facilities more efficient. The complexity of their multivariate non-linear systems have posed challenges for modeling and performance optimization. Building simulation tools have evolved over the years, increasing their capabilities to handle advanced approaches that integrate multiple aspects. Currently, these tools are widely accepted for building energy assessments with eQUEST being the most recognized simulation program in use. It has been validated in the public domain through its long history. Computational Intelligence (CI) techniques have been an emerging area of study providing powerful tools for predicting and optimizing complex systems. These techniques are concerned with the discovery of structures in data and recognition of patterns. They also embrace nature-inspired paradigms such as evolutionary computation and particle swarm intelligence. The recent advances in information technology and systems have enabled collection and processing of larger volumes of data. The contribution of the research reported in this thesis is the implementation of a hybrid model using eQUEST and particle swarm intelligence to replicate the current baseline energy performance of a fully occupied university facility and to optimize it in order to generate potential energy savings while maintaining comfortable indoor temperature. Another major contribution is demonstrating the accuracy of a hybrid energy savings model accomplished by modifying discharge air temperature and supply fan static pressure set point of the air handling unit.