Comparing the impact of different thermal comfort constraints on a model-assisted control design process

In the design of supervisory controllers for managing energy in buildings, model-based control design approaches have recently attracted significant attention. The control-design problem in these cases is typically posed as a constrained minimization problem: given a simulation model acting as a surrogate of the building, identify a controller that minimizes a cost function, say energy, subject to the constraint that thermal comfort stays within acceptable levels. The use of a thermal comfort model can be the means for estimating comfort so that the mathematical programming problem can be formulated. In the present paper, we investigate how the choice of thermal comfort model affects the quality of the resulting controller. We consider a building simulated in EnergyPlus and design, under the same conditions, controllers using three different thermal comfort models: the model of Fanger, the two-node Pierce model, and the KSU two-node model. A comparative study is performed to draw conclusions upon the effects that this selection has with respect to the performance of the resulting controller.