Predictive optimal control of fabric thermal storage systems

Hollow core ventilated slab systems provide an effective means of utilizing the building structure as a thermal store. The optimum control strategy for the system would be one that minimizes energy costs without prejudicing the occupant thermal comfort. This paper describes an implementation of one such strategy in which the optimum operation of the plant is predicted for the next day. The controller incorporates algorithms for predicting the ambient temperature and solar radiation over the next 24 hours. The predicted ambient conditions are used with a thermal network model of the building to evaluate the energy consumption and comfort conditions for a given control strategy, the optimum control strategy being found by exercising the thermal model with a Genetic Algorithm search method. The structure of the controller is described, together with algorithms for predicting the next days ambient temperature and solar radiation. The seasonal operation and performance of the predictive controller is also presented. The paper indicates that in relation to a conventional control strategy, occupant thermal comfort can be maintained throughout the year without excessive use of energy.