A predictive control algorithm for energy conservation and reduction of room temperature swings is presented. The algorithm is particularly suitable for massive buildings with large solar and other radiant gains and highly variable loads. It utilizes a detailed microcomputer thermal network program to predict the building response to expected weather inputs using several thermostat set-profiles (continuously varying setpoint). It then selects the optimum profile based on three criteria: minimization of the room temperature swing on a sunny day, maximization of total energy savingsl and a peak heating load that does not exceed the capacity of the heater. A case study resulted in two major conclusions: (1) Use simple night setback - square wave set-profile, may lead to a very high increase (40%) in the peak heating load on cold days, as compared to a constant setpoint. (2) Use of smooth set-profiles improves building performance, especially for a sunny cold day; the results showed a decrease of the maximum room temperature by 1°C and energy savings of 18%, while the peak heating demand was increased by only 11% as compared to a constant setpoint.

Units: SI