Demand -limiting control using building thermal mass in commercial buildings

Kyoungho Lee, Purdue University

Abstract

Peak cooling loads for buildings during critical demand periods can be reduced significantly using heat capacity in building materials by lowering thermostat settings prior to the critical period and then adjusting setpoints upward in an optimal way between the beginning and end of the critical period. In this study, four methods for determining building-specific demand-limiting control are developed. One of them, termed a model-based method, uses a detailed inverse building model that is trained using several days of hourly data and requires a non-linear numerical scheme for determining the optimal setpoint trajectory that minimizes peak cooling load. Two other methods, termed the semi-analytical (SA) method and exponential setpoint equation-based SA (ESA) method, employ simple inverse building models trained with short-term data and use analytical solutions from the models for the setpoint trajectories. The fourth method, termed the weighted-averaging (WA) method, exploits a locally linear relation between zone temperatures and cooling loads and uses short-term data. The model-based demand-limiting method was evaluated using both simulation and experimental testing. Experimental testing at a test facility representative of small commercial buildings showed about 30% peak load reduction for 5 hours in the afternoon with a comfort range between 69 and 76°F, which is consistent with simulation results. According to a comfort survey at the building, an acceptable comfort range was determined to be between 70 and 78°F. The SA, ESA, and WA methods were tested through simulation using high-order detailed inverse models for three building models representative of small, medium and large commercial buildings. The model-based demand-limiting control was used as a benchmark to provide the maximum possible peak reduction. Based on simulation evaluations, the WA method is recommended for demand-limiting control. The method requires no building description information and showed better performance than the other simplified methods for peak load reduction. Peak load reductions for the WA method with updating were within 12% of the maximum possible peak reduction and were between 30% and 55% compared to conventional control. Furthermore, the method can be used for demand-limiting of building aggregates.

Degree

Ph.D.

Advisors

Braun, Purdue University.

Subject Area

Mechanical engineering

COinS