linear models, predictive control, thermal space, state-space model
A study on system identification and modeling of thermal spaces in a large institutional building is presented. The main topic of this paper is how the optimum model order associated which each thermal zone depends on factors such as the location of the zone within the building, its orientation and its exposure to outdoor space. Thermal models are essential in predictive control since they are required to predict the thermal load of a single building zone, a collection of various thermal spaces, or a whole building. The results of this study will serve as a guideline for choosing the appropriate order of linear models in similar buildings. The case study building is a model of a two storey school with a floor area of 24,000 m2 (258,000 ft2). The detailed thermal model of the building is created in EnergyPlus. This building model consists of 46 thermal zones covering a large variety of spaces: small offices, classrooms, long hallways and two gymnasia. The EnergyPlus is used to generate yearly input and output data available at 10-minute intervals; this data is used in a methodical system identification exercise, resulting in a set of multi-input single-output (MISO) state-space linear models. The challenge in modeling the thermal zones is to develop a relatively low-order model such that the thermal response of each zone is calculated by incorporating the effect of diverse inputs, such as outdoor factors (solar gains and outdoor temperature) as well as indoor factors, e.g., internal gains and heating and cooling energy delivered to the zone. Moreover, in a multi-zone building, accounting for the thermal effect of adjacent zones on one another is also an important factor to be taken into account. It has been found that this additional complexity requires careful selection of the inputs to the linear models, e.g., it might be helpful to include the heating/cooling delivered to adjacent zones.