Building thermal model reduction using nonlinear constrained optimization

The short-time-horizon modelling of building thermal response is of relevance in situations where HVAC plant and control system analyses are of interest. In this precise area of building thermal modelling issues of model accuracy and computational efficiency become important. In this work a nonlinear constrained optimization method is used for reducing the model order of building elements. The approach involves minimizing the error between the step response of a high-order reference model whilst tuning the parameters of a lower order model in order to obtain an optimized reduced-order model. Results show that a reduced model based on a 2nd-order building element gives minimal loss of accuracy but significant improvements in computational effort when treating both high and low thermal capacity modelling problems.