Calibration and uncertainty analysis for computer models – A meta-model based approach for integrated building energy simulation

In energy and environment field models are constructed, in general, based on well-defined physical phenomena and properties. Calibration and uncertainty analysis hold a particular interest because models represent a simplification of reality and, therefore, it is necessary to quantify to what degree they are imperfect before employing them in design, prediction and decision making processes. Integrated building energy models attempt to describe the effect of various internal and external actions (weather, occupancy, appliances, etc.) through physical relations (both algebraic and differential) and they are being widely used to design and operate high performance buildings, which are an essential component of a global energy strategy to reduce carbon emission and fossil sources depletion. An approach oriented to systems and able to integrate effectively field measured data and computer simulations for calibration in the modeling process has the potential to revolutionize the way buildings are designed and operated, and to stimulate also the development of new technologies and solutions in the field. The research presented in this paper aims to represent an initial step towards this integrated approach.

► We monitor and model an office building before, during and after an energy retrofit.
► We describe the advantages and the drawbacks of different calibration strategies.
► Piece-wise regression, Gaussian processes regression and model fitting techniques are used.
► We cross-validate the calibration comparing indicators obtained with different techniques.
► We present the calibrated prediction values for a reference meteorological year.