Paper Conference

Proceedings of eSim 2018: 10th Conference of IBPSA-Canada

     

An approach to virtual sensing of cooling systems using state-space inverse modelling and gradient descent

Jayson Bursill, William O’Brien, Ian Beausoleil-Morrison

Abstract: Despite improvements in sensing quality and pricing, the installation of building automation integrated sensors often remains economically unjustifiable. Limited physical sensing in many buildings makes quantifying zone and room level energy flows challenging. A method to estimate cooling energy usage at a 15-minute resolution using state-space models and gradient descent parameter optimization is presented. Indoor air temperature, outdoor air temperature, and solar irradiation (with illuminance as a proxy) were used as inputs to inverse models of the indoor air temperature for estimating variable air volume cooling unit energy input. The cooling load was estimated for eight offices of an institutional building over a cooling season and maintained seasonal errors of 16% or less. The resultant parameters were found to not be universal, and varied significantly between offices within each zone. The presented methodology provides an approach for cooling quantification in previously unmetered buildings with applications to modelling and operations.
Keywords: virtual sensing, energy use estimation, building performance simulation
Pages: 105 - 113
Paper:
esim2018_1-2-B-1