Calibration de modèles d'équipement CVCA à l'aide de données colligées

Abstract

The calibration of heating, ventilation and air conditioning (HVAC) equipment models found in energy simulation software is a long, difficult task for which the calibration soundness depends on the analyst. Currently, the different proposed approaches are not usable because the results obtained are not always conclusive and the costs are considerable. Furthermore, the real performance of equipment and systems might vary significantly compared with the as-built drawings and manufacturers’ data. In this context, this master’s thesis presents a new calibration approach of HVAC equipment model found in energy simulation software that uses an optimization algorithm and measured trend data. The automatic calibration approach is presented in the form of three papers. The first paper explores the automatic calibration of HVAC equipment models under the TRNSYS software using synthetic data and a genetic algorithm. The second paper covers the calibration approach in the EnergyPlus software using synthetic data and the GenOpt software. The last paper supplements the calibration approach with an additional step to reduce the optimization domain and documents the application of the approach over measured trend data taken from a mechanical system of a university building. The estimation of the influential parameters of the variable volume fan model (Fan:VariableVolume) is performed with the minimization of the least squares using the GenOpt software. The calibrated equipment model provides an excellent prediction of the supply air temperature (NMBE = 1.80 %, RMSE = 0.53ºC) and fan power (CVRMSE = 4.30 %, NMBE = 2.02 %, RMSE= 0.46 kW). The results presented in the papers of this master’s thesis show that an HVAC equipment model of the EnergyPlus software can be automatically calibrated using the proposed approach.

Date	24 Oct 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Danielle Monfet (Supervisor) & Hervé-Frank Nouanegue (Co-supervisor)