An Automated Framework Based on Rc Model and Ga Optimization for Calibrating Coupled Residential Buildings and Hvac Systems

Abstract

Modeling and simulation of buildings and building air-conditioning (AC) systems have been widely used in various studies like energy optimization, optimal controls, fault detection, etc. A critical requirement for such models is proper calibration. However, the calibration process could be highly labor-intensive, especially when performed manually. Consequently, several studies have explored ways of automating the calibration process. But most of these studies are for industrial or commercial systems and use advanced search-based optimization algorithms. This study therefore proposes an automated calibration framework for a coupled residential building/AC system, based on Genetic Algorithm optimization and a simple RC (Resistance-Capacitance) model-based optimization. The proposed framework is validated using measured data from a residential home and an AC system, alongside simulation data generated from a co-simulation testbed developed using Modelica and EnergyPlus. The results obtained showed a good match between the calibrated model and the physical system, validated through comparisons of measured and simulated data. For the calibrated AC model, a maximum absolute error (MAE) of 0.617 ˚C and 0.757 ˚C was obtained for supply air temperature and degree of subcool, while the maximum Coefficient-of-Variation Root-Mean-Squared Error (CVRMSE) for power consumption was 7.5%. For the calibrated building model, the thermal properties of the envelope showed a difference of only 2.5% with those of the real building. These results demonstrate the prospect of the proposed automated calibration framework and can be adapted to other residential building modeling studies.

Keywords: Automated calibration framework, RC model, Genetic algorithm optimization, co-simulation, Modelica, EnergyPlus