Design of a short perturbation method for on-site estimation of a building envelope thermal performance

Abstract

On-site measurement of the actual thermal performance of a building envelope could considerably benefit to the building sector by securing an effective performance of newly built or retrofitted buildings. Current methods to estimate the Heat Transfer Coefficient, for example, suffer however from either excessive equipment downtime and therefore large costs, or fail to achieve a reasonable accuracy. This paper aims therefore at developing and assessing an original 24-h perturbation method for estimating a whole-building Heat Transfer Coefficient. Based on a binary heating signal, the proposed method exploits the data collected with stochastic RC models.The paper first presents a numerical comparison of 398 binary signals along with variable initial indoor temperatures to bring out common characteristics of signals that achieve highly accurate HTC estimations. An experimental campaign in a highly insulated wooden-framed house then implemented one particular signal and assessed its reproducibility under variable weather conditions. The paper shows that the proposed 24-h method indeed achieves a highly accurate estimation when performed in cold weather. In mid-season conditions, the method shows a better robustness from 30- to 36-h measurements. The results invite to further investigate other building types in variable weather conditions to validate the proposed method.