Assessment of data analysis methods to identify the heat loss coefficient from on-board monitoring data

Abstract

The past decade has seen the rapid development of sensor technologies. Monitoring data of the interior climate and energy consumption of in-use buildings, so-called on-board monitoring (OBM) data, offers the opportunity to identify as-built energy performance indicators, such as the heat loss coefficient (HLC) of the building envelope. To this end, it is important to advance the understanding of the impact of the OBM set-up and the applied data analysis method.This paper uses synthetic OBM data sets, generated from building energy simulations. The level of accuracy achieved with four data analysis methods for characterizing the HLC is investigated. The considered methods are the Average Method, the Energy Signature Method, Linear Regression and ARX modeling. Different cases, representing different building types, are considered in order to gain thorough insight into the physical interpretation of the results. By taking subsets of the original data sets, the sensitivity of the data analysis methods to the availability of specific data is assessed.This theoretical exercise illustrates how, under idealized monitoring circumstances, both linear regression and ARX models can accurately determine the HLC. The latter is able to assess the performance indicator within 5%. However, when subjected to practical limitations regarding the measurement of system inputs, such as unavailable solar or internal heat gains, the characterization results show large variations in accuracy and uncertainty.