< Back to previous page

Project

Assessing the Thermal Performance of building Envelopes based on Limited Onboard Measured Data: Determining the Heat Loss Coefficient on Large Scale

One of the first steps in reducing energy demand is accurately predicting energy requirements in urban areas and improving the thermal performance of buildings, especially in the residential sector. Determining the energy use in buildings and understanding the dynamics of the building stock depends on several factors such as climate, user behaviour, and the executed operation of building systems, and fabric properties. Research shows that the actual energy performance of the building fabric can differ significantly from what was initially predicted due to inconsistencies in building materials and poor workmanship. Therefore, the achieved thermal transmittance of building components often exceeds theoretical values, emphasizing the importance of evaluating as-built performances to gain insights into the actual building behaviour. This discrepancy between predicted and actual energy usage is known as the energy performance gap.

To bridge this gap and quantify the actual energy performance of building envelopes, the Heat Loss Coefficient (HLC) is introduced as a stationary performance indicator. The HLC quantifies the power required to maintain a 1 Kelvin indoor-outdoor temperature difference over the entire envelope and considers both thermal resistance and air tightness. Various methodologies and testing procedures have been developed to quantify this performance indicator on-site with the co-heating test being considered one of the most accurate methods for measuring the HLC. However, most of the prevailing methods rely on extensive measurement campaigns which are considered impractical by stakeholders due to intrusion, cost, and length.

Limitations of the existing assessment procedures make them unsuitable for large-scale implementation. Thus, non-intrusive assessment methods have been proposed as alternatives, utilizing available energy consumption data, weather data and on-board measurements to estimate the thermal performance. To complement that, the increasing availability of data, such as smart meters and IoT platforms, presents opportunities for non-invasive monitoring of energy requirements, occupancy, and indoor and outdoor environments. Nevertheless, they introduce challenges related to the metering of heat sources, data availability and management. Therefore, there is a need for research that can quantify the uncertainty of HLC estimated using on-board collected data paired with suitable statistical models. Based on the knowledge of the level of detail of collected datasets and methodological limitations, accurate assessment of thermal performance in an automated and unsupervised way should be enabled.

The research presented in this dissertation aims to develop reliable and fast methods based on verified statistical models for thermal performance assessment with limited data at the same time applicable to an unlimited number of dwellings. The work is divided into three main parts: generating a comprehensive artificial dataset, developing automated assessment procedures, and investigating the impact of monitored variables on estimation accuracy within realistic monitoring setups. The proposed methodology includes creating artificial datasets reflecting different building stock scenarios, which are found suitable to compare estimates calculated by the co-heating test, used as a benchmark, statistical models based on on-board collected data and a known reference value. The automated assessment procedure is based on AutoRegressive models with eXogenous input, which are purely data-driven and, thus, do not require previous physical knowledge of the observed dwellings. Overall, the research contributes to the state of the art by generating a large-scale artificial dataset, investigating the reliability of the co-heating test, extending existing statistical methodologies to unsupervised applications, quantifying estimation errors, and providing practical guidelines. 

Date:27 Sep 2019 →  13 Dec 2023
Keywords:heat loss coefficient, HLC, thermal characteristics, building district, neighbourhood
Disciplines:Building physics
Project type:PhD project