Multi-level Characterisation of the Building Energy Performance of Residential Buildings based on Analysis of Measurement Data.

In support of the energy transition in the built environment, knowledge on
performance characteristics of the building fabric and technical systems is
essential for various applications, such as quality assurance, the development of
renovation strategies, or the appraisal of the potential for demand side flexibility.
In practice, the available information on the existing building stock is often
limited and of poor quality. Furthermore, research shows that results of
calculations and simulations made during the design phase significantly differ
from the actual as-built performance.
Characterization techniques based on on-site measurements, provide valuable
alternatives for the assessment of performance indicators, but some of the
experimental setups can be perceived as intrusive and costly. Novel techniques,
such as 3D Lidar geometry scanning, aerial thermography, smart meters, and IoT
sensors provide new opportunities to develop less intrusive and faster
identification methods.
The first aim of this work is to elucidate the interplay of four aspects involved in
building energy performance characterization, namely (1) the applications of
characterization, (2) the performance indicators that need to be determined, (3)
the characterization methods that can be used and (4) the demands of
stakeholders involved. Hereto, an assessment framework is developed which
links all these aspects and presents them in the form of a three-dimensional
matrix. A potential application of this matrix is for example a tool that guides
stakeholders in selecting a suitable characterization method to determine a
performance indicator within a specific range of accuracy.
Subsequently, the research focuses on one particular performance indicator
included in the matrix: the Heat Loss Coefficient or ‘HLC’. The HLC describes the
insulation quality and airtightness of a building envelope in a single factor. This
work explores whether this performance indicator can be assessed for single-family
dwellings based on a combination of ‘in-use monitoring’ and data-driven
modeling using steady-state or dynamic analysis methods. In-use monitoring is
hereby defined as the monitoring of the energy consumption and interior climate
of occupied buildings via non-intrusive sensors. It is investigated how the results
of the characterization can be linked to the physical reality. Since several ranges
of input data are possible, going from solely smart meter data to a combination
of sensor data, data from Building Information Models, surveys and default
values, a sensitivity analysis is conducted of the HLC estimate to the amount and
accuracy of the input data. Furthermore, the influence of the building considered
(e.g. type, insulation quality, heating profile) and data analysis method used on
the obtained HLC estimate is analyzed.
This dissertation includes case study analyses on both actual measured data and
synthetic data derived from energy simulations. This combination allows to
examine various scenarios regarding the building type and interior climate,
without losing sight of the particularities of on-site collected data.

The research confirms the intrinsic capability of HLC characterization based on
in-use monitoring data. It is demonstrated that an accuracy of up to 2.5% can be
achieved. The characterization accuracy is however strongly dependent on the
investigated building and the methodological choices made during the collectionand analysis of the monitoring data, such as the number and position of the
temperature sensors installed, the measurement duration, selected data analysis
method, or the interpretation of the identified model coefficients. The work
concludes by exemplifying how by sensibly selecting the input data and analysis
methods, a wide range of in-use characterization methods can be developed,
suited for different applications, budgets and timescales.