Search eceee proceedings

Quality control routines for EPC Databases

Panel: 3. Policy, finance and governance

This is a non-peer reviewed extended abstract. No paper available.

Authors:
Iná Maia, TU Wien, Austria
Lukas Kranzl, TU Wien
Fabio Zanguirella, ENEA
Pasquale Regina, ENEA
Andreas Androutsopoulos, CRES
Lena Lampropoulou, CRES
George Koras, CRES
Casper Thielsen, DEA
Jacob Palm, DEA
Péter Gyarmati, TU Wien

Abstract

The EPBD emphasizes the role of EPCs as a policy instrument to assess the energy performance of buildings. EPC databases present the huge potential of providing building stock information, that can guide public authorities on the decision making process. Therefore, the stored data must be reliable, trustful and accurate. Currently, in most countries the quality of data stored in EPC databases is not known and systematically assessed. The EU-countries Italy, Greece and Denmark have a well-established EPC database, where building data in XML format is extracted from issued EPCs and stored in the database.

In general, an EPC quality control routine can be carried out in different phases of the EPC issuing process, as for example: emitting warnings and alerts in the EPC calculation software, or validating the EPC data before issuing the EPC. These mechanisms would avoid that “low quality EPCs” enter the databases. However, many EPCs in the databases have not passed through any quality control routine. And, in most countries, the selection of EPCs for themanual checks is based on a random selection of a very small sample of EPCs.

This paper sets-up a method to perform systematically a quality control routine in EPC databases. The method is applied in two selected countries (Italy and Greece) and builds on different types of rules for the quality check of the EPC parameters. Some of these rules (1st level verification) consist of verifying the completeness of the input data, e.g. if key data like the building cadastral field is empty. In addition, more complex rules were also developed (2nd level verification). They consist of defining the threshold values for specific parameters according to different building archetypes, where the threshold values have been defined based on a statistical evaluation of the database for the building archetypes. It was also part of the method to demonstrate how an analysis of inconsistencies (violated rules) can serve to improve energy auditors’ practice (and avoid that the same mistakes being repeated).

The method was implemented in a software programmed in Python language and different parts of it have been implemented and tested on the EPC databases. The software allows an interface with the national databases, for accessing the data from different databases. Then, a core module performs the computations using the data retrieved by the interface.

In Italy, almost 2 Mio. EPCs passed through the quality control routine. In the first level verification, 56.1% if the EPCs did not violate any of the rules, while 14.8% violated all rules. The remining 29.1% violated between 1 or 13 rules (from a total of 45 rules).

In the second level verification, 240 building type cluster were identified and for each an threshold values was fixed, using the statistical indicator 99th percentile. In this second level, 75,6% of the EPCs did not violated any rules, showing also no risk. 0,5% presented high risk by violating all rules. Finally 23,9% violated between 1 and 4 (from a total of 11 rules).

Learnings from the exercise showed that it is important to avoid parameter interdependence between the rules, especially in the first check. We can conclude that quality management procedures are highly needed, as they help on gathering more information about the quality of the EPCs and also inform assessors and trainers about common mistakes. Furthermore, trustful and reliable EPCs are an important source of building stock information. Prerequisites for the implementation of such routines are: machine-readable format of EPCs and a properly designed and well-established databases. This work has been developed in the frame of the H2020 project X-tendo.

Downloads

Download this display (photo) as pdf: 3-149-22-displayphoto.pdf