A data science solution for measurement and verification 2.0 in industrial buildings

Simple item page
dc.check.embargoformatEmbargo not applicable (If you have not submitted an e-thesis or do not want to request an embargo)en
dc.check.infoNot applicableen
dc.check.opt-outNoen
dc.check.reasonNot applicableen
dc.check.typeNo Embargo Required
dc.contributor.advisorO'Sullivan, Dominicen
dc.contributor.authorGallagher, Colm V.
dc.contributor.funderScience Foundation Irelanden
dc.date.accessioned2019-05-08T08:50:21Z
dc.date.available2019-05-08T08:50:21Z
dc.date.issued2019
dc.date.submitted2019
dc.description.abstractThe advent of advanced metering infrastructure has led to vast quantities of energy data becoming available. Despite this, the typical methods employed for the performance verification of energy efficiency improvements have not progressed, as they continue to rely on expert knowledge and simplistic statistical modelling techniques. This leads to uncertainty in the quantity of savings arrived at, with this uncertainty acting as a barrier to investment in energy efficiency. In response to this, the industry is evolving towards more advanced and automated methods known as M&V 2.0. This however presents the challenge of keeping the resources required to perform M&V at a minimum level, while also improving the accuracy, reliability and trust in the process. The research presented in this thesis can be largely classified into two prominent tasks. These are the development of a machine learning-based methodology for the construction of accurate baseline energy models and the establishment of a framework and final solution for M&V 2.0. It will be shown through theoretical and practical work that: - Machine learning techniques reduce the uncertainty introduced into the performance verification process by the baseline energy regression model. Additionally, the utilisation of a broader scope of analysis with respect to traditional methods is advantageous in further improving model accuracy. - Novel, computationally efficient data processing methods, including cleaning and feature selection, can be tailored for the industrial buildings sector to minimise the resources required to carry out performance verification. - The void in knowledge resulting from the established M&V protocols can be populated by a prescriptive methodology that utilises machine learning techniques to accurately and reliably quantify energy savings; thus, empowering performance verification practitioners in the use of advanced analytics on granular data sets and removing the need for expert knowledge. - The industrial sector requires a specific framework for the application of M&V 2.0 practices. An M&V 2.0 framework is developed to offer a solution to the challenge of persisting energy savings. A performance deviation detection system enables integration with ongoing monitoring and targeting practices. - M&V 2.0 does not have to increase the resources required to carry out performance verification. A novel, cloud computing-based solution, IntelliMaV, is capable of quantifying energy savings in near real-time with minimal uncertainty at high confidence levels. This thesis addresses some of the key challenges facing the performance verification industry including utilising the large quantities of energy data available in industrial facilities and evolving practices to a level of maturity that will enable it to support M&V 2.0. The implications of such challenges are shown to be significant beyond the individual project level, with the effectiveness of European energy policy dependent on accurate and reliable M&V. The methodology, framework and IntelliMaV application developed all address these challenges, while aiding the transition to M&V 2.0 practices. Despite these advancements, this is not the final solution for the industry. A collective effort must be made to continue to modernise performance verification practices to ensure M&V remains a valued and beneficial practice in energy management into the future.en
dc.description.statusNot peer revieweden
dc.description.versionAccepted Version
dc.format.mimetypeapplication/pdfen
dc.identifier.citationGallagher, C. V. 2019. A data science solution for measurement and verification 2.0 in industrial buildings. PhD Thesis, University College Cork.en
dc.identifier.endpage186en
dc.identifier.urihttps://hdl.handle.net/10468/7862
dc.language.isoenen
dc.publisherUniversity College Corken
dc.relation.projectinfo:eu-repo/grantAgreement/SFI/SFI Research Centres/12/RC/2302/IE/Marine Renewable Energy Ireland (MaREI) - The SFI Centre for Marine Renewable Energy Research/en
dc.rights© 2019, Colm Vincent Gallagher.en
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/3.0/en
dc.subjectEnergy efficiencyen
dc.subjectMachine learningen
dc.subjectData scienceen
dc.subjectEnergy savingsen
dc.subjectMeasurement and verificationen
dc.subjectIndustrial buildingsen
dc.subjectEnergy modellingen
dc.thesis.opt-outfalse
dc.titleA data science solution for measurement and verification 2.0 in industrial buildingsen
dc.typeDoctoral thesisen
dc.type.qualificationlevelDoctoralen
dc.type.qualificationnamePhDen
ucc.workflow.supervisordominic.osullivan@ucc.ie
Files
Original bundle
Now showing 1 - 3 of 3
Thumbnail Image
Name:
GallagherCV_PhD2019.pdf
Size:
48.51 MB
Format:
Adobe Portable Document Format
Description:
Full Text E-thesis
Download
Thumbnail Image
Name:
GallagherCV_PhD2019_Abstract.pdf
Size:
34.18 KB
Format:
Adobe Portable Document Format
Description:
Abstract
Thumbnail Image
Name:
GallagherCV_PhD2019_LatexSourceFiles.zip
Size:
165.27 MB
Format:
http://www.iana.org/assignments/media-types/application/zip
Description:
Latex Source Files
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
5.62 KB
Format:
Item-specific license agreed upon to submission
Description:
Download
Collections
Doctoral Theses
Centre for Marine and Renewable Energy (MaREI) - Doctoral Theses
College of Science, Engineering and Food Science - Doctoral Theses
Energy Engineering - Doctoral Theses