Thesis Defence, An Open Energy Data Lifecycle: Organisation, Methods, Tools

An open energy data lifecycle: organisation, methods and tools PhD defence 4 November 2025 © 2025. This work is openly licensed via CC BY 4.0. Presented by Seun OSONUGA Before a jury of: Supervisors •Dr. Frederic WURTZ, CNRS/G2ELab •Prof. Benoit DELINCHANT, Grenoble-INP UGA Invited guest •Aude de TOUCHET, Agence ORE Reviewers •Prof. Romain BOURDAIS, CentraleSupélec •Prof. Bruno FRANÇOIS, Centrale Lille Jury president •Prof. Stephane Ploix, Grenoble-INP UGA 2 The energy sector: a real enigma Sources: Raworth, K., 2017. A Doughnut for the Anthropocene: humanity’s compass in the 21st century. The Lancet Planetary Health 1, e48–e49. https://doi.org/10.1016/S2542-5196(17)30028-1 | Ge, M., Friedrich, J., Vigna, L., 2024. Where Do Emissions Come From? 4 Charts Explain Greenhouse Gas Emissions by Sector. https://www.wri.org/insights/4-charts-explain-greenhouse-gas-emissionscountries-and-sectors INTRODUCTION 3 What is this data used for? ▪System planning and forecasting at different levels ▪Energy system digitisation ▪Leveraging AI for operations Operational uses - Modelling of technical and human aspects of our energy systems - Observing trends in energy use (individual, regional, or global) - More open and reproducible science Knowledge and innovation Source: Images by BulentYILDIZ de Pixabay Data serves to meet our energy security goals globally while lowering the environmental, economic, and social costs of our global energy system INTRODUCTION 4 But what is energy data? ▪Commons: Precipitation data, Solar radiation data ▪Shared and large-scale: Data on the electricity grid, dam status ▪Private (highly variable): Individual building data, EV usage data INTRODUCTION TIERSDATA HOLDERS 5 Percentage of total datasets by theme on the European open data catalogue (data.europa.eu) The energy sector: a real enigma Source: "Data in figures | Data.Europa.Eu". 2017. Accessed 28 October 2025. https://data.europa.eu/catalogue-statistics/currentState/category?locale=en. 20% 18% 13% 11% 8% 7% 6% 6% 5% 3% 1% Agriculture, fisheries, forestry and food Justice, legal system and public safety Environment Government and public sector Economy and finance Science and technology Population and society Regions and cities Transport Education, culture and sport Health Energy Provisional data International issues INTRODUCTION Total number of datasets = 1 530 721 6 The energy sector: a real enigma 20% 18% 13% 11% 8% 7% 6% 6% 5% 3% 1% INTRODUCTION 7 So why isn't it currently shared? Technical barriers Social and institutional barriers Lack of common metadata Policy and regulation spaghetti Fears related to commercial competition Awareness of the usefulness of data Limited resources INTRODUCTION Limited sensors/meters Trust in data and its use Protection of privacy Data fragmentation SOCIO-TECHNICAL COMPLEXITY 8 Our questions for today... My thesis seeks to answer three questions HOW can energy data be made ACCESSIBLE? 1 Will this be VALID for the DIVERSITY of energy data? 2 How will this FACILITATE INTERACTIONS between stakeholders? 3 Presentation plan INTRODUCTION 1. Introduction 2. Formalisation of the data lifecycle 3. Experimentation based on six reallife case studies i. Experimental setup ii. Impact of data typologies iii. Supporting interactions between stakeholders 4. Conclusions, limitations, and future work 9 Agenda - Introduction -Formalisation of the data lifecycle - Experimentation based on six real-life case studies ▪Experimental setup ▪Impact of data typologies ▪Support for interactions between stakeholders - Conclusions, limitations and prospects ▪What a data lifecycle is? ▪Common steps in most data lifecycles ▪Suitability for the energy sector ▪Proposition of a new data lifecycle 16 Data Re-use Proposition: OpEnDaLe in full detail Mobilisation Data Valorisation Data sharing Feedback Data processing Tracking use & user interaction Data Reuse Self-Use Data Collection 1 2 3 5 7 9 8 4 6 Data Valorisation Resources Raw Data Versioned Data Public Data Useful Data Uses & Issues Reinforcement Impact of shared data New data requirements New processing steps Insights Data collection 2 Mobilisation 1Data processing 3Data sharing 5 Data Referencing 6a Data use incentivization 6b Use-case publishing 6c Feedback (Internal) 9a Feedback (External) 9b Tracking use & user interaction 8 Issues Private Data Self-use 4 Private Data Downloads & views Questions & doubts Responses & recommendations Data holders, researchers, funding organisations, general public, public administrations etc. Data re-users: Other researchers, Companies, General public Data holders and re-users Data Reappropriation 7b Data Access & Exploration 7a Model training 7c Other uses 7c Scientific articles 7c Data holder(s) only ??? ??? Citations LIFECYCLE FORMULATION 17 On formalising a datalife cycle... LIFECYCLE FORMULATION Introduction to datalifecyles and a generalised model The shortcomings of this model for energy data My proposition: OpEnDaLE with the modified stages OpEnDaLE in more detail flows & stakeholders 18 Agenda - Introduction - Formalisation of the data lifecycle -Experimentation based on six real-life case studies ▪Experimental setup ▪Impact of energy data typologies ▪Support for interactions between stakeholders - Conclusions, limitations and prospects ▪OTE-UGA as an experimental plaform ▪Domains of case-studies used to study OpEnDaLe 19 Shared Platform What is the Observatoire de la Transition Énergétique? AXIS 6 Shared Platform OpEnDaLe UGA OTE-UGA EXPERIMENTAL SETUP Moderation “Sobriete” Research - Chaire Sobriete-Resilience - ANR Satiable Flexibility Research - FlexTASE project Main advantages ▪Experimental structure I had easy access to ▪Access to UGA resources, including a Data Protection Officer (DPO) ▪Attractive non-profit status for panel ▪Enables interaction with end users of energy 20 The datasets used as case studies covered different domains EXPERIMENTAL SETUP Tertiary buildings Predis-MHI Thermal GreEn-ER weather Electric Mobility EVE Residential buildings EtudELEC Etude xKY In experimenting on the case studies, I worked on multiple aspects of the data lifecycles, including mobilisation, study design, data processing and sharing, and data valorisation, amongst others Sources: Electric vehcles - Image from Pixabay; Residential buildings - Image from Pixabay; GreEn-ER buiding: Architecte du bâtiment:Groupe-6, Bruno Hallé —Milky2, CC BY-SA 3.0, https://fr.wikipedia.org/w/index.php?curid=9386438 21 Agenda - Introduction - Formalisation of the data lifecycle -Experimentation based on six real-life case studies ▪Experimental setup ▪Impact of data typologies ▪Support for interactions between stakeholders - Conclusions, limitations and prospects ▪Definition of data typology dimensions ▪Mapping case studies to typology ▪Comparison of case-study lifecycles ▪Impact of typology dimensions on lifecycles 22 Proposed data typology dimensions and their attributes IMPACT OF DATA TYPOLOGY Typology Dimensions Typology attributes Data subject What does the data describe? ▪Individual ▪Organisation Temporality Is the data fixed in time or does it change? ▪Static ▪Dynamic/live Granularity How attributable is the data to a single data subject? ▪Granular (Meter/Entity) ▪Aggregated Prevalence of the concept Is the data about a common concept or subject? ▪Generalisable ▪Specialised Dimension asks… = A16-space typology grid 23 The case studies fell into six different data typologies IMPACT OF DATA TYPOLOGY Data subject Temporality Granularity Prevalence of the concept VS 24 Predis-MHI Thermal Data ▪Metering infrastructure provided by the organisation ▪Authorisations with one decision maker in the organisation (G2ELab) ▪Limited personal data (check that none of the offices were single-occupied) ▪Deposited on in a data repository with a DOI ▪Data paper+ needed to provide more context Comparison of the data lifecycle for two cases EtudELEC Mobilisation Data Valorisation Data sharing Feedback Data processing Tracking use & user interaction Data Reuse Self-Use Data Collection 1 2 3 5 7 9 8 4 6 ▪Metering infrastructure provided by 3rd party (DSO) ▪Authorisations with all participants individually and with DSO (ENEDIS) ▪Personal data removal ▪Aggregation to safeguard privacy ▪Deposited on in a data repository with a DOI ▪No data paper IMPACT OF DATA TYPOLOGY 25 GreEn-ER Live Weather Data ▪Metering infrastructure provided by the organisation ▪Authorisations with one decision maker in the organisation (G2ELab) ▪No personal data ▪Live data provided on a dashboard-style website ▪Link and documentation deposited in a data repository with DOI ▪Website metrics ▪Repository metrics (views, downloads, citations) ▪Forum for discussion Comparison of the data lifecycle for two cases: EVE ▪Metering infrastructure provided by the participants ▪Authorisations with all participants individually ▪Personal data removal (VIN and personal data from questionnaires) ▪Deposited in a data repository with DOI ▪Data paper+ needed to provide more context ▪Repository metrics (views, downloads, citations) ▪Forum for discussion Mobilisation Data Valorisation Data sharing Feedback Data processing Tracking use & user interaction Data Reuse Self-Use Data Collection 1 2 3 5 7 9 8 4 6 IMPACT OF DATA TYPOLOGY 32 Interactions between stakeholders on the forum Source: https://forum-ote.univ-grenoble-alpes.fr/ Roles/characters Data users Witnesses to interactions ▪Clarifications between producers and users ▪Producer-user corrections ▪Collaboration between users ▪Collaborations between interested parties ▪Co-creation of data by producers and users STAKEHOLDER INTERACTIONS Forum moderators and administrators Data producers Other interested parties 33 + House type Heating Time-series Co-creation of data with users: An EtudELEC example OTE private data metadata elec. cons. elec. cons. OPEN DATA Gas-heated apartments District-heated houses Aggregation filter elec. cons. Cluster 1 Cluster 12 - # of inhabitants - Type of appliances - Clustering of time-series OTE FORUM ... ... STAKEHOLDER INTERACTIONS 34 The metrics dashboard ▪Metrics for each dataset are collected daily via the recherche.data.gouv API, where valuation events are manually entered into a database. ▪Similar dashboards are created for downloads and citations. STAKEHOLDER INTERACTIONS Appreciable links between valorisation events and dataset metrics 35 VALENS indicators: data processing STAKEHOLDER INTERACTIONS Aggregated the metrics for all datasets as often valorisation activities affect multiple datasets Go from cumulative values to daily values to better quantify impact 36 Quantifying the impact of valorisation events: VALENS VALorisation Events Normalised Score Comparison of the daily metrics in between pre and post windows Post windowPre window STAKEHOLDER INTERACTIONS Lift percentage Absolute lift score __ _ %post window pre window pre window Avg Avg Lift Avg − = Difference in residuals between post and pre-windows Valorisation event Day J Day J+7Day J-7 _t DoW median res y y=− 1k ak + = _. tt post pre Abs lift res a res=−  37 VALENS results Events (Avg lift percentage) Data referencing (3.90%) Data use incentivisation (6.14%) Use-case publication (3.07%) OTE-xKy Newsletter - #14 Opening of the OTE Forum Presentation of work based on the datasets ▪Activities that encouraged data exploration were the most impactful for view increases. ▪The impact of use-case publication will most likely be more long-term than the others. ▪The same general trends were found for downloads but a larger sample size will be required to make more definitive conclusions. STAKEHOLDER INTERACTIONS 38 On the tools to support interactions... STAKEHOLDER INTERACTIONS Mobilisation Data Valorisation Data sharing Feedback Data processing Tracking use & user interaction Data Reuse Self-Use Data Collection 1 2 3 5 7 9 8 4 6 Tracking use & user interaction 8 ▪The OTE forum and the interactions around the forum ▪An example of how the forum was used for data co-creation OTE Forum ▪Measuring data metrics and valorisation events with a dashboard ▪Quantifying the impact of valorisation events with VALENS Metrics dashboard & VALENS 39 Agenda - Introduction - Formalisation of the data lifecycle - Experimentation based on six real-life case studies ▪Experimental setup ▪Impact of data typologies ▪Support for interactions between stakeholders -Conclusions, limitations and prospects 40 Overall, we discussed... - OpEnDaLe, a data lifecycle for the energy sector that addresses some of these barriers through additional steps and modifications - The experimental setup used in my thesis: OTE-UGA, and the various published datasets - Relevant data typologies and their impact on the data lifecycle, demonstrating the robustness of this cycle. - The tools and indicators we use in the data lifecycle to facilitate interactions between stakeholders Our questions for today... HOW can energy data be made ACCESSIBLE? 1 Will this be VALID for the DIVERSITY of energy data? 2 How will this FACILITATE INTERACTIONS between stakeholders? 3 CONCLUSIONS 41 But the work is neither perfect nor final… Limitations Outlook Centered around only individuals and research communities Extension of the implementation and testing to other organisations Lack of metrics to study the effects adequately (esp. citations) Insufficient control study on the individual lifecycle additions Battery of tests for various phases (interaction & valorisation) with improved metrics pipelines Deeper experimentation on more technical privacy-preserving techniques Exploring semantic data enrichment and leveraging AI in the data lifecycle process CONCLUSIONS Continue to use OpEnDaLe and the tools at OTEUGA on current projects (FlexTASE, Chaire Sobriete Resilience, ANR Satiable)