The project has received funding from the European Partnership on Metrology, co-financed by European Union Horizon Europe Research and Innovation Programme and from the Participating States.
Work Packages
Empirical Datasets of Time-Resolved Gas Flow Measurements
Time-resolved data of gas metering are scarcely available. Most of these data are obtained as part of the regular activities of TSOs and DSOs and perceived as commercially and strategically sensitive. A further issue with the data that are available is that they rarely contain all meta data to make them interoperable. Part of these meta data include the calibration status of the equipment, and context information about the changes in the signals of the instruments.
The aim of this work package is to create datasets for time-resolved gas flow measurements, including temperature, pressure and gas composition using state-of-the-art methods and techniques, mimicking changes in gas grid flow rate, pressure, temperature and gas composition typically seen in real world scenarios in order to enable studying and modelling of autocorrelation effects. In SmartGasNet, these datasets will be input for the time series analysis in WP2 and serve as training data for ML in WP3.
Leader: LEI
Methods for the evaluation of time-correlation and uncertainty in gas metering data
Gas grid data is obtained in the form of time series. These time series can show interdependencies, which are referred to as time correlations. These interdependencies can be assessed using statistical tests and modelled with various kinds of statistical models (e.g., autoregressive and moving average models). The interdependencies in the measurement data are called time correlations. These time-correlations cover both completely correlated uncertainty contributions that cannot be determined purely statistically, as well as partial correlations over time, that can be identified through a statistical evaluation of gas metering data for autocorrelation.
The aim of this work package is to develop analytical and numerical methods for identifying and evaluating time-correlations between different measurement uncertainty contributions, and to use this insight when propagating measurement uncertainties through accumulation and averaging processes. Synthetic datasets, created using time series models with known time-correlations of the measurand/true signal and known measurement uncertainties, will be used for developing and validating the methods. The synthetic datasets will also be used for testing in WP3 and the findings from WP2 will also serve as a support for the task on gas allocation uncertainty in WP4.
Leader: NORCE
The aim of this work package is to assess the feasibility and possibility of integrating ML as a specific and well-known kind of AI, in smart metering and monitoring systems used in gas grids with respect to its applicability to produce ‘virtual’ measurement results aimed at meeting the requirements of ISO/IEC 17025, OIML R140 and ISO 15112, and the Guide to the Expression of Uncertainty (ISO/IEC Guide 98) in ML model estimations. The following potential issues will be addressed in the development: evaluating uncertainty in the results of ML models, transparency and explainability as well as the trustworthiness of the ML models, auditability of the ML models, sustaining model performance over time.
In this work package, gas flow metering model with ML algorithms will be assessed and benchmarked using empirical and synthetic data set (flow, pressure, temperature, gas quality and topology) using the experimental data from WP1 and simulation data from WP2. The activities will provide guidance with a proposed workflow on how to setup, train, and validate ML models for gas grid monitoring and measurement in a way that the outcome can be integrated into existing guidelines and standards.
Leader: TNO
The aim of this work package is to develop novel evaluation methods for fiscal metering data of natural gas, HENG, biomethane, hydrogen and other energy gases that are traded on either quantity (volume or mass) or energy. These evaluation methods will build forth on those developed in the Partnership project 21GRD05 Met4H2 and the EMPIR project 19ENV09 MetroPEMS. They take up the models developed for serial correlation in data sets in WP2, the effects of measurement errors due to instrumentation and where appropriate the outputs from the work on ML from WP3.
The evaluation methods will propagate the uncertainty from the measurements, including the dependencies that exist between the measurement results used in the totalisation of quantity and energy, in accordance with ISO/IEC Guide 98. The developed methods will be provided in the form of stand-alone guidance documents that can be used alongside OIML R140, ISO 15112, EN 1776 and standards referenced to, such as ISO 6974 and ISO 6976.
Leader: VSL
The aim of this work package is to create impact from the project to support all relevant stakeholders from industry, measuring equipment manufacturers, standardisation committees, metrology institutes and the scientific community.
It is a cross-cutting work package, led by GERG, in coordination with all project partners, to facilitate the dissemination of project results and the communication of project activities, public events, workshops, etc. To achieve greater impact, we will develop and implement an exploitation strategy to promote the rapid uptake of SmartGasNet results, by key target groups such as gas infrastructure operators.
Leader: GERG
The project will be managed by the coordinator from VSL, who will be supported by the project management board, including the leaders of each work package.
Leader: VSL