Data-driven algorithms optimize gas pressure, detect faults, and enable low-carbon injection in networks for efficient regulation and enhanced safety

The Intelligent Gas Grid project seeks to address these issues by developing machine learning and artificial intelligence applications to create optimised set-point schedules for automatic application at governor stations.

SGN owns one of the UK's largest gas distribution networks (GDNs) in the UK, operating across Scotland, southern England and Northern Ireland. They look after 74,000 km of pipe network, keeping their customers safe and the gas flowing to 5.9 million homes and businesses. All GDNs are facing massive change as they develop strategies for net-zero. SGN are taking a proactive approach with their Ofgem Strategic Innovation Fund-funded Intelligent Gas Grid project by aiming to reduce their methane emissions and improve operational efficiency and customer service further.

Addressing the industry challenges

Effective pressure management is crucial to reducing methane emissions. Current management practices rely heavily on human intervention to provide the necessary reductions. While the networks manage pressures at the lowest feasible levels with the technology at hand, new and innovative solutions need to be explored and adopted to reduce pressures further and consequently minimise emissions. All networks face challenges in managing anomalies, such as water ingress, gas escapes, low-pressure events and malfunctioning governors. These anomalies require significant manual intervention for diagnosis and resolution, which can be time-consuming and costly. As the number of entry points for green gas into the lower pressure tiers of the network is increasing, the current manual pressure control can prevent biomethane plants from feeding into the network during certain times of the year, leading to waste.

The Intelligent Gas Grid project seeks to address these issues by developing machine learning and artificial intelligence applications to create optimised set-point schedules for automatic application at governor stations. The aim is such that the average pressure in the network, and therefore associated methane emissions, are always minimised while ensuring security of supply. These applications will also monitor measured pressure data for early signs of potential upcoming issues on the network, with the potential to enable faster diagnosis and facilitate remote or automated resolution of problems where possible. Expanding the automated network control approach will also maximise biomethane plants' feed-in potential.

Building Confidence in Digital Transformation

DNV is working with the project partners to independently assesses the project outputs for compliance with safety regulations, security of supply, and relevant AI policies and best practices. Building on the relationships from the earlier project phase, DNV is providing digital trust via assurance services on the AI/ML solutions developed during the project. A claims-based approach to assess the components’ capability and compliance based on evidence provided by the project partners is being used, based on the following DNV Recommended Practices:

• DNV-RP-A204 Assurance of digital twins
• DNV-RP-0510 Framework for assurance of data-driven applications
• DNV-RP-0497 Assurance of data quality management
• DNV-RP-0671 Assurance of AI-enabled systems

In addition to RPs, DNV is also providing:

• Expert support during field trials (PS5/6, G23), user research, data collection and analysis tasks
• Chair formal HAZOP process to assess safety of proposed field trial activities
• Assure and independently assess the project’s outputs to comply with safety, security of supply, AI policies and best practices

Impact

1. Reduced Methane Emissions: By implementing AI-driven pressure control, SGN significantly reduces methane emissions, contributing to environmental sustainability.
2. Improved Network Response: The project enhances the network's ability to detect and respond to anomalies swiftly, ensuring operational resilience and customer satisfaction.
3. Increased Biomethane Feed-in Capacity: Automation enables seamless integration of biomethane into the network, maximizing its feed-in potential and reducing waste.
4. Project Perspectives Metrics:
   • Data-driven AI algorithms process network data efficiently.
   • Integration of external data, such as weather forecasts, enables predictive demand estimations.
   • Actuators facilitate rapid incident response and automated optimization.
   • Distributed sensors measure and transmit pressure and environmental data at critical network locations, enhancing monitoring capabilities.