Deep Dive on Data and Information for Critical Infrastructure Management and Maintenance

This report examines the role of data and information in managing and maintaining critical infrastructure in the UK. It focuses on four sectors with large networks of physical assets: energy, transport, water, and communications. A report from an earlier study into Critical Infrastructure Management & Maintenance for Safety, highlighted the need for a deep dive on the role of data and information, posing four questions. These questions acted a catalyst for the study, and were addressed during this research. This report reviews physical commonalities through categorising critical infrastructure assets into nodes, links, and supporting structures, and highlights the importance of data and records, both analogue and digital, in asset management. The UK’s critical national infrastructure (CNI) sectors share common challenges such as ageing assets, safety concerns, and a skilled workforce shortage. While each sector has its own history and terminology, there are sufficient commonalities to develop transferrable approaches for data collection and analysis. There exists opportunities for innovation in the whole asset lifecycle, from the design phase, through operations and maintenance, upgrades and decommissioning. While cost and safety drive asset management strategies, sustainability is increasingly desired. How to best incentivise this, when any one individual gain is small, is an open question. Run-to[1]failure (RTF) is the cheapest asset management strategy but unsuitable for critical assets. Time-based maintenance, driven by regulations and traditional inspection methods, remains prevalent despite aspirations for condition-based methods. Predictive maintenance, encompassing condition-based and risk-based approaches, relies on data analysis to predict asset degradation, but offers the greatest promise. Leveraging artificial intelligence (AI) will enable degrees of automation in anomaly detection, classification, prediction, and maintenance scheduling tasks. However, challenges exist in balancing data-driven approaches with explainability, aligning rapid technological innovation with slow[1]moving engineering standards, and ensuring data integrity and risk evaluation for safety-critical applications. Going forward Standards and ontologies will play a vital role in curating and organising data, enabling both human and AI interpretation. This codification of both explicit and implicit data, information, and knowledge (DI&K) is crucial for implementing AI-enhanced decision-making processes in critical infrastructure. This includes embedding expert knowledge through training programmes and mentoring schemes, as well as leveraging multimodal AI and neuro-symbolic approaches to capture and codify implicit knowledge. While data aggregation is important, the focus should be on data provenance, integrity, and ensuring it represents all expected asset conditions, both healthy and degraded, to avoid bias and data overload. To realise the vision of a truly data[1]augmented management of critical national infrastructure, four key areas must be tackled: Improved capture of data and information at point of inspection, Improved management and retention of digital data, information and knowledge, Improve shared understanding of asset degradation, and Maximise utilisation of non-asset management data. We present a roadmap outlining the actions required to address those areas.