Learnings

Outcomes

Within the WWU and SGN regions alone, it is anticipated that there are several dozen existing data centres sites where a gas connection might be considered techno-economically suitable, and where a detailed feasibility study could be undertaken. Connection of these or future data centres to the gas network could:

• Improve timescales for energisation of new data centre facilities facing delays due to electricity network constraints;
• Allow data centre development in new regions near suitable gas infrastructure that have been limited by the availability of, proximity of, or cost to connect to the existing electricity network;
• Provide alternative industrial-scale off-takers for gas demand ahead of electrification, reducing the risk of having stranded assets;
• Utilise existing and well-established ‘off-the-shelf’ SOFC & ICE technologies;
• Make futureproofing allowances for replacing the incoming gas network fuel supply with a partial blend/100% hydrogen fuel mix to address operational carbon emissions;
• Co-locate the on-site gas-fired generation plant with suitable infrastructure for further carbon reduction or commercial benefits (e.g. carbon capture and storage, waste heat recovery and distribution, hydrogen electrolysis, electricity grid- balancing services). Several items were raised as part of this study and stakeholder engagement, identifying issues that should be clarified before formally progressing with further feasibility studies of particular sites:
• The regulatory and planning landscape is very unclear, with uncertainty of how a gas connection would affect environmental impact assessments, electricity generation permitting, resilience/uptime certification, Health and Safety Executive consultation, and insurance requirements.
• The current carbon intensity of the gas network fuel is significantly higher than the electricity grid’s carbon factors. Although a 100% green hydrogen fuel mix would address this, there are currently no firm plans from the gas network operators confirming when this would be completed, risking a significant increase in data centre operational emissions if carbon capture and storage plant is not incorporated.
• ICE and SOFC solutions are less suitable for Hyperscale data centre loads due to their land requirements, and whilst suitable modular solutions can be scaled up, the land area required for these options become significant for high power demands. Gas-fired turbines can provide significant power volumes with relatively smaller footprints, but these were not considered extensively in this study.
• Data centres require highly stable and consistent power supplies throughout the day. This therefore creates a steady base-level gas demand, meaning that fluctuations in gas network linepack volumes must be carefully managed to maintain supply stability. This ensures the data centre's control system can keep power delivery stable.

Lessons Learnt

This feasibility study highlighted the potential to supply the energy data centres required using onsite power generation using gas (or hydrogen/biomethane in the future) from the gas network. All the technology required to do this is commercially available and some data centres out with the UK are already generating their own power on site instead of relying on the electricity networks.

Although this study identified that it is technically feasible to power a data centre using onsite power generation from the gas network, it was highlighted in this work that the gas volumes that can be supplied are highly location specific. This is one area where this study could be built upon, assessing the capacity headroom of the gas networks to identify potential suitable locations for future data centres.

Another area where the work in this study could be built on would be around the carbon impact of using the gas network and onsite power generation instead of using energy from the electricity network. SOFC and ICE vendors who were engaged during this work all mentioned that their solutions could be fitted with carbon capture equipment. The economic and technical impact of this was not fully explored as part of this work.