Power Systems for Data Centres – Data Centres Are the New Grid Challenge
Data centres have become one of the fastest-growing and most power-hungry infrastructure types in the modern energy system. Fuelled by the expansion of cloud services, artificial intelligence, and hyperscale data platforms, their development is surging across the UK and Europe. Unlike traditional loads, data centres are highly concentrated and inflexible, often requiring between 30 and 100 MW of firm capacity in a single location. This presents a new set of technical and regulatory challenges for developers, utilities, and grid planners.
While renewable generation, battery energy storage systems, and flexible industrial demand are now commonplace in network studies, data centres are a different type of challenge. They are not designed to export, shift, or reduce load in response to network conditions, and their uptime requirements are stringent, with continuous load profiles. This makes the electrical design and connection process of data centres both critical and complex.
At Blake Clough Consulting, we support clients with power systems for data centres design, integration, and feasibility planning required to deliver successful data centre projects. This blog explores what makes data centres different, why detailed electrical system design matters, and how grid connections for data centres can be optimised.
Designing for Availability, Resilience and Efficiency
The power system inside a data centre must ensure high availability while maintaining performance, efficiency, and reliability. This is typically achieved through a layered, modular approach. Primary grid connections often connect at medium or high voltage, typically 11 kV or 33 kV depending on the size and location of the site. Redundancy is provided by multiple feeders or diverse substations which deliver independent supply paths to ensure continuity. Step-down transformers and main switchgear are arranged in configurations such as N+1 or 2N to meet tier-based reliability standards. Generator sets are installed to support full-site operation during extended grid outages, and Uninterruptible Power Supplies are used to ride through power dips and maintain continuity during transitions. Power paths are carefully segregated to differentiate between critical loads (IT servers, storage, networking) and non-critical loads (HVAC, lighting, ancillary systems), with all components specified and coordinated to align with Uptime Institute Tier requirements and modular construction models, ensuring scalability and consistent fault handling as the site expands.
From the main switchgear, electrical power is distributed through remote boards and eventually to individual racks. This internal distribution system incorporates Power Distribution Units with intelligent monitoring, enabling operators to track real-time values for voltage, current, and energy use. Metrics such as active and reactive power, power factor, harmonic distortion, and phase load balance are monitored closely. Operators also rely on breaker status and alarm histories to ensure early fault detection.
Environmental Performance and Energy Use
Data centres are under increasing pressure to demonstrate energy efficiency and environmental responsibility. As their scale and strategic importance grow, so too does their contribution to global electricity demand. According to the International Energy Agency (IEA), global data centre power consumption reached around 460 TWh in 2022 and is expected to rise to between 620 and 1,050 TWh by 2026, depending on the pace of AI development and data usage. This positions data centres as one of the fastest-growing electricity-consuming sectors globally, as illustrated in the figure below.
Several factors contribute to this sharp rise in energy use. The rapid increase in artificial intelligence workloads, such as large-scale model training and inferencing, has significantly raised computational intensity and hardware requirements. These systems tend to run continuously and require much higher power density per rack, often reaching 30 kW or more. In comparison, traditional enterprise data centres typically operated within a 5 to 10 kW range. The steady expansion of hyperscale cloud platforms, real-time applications, and edge computing is also adding to base demand across the sector.
Various academic and technical literature highlights the increasing importance of sustainability in data centre operations. While power usage effectiveness (PUE) remains the most widely used efficiency metric, water usage effectiveness (WUE) and carbon usage effectiveness (CUE) are becoming increasingly relevant, particularly as climate-related disclosures expand and public scrutiny increases. Water consumption, in particular, is emerging as a critical environmental concern. Traditional evaporative cooling systems can use over 25 million litres of water per MW of installed IT capacity per year. For a single 10 MW facility, this could mean 250 million litres annually, which is equivalent to the water use of several hundred households. Industry-wide, total global water withdrawals for data centres were estimated at approximately 560 billion litres in 2022 and could exceed 1.2 trillion litres by 2027. In the UK, partial data suggests that the sector already consumes at least 10 billion litres of water annually, and possibly as much as 20 billion litres when all facilities are considered. These increasing numbers are driven largely by AI-related workloads.
These rising figures are prompting operators to invest in more efficient and sustainable cooling technologies. Solutions include air-cooled or adiabatic systems that dramatically reduce water consumption, closed-loop designs that recycle water on site, and the use of treated wastewater in regions where freshwater is scarce. Some facilities are also shifting to dry cooling in temperate climates or using hybrid systems that switch cooling modes depending on the season or time of day.
At the electrical level, data centres are increasingly adopting high-efficiency uninterruptible power supply systems using lithium-ion or flywheel storage to reduce conversion losses and improve load response. Air flow is managed using hot aisle containment and intelligent fan control, allowing dynamic adjustment to thermal demand.
IT utilisation is being improved through server virtualisation, container orchestration, and dynamic workload allocation. These methods increase the amount of computing capacity delivered per unit of energy consumed and reduce the prevalence of underutilised infrastructure. Real-time monitoring via data centre infrastructure management platforms provides detailed visibility into power usage, power factor, rack-level loading, and cooling performance, enabling more efficient planning and operation.
As artificial intelligence and digital infrastructure continue to scale, the pressure to deliver both performance and sustainability will intensify. Data centres must now be designed not only to meet resilience and uptime targets but also to demonstrate responsible use of energy, water, and carbon. These environmental considerations are becoming just as important as technical performance in the future viability of the sector.
Power Quality and Protection
Maintaining power quality is essential in environments where high-density computing introduces nonlinear loads and sharp demand variations. These factors increase harmonic distortion, introduce phase imbalances, and heighten sensitivity to transients. To address this, harmonic filters are integrated at various points within the data centre’s internal electrical distribution network, and voltage regulation devices are deployed to stabilise supply under dynamic loading conditions. Fast-acting protection relays are time-graded to isolate faults selectively and prevent cascading outages, and fault-level analysis supports the specification of protection devices to handle worst-case scenarios while maintaining coordination. Routine testing of switchgear, generator synchronisation equipment, and UPS transition capability is standard practice and is necessary to ensure that the systems within the data centre are operating as efficiently as possible. Compliance with local protection standards and interface protocols is also necessary to secure safe and stable integration with the utility supply.
Planning Risks and Strategic Grid Advice
For distribution network operators and transmission system planners, the connection of a large data centre represents a significant infrastructure event. These developments introduce concentrated demand loads that can exceed 50 MW, often triggering upstream reinforcement or network reconfiguration. Blake Clough provides engineering and strategic support across the entire connection process for data centres. This includes load flow and fault current studies, capacity analysis at primary substations and GSPs, and coordination with DNOs and National Grid. We regularly support developers with technical submissions, including G99 documentation and connection offer reviews. On-site infrastructure can be optimised using information produced by our studies, which can minimise cost and time while ensuring alignment with resilience and expansion goals. Our approaches reduce the uncertainty of the connection process and allow for earlier decision-making and more strategic planning.
Power Systems for Data Centres – The Future of Data Centres and the Grid
As digitalisation accelerates, the demand for data centre infrastructure will continue to rise. Forecasts suggest significant growth in both hyperscale and edge facilities, with regional clusters forming around major cities, technology corridors, and renewable energy hubs. This has major implications for grid planning, GSP capacity, and local reinforcement costs, but it will also open new opportunities for innovation.
Flexible demand technologies may allow certain types of data centre activity to shift in response to real-time signals which could reduce pressure on peak demand and support system balancing. Backup power systems, if correctly configured, could contribute to fast frequency response or capacity services. In areas of grid constraint, demand side flexibility may help unlock new capacity and improve connection viability.
Blake Clough continues to support data centre developers, utilities, and investors with feasibility studies, electrical design, and strategic grid advice. If you are planning a data centre project and need support navigating the technical or regulatory landscape, our power systems and connections team is here to help.