The data centre sector is expanding at an unprecedented rate, and Building Services Engineers will be at the forefront of developing designs, undertaking research into sustainable solutions, providing energy efficient facilities and new distribution topologies for providing resilient power networks, controls and cooling systems.
They may include harnessing low-grade waste heat from the data centres, using distributed heat networks, providing alternative cooling solutions such as immersed liquid cooling systems.
The digitalisation of our society relies upon the data centres, from hyper-scale to edge facilities and their power consumption and distribution arrangements will need to be reviewed and research undertaken in the context of reducing power and providing off-grid solutions for the 21st Century using a mix of onsite renewables, static and dynamic uninterruptible power supplies and demand side management and critical controls.
The following matrices currently aid developing efficient designs for data centres:
- Power Utilisation Effectiveness (PUE).
- Water Utilisation Effectiveness (WUE).
- Infrastructure Utilisation Effectiveness (IUE).
There have been various sector reviews and white paper publications on the subject over recent years and although they all have their merits, many data centres have differing and varying power densities and cooling demands that may sometimes lead to inaccuracies when predicting load growth relative to power network losses, control strategies and cooling plant optimisation, complying with ASHRAE standards (1) and IEEE recommendations (2).
Scalability and modulisation of electrical systems and cooling plant to reduce capital expenditure over the life cycle of the data centres is beneficial and Building Services Professionals will need to understand the nuances of system specification and load growth in relation to resilient and efficient designs and associated operational costs.
Although the design of power and cooling systems have been based on resilient Uptime Institute tier clarifications (4) namely Tier I, Tier II, Tier III and Tier IV, inherent design topologies closely following the plants and systems specified must align with system capacities without compromising the resilient critical power and cooling systems needs while considering operational costs over the life of the data centre.
Data centre organisations like Stack (3) are now quoting 300kW per server rack for Generative Artificial Intelligence (AI) digitalisation using “closed loop” water cooling systems will require a fundamental change in power delivery at rack level and upstream within the power distribution network and at electricity generation level.
Major standing organisations have been developing and offering microgrid solutions in various forms. However, there are practical design considerations and operational aspects that require Building Services Engineers, Consultants, Contractors, Manufacturers, Operators, Maintenance and Developers each need to consider for system designs that are modelled correctly and come to fruition for loads now predicted.
Such key considerations are listed below for clarity. Certain items will be interdependent:
- Actual power consumption of servers relative to power loadings over a given area.
- Capacity planning and future proofing.
- Cooling strategies and topologies.
- What are the most efficient cooling solutions over the life cycle of the data centre
- Cooling systems plant specification power consumption throughout varying loads.
- Scalability of systems without compromising resilience.
- Heat network design.
- Utilisation of low-grade district heating networks.
- Interfacing with district heating networks and analysing suitable applications.
- Availability and capacity of electrical and water utility supplies and capacities within the locations of the data centre centres and incorporation of renewables and battery storage systems.
- Theoretical peak electrical demand including parasitic loads and distribution losses.
- Plant efficiency and selection.
- Embodied energy and thermal modelling of data halls and plants.
- Total cost of ownership.
- Power system efficiency and equipment specification.
Microgrid design, power network integration, use of static and dynamic UPS and battery storage systems for balancing plants during grid transients or outages, protection, synchronisation and analysis. - Design and application of power system islands and connectivity.
- Alternative sources for off-grid embedded power generation and HVO fuels.
- Export of power to support the grid during peak demand periods.
- Voltage levels for large scale power infrastructures and balance of plant.
- Power network contingency analysis.
- Load flow and system harmonics.
- Power exchange.
- SCADA / PMS and BMS architectures and system resilience.
- Application of Generative Artificial Intelligence (AI) and Machine Learning (ML) for data servers, building services systems and plant maintenance.
The above is not exhaustive in relation to the growth of the data centre sector and the requirement for Building Services Professionals across the industry spectrum is to enhance their knowledge on how to deliver efficient practical design solutions; providing sustainable networks that can be constructed, traditional, modular or prefabricated that are maintainable and build upon this knowledge.
There are specialisms needed for each of the above to develop data centres. The role of the Building Services Engineer will need to drive the design and operation of these facilities, and their supporting infrastructures, forward in a holistic way embracing future technological trends and innovations developing alterative ways to design sustainable and resilient networks to meet the power and cooling loads now manifesting within this ever-evolving sector.