Using whole life carbon assessment to optimise design for data centres
Design of building services for data centres is a crucial aspect of reducing both embodied and operational energy emissions. This is particularly true in locations where geography and climate conditions combine to make it virtually impossible to source 100% locally-generated renewable energy to supply data centre operations.
For a recent data centre project, our team undertook a detailed study into the carbon associated with the whole lifecycle of the asset, including materials, energy use for construction/commissioning, MEP & electrical equipment, operational energy use, materials/energy associated with repair and maintenance, and end-of-life emissions from decommissioning.
The development was to commence with the demolition of an existing obsolete data centre followed by reconstruction of a new facility.
The calculations for building fabric showed that re-use of materials obtained during demolition would reduce new material embodied carbon. These calculations are likely to become more standard for most construction projects as the industry addresses embodied carbon in materials such as concrete, steel and glazing.
The building services carbon data gap
Where there is a significant industry embodied carbon data gap is in the carbon implications of building services materiality and related lifecycle energy and water consumption.
Data centre equipment consumes substantial amounts of electrical energy, and conventional data centre cooling approaches also consume large amounts of water. Water is an issue both from the perspective of constraints on its supply and availability, and also from the perspective of the energy used and emissions generated by the systems that supply, purify and reticulate it.
Energy modelling is a key tool we use to understand the operational emissions of the facility and its building services and systems. But there are generally many assumptions and estimations involved where we are calculating energy at design stage. Over the average 25-year lifespan of a data centre, many things can change, including equipment replacements and upgrades, changes to the density of data racks and changes to the emissions footprint of the energy supply and the water supply.
Lifecycle analysis of an Uninterrupted Power Supply (UPS)
One of the important decisions made in designing building services is the system used for ensuring continuity of power supply. It is critical to maintain data centre operations in the event of grid disruptions.
We used carbon as a unit for comparison to calculate which of two types of system would deliver the best outcome for the project from a lifecycle perspective.
The first was a diesel rotary uninterruptible power supply (DRUPS). This system comprises rotary flywheels that continually spin generating kinetic energy which is transformed into electrical energy if the main power supply from the grid fails.
The flywheel-generated power will maintain continuous operations of the DC systems until the main diesel back-up generator comes on-line. Another aspect of a DRUPS system is it can provide power factor correction, that is, smooth out irregularities in the mains voltage supply, and this has benefits for equipment life and reduced maintenance.
The alternative we modelled was a static uninterrupted power system (SUPS). This type of system will often comprise energy storage batteries which are kept charged via the mains grid supply. Smart switching as part of the SUPS enables the batteries to discharge and provide for DC energy needs until the back-up generator kicks in, if there is a mains grid power outage or interruption.
Our analysis showed that over a 25-year lifetime DRUPS will have a carbon footprint around 54.3% less than a comparable SUPS system, assuming the engine block, alternator and other major components are all fully recyclable with established technologies.
Further, a comparison between flywheels, Li-ion (Lithium-ion) batteries and lead-acid batteries was conducted, showing that total whole life carbon emissions of Li-ion and lead-acid batteries are both 97% greater than the flywheel. So, a flywheel is more environmentally sustainable, and also avoids the human health and safety risks associated with lead and Li-ion technologies.
Lifecycle analysis of cooling solutions
We also examined the carbon impact of cooling technologies for both operational energy use and for water consumption.
By analysing the water consumption per year of three combinations of chiller plants, we found that a chiller plant design schematic comprising water cooled chiller, precool and air cooled chiller can achieve 36.2% and 35.2% water saving compared to a water-only cooled chiller plant or a water cooled chiller plant with precool combination. This water saving can provide enough water for around 500 families for the next 25 years of the data centre operations!
The water savings also offset the power use efficiency carbon gain by reducing the pumping power and reverse osmosis power in the order of 340 tonnes CO2e over a 25-year timeframe.
As a third piece of research, we undertook whole life carbon analysis of three combinations of air-cooled system and liquid-cooled system over a 30-year timeframe. The data showed that using liquid cooling on all technical suites can achieve a measurable reduction in carbon footprint than a baseline conventional cooling approach.
Ultimately, the final design decision is a matter of finding the optimal balance between solutions that will use more energy, but less water; and systems that will use more water, but less energy; and factoring in whole life carbon as a lens to validate overall sustainability gains.
By undertaking this detailed analytical work, we help both the building services design team and the client feel confident the most responsible and efficient outcome will be achieved.