Refurbishing data centres for sustainable efficiency

We sit amidst the buzz of artificial intelligence and liquid cooling. However, a quieter, high-impact trend is unfolding – the pressing need to refurbish ageing data centres. With every leap forward in technology came infrastructure that served as a foundation for it. The data industry, born in the mid to late 2000s alongside the boom of the internet. And subsequently, cloud-hosted business software, social media and streaming, catapulted data into the limelight as a valuable resource. But the infrastructure built to meet these demands is ageing, in many cases reaching its second if not third decade. Upgrades are required, and the need for refurbishment is not a minor concern; it is a steadily increasing demand organisations must acknowledge to manage the industry’s exponential growth.

The factors demanding refurbishment

Refurbishing an older data centre over constructing a new one offers advantages for customers, especially those in colocation – organisations who rent data space to others. Existing colocation sites typically operate at very low utilisation levels, with the actual kilowatts (kW) consumed significantly less than the kW sold, often by about 50%. Investing in upgrades can unlock trapped capacity within the mechanical, electrical, and plumbing (MEP) infrastructure or enable data centre operators to tap into secured yet unused utility power, a precious commodity in all developed markets.

In addition to these operational gains, refurbishing existing sites has financial benefits. It incurs much lower Capital Expenditure (CapEx) than constructing new facilities. Furthermore, this approach also contributes to reducing embodied carbon by avoiding the demolition and re-construction associated with a new data centre build. Addressing embodied carbon is crucial in the fight against climate change, as it reduces the upfront emissions generated before a building becomes operational.

Refurbishments also facilitate broader sustainability benefits. Lifecycle upgrades of equipment enable data centre operators to adopt more efficient MEP systems. This subsequently lowers the Power Usage Effectiveness (PUE) and overall wasted energy on the site. For instance, choosing modern free cooling chillers and transformer-less Uninterruptible Power Supplies (UPS) over traditional cooling systems and legacy UPS significantly enhances energy efficiency through design.

Lastly, a pivotal benefit lies in the importance of power density. Converting data centres from the older system designs to contemporary approaches can achieve higher power densities, thereby delivering superior performance. By way of example, in the UK, Cundall successfully planned to refurbish some existing 2MW data halls with 2N mechanical and electrical infrastructure in collaboration with a client. The transition to an N+1 (3+1) system increased IT capacity from 4MW to 6MW while retaining most of the existing infrastructure. An N+1 system, denoting additional infrastructure such as another generator for reserve operation, enhances both the data centre’s performance and resilience, providing a valuable safeguard against unexpected issues.

Overcoming complexities

Data centre operators must understand the site’s operations under every scenario to design a large-scale upgrade that integrates with existing systems. This includes day-to-day usage where the system is operating normally, system adjustments during planned maintenance, and extreme events like emergencies or failures. Obtaining this information on older sites where records and construction specification information is sparse can be challenging, requiring teams to invest significant time on site investigations with the facility’s technical operations teams. These discussions often involve reviewing hard-copies of Operation and Maintenance Manuals and As-Constructed Drawings, which can be thousands of pages long and add complexity to the task.

Another complexity arises when upgrades are necessary on sites that must remain operational. This demands careful planning and sequencing of works to ensure a continuous power supply and cooling remain available to the whitespace to sustain full system functionality throughout the upgrades. Given the extended duration of such projects, often spanning months, temporary power and cooling solutions become essential. For high-end tenants and operators, the inherent risk of relying on a single N infrastructure (N referring to the number of generators as a level of data centre redundancy) for an extended period is not acceptable and may limit the opportunity to conduct planned preventative maintenance.

Moreover, the challenge extends to power density in older sites, where the norm was 2-5kW/rack, in contrast to the current industry standard of 8-12.5 kW for air-cooled data halls. This discrepancy is further exacerbated given the emergence of liquid cooling technologies which can push rack density significantly higher. Consequently, operators must drastically increase power and cooling capacity within the existing spatial constraints if they are to make the business and sales case stack up.

The issue of global warming is also categorically a concern that needs to be addressed when it comes to renovating data centres. During the construction of many older data centres, the highest external design temperature was around 35 degrees Celsius. However, London temperatures in 2023 reached as high as 40 degrees Celsius due to global warming. This increase in temperature affects the cooling systems within data centres and necessitates refurbishment to prevent these old data centres from becoming prone to failure.

Sometimes it’s better to fix than rebuild

Despite the hurdles outlined, retrofits and refurbishments are becoming core business for the data centre industry. For many organisations, establishing new data centres on greenfield sites means navigating issues including finances, sustainability, and time investments. So, while these upgrades are not a novel concept, the increasing commercial value of data and the growing competition for space within key locations, combined with the quantity of ageing infrastructure make them increasingly necessary.

Given all the challenges, particularly the chronic shortage of skilled professionals in the industry, there is a growing recognition that recycling existing infrastructure may be the most pragmatic approach. Through meticulous planning and leveraging foresight from industry experts, an opportunity exists to expand the sector’s footprint in a sustainable and economically viable manner. This strategic move aligns with the industry’s evolving needs and the imperative to balance growth with resource efficiency.