Can data centres heat homes?

Why are so few data centres used to supply residual heat to district heating networks, and what can we do to change this?

According to the EIA (Environmental Investigation Agency), data centres are estimated to account for around 1.0-1.5 percent of global electricity consumption. Most of this power is supplied to the information technology equipment and is then dissipated as heat.

Subsequently, much of the remainder is consumed by the cooling system, which is focused on removing this heat from the technical space and conveying it to atmosphere by the most reliable and energy-efficient means possible.

What if, rather than being treated as a troublesome waste product, this heat was to be regarded as a valuable commodity, to be harnessed and reused?

Aligned with its commitment to the COP21 Paris Agreement, the UK Government has pledged to achieve net zero carbon by 2050. To realise this goal, decarbonising the UK’s housing stock must be a priority as this alone is responsible for approximately 14 percent of the UK’s carbon emissions. The primary contributor to this carbon footprint is the combustion of fossil fuels for heating.

There has been much discussion about phasing out the installation of gas-fired boilers in newly constructed houses, and there are propositions to discontinue their sale. Originally slated for implementation by 2025, the ban on new boiler sales has been extended to 2035, affording homeowners additional time to transition away from fossil fuels. As a result, finding affordable, low-carbon alternatives to traditional gas-fired boilers has become a priority.

Harnessing residual heat for enhanced district heating

The proposed sustainability targets, as well as the move away from traditional heating systems provide an enticing prospect. Redirecting residual heat from data centres to district heating networks offers advantages to be claimed by various stakeholders.

Data centre operators

Harvesting residual heat proves instrumental in reducing operating costs. This eliminates the need for a heat rejection plan, allowing resources to be allocated to critical construction elements like cooling.

Moreover, selling residual heat to Energy Service Companies (ESCos) provides an additional avenue for offsetting operating costs. Sustainability gains are also substantial. Shutting down parts of the cooling plant only modestly affects the data centre’s energy consumption and carbon emissions.

The true environmental impact lies in the reduction of offsite carbon emissions from district heating network users, previously reliant on fossil fuels. To improve the likelihood of this, data centre operators need to highlight the potential of Municipalities and Planning Authorities. This would bolster the facility’s environmental benchmarking and help secure Planning Approval.

Energy Service Companies (ESCos)

Data centres serve as a reliable and plentiful source of heat, offering a solution to customer needs while concurrently reducing carbon emissions. Utilising low-grade residual heat from data centres as the primary source for heat pumps enables ESCos to supply hot water to their networks without requiring centralised boiler plant. As data centres increasingly derive their energy from renewable sources, the residual heat becomes truly zero carbon.

Overcoming adoption challenges

Despite the advantages, the integration of data centres with district heating systems remains limited in practice. Notable examples are in the Nordics and The Netherlands rather than the UK. However, the primary hurdle lies not in the technical constraints but in economic considerations.

The challenge originally stems from the practical complexities of collecting and harnessing residual heat from data centres. Planning authorities actively encourage heat reclamation, but the lack of existing infrastructure poses a significant obstacle.

While planning conditions that mandate developers to allow for connections to ‘future’ heating networks is a positive move, this becomes futile where there is no corresponding plan for heat network development. Developers comply with the condition out of an obligation to meet regulatory requirements rather than in genuine expectation of the infrastructure ever being used.

From the perspective of data centre operators, investing in the infrastructure only makes sense when it generates Operational Expenditure (OpEx) savings through the reduced power and water consumption. However, the misalignment in load profiles complicates this matter. As the heating network’s demands peak in winter whilst reducing in summer, the data centre operates the opposite way, as it can take advantage of ‘free cooling’ during the colder months.

The misalignment in load profiles also impacts the ESCos. The viability of investing in heat pumps and pipework to connect data centres to their network hinges upon the assumption that the infrastructure will be fully utilised.

However, in practice, data centre operators will be hesitant to guarantee this, as the availability of residual heat is contingent on operational factors that may lie beyond their control. This uncertainty, linked to factors like speed of IT equipment deployment and maintenance shutdowns, makes committing to an agreed-upon amount of heat challenging.

Even when these factors can be solved, these is another large challenge. How much should the energy cost? The disagreement forms as data centre operators look to sell heat at prices that cover their infrastructure costs, while the ESCos aim to offer competitive prices to customers, ensuring their profitability and that their infrastructure and waste heat costs are met. Finding a mutually agreeable cost remains a key challenge in fostering a collaborative partnership between the two parties.

What are the solutions?

As a result of these challenges, devising a strategy that can solve these problems is imperative.

The UK confronts a critical need to expand its district heating network infrastructure, which significantly lags counterparts in the Nordics and other continental countries. Government intervention is needed, which could emulate the successful models implemented elsewhere.

Providing grants and tax incentives that can fund the cost of the new infrastructure would mean that planning conditions requiring data centres to connect to district heating networks might be realised. Building the necessary infrastructure would be challenging, but no more so than the creation of the UK’s broadband fibre network, or construction of the natural gas distribution network during the late 1960s.

The governmental support could also play a role in fortifying the business case for both data centre operators and ESCos. Implementing tax breaks on energy costs can be instrumental in this regard. This would incentivise data centre operators to contribute residual heat, coupled with subsidies for ESCos to harvest the residual heat from the facility. This would further push the two into a mutually beneficial, symbiotic relationship.

Addressing the challenge of the load mismatch requires a solution by building big. By developing extensive networks that serve various user types, both commercial and domestic, the demand profile is smoothed out.

This approach enables ESCos to capitalise on multiple heat sources based upon season, unit cost, and heat availability. Data centres could be selected to meet a summertime heating demand benefitting the operator in terms of energy savings and carbon reduction.

To minimise transmission losses and pumping energy, data centres – like Edge facilities – would need to be situated close to urban centres where the heating network’s users live. This might create planning obstacles, but if municipality planners came to see data centres as a source of low-grade, low-carbon heat, that could support the local community, then this consideration might add weight in zoning decisions.

The issue of reducing operating temperatures to promote decarbonisation is also paramount. Many district heating networks operate in the medium temperature range (120-175⁰C), and most are designed for a supply temperature above 80-100⁰C. However, these temperatures can only be achieved by burning fossil fuels and preclude the use of ammonia heat pumps combined with low-grade heat sources. To achieve de-carbonisation, it will be necessary to adopt much lower operating temperatures.

An example of success

These insights are supported by Cundall’s hands-on experience in Odense, Denmark and underscore the potential of effective residual heat utilisation. In a close partnership with the client, and local ESCo, Fjernvarme Fyn, we were responsible for the design of a hyperscale data centre that exports residual heat into the local district heating network.

Fjernvarme Fyn’s expansive network services 65,000 metered connections. It is linked to several different heat sources and user types, which smooths the load and demand profiles. The data centre’s integration into the district heating network is facilitated by multi-stage ammonia heat-pumps housed in a dedicated energy sector.

These heat pumps leverage low-grade residual heat from the data centre, boosting the temperature of the district heating network by 70-75⁰C. When fully built out, the data centre will provide 165,000MWh of heat per year. Notably, this heat is inherently free from carbon emissions, given the data centre’s reliance on renewable energy sources.

While replicating this model in the UK will require substantial investment and some political manoeuvring over several years, the demonstrable success of Odense signifies its potential. Harnessing residual heat from data centres to energise district heating networks is not merely a theoretical technical concept but is something that is being done elsewhere. If replicated in the UK, it would mark an important contribution to achieving 2050 net carbon zero targets for the UK.

Originally published in Data Center Dynamics in December 2023.