From compliance to commitment: rethinking biodiversity in project planning
How engineering and construction projects are evolving to champion biodiversity, through early planning, advanced technologies, and ecosystem stewardship.
Change your location to view content tailored for you.
Information and Communication Technology (ICT) and the associated storage of data in data centres is anticipated to grow and consume higher volumes of the world’s energy. In part, this is propelled by society’s insatiable need to stay connected, including the increase in processing capacity required to match artificial intelligence computing loads.
Cognisant of the need to decarbonise, data centre developers, owners, and operators understand the importance of considering sustainability and using energy and materials wisely.
In addition, the increasing demands from tenants to meet their sustainability goals is placing extra pressure on the need to decrease greenhouse gas emissions through the full asset lifecycle, from design and construction to reuse/repurposing, end-of-life and resource recovery.
All regions across Asia Pacific have their unique challenges, and opportunities, so the ways of designing data centres to reduce operational, as well as embodied, carbon is different for each region (Table 1).
| Country | Potential risks | Potential opportunities |
|---|---|---|
| Australia |
|
|
| Hong Kong |
|
|
| Indonesia |
|
|
| Malaysia |
|
|
| New Zealand |
|
|
| Philippines |
|
|
| Singapore |
|
|
| Thailand |
|
|
| Vietnam |
|
|
Table 1. The differing challenges and opportunities for sustainable data centres in Asia Pacific.
In some regions, there are regulatory and market challenges to accessing renewables, but in others, such as New Zealand, the grid is already powered almost entirely by renewable energy.
On the other hand, the sun doesn’t shine overnight, and the wind doesn’t always blow, hence renewable energy supply varies (or in other words, the emissions intensity of grid energy can vary hour by hour). This challenge is compounded by a lack of energy grid connection opportunities for renewables because networks – for instance in Australia – are predominantly the strongest where centralised fossil fuel production has traditionally dominated.
Furthermore, volatile climatic conditions and energy market conditions, which is the case for many of the emerging markets in the Asia Pacific region, like Indonesia, are increasingly affecting networks. It all adds pressure to data centre operations which are exposed to these volatile conditions yet absolutely cannot be impacted by loss of supply.
As a result, many data centres set up power purchase agreements (PPAs) with generators of renewable energy to underpin their renewable energy supply.
A range of complementary energy generation and storage technologies are also emerging for data centres, including:
Hydrogen – While hydrogen is everywhere, its cost-effectiveness is in its infancy as the distribution and storage challenges need to be solved, in particular the associated safety concerns attached to storage.
Biodiesel – As long as biodiesel comes from a sustainable source, greenhouse gas emissions may be decreased. However, compared with conventional diesel, biodiesel doesn’t have a long shelf-life and when unused, it would need to be turned over more regularly.
Batteries – Batteries may provide an alternative to diesel generation for short durations, and when charged with renewable electricity their operational carbon emissions are minimal. Although, if they’re not regularly used in a manner that reduces fossil fuel use, their relatively high embodied carbon emissions (around 100tCO2e/MWh of storage) do not get ‘paid off’ over time. For long-duration storage, it is especially significant given the capacity required for a large datacentre. Additional limitations and risks include the physical size of the battery required, and the risk of fire from thermal runaway for higher-density battery chemistries.
Beyond renewable energy, reducing overall energy consumption during operations is key for achieving sustainability outcomes. For countries with limited access to renewable energy, as is the case in Thailand, this is particularly pertinent for reducing carbon emissions during operations.
Five trends for designing energy systems to reduce overall consumption or reuse energy include:
The modelling of higher operating temperatures within a data centre is indicating that overall energy requirements would reduce, subject to the ability of equipment and microchips to withstand warmer temperatures while still retaining the comfort and well-being of building staff that maintain the server rooms.
Data centre servers produce significant volumes of waste heat expelled during the cooling of equipment. There are instances of data centres selling waste heat to neighbourhood district heating networks[1].
Although the heat from data centres is constant, its energy value can be difficult to extract. The two main reasons for this are: district heat systems would need upgrading to connect into the data centre, and waste heat from air-cooled server technology traditionally isn’t hot enough for direct transfer to occur (without additional equipment such as heat pumps to raise the temperature).
Displacing the heat elsewhere is an option being explored by some data centres, such as for commercial laundries, or even community swimming pools. It does depend on which technology (liquid or air) is used to cool the rack, but doing more with the waste heat is certainly a future worth working towards given the potential environmental and financial benefits.
Air as a cooling medium is beginning to reach its limits since its technology has been optimised as much as practicably possible over the past decade. With the wattage of graphic processing units (GPUs) increasing exponentially (there are trials of some exceeding 500 W per chip), liquid cooling, which involves using water or a refrigerant, rather than air, to cool and draw heat away from the IT hardware, is no longer a distant future choice but rather a reality today.
Liquid cooling removes the need for air-cooled systems, even in the tropics, and lowers energy usage as less energy is drawn than with air cooling. It also provides operators with the flexibility to possibly condense a higher number of racks into data halls, or reuse some of the waste heat more effectively, because it is expelled at a higher temperature than air-cooled technology.
Liquid cooling taken to the extreme is the notion of submerging entire data centres underwater (immersion), an idea already being piloted by Microsoft[2]. More common today in newly built data centres is the concept that liquid, typically water, cools electronic components of servers, at the same time increasing the processing capacity of racks without exponentially increasing energy use.
As energy grids and markets shift to be more dominated by Variable Renewable Energy (VRE), both the cost and emissions intensity of energy consumption will vary more strongly than it does at present through the course of the day, week, month, season or year.
Data centre designs that unlock value with consideration beyond Scope 1 and 2 emissions will reap benefits, which can be achieved by enabling grid consumption to be flexible rather than a constant 24 hours a day.
An emerging concept is temporal matching which allows data centre operators to determine ‘fraction of matching’ or ‘time matching’ for their renewable energy input. The concept refers to demand (energy drawn from the grid) being matched in real-time with specific renewable generation operation to demonstrate that the consumption, at that point in time, was powered by renewables.
Physical infrastructure options to achieve this include energy storage (a typical example being batteries), onsite backup generation (renewable, where possible), and thermal management (including chilled water storage or other ‘cold’ storage).
The aim is to decouple the operation of the asset (the server demand) from energy grid conditions (physical reliability and also real-time cost) to provide the asset owner with greater operational control.
Machine learning can be used in data centres for smart monitoring and management to continuously optimise energy usage and identify areas for improvement. Having optimum preset points for usage provides marginal energy gains. By way of illustration, current research indicates that, when aggregated, a machine learning centralised monitoring system may provide up to a 10 to 15 per cent energy saving on an installed cooling system. Artificial intelligence already helps some data centres to manage their intermittency of renewable energy by regulating and shifting the usage of energy, and maximising demand flexibility.
Rethinking the design of data centres with a focus on decarbonisation entails three key strategies:
An increasing focus is to reduce the embodied carbon of structures by considering multiple construction techniques, from traditional concrete and steel, to novel mass timber designs. If liquid cooling is adopted in new construction, developers may choose to reduce the overall size of the data centre as the technology allows for a greater density of racks in each data hall.
One challenge regarding building materials is that they must be able to support the incredibly heavy weight of data centre equipment. Consequently, the scope for use of lighter, renewable materials such as timber may be less than traditional materials, especially for multi-storey facilities. This increases the importance of using recycled steel and low carbon concrete either as stand-alone materials, or in a hybrid design with mass timber.
The location of a data centre significantly influences its performance, resilience, and compliance with regulatory requirements. By considering various factors, such as proximity to users, connectivity, environmental conditions, power infrastructure, regulatory compliance, and disaster risk mitigation, organisations can make well-balanced decisions when selecting a suitable location for their data centre facilities.
For any asset owner, running real-time feasibility studies to find suitable sites for locating a data centre (either greenfield or brownfield development) is crucial in the current climate of increasing data use and subsequent demand for storage.
To support this pressure on asset owners, Aurecon, with partner Archistar, developed a world-first data centre site selection tool. It provides clients with options for choosing the best site for a data centre to meet their needs, while also responding to local market, country, and demand conditions.
Situating data centres in cooler climates and accessing outside ‘free’ cool air, or optimising pre-cooling in lower temperatures, is being implemented in areas like New Zealand’s South Island and the southern regions of Australia. Needless to say, demand for data centres is growing significantly in hotter, more humid regions, such as South-east Asia, where this approach is not feasible. One challenge to note is that as global temperatures rise due to climate change, today’s new data centre assets using this type of air-cooling methodology may not be able to maintain such an approach in the future.
As data centres must be large enough to serve customers yet condensed enough to fit into tight urban spaces, modular design and construction may be a viable solution.
The approach provides fit-for-purpose spaces for data centre operations, together with reduced embodied emissions during the construction stage. As modularisation involves offsite fabrication of large components, with onsite assembly, it can result in less waste, fewer truck movements, less use of hot-powered tools on site, less noise and dust, and a safer, more productive construction site.
One such case is in space-constrained Singapore, where the Global Switch Data Centre was the first large-scale project in the country to use prefabricated modular techniques for its building services.
A higher efficiency in use of materials, and therefore less wastage, can also be achieved using digital engineering tools. A good example was the 3D model developed for the design of Equinix ME1 in Melbourne, Australia, which allowed stakeholders to review the project in the virtual world and make critical design decisions before construction commenced. In Brisbane, a former bank building was converted into NEXTDC’s B2 project, requiring an unconventional design of configuring the modularised plant equipment external to the building, to allow the necessary number of data halls to be accommodated internally.
Some of the strategies and innovations highlighted in this insights paper may be short-term applications, albeit others will require long-term planning and reliance on technology advancements. But all factors contribute to the process of identifying efficiencies, formulating the optimal energy strategy, and ensuring the sustainability of assets for their full life cycle.
Looking forward, digital trends will continue to drive data centre growth – artificial intelligence, autonomous vehicles, robotics and the 'Internet of Things', so whatever we can do now to decarbonise this mission critical infrastructure will be beneficial to future operations.
[1] Local community buildings in Ireland to be heated by Amazon data centre.
[2] Microsoft finds underwater datacenters are reliable, practical and use energy sustainably.
Client Director – Australia
Industry Director, Infrastructure & Data Centres – Southeast Asia
Senior Consultant, Sustainability – Australia
Design Director – Australia
Industry Director, Digital Infrastructure – APAC
How engineering and construction projects are evolving to champion biodiversity, through early planning, advanced technologies, and ecosystem stewardship.
Aurecon shares insights on how the 2032 Brisbane Games can deliver legacy infrastructure and lasting economic, environmental and community value.
This fact sheet compares the community and environmental impacts of wind farm projects, and how ecologically sensitive design can deliver better outcomes.
