Q&A with ABB’s Data Centres Portfolio Manager Danel Türk

The recent wave of AI services is putting stress on data centres. AI is a double-edged sword as although its capabilities can help promote and facilitate energy efficiency, AI itself is a power-hungry beast and is putting an increasing pressure on the infrastructure that sustains it.

Danel Türk, Solution Portfolio Manager, Data Centers at ABB, is passionate about pioneering new and ever-improving technologies to support a more sustainable and better world.

Empowered to drive the change, in this Q&A with Energy Digital, he suggests actions data centre professionals should take to meet the increase in power demand in a cost-effective, sustainable way.

Q. What is causing an increase in power demand in data centres?

Since the rise of ChatGPT in 2022, there has been an exponential boom in AI-based services. AI and machine learning models, especially deep learning algorithms, require substantial computing power. So, demand for data centre power is rising and will continue to do so.

In fact, the International Energy Agency (IEA) recently predicted that global data centre electricity consumption would reach 1000TWh by 2026 because of AI — that’s equivalent to Japan’s annual electricity consumption.

We’ll need new, larger and more energy efficient data centres to accommodate this demand. So, it’s important that operators and managers know how to grow their facilities in a cost-effective and sustainable way that continues to support the global energy transition.

Ironically, while AI is driving the increase in power demand, it might also help solve it through optimising efficiency and balancing demand.

For example, employing an automation system running an AI suite can help optimise a data centre’s cooling system. It’s common for operators to see the cooling system as two separate entities — the upstream chiller and distribution system.

However, this means that efficiency improvements made to one can negatively affect the other, doing more harm than good to the overall system. Instead, AI provides a holistic overview so that you can ensure any changes made will be for the greater good.

It’s not just limited to cooling though, AI can be deployed across the whole data centre, monitoring hundreds of thousands of data points in a facility. It can then track performance, energy use and asset health, so that more informed decisions can be made to improve all those aspects.

Q. How can data centres capitalise on this increase in demand?

The most successful data centres will be those that embrace a total cost of ownership (TCO) mindset to investment and growth.

TCO goes beyond the initial purchase price of an asset to consider all the costs over its entire life — both direct and indirect. These include acquisition, operation, maintenance and disposal costs. In this way, it sees future operational cost savings as net present value.

Take the example of a UPS. They run for 15 years or more, so investing now in a more energy efficient model will result in long-term energy and cost savings, improving TCO. This can recuperate the higher initial cost in a few years and lead to massive savings over the lifetime of the UPS, especially when energy prices are high.

Of course, energy savings also mean emissions reductions, so it’s both a cost-conscious and sustainable decision.

The same principle applies to investing in more reliable equipment. Less maintenance means fewer spares and replacements and significantly reduces the risk of costly downtime. According to the Uptime Institute, 70% of data centre outages cost US$100,000 or more, with 25% costing more than US$1m.

Investing in reliability also helps meet sustainability goals because reliable equipment is inherently circular — it reduces waste, extends product lifecycles and promotes resource efficiency. Fewer service callouts means less travel emissions as well.

Q. What design considerations need to be taken?

Securing power from a utility is often the biggest challenge before expanding or building a new data centre. By building up in smaller blocks over time, rather than doing it all at once, you can speed up the process because the local utility and government will be more likely to approve incremental growth. It’s easier to agree to 10 20MW blocks over time than 200MW all at once.

This modular approach is also cost-effective and good for TCO because it avoids wasted capacity. Building up in 20MW blocks means you onboard enough customers to fill that capacity before growing further — optimising revenue in line with demand.

It also makes the specification and installation process much simpler, especially when using prefabricated skid or eHouse solutions that are pre-assembled and factory-tested before they arrive on site.

Q. Can you use the same equipment at higher power levels?

It depends on the situation, but data centres are becoming bigger and will continue to do so. Nowadays, 100+ MW data centres are commonplace. This has made a shift from low-voltage (LV) to medium-voltage (MV) equipment more attractive. 

Take a UPS as an example. At higher power levels, a single MV UPS can replace multiple LV UPSs and can often provide power protection for the whole data centre rather than just the server racks.

Furthermore, relying on a single UPS means fewer points of error, reducing the risk of costly downtime. MV UPS are also more energy efficient than LV and their lower currents mean you can use cables with a smaller cross-section. These benefits all add to a lower TCO.

Q. How can data centres grow sustainably?

Growing is one thing, but doing so in a way that meets sustainability goals is its own challenge. 

Finding ways to reduce your reliance on diesel generators is a good way to go green. Either investigate alternative fuels or incorporate battery energy storage systems (BESS).

A BESS/UPS combination can delay genset start-up and facilitates the integration of renewables into your energy mix, like wind and solar. A BESS can store excess energy from solar panels on the roof and use it at another time, like when prices are high during peak periods. This optimises renewables integration as well as TCO.

Incorporating a BESS can also help negotiations with the local utility, as batteries can provide load shifting and frequency response services, which will become increasingly important as the grid moves to renewables.

Lastly, it’s worth considering SF6-free equipment sooner rather than later, as regulations will be coming to most countries in the next few years against the use of SF6.

This insulating gas — used in some electrical equipment in data centres, like switchgear — can leak to the atmosphere, where it is 25,000 times more damaging than CO₂. SF6-free equipment is easier to maintain and makes sustainability reporting more straightforward and less time-consuming.

**************

Make sure you check out the latest edition of Energy Digital Magazine and also sign up to our global conference series - Sustainability LIVE 2024

**************

Energy Digital is a BizClik brand.