Data centres are the backbone of the digital world – and their technical infrastructure must keep pace with growing requirements. But what steps do operators need to take now in order to work efficiently, reliably and flexibly in the long term? In an interview with eco, Karsten Paeth, Marketing & Sales Director Central Europe at ABB, provides insights into current challenges and possible solutions – and explains what operators should look out for today in order to be optimally positioned for the next ten years. At the eco Data Centre Expert Summit on 5 and 6 May, he will give the keynote speech “Strong Infrastructures as Enablers for Future-Proof and Scalable Data Centres”.
What are the key challenges and drivers for ABB for future-proof data centres?
In the energy industry triangle with the cornerstones of supply security, sustainability and economic efficiency, supply security is already a key focus for data centres. As a solution provider in the energy infrastructure of data centres, our central task is to implement the increasing requirements for efficiency and sustainability – which range from energy consumption in operation to product regulations – in an economically viable manner. This must be achieved in a market that is experiencing sustained high growth due to the exponential increase in data and the growing use of cloud services. It therefore also requires the technical capacity for future investments at a given location. To address this overall context, we consider the network infrastructure in terms of future expandability, develop concepts for implementing redundant structures in a scalable manner, and integrate sustainability goals into our product development to minimise CO₂ emissions during both the production and operation of the products used.
In your opinion, which infrastructure components are crucial for scalability and reliability?
One potential advantage of scalability is the ability to expand with established concepts in a unified system, thereby enhancing redundancy and reliability. This requires early-stage network analysis. For example, the combination of scalable solutions can enable short-circuit currents that may not be achievable in an individual system. By using appropriate equipment, such conditions can be limited so that the underlying equipment can be standardised in terms of performance metrics such as current, voltage and short-circuit current – regardless of connection location. This enables replicable solutions in the medium-voltage grid, which can be pre-engineered in combination with low-voltage switchgear, transformers and UPS systems in the form of skids, compact stations or E-Houses. These energy infrastructure components can then be produced in the required quantities with defined engineering and shorter manufacturing times, while offering maximum protection in the event of a fault – minimising downtime.
What technical requirements does the increasing modularisation of data centres entail – and how does this affect infrastructure planning?
Modularised solutions, such as the previously mentioned skids, CSS or E-Houses, are already widely used in the data centre market. Compared to many other industries, data centre customers are particularly forward-looking, meaning these solutions are often highly mature. The basis for using pre-engineered modules is early agreement between the customer and manufacturer on the data centre concept and future plans. This ensures optimal selection of the equipment used in the modules for both current and future requirements. For instance, increased power demand can be met by increasing voltage or current – thereby influencing the choice of primary or secondary switchgear in the module concepts. The previously mentioned short-circuit current analysis, which is crucial in infrastructure planning, also plays a role here.
What are typical weak points during operation when it comes to the energy and supply infrastructure?
Errors most frequently occur during installation or are triggered during operation. Any assistance that supports personnel during the installation and operation of data centres can therefore be beneficial. A simple, proven example is the use of bus systems for signal transmission. While this is already standard in industrial automation, direct wiring is still often used in medium-voltage systems – including in data centres – which allows more room for error. Condition monitoring can also help identify weaknesses before they become critical: Data centres generally represent a constant load (unlike industrial processes, where loads vary greatly due to motor start-ups). Nevertheless, a control system can provide early warnings when analysing operating data. Finally, due to the consistently high power consumption and associated heat loss in data centres, early detection of conditions exceeding normal parameters increases availability. The future development of AI data centres, in which power consumption may fluctuate more as the AI learns, makes predictive maintenance systems increasingly relevant.
How do you deal with the tension between ease of maintenance and maximum availability in practice?
At first glance, the contradiction is not particularly great. For example, in medium-voltage systems, air-insulated switchgear is primarily used at the primary level, while gas-insulated systems are often used at the secondary level – with different maintenance requirements. In both systems, it is ideal to know precisely what to focus on during maintenance, and to minimise the load in the event of a fault to allow for quick reconnection. This can be achieved by using fast-action earth leakage systems, which switch off within milliseconds in the event of a short circuit, thereby reducing the strain on the entire system. Preventive maintenance, on the other hand, indeed involves a trade-off – and not only in data centre applications. Every intelligent algorithm requires sensors for data collection and evaluation technology – that is, electronics – to assess operating states. Compared with the conventional electromechanics of medium or low-voltage technology, electronics are inherently more prone to faults and need replacing earlier. This is already known from relay technology. The trade-off that manufacturers and customers will increasingly need to evaluate is how much more operating data and support for action and system availability can be gained, while recognising that the support system itself is another potential point of failure. Therefore, concepts are needed that provide the benefits and eliminate the drawbacks by ensuring that, in the event of a failure, they no longer provide maintenance data but have no effect on system operation. In addition, they should be easy to upgrade on a regular basis, for example at similar intervals to relay systems.
How can strong infrastructures be combined with sustainability goals?
Four points come to mind: Firstly, using sustainably designed products such as low-loss transformers or air-insulated switchgear. Secondly, material savings through targeted system downsizing. A third key aspect is the use of intelligent and flexible digital systems – for example, using sensors instead of conventional converters – which both reduce material usage and enhance flexibility. Finally, active fault prevention, such as through predictive maintenance or the use of high-speed transmitters, helps avoid damage to systems at an early stage.
What long-term infrastructure decisions do operators need to make today to avoid reaching a dead end in ten years' time?
This question effectively summarises the above points. A well-thought-out system should meet several key requirements: Firstly, the system must remain expandable, and the modules used should still be technically viable even in a scaled-up system. Understanding both current and potential future network topology is essential. In addition, the design should be as location-independent as possible. Finally, a balanced combination of proven and innovative technologies is recommended – this allows system availability, sustainability and cost-effectiveness to be optimised in the long term.
Thank you very much for the interview!
