Increasing data centre power consumption in an energy crisis: the role of generator maintenance in ensuring power resilience

The global energy crisis is affecting the profitability and operational efficiency of businesses from all sectors, but for data centres and their inherently higher power consumption levels, the impacts are polarised. For the data centre market to achieve its projected growth trajectory, it must continue to adopt power-intensive next-generation IT technologies. And, it must do that in the face of an unstable grid, potential planned blackouts and suggested power rationing. That means rigorous investment in backup power and generator maintenance, at a time when energy bills are consuming investable profit. 

Hiked power prices as a result of Russia’s fuel sanctions, and the threat of planned blackouts or power rationing to better manage an unstable grid, are forcing data centres to redistribute investment away from key business functions like training and recruitment, and towards improving energy efficiency and securing power resilience.

The current geopolitical challenges and their far-reaching implications are recognised in the 2023 Uptime Institute Report – Five Data Center Predictions for 2023 – in which the authors acknowledge that the Russian fuel sanctions, along with technological challenges “will make the planning of data centre development and operation more difficult.”

Powering next-generation data centres

According to the Uptime Institute, IT hardware has been fairly standard in terms of its draw on mainstream server capacity for a few decades now, creating technical stability and relatively constant power and cooling requirement.

This has enabled data centre designers to accommodate several IT refreshes without major upgrades to server technology and the associated hike in power consumption that would come with it. Essentially offering the latest technologies, without an increase in energy requirement.

This grace period is now over. Power requirements for next-generation IT hardware are far higher, rack power density is increasing, and “hotter” processors are putting pressure on the performance parameters of existing infrastructure.

This rapid rise in IT power density means server power consumption is on a steep climb, creating a need for more power, at a time when the grid is unstable, and energy prices are at their highest in decades.

Extreme-density racks are now commonplace in technical computing too, as well as high-performance analytics and artificial intelligence training. If data centre operators want to successfully penetrate these niche markets, they will need to foot the heightened energy bill and maintain a robust power resilience regime until the situation stabilises.

Data centres take action

Demand for next-generation IT technologies will not slow. As a result, data centres are forging ahead with the essential upgrades required to UPS, batteries, switchgear and generators to accommodate increased power density.

This is a risk. With an unstable grid and planned blackouts still on the agenda for many governments, upgrades will need to be bolstered with a watertight energy resilience plan to protect against power fluctuations and total outages.

The importance of a power resilience strategy in an energy crisis

According to research from the Ponemon Institute, in its third “Cost of Data Center Outages” report, the total cost of downtime has continued to rise over the last six years – rising a staggering 38% to $740,357 per incident. That equates to nearly $9,000 per minute – a figure that many data centres will have to swallow if the grid becomes too unstable to sustain their increasing power consumption.

Many data centres will already have a robust generator testing and maintenance regime in place, but the use of load banks is often overlooked, especially when budgets are stretched. At a time when power outages are more likely, load bank testing should play an integral role in a data centre’s energy resilience strategy and it would be prudent to evaluate your strategy in line with the current landscape to ensure it is robust enough. With this in mind, what would be considered best practice for testing a backup power system?

Best practice load bank testing in times of crisis

Ideally, as a minimum all generators should be tested annually for real-world emergency conditions using a resistive-reactive 0.8pf load bank. Best practice dictates that all gensets (where there are multiple) should be run in a synchronised state, ideally for 8 hours but for a minimum of 3.

Where a resistive-only load bank is used, testing should be increased to 2-4 times per year at 3 hours per test. In carrying out this testing and maintenance, fuel, exhaust and cooling systems are effectively tested and system issues can be uncovered in a safe, controlled manner without the cost of major failure or unplanned downtime. The alternator is not thoroughly tested though, with a resistive-only test and therefore a resistive-reactive test would always be recommended.

It may be advisable to test more frequently during times of crisis for added peace of mind.

Why is resistive-reactive the best approach?

Capable of testing both resistive and reactive loads, this type of load bank provides a much clearer picture of how well an entire system will withstand changes in load patterns while experiencing the level of power that would typically be encountered under real operational conditions.

Furthermore, the inductive loads used in resistive/reactive testing will show how a system will cope with a voltage drop in its regulator. This is particularly important in any application which requires generators to be operated in parallel (prevalent in larger business infrastructures such as hyperscale data centres) where a problem with one generator could prevent other system generators from working properly or even failing to operate entirely. This is something which is simply not achievable with resistive-only testing.

Navigating growth when power is scarce

No matter the geopolitical challenges and the effect it is having on power availability, data centres have no choice but to grow.

Demand will not cease, and power-intensive next-generation technologies are unavoidable. Ensuring power resilience via a watertight backup power supply and a robust testing and maintenance regime will enable data centre designers and operators to grow, safe in the knowledge that, should planned blackouts, power rationing or grid fluctuations happen, the power will always remain on.