Battery storage at data centres: peak load, grid, bridging
What a battery does at a data centre: capping peaks, stretching a limited grid connection, providing grid services, bridging. And what it does not replace.
A battery at a data centre is neither a standby generator nor a larger UPS. It is an operating asset that makes better use of the grid connection, absorbs load peaks and earns revenue from the grid. Planned well, it is sized together with the connection, not after it.
What does a battery do at a data centre?
Four things. It caps load peaks, it stretches the grid connection, it provides grid services and it bridges short periods. Which of these leads depends on the site, above all on how much connection capacity is available and when.
Capping peaks. The IT load of a data centre fluctuates, and cooling and auxiliary systems add to it. The highest quarter-hour sets the demand charge in the grid fee and the capacity the connection must carry. The battery covers the peak, and the grid sees a flatter load. The principle is the same as in industry; the basics are in our article on peak shaving.
Stretching the connection. Limited connection capacity is now the norm. The battery allows more IT load behind the same connection because it carries the difference between average and peak.
Grid services. A battery that is there anyway can offer balancing power, supply reactive power and charge when electricity is cheap. That earns revenue and relieves the grid.
Bridging. If the grid drops briefly or the connection has to be interrupted for works, the battery carries the load until the grid is back or the generators have taken over.
How does a battery help when the grid connection is limited?
It keeps offtake from the grid below the contracted capacity even when the load inside the building is above it.
A connection agreement states a maximum offtake; exceeding it breaches the contract and costs money. Without a battery, the IT load has to stay well below that limit. With a battery it can approach it: the battery charges when load is low and discharges at the peak, and a margin that would otherwise lie idle becomes usable capacity.
The same applies to phased build-out. A campus grows in stages and so does the grid connection, just not in step. Until the next substation or the next connection stage is in place, the battery covers growth that would otherwise wait. It also makes flexible connection agreements under § 17(2b) EnWG workable, in which the network operator may cap offtake at times: what the grid does not deliver in those hours, the battery does.
The limit is energy. A battery holds power for hours, not days. It smooths and shifts; it does not generate. Where the average load sits permanently above the connection capacity, no battery helps. What is missing then is grid or generation.
Which grid services can a data centre battery provide?
Balancing power, voltage support and load shifting, as far as the reserve for the data centre allows.
Balancing power stabilises grid frequency. Batteries are well suited because they respond within seconds; participation requires prequalification with the transmission system operator. Reactive power holds the voltage at the point of connection, which network operators require for large loads in any case. Load shifting uses price differences on the power market: charge when wind and sun fill the grid, discharge when power is scarce.
The conflict is obvious: every kilowatt-hour committed to the market is not available to the data centre as reserve. So the capacity is split, one part reserved for operations, the other working in the market. The split is a decision by the operator, not a property of the technology, and it can be adjusted in operation.
Does the battery replace the UPS and standby generators?
No. It complements both and can take over parts of their role, but the availability of the data centre still rests on the generators.
The UPS bridges seconds to minutes until the generators are running. The generators carry the load as long as there is fuel: hours to days. A battery sits in between. It carries the full load for a limited time and can therefore take over the UPS function or soften the generator start. For an outage that lasts days, diesel remains indispensable, because energy is expensive in batteries and cheap in tanks.
Two points to keep in mind. First, permitting: above a certain rated thermal input, standby generators require a permit under the German Federal Immission Control Act (BImSchG) and form a procedure of their own. Second, the availability class: whether the battery may sit in the UPS chain is decided by the operator's redundancy concept, not by the developer.
What does the battery not replace?
The grid connection, generation and standby power for long outages.
A battery is not a source. It shifts energy that comes from the grid or from on-site generation. It does not turn an undersized connection into an adequate one if the base load sits above it. And it does not replace diesel where the requirement is to survive days without grid. Anyone expecting that is planning with the wrong technology.
How is the battery planned together with the grid connection?
Together, from the outset, out of the load profile, the connection capacity and the redundancy concept.
- Load profile. How high is the IT load in each phase, how much does it fluctuate, how much cooling comes on top. From this follows which peak the battery must cap and for how long.
- Connection capacity. What capacity does the network operator commit to from when, is there a flexible agreement, when does the next stage arrive. From this follows how much energy the battery must bridge.
- Point of connection. The battery usually sits behind the data centre's grid connection, on the same switchgear. The network operator assesses it as part of the grid compatibility study, including export if it is marketed.
- Permitting. A battery system is a structure with requirements on fire protection, clearances and noise. It belongs in the building application and in the discussions with the municipality, not in an amendment.
- Operation. An energy management system decides every second what the battery does: cap a peak, hold reserve, trade. The rules are agreed with the operator.
Bright & Green develops, builds and operates battery storage and data centres itself. We do not plan storage as an accessory but as part of the energy concept, together with the grid connection and, where it makes sense, with on-site generation. More on the pages battery storage and energy and grid.
Frequently asked questions
How large does a battery at a data centre need to be?
The task decides. Capping peaks takes power equal to the peak and energy for its duration. Bridging a connection stage depends on how long the battery has to carry which difference. Grid services add capacity that is marketed. Sizing therefore starts with the load profile, not with a number.
Can the battery earn revenue while securing the data centre?
Yes, as a shared battery. Part of the capacity stays locked as reserve; the rest works in the market. The energy management system holds the reserve at all times. The revenue is lower than for a pure merchant battery, but a battery the data centre needs anyway pays part of its own way.
Does a battery at a data centre need a permit?
Yes, as a structure within the building application, with evidence on fire protection, noise and clearances. The network operator also assesses the battery in the connection procedure. Both run in parallel with the data centre's own procedure if the battery is part of the plan from the start.
Which battery technology is used?
Lithium iron phosphate is the usual choice for stationary storage today: high cycle life, good thermal stability, no cobalt. The choice follows the task and the site, not a catalogue.