AI and growing demand for computing capacity are pushing more power into data centres. This increases cooling requirements and makes continuous optimisation more important.
A small deviation in a cooling system does not necessarily cause an outage. The facility may continue to operate, but with higher electricity or water consumption than necessary.
That is precisely why inefficiency can be difficult to detect without data from the individual parts of the system.
“If you don’t measure, you can’t know how efficiently the facility is running. Metering also makes it possible to see whether a change has actually improved operations,” says Max Noakes, Business Developer – Heat & Cooling at Kamstrup.
For Kamstrup, the point is therefore that metering should not only be used for billing.
In a data centre, it can also serve as an operational tool that makes deviations visible and allows operators to verify whether an adjustment has had the intended effect.
One KPI does not show where the problem is
Data centre operators already use overall metrics such as PUE to monitor the facility’s total energy efficiency.
But a single KPI does not necessarily reveal which part of the system is causing a change.
Kamstrup sees relevant metering points at coolant distribution units (CDUs), on the primary and secondary side of heat exchangers, in heat recovery modules and at rack manifolds.
The key is therefore not simply to have data, but to have data with enough granularity to make it operationally useful.
If a change is only visible in the building’s total energy consumption, it can be difficult to determine which component or process caused it.
Metering closer to individual parts of the cooling system can make it possible to track temperatures and energy flows and narrow down where a deviation occurs.
The purpose is not to collect as much data as possible, but to place meters where the information can be used to identify changes in temperature conditions or system control.
Max Noakes points, for example, to control drift and problems related to low delta T syndrome as conditions that can be difficult to detect at whole-building level.
With more detailed data, the operations team can first identify the deviation and then verify the effect once action has been taken.
This becomes more important as power densities increase. A percentage deviation that previously had limited significance can represent substantially greater resource use in a much larger facility.
Excess heat also needs to be measured
The same metering takes on a different role when energy moves beyond the data centre.
The electricity used by servers is largely converted into heat, and interest in connecting data centres to district heating systems is growing.
But potential excess heat and heat actually delivered are not the same thing.
When heat is to become part of an external energy system, it becomes important to document both the quantity and timing of delivery. That gives the data centre and the heat utility a shared data basis.
“With metering, you can document precisely how much excess heat has been delivered to the grid and when it was delivered,” says Max Noakes.
For the heat utility, timing also matters.
Heat demand varies with weather, season and consumption, while the data centre produces heat according to its own load.
Metering data can therefore be used to view the data centre’s contribution alongside other heat sources and give the utility a better basis for adjusting production.
This has both technical and communicative value.
The utility gains a better opportunity to understand the data centre’s contribution in relation to overall heat production, while the operator can document what is actually being delivered back to the energy system.
Experience from district heating moves into the data centre
It is precisely the connection between data centres and utilities that Kamstrup highlights as an area where its existing experience can be applied.
For many years, Kamstrup has supplied metering solutions to heat utilities, where data is used, among other things, to monitor the relationship between production and demand.
At the same time, the utility sector has long worked with strict requirements for documentation, metering accuracy, compliance and cybersecurity.
According to Max Noakes, some of the same requirements are now becoming increasingly relevant for data centres and other energy-intensive commercial and industrial buildings.
Denmark also has a particular advantage because district heating networks are already widespread.
That does not mean that excess heat from every data centre can automatically be utilised. But where location, temperature and infrastructure align, the data centre can become a heat supplier in a larger system.
Here, metering becomes the link between the two sides of the energy equation.
First, the data centre needs to see how efficiently it uses energy internally. Then it needs to document how much of the heat produced is actually delivered onwards.
For Kamstrup, this is where metering data gains its broader value—not as an end in itself, but as the basis for optimising operations, verifying the effect of changes and making the data centre’s interaction with the surrounding energy system measurable.
BREAKING