Is there enough load on each side of a dual-fed (2N) data centre for a meaningful thermography survey?

Usually yes, with planning. In a dual-fed (2N) data centre each path is built to carry the whole load alone, so in normal running each carries only part of it. A survey still finds real problems at that load, provided the load is recorded with every reading and the A and B sides are compared like for like.

Why each side runs lightly loaded

In a 2N design the A and B paths share the IT load between them, and either one must be able to take all of it if the other fails. So under normal conditions each UPS, switchboard, busbar run and PDU is working well below its rating. In a hall that is still filling up, the load can be lower again.

That matters for thermography because a poor connection only shows itself when current flows through it, and heat at a resistive joint rises steeply with current. At light load a developing fault can look unremarkable. See what load a survey needs.

How to get a survey you can rely on

Record the load with every image. The current on the circuit or device, read from the UPS, PDU or meter at the time, goes in the report beside the measured temperature.
Compare A against B. Mirrored equipment on the two sides usually carries a similar share of the load. That gives a fair reference: the same component, same model, similar current. A connection that stands out against its twin on the other side is worth a closer look.
Survey when the load is highest. IT load runs around the clock, but it is not flat. We plan the visit around the site’s busiest period and phase it hall by hall where that helps.
Say what a clean result means. Where load is light, the report says “nothing found at this load”, not “nothing wrong”.

Using the windows the site already has

There are times when one side carries much more than its usual share: planned maintenance on the other path, or the site’s own transfer tests. If those are already scheduled, the survey can be timed to image the side that is taking the load. We fit around them; nothing is switched just for thermography.

Commissioning is the other window. Load banks put a known, steady load through each path before the IT arrives. On a London data centre build, commissioning thermography by our team found 16 busbar sections reading above 80 °C under test load, the highest 103.5 °C at a tap-off unit, each located and imaged before handover. See busbar thermography at commissioning.

Re-survey as the load grows

A 2N site that was lightly loaded at its first survey will not stay that way. When new customers go live or a hall fills up, connections that looked fine start carrying real current. A follow-up survey at the higher load, compared against the first report, shows what has changed. See re-surveying after a load increase.

Ti inspects and reports. Any finding is investigated and put right by the site’s electrical team or contractor.

Frequently asked questions

Should one side be switched off so the other carries the full load?

No. Thermography never asks a site to reduce its resilience. We use load the site already has, or planned maintenance and test windows it already runs.

Can a low-load reading be corrected to full load?

There are methods to estimate it, but the result is an estimate. The report always gives the measured value and the load, and labels any corrected figure as an estimate.

Is a 2N site surveyed in one visit or two?

Usually one, with the A and B chains surveyed in turn, hall by hall. Larger sites are phased to suit change control and load.