Blog · Grading findings

ΔT Severity Criteria for Electrical Thermography

How to turn a hot spot into a grade that holds up: NETA Table 100.18, the two comparison methods, load and criticality.

A thermal camera will put a number on every hot spot. Turning that number into a grade the client can act on, repair now, repair soon or keep an eye on it, is the part of the job that has to hold up when someone asks why.

In short
  • Grade by temperature difference (ΔT), not by absolute temperature.
  • Compare with a similar component under similar load first; use ambient air only when you have to.
  • NETA Table 100.18 is the common reference: over 15 °C above a similar component means repair immediately.
  • Record the load, weigh the grade against criticality, and keep the same scale every year.

Why ΔT and not temperature

A breaker terminal at 45 °C means little on its own. In a warm electrical room beside identical terminals at 42 °C, it is normal. Beside identical terminals at 25 °C, carrying the same current, something is wrong with that connection. The difference is the finding.

Two ways to measure it

Against a similar component under similar load

The preferred method. Compare phase A's lug with phase B's and C's on the same breaker, or one breaker's terminal with its neighbour carrying a similar load. Because both parts share the same room, the same load pattern and usually the same surface, most sources of error cancel out. Make sure the reference really is similar: same component type, similar current, similar surface.

Against ambient air

Used when there is nothing comparable to measure, such as a single-phase component or a lone connection. It is less forgiving: the reading now carries the full effect of load, emissivity and the air temperature you chose as ambient. Measure ambient near the equipment, not at the thermostat, and note in the report that the grade is against ambient.

Thermal image of a motor starter: the overload heater on one phase reads 77.1 degrees Celsius, the matching heater on the next phase 47.0 degrees
Similar-component comparison on a motor starter: the overload heater on one phase reads 77.1 °C (Bx1), the matching heater beside it 47.0 °C (Sp1). ΔT 30.1 °C, past NETA's 15 °C “repair immediately” band. HIKMICRO photo, DTI Suite SuperRes picture.

NETA Table 100.18

The ANSI/NETA Maintenance Testing Specifications give the temperature-rise guidance most electrical thermographers in North America work from. It has a column for each comparison method:

ΔT vs. a similar component, similar loadΔT vs. ambient airSuggested action
1 °C to 3 °C1 °C to 10 °CPossible deficiency; warrants investigation
4 °C to 15 °C11 °C to 20 °CProbable deficiency; repair as time permits
—21 °C to 40 °CMonitor until corrective measures can be accomplished
Over 15 °COver 40 °CMajor discrepancy; repair immediately

Summarized from ANSI/NETA MTS Table 100.18. Wording varies slightly between editions; check the edition your program or client specifies.

Note the third row: the ambient column has a monitor band (21–40 °C) that the similar-component column does not. Many reports collapse the table into the familiar three comparative bands, which is fine as long as the report says which column a finding was graded on.

Load changes the reading

The heat made at a resistive connection is the current squared times the resistance, so doubling the current makes four times the heat. A loose lug that is a mild finding at 30% load can be a serious one at the afternoon peak. That is why load matters as much as the camera:

  • Scan at normal operating load where you can. NFPA 70B guidance is no less than 40% of normal circuit loading.
  • Record the load with the finding, ideally amps on each phase.
  • Treat a reading taken at light load as a minimum. If the finding matters, re-scan at a higher load rather than estimating.

Severity is not the same as priority

A ΔT grade says how bad the connection is. It does not say how much the client loses if it fails. A probable deficiency on the feeder to an operating room, a process line or a fire pump is scheduled sooner than the same reading on a spare circuit. Good reports keep the two apart, a severity from the measurement and a criticality from the asset, and rank the repair list on both.

DTI Suite asset inventory list showing every piece of equipment with its rating, severity and criticality
Severity and criticality side by side for every asset, in the DTI Suite inventory list.

Pick a scale, publish it, keep it

Some programs use the NETA bands as they are. Others use a three-level traffic light, or a five-level scale like the one in our guide to writing an electrical IR inspection report. Any of them works if three things hold:

  • The criteria are printed in the report's methodology, so the client and the insurer can see how a finding was graded.
  • Every finding is graded on the same scale, by every thermographer on the job.
  • The scale stays the same next year, or the year-over-year comparison stops meaning anything.

How DTI Suite applies it

In DTI Suite you place spot and box markers on each radiometric photo and read the ΔT on the photo. Every edited photo is rated against your severity scale: the default, or your own, with your names for each level, the temperature-rise range for each and your colours, including a three-level traffic light. Findings are ranked for repair using NETA temperature-rise benchmarks and plotted on a severity × criticality matrix, and the report carries the same scale from one year to the next.

Questions

What is a ΔT in electrical thermography?
ΔT (delta-T) is a temperature difference: usually between a suspect component and a similar component under similar load, or, when there is nothing to compare with, between the component and the ambient air.
What ΔT is considered serious?
Under NETA Table 100.18, more than 15 °C above a similar component under similar load, or more than 40 °C above ambient air, is a major discrepancy to repair immediately. 4–15 °C over a similar component is a probable deficiency.
Should I compare with ambient or with a similar component?
With a similar component whenever you can. It cancels out most of the effects of load, ambient temperature and emissivity. Use ambient air only when there is no comparable component, and say so in the report.
How much load does a ΔT reading need?
As close to normal operating load as possible. NFPA 70B guidance is to scan at no less than 40% of normal circuit loading, and to record the load, because heat at a resistive connection rises with the square of the current.

Build the whole report in minutes

DTI Suite reads your radiometric photos, pairs them to your Excel asset list and writes the full Word report: summary, findings pages, severity matrix, re-inspection schedule and year-over-year trending. 30-day free trial, no credit card.

Start your free trial →