Do any of you take "Nominal measurements"?
While I am in the field, when I see well- loaded normal circuits, I take a thermograph as a "nominal shot". I have been compiling the data in a spreadsheet for a while. Sometimes time constrains, but I do have some data however not yet enough to claim statistical significance.

Anyway, I have arranged it in a spreadsheet according to wire gauge, wire temperature, and ambient reference in the same thermograph.

I have looked on the internet for a large set of such data and have not seen any. I have seen board level trace temperature calcs, but really nothing on the industrial scale.

Unfortunately in the past many have used, actually in my opinion MIS-used, the NEMA and IEEE specifications that exist for all electrical equipment. These are intended to represent the "not to exceed" temperature for an entire component.

What thermographers face is not uniform temperatures over the whole component but extremely high temperatures over a very small area, i.e. high-resistance heating. There are no standards that I know of that can help us with this reality.

What "alarms" might be useful. Other than melting temperatures, a useful alarm is the annealing point of any metal that depends on spring tension to function; this includes all switches, disconects, clips, etc.

John:
I am looking for a statistically significant data set of real world measurements representing "nominal" conditions where specific wire gauges, loads, and ambient temperatures are recorded, not for a theoretical model based on physics.

i.e., for a given wire gauge, load, and ambient temperature, what is the standard deviation of the measured temperature of the wires? How influential are the remaining variables?

Within the data set we could look for conditions or outlyers where anomalies could either be hidden by typically less influential variables or made to look worse than they really are. There is a lot of information that can be gleaned from such data, and I will continue to collect it. When I get 1 wire gauge with a good sample set, I'll post the data so you guys can see it.

Anyway I was hoping to encourage people to take extra nominal Thermographs when they have time in the field, and compile the data somewhere on the net. This sounds like it would be a cool project for a website, but I don't know exactly how it would be set up..

Martin - you bring up an interesting point.

In vibration, even if we don't know exactly what's going on in a machine, we always have a very easy comparison with the sister components and the trend over time for the individual component (I'm talking about the plant vib guys, not a contractor brought in for one-time survey). Most vibration guys would look down on any attempt to try to make a call based on a snapshot and absolute limit, ignoring trends.

But that's pretty much what we do all the time in thermography.

In thermography the way most plants do it (ours included), images are only saved "by exception". Even if you did save images, it would take a lot of work to go compare invidual locations on the images. This might be faciliated by extracting relevant temperautres into a spreadsheet as it sounds like you have done.

For connections, I don't feel that bad about this state of affairs. Almost always we have sister connections to compare to as well as the wire far away from the connection (connection shouldn't be hotter than the wire except for heat conducted from what it's attached to).

But there are other components, particularly relays and small transformers inside panels where we often don't have anything to compare to.

We had an auxiliary transformer inside of an inverter overheat to the point of failing the internal windings which resulted in inverter failure which was the initiating event for a nuclear plant trip (there were other factors that contributed to the trip as well) with a lost-generation cost of several million dollars, and other undesirable consequences. Post-mortem indicated it had been heating for a long time. We don't have any easy way to see that unless the guy behind the camera is alert enough to recognize that it looks different than it did last year or in th other panel, or unless we take special trending as you mention.

Maybe someday when computering power and memory is super-cheap, we can save all images into a route (similar to vibration route) and the software would use image processing and expert rules extract the relevant features for trending and give us statistical alarms from the historical database. The machine wouldn't make the decision... it would just flag the change to the predictive maintenance guy who would review further considering other factors including load etc. Someday...

Pete:
That's not really my point though..I don't know that it will ever be practical to save all images, and that is not my intent. My intent is to use nominal images to establish a mean for given conditions, around which deviation sets can be calculated. For example:

For a given wire gauge, at a given load, the circuit is expected to be "X" degrees over ambient within a certain margin of deviation.

I've also been tinkering with "Watts used by the Anomaly" as a measure of severity, but this can be a little tricky, based on voltage drops across the anomaly.

Pete,

Didn't you post a logic discussion similar to what Martin is proposing a while back here - maybe the old reliability magazine boards???

I like the idea and I have also thought of providing some type of go-no go criteria for the guys performing inspections...

We have baselined most of our MCC breakers and having this additional info could only help in increasing early warning of a pending issue.

I like the comparison to vib criteria... it's similar in philosophy.

8g
Amps Temps© amb diff
18 28
13.7 28.7
33.2 44.3 23.2 21.1
35.1 46.6 23.2 23.4
34.2 43.9 23.2 20.7
14.8 32.4 26.2 6.2
13.8 33.9 26.2 7.7
14.1 32.9 26.2 6.7

Above is a sample of some of the data I have collected for 8gauge wire. The first 2 readings I took no ambient temp. It takes 30 seconds to take the thermograph and another couple of minutes to enter the data from the thermograph into the spreadsheet.