2

I am trying to understand a temperature rise test that is being conducted over a switchboard. The test conforms to IEC 61439-1. The clause 10.10.2.3.1 (Method of test - General) regarding the temperature rise states about the constant value. It says,

The constant value is reached, in practical condition, when the variation at all measured points (including the ambient air temperature) does not exceed 1 K/h (1 kelvin per hour).

By this statement, I can infer that, if the temperature rise at a specific joint at 15:00 is 78.5 K and at 16:00 is 80 K, then the total variation is 1.5 K over the period of an hour. Thus this is not a constant value. At 17:00 it reaches 80.5 K. Thus now, the total variation is 0.5 K over the period of an hour. Thus this can be inferred as a constant value as per the clause, and we can stop the test. To simplify the scenario, refer below chronological sequence

  1. 15:00 - 78.5 K
  2. 16:00 - 80 K
  3. 17:00 - 80.5 K

The question arises is, whether my inference is correct? I am curious since, the test coordinator insisted to extend the test till 18:00 to check if the variation is below 1 K/h, which, as per my opinion, against the clause in the standard. Requesting all to please clear my thoughts if I am wrong.

For reference I am attaching a snap of the sentence pertaining to the clause.

enter image description here

winny
  • 13,064
  • 6
  • 46
  • 63
DaSnipeKid
  • 253
  • 1
  • 10
  • While I would buy your interpretation (assuming this is the last measurement point to meet the condition) I would ask : what is the cost of exceeding the test requirement by one more hour? Can that cost be justified by the assurance the 18:00 reading would give? Going beyond the absolute minimum is often a good idea. –  May 02 '23 at 14:31
  • @user_1818839, I agree that going beyond the absolute minimum is often a good idea, but let's assume I am very stringent to the test clause, and I shouldn't allow to go beyond the '1st passed slot hour' (provided my interpretation is correct). Is there any issue in that ? Also, it would be good, if you could elaborate what benefit I would get if I go beyond 17:00, when my test was passed at 17:00. Temperature would be fairly constant (provided my design is good). – DaSnipeKid May 02 '23 at 14:36
  • @user_1818839 I am just curious if my interpretation is correct. – DaSnipeKid May 02 '23 at 14:36
  • Re-read the first clause in my comment, but bear in mind I am not your independent reviewer. –  May 02 '23 at 14:38
  • @user_1818839 To answer your question, there is no such "cost of exceeding". We can always exceed. But I am unaware of the after effects of an assembly supplied with the FLA even after getting a constant temperature value. There are chances that there might be a permanent damage to the assembly. – DaSnipeKid May 02 '23 at 14:43
  • I think your interpretation is correct. What I think the spec writers are trying to do is to quantify what is meant by a steady state temperature, as it relates to this test. – SteveSh May 02 '23 at 14:47
  • 1
    The "cost of exceeding" is increased test time. This may or may not matter in your application/industry. We do temperature cycle testing on most of our products - cycling between hot and cold and dwelling at those temperatures long enough to ensure that all parts of the assembly are at the target temperature. The we test. The cost of adding a hour or two to each temperature plateau (dwell), could stretch out temp cycling by a day or two. – SteveSh May 02 '23 at 15:25
  • 1
    Thanks for writing. I understood well your question. The testing laboratory is correct to wait to have <= 1K because this is what is requested in the standard. The standard text says “constant” but should be better written “stable” . You only got <=1K in the last hour and not when was <=1,5K Using the opportunity, I will ask you a question: What was the maximum limit of temperature rise considered by the laboratory for the connection between the busbars and the main circuit breaker ? This value was informed by you, or the testing lab stated it ? Is the connection bare or plated ? – Sergio Feitoza May 03 '23 at 14:44
  • Thanks for writing Sergio. However, I am still confused about the "constancy". When the standard says that a constant value is reached when the variation is under 1K/h then is it necessary to extend the test for next hour as well ? It is 1K/h and not "1K/2h" ? Please correct me if I am wrong. Also, let's assume, even if the IEC would have stated, "stable" than "constant", then it should mention the number of stable hours. – DaSnipeKid May 03 '23 at 14:51

1 Answers1

1

You are correct in your interpretation.

Of course, if all you got is 1.0K/hr in the last 60 minutes, then you'd want to wait a bit longer for the rates to drop to say 0.8K/hr, to be sure you're below the "constancy" threshold even if measurement errors are included.

the test coordinator insisted to extend the test till 18:00

That's silly anyway, since what matters is the change in a 60 minute sliding window, not between whole hours or anything like that. Once you get abs(max-min) below say 0.8K/h on all points, you stop the test, on the reasonable assumption that the range will only be asymptotically shrinking down to some thermal noise level.

Now, of course this requires on-line data processing that constantly displays the range on all channels in the last 60 minutes. But that's super simple to get with any modern piece of temperature instrumentation, so I don't see a problem here. They should have software that shows the largest (worst case) range from all channels in a given sliding time window, and when that number falls to your acceptance level, you stop.

It definitely is not enough to just compare temperatures once every 60 minutes. They need to be sampled often enough that the changes between adjacent samples are much smaller than the acceptance criterion (say 10x smaller), and the temperature range on each channel needs to be determined in a 60 minute window, across all samples taken at the given sample rate.

You can have 80.0K at 17:00, 80.5K at 18:00, but 81.6K at 17:30. With the sliding window, you'll catch that. But looking at numbers once an hour, it will be missed, and such results are next to useless.

Kuba hasn't forgotten Monica
  • 32,734
  • 1
  • 38
  • 103