SRC Forum - Message Replies
Forum: Reliability & Maintainability Questions and Answers
Topic: Reliability & Maintainability Questions and Answers
Topic Posted by: Reliability & Maintainability Forum
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Posted by: Andrew Rowland
Date posted: Mon Jul 7 10:51:53 US/Eastern 2003
Subject: ALT Misunderstandings
Here's the scenario...electromechanical relays that are normally energized at ~262VDC except when charging the battery then they are exposed to ~277VDC. In one year these voltage levels are split about 8616/150 hours. The manufacturer rates the coil for 285VDC.
Several of these relays have failed with open coils after about 12-14 years in service. Two were sent for FA to two different companies. Both found the same "body of evidence". First company concluded cause of failure was being continuously energized. Second lab concluded that wire opened due to stress cracking corrosion.
In our attempt to determine the appropriate replacement periodicity, we "cooked" one relay at 300F, energized at ~260VDC for 480 hours and the relay DID NOT FAIL. From this we concluded that 4.7 years is the "right" periodicity. Normally energized relays is on our "Top 20" issues list, so we're going to spend about $175,000 every 4.5 years to replace normally energized relays.
Before I rain on the plant's parade, I want to make sure I'm correct on a few points.
First, the 300F test we did was based on the premise that the coil insulation would fail. This is not the failure mode/mechanism that we have experienced in the field. It seems to me than, this test was not applicable to what we know and the results cannot be used to determine an appropriate replacement periodicity to preclude open coils.
Second, the Arrenhius model was used to determine usage time. Would a temperature-voltage model have been more appropriate?
Third, since the relay we tested at 300F did not fail, I could reasonably argue that 6 years is also appropriate (or any other period of time).
Fourth, with only one relay under test, the range would be so great at any reasonable confidence level that the results are essentially worthless.
My opinion is that any replacement periodicity based on the work conducted to date would be based on poor engineering, at best. However, I would like the experts to weigh-in before I point this out.
Thanks for reading my long-winded account.
Subject: Relay Testing
Reply Posted by: BWD
Date Posted: Wed Jul 9 13:04:46 US/Eastern 2003
Performing a test with only one sample and then not testing to failure is not the way to make critical decisions. The data collected is not sufficient to decide that the wear-out is 4.7 years. As you stated, it could be much higher or lower as this one sample is not statistically significant. I would recommend that at least 20 or more sample be tested to FAILURE. From these data one could determine the start of the wear-out or L-10 life point. This information could then be compare to the operating data previously collected and a replacement time determined.