When a site inspection report states that the breakdown voltage is qualified, it often means a binary result: above 65 kV is recorded as a pass, below as a fail, and no information is left in between. This way of dividing the value into one box is convenient for filing, but it compresses a large amount of useful information into a single word. When the same oil is processed over several consecutive batches and every batch passes, there appears to be no difference, yet the values actually measured each time may differ widely.
The margin carries the diagnostic information
The margin is the amount by which the measured value exceeds the pass line. It measures how far the current state of the oil is from the failure boundary. A batch that only just clears the pass line and a batch that stands clearly above it have completely different capacities to withstand subsequent contamination, rising water content and changes in gas content. The former may fall below the line if the operating condition moves slightly, while the latter tolerates fluctuation much better.
The second function of the margin is to connect the current re-test with the baseline of the previous one. A single pass is an isolated event; it is the sequence of margins that forms a trend. Only a string of values measured consecutively on the same oil, at the same sampling point and on the same machine can be compared with one another and can legitimately be used to talk about rises and falls; otherwise, putting two sets of figures together yields nothing but a half-true conclusion.
Three sources of a falling margin
- The condition of the oil itself is deteriorating: accumulated running time, incomplete removal of water and a slow rise in cleanliness all show up in the breakdown voltage.
- Contamination introduced during sampling: an unclean sample bottle, an imperfectly sealed bottle mouth, dripping or a second transfer can all make the particles or the water content of the sample higher than the actual state in the tank.
- Inconsistent sampling conditions: a different sampling point, a different residence time and whether or not the sample has been left to stand all introduce differences unrelated to the equipment into the comparison between two results.
These three sources look similar in the figures, all of them showing as a reduced margin, but the handling is completely different. One requires going back to the oil and the process, one requires revising the sampling procedure, and the third requires first confirming whether the other two apply before going any further. If the source is not identified and the machine is adjusted straight away, a problem in the sampling step is often turned into a process problem, the machine is adjusted repeatedly, and the cause of the fault is buried deeper instead.
A comparable trend requires fixed conditions
The precondition for reading a trend is that the data are comparable, and comparability is not automatic. Several items should be fixed across consecutive batches: the same oil grade, the same sampling point, the same machine and the same sampling method, including the same container, the same sample volume and the same standing time, with the test method recorded on the same record sheet. If the oil, the machine or the sampling point has to be changed midway, the boundary should be marked clearly in the record.
Data on either side of the break remain comparable within themselves; mixing them into a single curve and calculating an average over it is meaningless. Fixing the table columns has another use: when the record is reviewed six months later, it is possible to judge directly whether a given drop in the margin was caused by a change in the process or by a change in the recording conditions, instead of first trying to recall the operating details of the time.
Four indicators point to different failure directions
The breakdown voltage reflects only one dimension. Placing several indicators together is what makes the direction of the failure visible. A high water content points to insulation performance being impaired by water, a high gas content points to the risk of bubbles and discharge, a rising cleanliness class points to particle contamination, and the breakdown voltage is the combined result of all of these. The four limits come from different test methods and cannot substitute for one another.
| Indicator | Limit | Main indication |
|---|---|---|
| Breakdown voltage | ≥65 kV (GB/T 507) | Overall insulation capability |
| Water content | ≤3 ppm (GB/T 7600) | Impairment of the insulation by water |
| Gas content | ≤0.01% (GB/T 423) | Risk of bubbles and discharge |
| Cleanliness | NAS 1638 class 5 | Degree of particle contamination |
| Filtration accuracy | ≤1 μm | Filtration capability |
Stopping work as soon as the breakdown voltage is judged qualified can easily miss one direction: the gas content in the oil has already risen while the breakdown voltage has not yet fallen below the line. This combination indicates that the degassing step has to be examined, rather than continuing to treat the oil as qualified and raising the flow rate further to cover the problem, because what is being covered is the gas content itself and not the measured number.
What to do on site
Putting the above together, the handling sequence on site can be divided into three levels. The first level is re-testing. When the margin has dropped but the data itself is in doubt, take a new sample under the same conditions and run the test again, to confirm that this is not a fluctuation caused by the sampling and the operation, nor instability of the test procedure itself, but a single incidental reading mistaken for a real change in the state, which then leads to adjusting process parameters that were normal in the first place. This step is the easiest one to send the diagnosis in the wrong direction, so it is safer to stop and go through the original record once more.
The second level is going back to check the machine. Once the re-test confirms that the margin is indeed falling, and the water content, the gas content and the cleanliness have moved in a consistent direction, the operating state of the filtration and degassing steps should be examined. First look at whether the parameters have deviated, then at whether the machine itself shows an abnormality, and only after the phenomena and the parameters have been matched should a decision be made to open the machine for inspection, which avoids unnecessary shutdowns and dismantling.
The third level is treating the oil itself. When all the indicators are still within the qualified range but the margin for the same oil keeps falling across consecutive batches while the operating parameters of the machine show no corresponding abnormality, the problem is most likely in the oil. The oil then needs further treatment or replacement, after which the baseline has to be re-established, so that the batches that follow again have a comparable starting point.
The pass line is used to judge whether the oil can be used; the margin is used to judge how much longer it can be used. The former is a one-off conclusion; the latter is a continuous diagnosis.
