The target water content for insulating oil is ≤3 ppm (GB/T 7600) and the target gas content is ≤0.01% (GB/T 423). How low are those two figures? Low enough that a film of water left on the inside wall of the container, or a slight deposit on the sealing face of the bottle cap, is enough to change the result. A shaky sampling action can still be put right, but a bottle that has been chosen wrongly or a cap that has not been seated correctly gives the same reading whichever instrument is used afterwards.
When the target is narrow, the container and the seal become part of the measurement
The sample sent for testing travels in a container, and the container is not a means of transport but the last link in the measuring chain. The water content is determined to GB/T 7600 and the reading has to be accurate to the ppm level, and that precision requirement falls on the container as well: whether the bottle material lets moisture through, whether a fibre has been trapped between the cap thread and the sealing face, whether the container has been thoroughly dried after use, and whether a bottle that has held a wet sample has been reused as it stands all enter the result.
As far as the material is concerned, glass and plastic differ in how well they bar water vapour, and a bottle that lets moisture through allows the water inside it to re-equilibrate with time; a plastic container can also react with the oil, showing surface swelling, softening or deposits, which contaminate the space inside the bottle in the same way. Choosing a bottle is therefore not a matter of picking whichever is convenient, but of checking whether the material is compatible with the oil and whether the material of the cap seal is compatible with the oil and the temperature range.
The sealing method decides where the water and the gas go
The first requirement for a seal is isolation, the second is stability. Poor isolation allows exchange in two directions: moisture from the air dissolves into the oil, and water and dissolved gas in the oil escape outwards. Poor stability shows up differently: the container itself is closed, but the stopper has not been pushed home, an oil film hangs on the thread, and the cap looks closed without actually pressing on the sealing face, so the exchange still happens, only slowly enough to go unnoticed.
The two cases point in opposite directions. With an open container or a failed seal the reading may be high or low, depending on the air humidity, the temperature and the waiting time at the time, and the result cannot be reproduced; when the container is genuinely closed the water does not appear or disappear on its own, and only its form and distribution in the oil change. So before deciding whether a result is trustworthy, the first question is what state the container was actually in.
Dissolved water and emulsified water are not the same thing during the waiting period
Water in the oil exists in at least two forms. Water dissolved at molecular level is bound fairly tightly to the oil molecules and moves between the dissolved and the released state as the temperature changes; emulsified water dispersed as fine droplets is not stable, and standing it gradually coalesces and grows until it collects at the bottom of the container or in one layer. In a closed container the first redistributes with the temperature, while the second shows up during the waiting period as a continuous change of position and droplet size.
The effect on the test is not in the same direction for the two. When dissolved water is released as fine emulsified droplets, whether the test method counts them has a direct effect on the reading; after emulsified water has coalesced and settled, a different sampling position and different amounts of shaking and standing give different readings. The case of gas content is similar: once dissolved gas has escaped into the air phase at the top, GB/T 423 measures a state of equilibrium between the two phases, and the ≤0.01% limit is no longer representative.
A checklist to confirm item by item before the sample is sent
The checklist does not have to be long, but it has to be completed on the spot before the sample is sent and it has to leave a record. The items below are the most common places where marks are lost, and each of them can be seen on the bottle or asked about.
| Check item | What it looks like when it is not met | Which figure it affects | How to confirm it |
|---|---|---|---|
| The bottle has not been dried, or has not been treated after reuse | The water content is high across the board | Water content (GB/T 7600) | Confirm that the bottle, the cap and the seal are dry as well |
| The material is not compatible with the oil | The bottle softens or deposits appear | Water content, gas content | Check the material and the contact history, and do not use one container for different oils |
| The cap is not pushed home and a fibre is caught in the thread | The reading drifts and cannot be reproduced | Water content, gas content | Check the sealing face and the thread, and let the same operator carry it out |
| The temperature changes considerably during the waiting period | The dissolved and the emulsified states migrate into each other | Water content | Record the interval between sampling and sending, and avoid direct sun and heat sources on the way |
| The bottle is shaken violently or turned upside down in transit | The emulsified water coalesces again and liquid clings to the inner wall | Water content, cleanliness | Transport the bottle upright and handle it gently |
- One container carries one sample; do not mix them, and do not reuse a bottle that has held a wet sample.
- On the test request, in addition to the sampling point, add the time of sampling and the time of sending; these two determine what happened during the waiting period.
- When a re-test is needed, keep a reserve sample from the same batch, sealed in exactly the same way as the original.
- When the result differs considerably from the site judgement, check the container and the seal first, and look at the instrument afterwards.
An oil sample is not measured at the moment it leaves the tank; it only becomes data once it reaches the laboratory, and the container and the sealing method decide what it becomes in the meantime.
