A level indication on the machine that does not agree with what is seen on site is a fairly frequent class of feedback. The first step in dealing with such feedback is not to dismantle the machine but to separate two situations of a completely different nature: one where the indication is on the wrong side and the oil has not actually been lost; the other where the oil really has been lost and has to be investigated as a leak or as draining that ran away. The order of investigation differs, and so does the cost of dealing with them.
First establish whether the deviation is on the indication side
On oil treatment equipment the common level detection is infrared level detection and foam detection, with an automatic defoaming device working with it. When the indication does not agree with reality, three directions are usually checked first. The first is that the detection point and the effective oil surface are not in the same place: if the oil in the tank has not settled completely, if there is disturbance or a foam layer near the detection port, the reading will not match the true oil surface after the oil has settled. The second is the difference caused by temperature: the volume of the oil expands as the temperature rises, so the level of the same tank of oil is different at different temperatures in the first place, and comparing a reading taken in the 45–80 °C range with the visual height of cold oil gives a normal difference. The third is that the display and the object being detected are not the same thing: a level signal and a foam signal do not mean the same thing in the control logic, and their action thresholds also differ.
The way to check it is to use the same tank at the same moment and compare by several methods that do not depend on the display: use a sight gauge or the oil drain to establish the actual oil surface; check the operating record for the batch to see whether the oil in and the oil out balance; look for alarm records, or for defoaming actions that are more frequent than usual.

Then establish whether the oil really is missing
If the check shows that the display agrees with reality and the oil in the tank really is falling, then it is a different class of problem. There are three common sources: first, seepage at a sealing face, where static and dynamic seals behave differently and an oil trace at a static point has to be traced along the pipework and the flanges; second, an oil drain valve or a bypass valve that has not been reset, which leaks oil slowly when the valve is left in an intermediate position; third, sampling or draining on site that has not been recorded, in which case the machine itself is normal.
This class has to be dealt with early. A loss of oil changes the level and the atomisation space inside the tank directly, and the degassing performance with it, so it is not only a question of loss but also a question of performance. When the oil surface is found to be falling continuously, stop the pump first and then see whether the level stops changing; this single action narrows the range down quickly to whether the loss is inside or outside the machine.
| What is observed | Which class it more likely belongs to | The check to make |
|---|---|---|
| The display reads low or high and comes close to reality after the machine has stopped and settled | A deviation on the indication side | After settling, establish the actual oil surface at the oil drain |
| The display reads the same whether running or stopped and differs from reality by a fixed amount | A difference in detection position or temperature | Keep the same measuring point and compare again at the same temperature |
| The level falls continuously with time and does not recover | The oil really is missing | Stop the pump and observe whether the level stops changing |
| The level still falls after the pump is stopped | A valve that has not been reset, or continuous seepage | Check the oil drain valve, the bypass valve and the pipework flanges one by one |
| Defoaming actions are more frequent than usual and the level indication fluctuates at the same time | Foam affecting the detection | Check the water content and the oil temperature of this batch |
The order of handling it on site
Here is an order that does not require dismantling the machine. First, treat the display and reality separately: measure the actual oil surface once by a method that does not depend on the display, which separates an indication deviation from a real loss. Second, if the oil has been confirmed missing, stop the pump and read the level once more; the range becomes clear at once. Third, separate seepage from draining: valve positions and sealing points each have their own order of checking, so look first at what can be reset and what can be seen, and only then start dismantling pipework. Fourth, write down the process of this check, including the oil temperature, the operating state and the number of defoaming actions at the time, so that the next occurrence of the same symptom can be compared directly.
A level indication is not a precise measuring instrument; treating it as a signal for detecting change is more appropriate than treating it as a ruler. Where the exact quantity of oil in the tank really has to be known, draining a sample and weighing it remain the most reliable method.
- Record the oil temperature at the time of every level check; figures taken at different temperatures cannot be compared directly.
- Read the level once more after stopping the pump and letting the machine settle; this one step separates most display deviations from a real loss of oil.
- When an oil trace is found on the outer wall of the tank, follow the pipework from top to bottom rather than judging it on the spot.
- Record the reset state of the valves in the handover notes, so that a problem from one shift does not carry over to the next.
- When the level signal is abnormal, record it before resetting; repeated switching destroys the basis for the judgement.
The display is a signal, not a ruler. Once an indication deviation is separated from a real loss of oil, no later step goes in the wrong direction.
