Why an isolated value says little
A result is interpreted with general criteria, but each unit has its own baseline. A value that is acceptable according to a guide can be alarming if the equipment had been far below it for years, and one that appears out of limits can be normal for equipment with many years of service and good behavior. History turns a data point into information.
What to gather
- Results of earlier samples for the same tests, with dates and laboratory.
- Operating data: load, oil temperature, trips, overloads, past short circuits.
- Interventions: filtering, regeneration, oil change, repairs, replacement of seals or desiccants.
- Sampling conditions: sampling point, temperature, time, type of container and method.
How to make the comparison
- Verify that they are comparable: same sampling point, similar temperature, same test method. A change of laboratory or method introduces jumps that do not come from the equipment.
- Calculate the trend: the slope between samples matters more than the latest value. A sustained increase is a signal, even if none of the values exceeds a limit.
- Check against events: a sudden change after an overload, a repair or the opening of the equipment has a probable explanation. With no apparent event, suspect a developing problem or a sampling error.
- Correlate tests: water, dissipation factor, resistivity, acidity and dielectric strength move together in characteristic ways. An inconsistent combination points to a measurement error.
- Confirm with a second sample before acting on an unexpected result.
An example
If dielectric strength falls and water in oil rises relative to the history, and the equipment also ran hot during the period, it is reasonable to think of moisture ingress or release from the paper. If the strength falls but the water does not change, particles or the sampling should be reviewed.
How to interpret the result
The conclusion must state the condition, the trend and the suggested action, and the next sampling date must be adjusted to the observed rate of change, not to a fixed calendar.