What’s the best and least disruptive way to go about check the condition of the CTs in our installations?
The traditional answer would be to sample the oil from the transformers, and to determine the moisture content of the sample either by dissolved gas analysis (DGA), or by the Karl Fischer titration method. However, this approach has some shortcomings. One is that the oil content of a typical HV CT is small, so repeated sampling to monitor moisture ingress into the CT over a period of time is not really practical. Another limitation is that the DGA and Karl Fischer tests determine the moisture content of the oil, but cannot be depended upon to provide accurate information about the moisture content of the solid insulation (usually papers) in the CT, which is often implicated in thermal runaway leading to catastrophic failure. A better option for determining the moisture content of HV CTs is to use frequency domain spectroscopy (FDS) testing.
How does FDS testing work?
The test set up for an FDS test is virtually the same as that used for conventional tan δ testing. The big difference is that conventional tan δ tests are carried out only at power frequency, whereas as FDS tests involve injecting a test signal that sweeps over a wide range of frequencies, typically from a millihertz or so up to around a kilohertz. By recording the response of the CT to this test signal during the frequency sweep and analysing the results, it is possible to reliably estimate the percentage moisture in the solid insulation. No oil sampling is of course required with this method, so not only is more convenient, it also means that testing can be carried out regularly with no risk of depleting the volume of oil in the transformer.
Is test equipment readily available for carrying out FDS testing?
Although FDS testing is a relatively new technique, easy-to-use transformer test equipment is now readily available. A good example is Megger’s portable IDAX300 test set. This performs FDS tests on standard CTs and captures the test data fully automatically, using pre-stored proven algorithms. It is also fully client configurable to meet any bespoke testing requirement.