Mexico already has over 3 GW of installed wind power capacity and plans are in place to increase this to more than 9.5 GW by the end of 2018. This ambitious target can only be achieved if the installation and commissioning of a new plant proceeds smoothly and the plant proves reliable in operation.
To help achieve the necessary high levels of reliability, partial discharge (PD) analysis is routinely carried out on newly installed cables and has proved invaluable in detecting poor workmanship and splice defects. This was amply confirmed by a recent experience with a 6.3 km XLPE cable.
The cable was tested with a Megger TDS NT 60 kV test set using standards-compliant VLF cosine-rectangular test technology operating at a frequency of 0.1 Hz. This test technology is ideally suited to testing the long cables often encountered in wind farm applications, as these are difficult or impossible to charge with VLF sinewave equipment unless the frequency is reduced below 0.1 Hz, which then makes the testing non-compliant with IEEE 400.3.
For the cable at the Mexican wind farm however, the test set was configured to operate at 20 kV. Partial discharge activity was immediately detected on all three phases at the near ends of the cables, and at 1.8 km on phases L1 and L2. The near-end activity was of little concern because the connection between the test set and the cables is rarely PD free. The PD activity at 1.8 km was much more puzzling as the cable operator had no knowledge of any cable feature at this distance.
Fig 1. : PD activity shown by the TDS NT software in circuit 7 of the wind farm
Despite this, the indication provided by the TDS NT test set was clear and unequivocal, as is shown by the screen capture in Fig. 1. With this in mind, the cable operator checked the documentation for the installation and discovered that this had been updated. The new version revealed that there was a joint in the cable at 1.8 km, in exactly the location where unexpected PD activity had been detected.
The splices were removed and opened, and a visual inspection revealed that no sealing mastic had been used in the mechanical connector screw housings on the L1 and L2 phases. (See Fig. 2). This shortcoming was quickly remedied and, on retesting, the cable was found to be free of PD activity.
Fig. 2: Cicuit 7, L1 joint without seal mastic in the mechanical connector housing
This exercise in preventive and predictive maintenance revealed a problem that was possible to address relatively easily and inexpensively. Had the lack of mastic remained undetected and the cable been put in service, however, it would almost certainly have failed in a relative short time, causing costly disruption and damage.
With modern test equipment that uses cosine-rectangular waveform technology, experience on Mexican wind farms has clearly shown that PD analysis, even on long cables, is a convenient and cost-effective form of insurance against premature failure.