John Allen
Sheppard Engineering
Clachan Hydro Power Station (HPS) went into service in 1956. Clachan HPS is located on Scotland’s west coast about 40 miles north of Glasgow. The underground power station is at the head of Loch Fyne sea loch. See Figure 1. The tailrace discharges into the Fyne River, a salmon fishing river. Loch Fyne has a renowned fishery and seafood restaurant within a mile.
The 900 ft (275 m) head vertical shaft Francis turbine driven 50 MVA, 40 MW 428.6 rpm 11 kV generator was designed by English Electric. The generator stator was recored and rewound in 1984 by Peebles Field Services (acquired by Dowding & Mills in 1998).
During the 1984 rewind, the original split core stator was rebuilt as a complete annulus. And the Class B winding was replaced by a resin rich Class F epoxy winding. The turns were insulated with Samicaflex insulation tape and the slot cell would typically have been an S5 mica tape; this is a 180g/m2 epoxy mica tape on a glass fabric. The dielectric stress for the slot cell (wall) insulation was very conservative at 41.9 v/mil (1,650 v/mm).
The rewind used 5 mm (0.197”) epoxy glass wedges with Nomex 410 packing and 4 mm (0.157”) phenolic glass coil separators. The punched slot width was 22.15 mm (0.872”) which would typically have resulted in a built slot width of 21.8 mm (0.858”) and the specified slot cell width was 21.1 mm (0.831”).
As part of Scottish & Southern Electric (SSE) program of power station refurbishment, Clachan HPS was refurbished in 2000. The program included refurbishment of the generator stator with an option to rewind if the partial discharge levels could not be improved.
Dowding & Mills refurbished the generator stator, re-insulated the rotor and replaced the DC exciter with a brushless excitation system. They also replaced the complete station control systems together with all the low voltage (LV) and high voltage (HV) electrical installations.
Condition of generator stator
SSE had used an independent organization to assess the condition of the generator stator. This report included the following:
- Results of a very limited visual examination of the stator indicated that the stator end windings, although contaminated with an oil/carbon film, appeared in generally satisfactory condition for its age/operational life.
- Partial discharge (PD) signs found in the region of the winding conductor slot emergence in the bottom of the stator core end winding was a concern. It indicated possible signs of early-stage development of slot discharging/stress grading discharging which is known to be progressive.
- Results of the very limited wedge tightness survey indicated approximately 50% of wedges surveyed had developed some degree of slackening, ranging from complete to 25% of wedge length.
- The consequences of which are that slackening will continue in service, permitting bar vibration and resulting damage to the corona shield leading to the development of serious slot discharging and finally (if not corrected) insulation failure. Experience indicates that the time to failure following development of slot discharging may be less than a year.
- Results of HV PD tests indicated low levels of PD activity indicative of a satisfactory winding main insulation, but with isolated areas indicating early stage initiation of possible slot discharge and/or stress grading related discharging.
- If refurbishment is carried out (requiring the removal of the rotor), supplemented and supported by rewedging and Tennessee Valley Authority (TVA) probe tests, a probable future operational life of 15 – 20 years should be possible (providing a program of regular inspection and HV insulation testing are implemented).
- On the basis of the stator wedge tightness condition, it is considered that if a stator rewind is not carried out, the stator should be rewedged as soon as possible to prevent development of slot discharging. On-line condition monitoring should be fitted to monitor the ongoing condition of the stator winding.
This generator condition report provided guidance when the generator stator was refurbished. During disassembly, evidence was found supporting the generator assessment report.
Due to bearing seal leaks there was extensive contamination and together with poor exciter commutation, carbon dust was sticking to the oil contaminated winding. See Figures 2 and 3.
The wedges were loose, and it was found that the fit of the coils in the slot width was unacceptable, with clearance between the coil side and slot wall varying from 0.20 to 0.46 mm (0.008” to 0.018”). Also identified was some damage to the corona control surface within the slot and at the interface between the corona control surface and the stress grading tapes.
As a result of failure to fully cure epoxy, impregnated felt packing around the throat blocks during the original rewind had absorbed oil. See Figure 4.
cause of environmental concerns, the stator was hand cleaned using a citrus based cleaner and paint brushes, lint free cloths, and hand spraying of citrus cleaner from bottles.
Slot voids were packed with conducting Vetronite with the minimum acceptable gap of 0.20 mm (0.008”), which has been accepted practice in UK HV rewinds.
The oil impregnated felt across the top of the throat blocks was cut away, and any damage to the corona control surface in the slot was repaired using a suitable graphite impregnated paint.
Areas at the slot ends with indication of PD activity or overheating were repaired using graphite impregnated paint and silicon carbide impregnated stress grading paint.
The stator winding was re-wedged with Nomex and epoxy glass strip packing under the solid G11 epoxy glass wedges.
Any defective end-winding bracing was replaced and locked with epoxy resin cured with hot air blowers. IRIS (now part of Adwel) PDA (partial discharge analysis) capacitive couplers were installed and connected to an external terminal box.
The stator was air sprayed with an anti-track varnish and the winding enclosed and dried with hot air blowers.
The rotor field coils, which had not been rewound in the 1984 rewind, were re-insulated with an epoxy impregnated Nomex inter-turn insulation and cured under heat and pressure.
The original DC exciter was converted to a 3 phase 21.5 Hz brushless exciter with a static DC field and a rotating diode assembly. This was to eliminate carbon dust from the exciter, and also to facilitate field forcing to give better response to step load changes with a new digital automatic voltage regulator (AVR).
Although the generator rating was not changed from its original 50 MVA 40 MW pf 0.80, the operating mode was changed to multi start peak lopping (peak shaving).
The sheet steel winding covers (air baffles) were replaced by non-magnetic winding covers manufactured from high temperature (1100°C) flame retardant GRP Balsa composite (12 mm thick). This reduced stray losses by around 50 kW; and also made maintenance much easier because of the greatly reduced weight of the winding cover segments.
The generator was returned to service in 2000 and has been operating satisfactorily ever since.
Test results
The third-party pre- and post-refurbishment assessments included off-line partial discharge measurements using a Robinson DD5 analogue partial discharge detector.
The test results after refurbishment were dramatically reduced, so at phase voltage (6350v) no PD could be recorded. The pre-refurbishment test recorded up to 5600 picocoulombs (pC) at 7 kV whilst the post-refurbishment test only recorded up to 500 picocoulombs at 8 kV, a very significant reduction as shown in Table 1 and Charts 1 and 2.
The TVA Probe test showed that apart from an area at the core end regions the readings were in the 3-6 mA range. The worst case for the end of core readings was 28 mA. No evidence of slot discharge or significant insulation delamination was found.
During the 1984 re-core, the end core pack laminations had been bonded together with an adhesive. It is possible that some of that adhesive had migrated into the slot, insulating the slot wall from the graphite surface of the coil resulting in the higher TVA probe readings. Or there may have been some delamination or voids where the coil exits the slot.
IRIS PDA calibration tests were carried out July 2000 before the post refurbishment tests in August 2000.
The IRIS PD monitoring system evaluates Qm and NQN. Qm is defined as PD magnitude corresponding to a PD rate of 10 pulses per second whilst NQN is the total PD activity per half cycle.
Since Qm is more indicative of winding condition Qm is used for assessing winding condition.
PDA records have been intermittent since the refurbishment in 2000, but some records are available from 2006 through to 2014. For Table 2 and Table 3, a straight-line extrapolation between data records has been assumed.
The statistical distribution for Qm for air cooled hydro-generators with 80 pF capacitors, as published by IRIS Power, would indicate that the condition of this winding is reasonable for the age of the winding.
The trends on NQN and Qm are not showing a significant gradual increase; and although there had been a high excursion in 2011, in the next year’s test the result was more normal.
When comparing the Qm results from Table 3 to Table 2, they are all within the 75% (111mV at 10-12 kV in Table 3) of machines. When Qm levels move above 90% of machines or when the Qm doubles in a 6-month period and continues to rise or stabilises at the higher level, there is a clear indication of a problem.
Apart from Phase B (yellow) in 2011 the PD readings were reasonable with no significant rates of change. See Chart 3. Considering that these records are starting with a 22-year-old winding through to a 30-year-old winding, they show acceptable performance.
In the pre-assessment report Phase A (Red) had the highest PD levels recorded; when energised at the line end, this phase now has the lowest level.
High PD levels were recorded, in the pre-assessment report, in Phase C (Blue) which currently is producing much reduced PD activity. The higher PD figures were when measuring from the star (wye) point end of the winding which in service sees much lower dielectric stress.
It is not possible to have a direct comparison of PD results when comparing off line and on line testing, but it would be interesting at some stage in the future to assess the winding before its next rewind with off line testing and have an assessment during winding removal to identify sources of PD activity. See Chart 4.
However, with the condition of this winding that will be many years away and unlikely to be before 2024.
Conclusion
This paper demonstrates the suitability of resin rich insulation systems for hydro generator service and highlights the advantages of being able to introduce remedial measures during the life of the winding to address issues and ensure that acceptable winding life is achieved.
AVAILABLE IN SPANISH
ANSI/EASA AR100
More information on this topic can be found in ANSI/EASA AR100
Print