The Unit 6 generator (original Canadian Westinghouse) at the 450-MW (420 MVa) Isle-Maligne powerhouse began operating in 1923 and still has the original stator core and frame. The stator winding was replaced at least twice, and the last winding installed in 1988 started to show signs of degradation. Rio Tinto’s maintenance team assigned this stator winding the second lowest reliability rating of its entire fleet (52 hydro units).
In January 2014, the unit failed. Returning the unit to service required taking three coils out of the circuit, for a total of 12 coils cut out on a 360-coil winding. This breakdown was the fourth in four years; all the defective coils were located in the first four electrical positions close to the main connections. All showed a voltage to ground of 6.6 kV or higher. Partial discharges (PD) were very high and growing since 2010.
Based on the occurrence of each failure, the breakdowns are due to degraded coil insulation. The winding has only 26 years of usage (some similar ones have about 50), but the original manufacturing quality was reported as questionable based on the experience of other utilities with the same equipment. The cores show some damage in the direct vicinity of the faults, but it is impossible to quantify it accurately without dismantling.
The probability of another failure within a year was evaluated at 99% by the engineering team at Rio Tinto. If a failure happened in the last untouched circuits of Phases A or C, it would be impossible to generate the required nominal voltage and the stator would possibly be stopped for good. Engineering proposed to invert the main and neutral connections to extend the remaining life of this winding. The group evaluated the success probability at 50%, and that was enough to convince management.
The work had to be done before spring freshet to reduce production losses. It was scheduled from March 17 to April 17 (four days a week, no overtime planned). An inversion plan was completed, after discussion between Rio Tinto and Voith Hydro, that involved installing 24 new jumpers (four groups of six jumpers and 48 welding joints). The shape and length of these jumpers needed to be quite different from the originals and from each other because of the large span of the new jumpers. To do the inversion, the jumpers had to span about seven slots (from slot 10 to 17, 100 to 106, etc.).
On Feb. 26, a last PD reading collected data to be compared with after-work data. Material needed for the modifications was ordered and the magnet wire was substituted, from 10 (0.310 x 0.076) to 5 (0.087 x 0.548). Using bigger and wider wire made the new jumpers difficult to shape. It would have been a wiser decision to use the same wire, but it was the fastest available material at the time.
Start of the reconfiguration
Work started on March 19. The jumper forming work was found to be harder and more time-consuming than anticipated (in part due to wider wires and lack of space), so it was decided two teams would work together instead of one. To avoid any mistakes, all the new connections affected (leads and coils) were color-coded before cutting.
The connections to be changed were unwelded and removed. A template of the future jumpers was created using wire rigid enough to keep its odd shape during the forming of the future jumpers. This exercise was also needed to select the order of installation for welding purposes. Each jumper was also color coded.
The jumpers were formed and partially insulated, using 10 layers of silicone tape half lap, then positioned and welded (using copper sleeves and silver rods) inside the stator. A bending tool was designed to hold the copper wire. Due to the odd geometry of the new jumpers, additional space had to be taken on top of the regular jumpers, keeping the assembly as compact as possible (lower than the circuit ring support top). After the welding was completed, the joints had to be insulated with 10 half lap of silicone tape. Finally, the completed jumpers were covered by two layers of half lap glass tapes soaked with epoxy.
|This new insulated jumper was created to solve a problem with winding degradation at Rio Tinto’s 450-MW Isle Maligne powerhouse.|
Test and inspection
Insulation and inspection procedures were developed for this work. Strict inspection and test procedures needed to be put in place to proof soundness of the work at each stage. Polarization index (PI) measurements and insulation resistance were used to confirm insulation quality and temperature reading (thermography) to confirm welding work. Upon completion of the work, PI and resistance were found acceptable at a PI at 10 kVcc of 3.2 and 1 min resistance of 538 Mohms.
The jumpers then were fixed in place using fiber blocks of soaked felt to keep minimum air distance and to stiffen the entire set of jumpers. Glass roving was also used to attach everything solidly. PI was taken after the blocking process and the result was surprising. Phases B and C got results under 2.0 and it was impossible to get results on phase A (leakage too high).
Despite abnormal PI results, a DC hipot ramp (manually controlled) test was done on the winding, showing acceptable results. As phase A presented a lower PI than the two other phases, the 21 kVdc was left on it for 10 minutes to get further safety. The leakage current increased slightly but the insulation resisted correctly to the test. The only explanation for the lower PI on phase A is the potentially uncured epoxy.
Back to the grid
Going back to the grid was performed to gradually test the winding and collect as much data required, validating the intervention is adequate. First, a gradual rise of the voltage was done at speed no-load: 6 kV, 8 kV, 10 kV and nominal (13.4 kV). Each step was maintained for a couple of minutes, looking at the wave shape of the voltage and any PD abnormality.
The unit was then put back on the grid by gradually increasing the load, always checking for PD. An infrared camera was used to confirm that all the new welding joints were good.
The results of all measurements did not show evidence of abnormality and the unit performed as expected. The unit was called “ok for use without restriction.” The only requirement, issued in 2010, is to keep Unit 6 in service as much as possible, to limit stops and duration, and avoid sudden and frequent load changes.
The PD measurements taken after the inversion were compared with those taken on Feb. 26 and the results were: PD for all three phases decreased substantially and beyond expectations. The situation of Feb. 26 placed Unit 6 in the red zone among the Rio Tinto Energie Electrique fleet with a higher PD activity, while the situation of April 10 placed it in the green zone with the lowest PD. On April 14, Unit 6 returned to service.
To state that this work was a success, we must demonstrate the remaining life of the stator winding was increased. Taking PD readings at regular intervals establishes a trend to assess the risk of failure and predict future performance of the stator winding. Four measurements have been performed after the inversion, the last one on Feb. 18, 2015, and data collected clearly shows that the reconfiguration is a success. Even if PDs are slowly increasing, they are well beyond the level before inversion, and there are no expectations of a short-term failure.
– By Denis Bussieres, senior engineer (retired), and Pierre Fortin, engineer, Rio Tinto; Mario Dumouchel, AMB service specialist, Voith Hydro Inc.