Mitsubishi SF6 Gas Circuit Breaker Type 100 SFMT-40E. Breaker Failure Report

Transcription

1 Mitsubishi F6 Gas Circuit Breaker ype 100 FM-40E Breaker Failure Report Prepared by: Jeffery M. Wright Manager ransmission Field ervices Vermont Electric Power Company, Inc.

2 April 6, :08 VELCO experienced a breaker failure in the Essex FAC switchyard. he 115 kv F6 gas circuit breaker is a Mitsubishi 100MF-40E. Although there were no visible external damages, relay and protection engineers analyzed the relay information and all indications led to an internal failure of the K98 circuit breaker. EXAMPLE ONLY Mitsubishi 100MF- 40E Circuit Breaker Location: Georgia C80 Cap Bank Breaker

3 DFR analysis indicates the fault lasted nearly 4 cycles with phase 3 volts dropping to.02 p.u.

4 IBM, the largest manufacturer in Vermont is electrically connected to redundant VELCO 115 kv sources. hey are located.7 miles from the Essex substation. IBM 115 kv phase 3 bus voltage dropped to approximately the same as what VELCO recorded at Essex. Approximately.02 p.u. for nearly 4 cycles. he power quality interruption cost IBM in excess of $1.5M in lost tooling hours and production.

5 F6 arc byproducts An internal inspection directed by Mitsubishi technical engineers verified that an internal fault had indeed occurred. he breaker was shipped back two days following the failure to Mitsubishi (Pittsburgh, PA) for a thorough failure analysis.

6 Formal Mitsubishi Response Friday, April 16, 2004 Conference Call Failure Analysis Report 1) Repair Efforts & Flashover Location he tanks from Breaker G were dis-assembled and inspected. he flashover was noted to have occurred at Bushing Assembly #4, center phase, at the moving end. he evidence indicates that the arc struck from the Bushing #4 conductor to the bushing adapter plate. Fotos are attached. he quantity of arc byproduct powder in the center phase tank was moderate, indicating the flashover was not severe. Arc byproduct powder was collected and the F6 which came inside the breaker was also collected, for possible future analysis. Breaker components which were damaged or suspected of being damaged were replaced; major replaced components included center phase interrupter, Bushing Assembly #4 (conductor & insulator), all 6 bushing adapter plates, the F6 gas system, desiccant, and associated hardware, O-rings, etc. he other 2 phases were also disassembled; all major components from these 2 phases were re-used. All 3 tanks were thoroughly cleaned; all re-used components were thoroughly cleaned. Breaker G was re-assembled and re-tested following written instructions. Breaker G was shipped on 4/15/04. 2) Investigation into Cause of Flashover Our investigation indicated that randomly-occurring water was the likely cause of the flashover. Water was detected using our ultrasonic (U) inspection technique, which can detect water in boltholes without dis-assembling the breaker. Boltholes in every nozzle were evaluated for presence of water by the ultrasonic method, and every bolt was removed for visual examination. his gave us an opportunity to visually confirm the ultrasonic indications. his investigation indicated that water was present randomly, at different levels, in 3 of the 6 bushing nozzles, in boltholes Bushing Nozzle # 2 (2 of 12 boltholes, very minor <0.5 mm) Bushing Nozzle # 3 (2 of 12 boltholes, very minor < 0.5 mm) Bushing Nozzle # 4 (6 of 12 boltholes, minor & moderate levels) mall traces of a water-and-oil liquid were noted on the undersides of 4 of 6 bushing adapter plates, where they mate with the tank nozzles. Foto attached. Visual observations of removed bolts & boltholes also indicate that some of the water in the boltholes does evaporate, and is likely absorbed by the desiccant in each tank

7 he internal inspection conducted by Mitsubishi uncovered evidence of the internal fault between the bushing conductor and bushing adapter plate inside the bushing throat.

8 Further investigation uncovered water droplets in the breaker bolt holes of four out of six bushing adapter plates. Water Droplets Present Following the failure analysis, the major components damaged were replaced, the breaker was rebuilt, shipped back to the Essex substation and commissioned by VELCO technicians on April 21 st at 13:24.

9 Formal Mitsubishi Response Friday April 16, 2004 Conference Call Mitsubishi admits knowledge of three other similar failures of the same type of breaker one in Illinois and two in Nebraska. Mitsubishi admits to a known process control problem which was identified and corrected in June 2001, affecting breakers manufactured between he manufacturing process began with pressure washing the cast breaker tanks with water. he tanks were set outside to dry and the failures are attributed to trapped water in internal bolt holes that eventually migrates into the tank providing a point of discharge, ultimately faulting the breaker internally at the point of the water droplets on the tank wall. Mitsubishi did not issue a service advisory to all of their customers notifying them of this problem. Mitsubishi has since developed a method for detecting the water in the bolt holes using an ultra-sonic method. hey agreed to send a crew to Vermont on Monday April 19 th to begin testing other subject breakers yet to be identified.

10 Mitsubishi Response Mitsubishi identifies the subject breakers that VELCO owns. Out of the thirty (30) 100FM- 40E type breakers VELCO owns, seventeen (17) are potentially subject to the same problem FM-40HE G C82 ANDBAR G K23 WILLION K43 WILLION G01316 K34-40 WE RULAND G01312 K37-40 WE RULAND G01313 K39-37 WE RULAND G01315 K34 WE RULAND G01314 K30-39 WE RULAND C71 NORH RULAND C72 MIDDLEBURY G C80 GEORGIA G C36 G C35 G K98 G K97 G C39 G C38 G C37 G C34 G K25 G K25-22 G K22-23 G K23 G K24 G K20-24 G K21-20 G K21 G01311 C92 COOLIDGE G01318 C91 COOLIDGE G01319 K32 COOLIDGE YEAR YPE ERIAL NUMBER MANUFACURER POIN AION

11 he esting Process he breaker is removed from service and C shrouds are removed, exposing the bushing throats. he bushing throat is filed clean, providing a clean surface to apply the ultra-sonic device to. By aligning the test probe directly with the bolt hole behind the test area, the test device indicates a waveform. A non-distorted waveform indicates a dry bolt hole, whereas a distorted waveform can be an indication of a contaminated bolt hole. Interpretation of test results is done on site by the technician. An inspection can be completed within four hours.

12 he bushing throat is filed clean, providing a clean surface to apply the ultra-sonic device to.

13 By aligning the test probe directly with the bolt hole behind the test area, the test device indicates a waveform. A non-distorted waveform indicates a dry bolt hole, whereas a distorted waveform can be an indication of a contaminated bolt hole. Interpretation of test results is done on site by the technician.

14 he Internal Inspection / Rebuild Process If water is suspected in the bushing adapter plate bolt holes, the breaker is drained of gas, bushings are removed and all mounting hardware is removed, cleaned and reassembled. he rebuild process typically consumes a twelve to fourteen hour day. Four breakers were rebuilt.

15 Once the bushing adapter plate is removed, the holes are inspected for water, blown dry and prepped for reassembly.

16

17 Mitsubishi Response Mitsubishi s response was timely. M APRIL W F he failed breaker was repaired and returned to VELCO in seven (7) days BREAKER FAIL BREAKER HIPPED O MEPPI 3 10 Inspections followed immediately afterward INPECION AR BREAKER HIPPED O VELCO FAILURE ANALYI CONF CALL

18 Mitsubishi Response All work was wrapped up by the end of June. M APRIL W F M MAY W F MEPPI was very cooperative and worked weekends to accommodate outage requirements Of the seventeen (17) breakers that were ultrasonically inspected, four (4) required internal inspection and cleaning M JUNE W F M JULY W FINIHED F

19 VELCO Response VELCO has kept all of its owners and IBM informed of the results of the failure analysis report and the progress and results of the testing. VELCO has gone forth and publicized the failure on the Doble Engineering upport Group forum to inform other utilities of the facts related to the VELCO failure. VELCO is satisfied with the Mitsubishi inspection process and rebuilds. VELCO intends to hold MEPPI accountable for remedying any further problems identified and is currently negotiating the compensation for VELCO labor and materials. VELCO s labor and material costs exceed $33,000.