You are on page 1of 10

OFF-LINE PARTIAL DISCHARGE TESTING

REACTOR COOLANT PUMP RE-WIND USING HARD COILS


Ken Jackson
Duke Power Company

ABSTRACT
A stator from a Catawba Nuclear Station Reactor Coolant Pump Motor was rewound because of
high partial discharges (PD). Reactor Coolant Pump Motors recently rewound at another Duke Power plant,
McGuire Nuclear Station, have also been exhibiting high PD. As a result, Duke Power Company Engineers
required off-line PD testing for the Catawba rewind as part of acceptance testing. The maximum
acceptance level for PD was to be 200 mV.

The selected vendor made two attempts to obtain a winding that would pass AC high potential
tests, in addition to meeting the PD criterion. The first set of coils consisted of hard coils with dedicated
turn insulation and flexible end-windings, and received global VPI after insertion into the slots. PD testing
would be performed after installation of the coils, but prior to vacuum-pressure-impregnation (VPI). PD
testing would then be repeated after the stator is VPI’d, and after successfully passing AC high potential
testing. The engineers at Duke Power Company believed that this process might have allowed personnel at
the repair facility the opportunity to correct any problems prior to VPI. PD results were very favorable prior
to insertion into the VPI tank, but the stator failed high potential testing after the resin treatment was
complete.

The motor was rewound using differently designed coils. Hard coils with flexible end-turns would
still be employed, but the coils would no longer have dedicated turn tapes. Each strand was to be insulated
with a wrapper, folded parallel to the copper conductor. The stator met the PD requirements prior to being
VPI’d. After VPI, the stator successfully passed AC high potential testing, and the post-VPI PD values (in
mV) were all in the single digits. As a result Duke Power Company engineers have a high level of
confidence in the stator’s reliability.

INTRODUCTION

An off-line PD test was performed on a Reactor Coolant Pump Motor stator, which had been
recently rewound. The motor was removed from Catawba Nuclear Station’s 2D Reactor Coolant Pump
during the October 1998 refueling outage. The stator was rewound because previous off line PD tests
suggested that the insulation system had significant voids.

This test was performed as part of acceptance testing, for Duke Power to receive the stator from
the vendor.

THE REWIND

The previous windings were green coils, and were globally VPI’d after installation. Mica paper
tapes were wrapped around the conductors to make up the groundwall insulation. The stator would be
placed in a tank where air would be extracted from the tapes. The tank would then be pressurized to force
resin through the tapes. The resin would fill any voids within the new insulation system.

Engineering and Maintenance personnel within Duke Power Company considered re-winding the
stator with hard coils because of the following: The two motors, removed previously, from Catawba
Nuclear Station had exhibited high PD during off-line tests. Both stators had windings that were VPI’d.

1
Motors at McGuire Nuclear Station, which were rewound as recently as two years ago, exhibited high PD
during on-line testing. These stators were also VPI’d.

Industry experts believe the quality of hard coils and coils employing the VPI process to be equal.
VPI’d windings have been successfully used in the industry for many years. But the integrity of the
winding is more process dependent. An important part of the process is the preheat and curing cycles that
effect the viscosity of the resins used during VPI. If the resin is too thick it may not be able to penetrate the
winding, and fill the voids. If it is too thin, it may run out of the stator.

For a motor of this size, the opportunity of achieving a high quality winding may be increased by
using hard coils. Hard coils employ stage-B tapes, which have already been backed with resin. The slot-
section of the coil is hot-pressed for compaction, and to allow the resin to spread evenly through out the
insulation system. The coils are inserted into the slots and secured using packing and wedges. A hard coil
can be tested after assembly. If a problem is detected, a hard coil may be removed and repaired. A coil
which has been VPI’d is bonded to the slot, and cannot be removed without significant damage.

There were some concerns about using hard coils on a critical motor. A major concern, especially
for vertical machines, was that coils might vibrate as the packing or wedges shrink due to thermal aging.
The desired winding system would employ hard coils, but also be VPI’d to seal the packing, wedges, and
blocking in place.

The vendor offered two coil designs that met Duke Energy’s requirement. In one design, each
strand was insulated, and the fold was parallel to the copper conductor. Dedicated turn tapes were not
required. The other design employed copper strands with heavy film and dedicated turn insulation. With
either coil design, a semiconductor layer and voltage gradient tapes would be used to repress slot
discharges.

The coils using strands with heavy film and dedicated turn insulation were manufactured in the
United States, and were selected for the rewind. The stator had to be rewound within a few months for
installation in Catawba Unit 1 during its 1999 refueling outage. The other coils were made in Europe, and
had a longer delivery time.

Additional changes were made to the stator. The stator was radiologically contaminated. The
windings were removed, but the core was still contaminated. The rewound stator could not be inserted into
a VPI tank because of the risk in contaminating such a large quantity of resin. Therefore, the core was
restacked with new laminations. The thickness of the groundwall insulation would be slightly reduced to
allow room for packing. The design should be more than adequate for the rated voltage.

TESTING THE FIRST WINDING PRIOR TO VPI

Hard coils were manufactured, and in-process high potential testing of each coil was performed at
18 KV. These coils were inserted into the slots of the stator, and all connections were made. The stator had
not been VPI’d. Duke Power Engineers would perform a PD test of the new windings on March 24, 1999.
The objective was to permit corrections as necessary to lower PD to acceptable values. It would be difficult
enough to lift a coil from the slot, and this would be almost impossible after VPI.

The neutral terminals, identified as T4, T5, and T6, were connected together. The bus coupler and
power cable were connected to the line terminal identified as T1 (“A” phase), and the stator energized to
obtain PD data. The bus coupler and power cable were then moved to the T2 (“B” phase) terminal to obtain
data, and then placed at the T3 (“C” phase) terminal. Duke Power engineers and maintenance personnel
were very pleased. The results, as shown in Table I, revealed that little activity was occurring within the
windings in the slot area.

Each phase was tested with the other two phases, not under test, grounded. PD data was obtained
with the bus coupler and power cable placed, alternately, at each of the line terminals (T1, T2, and T3).

2
Any increase in activity under this configuration can be attributed to endwinding discharges. Negligible
change was expected for a new winding connected in the configuration. The endwindings were adequately
spaced, and the coils were clean. PD data was obtained with the bus coupler and power cable placed
alternately at the neutral terminals (T4-“A” phase, T5-“B” phase, and T6-“C” phase). As seen from Table
II, little change was observed between line and neutral end of “A” and “C” phases. However, there was a
significant increase in PD on the neutral end of “B” phase, which was measured at approximately 100 mV
above the acceptance criteria.

The last few coils on the neutral side of “B” phase were the suspected culprits, and thus identified.
The windings were energized, and an ultrasonic probe was used as an attempt to locate the specific area of
activity. The ultrasonic probe revealed PD activity at the endwinding area, where the bus from terminal T5
connects to a coil. All three line end terminals were connected together. The bus coupler and power cable
were connected to the T5 terminal, and the windings energized. If the identified area was the cause for the
increase in PD activity, it should disappear when examining “B” phase for slot discharges at the neutral
end. The activity still appeared, and thus concluded to be occurring in the slot area.

As shown on Table III, the PD was much lower than for the previous windings, and Duke Power
engineers accepted the windings as is. The neutral end of a winding should see little voltage stress during
actual operation, and PD activity should be much lower than what was observed during the test. Further, it
was hypothesized that the VPI process should fill any voids within the insulation system. PD activity
should decrease, hopefully to acceptable values, after the stator has been VPI’d.

TABLE I: PD Data, All Three Phases Energized

3/24/99 Line
Phase NQN+ NQN- Qm+ Qm-
A 104 100 58 55
B 112 109 60 58
C 158 129 77 69

TABLE II: PD Data, Each Phase Energized Separately

3/24/99 Line Neutral


Phase NQN+ NQN- Qm+ Qm- NQN+ NQN- Qm+ Qm-
A 99 81 57 46 118 139 61 72
B 140 128 71 66 652 432 297 199
C 157 137 76 72 205 189 94 83

TABLE III: Comparison Between Old and New Windings

Date: 3/24/99 Date: 1/21/99


Phase Qm+ Qm- Qm+ Qm-
A 57 46 541 600
B 71 66 2380 2319
C 76 72 224 245

3
TESTING THE FIRST WINDING AFTER VPI

The stator was VPI’d, and follow-up PD testing was to be performed on April 5, 1999. Prior to the
PD test, acceptance AC high potential testing would be performed. For AC high potential testing, the coils
of each phase would be energized to 14.2 KV, line-to-ground, for one minute. The acceptance criteria are
that leakage currents do not exceed the specified value of the test set. In other words, the test set does not
trip.

“A” phase was energized, but the test set tripped shortly after reaching the target test potential of
14.2 KV. The leakage current just prior to the trip was about 1.5 Amps which was not considered to be
unusually high. Insulation resistance measurement indicated that a coil had faulted to ground. No visual
indications of the fault such as discoloration or blistering of the insulation could be observed. To locate the
specific area of the fault, low voltage ac (approximately 110 volts) was applied to the coils. But just enough
power was used for smoke to emanate out of the faulted area, as seen in Figure 2. The failure occurred on
the top bar a few inches from the edge of the iron on the connection end of the stator. The bar was in slot
50, and was identified as coil YA13. As shown on Figure 1, the coil was towards the neutral end of the
circuit.

“B” phase was then energized to 14.2 KV, and successfully passed the one minute high potential
proof test.

Voltage was applied to the “C” phase coils; however, the test set tripped at 10.5 KV. Arcing was
observed to a finger plate. This occurred at the end of the stator opposite to the connection end. Again, the
fault left no visual indications on the insulation. Application of low voltage revealed the area of failure at
the top bar between the “finger plates” as shown in Figure 3. The failure occurred at the edge of the core, at
slot 45, and the coil was identified as XC17. As shown on Figure 1, the coil was towards the neutral end.

Groundwall insulation was notched out from the area of both failures. A compound consisting of
air-curing epoxy and ground mica dust was applied to fill in the notches, and tape with epoxy was applied
over the compound. Heat lamps were used to accelerate the curing of the compound overnight. The
repaired area is shown in Figure 4.

The high potential test was repeated on “A” and “C” phase coils. “A” phase coils passed the one
minute test at 14.2 KV. However, the test set tripped at 13.5 KV while testing the “C” phase coils. It was
determined that the faulted area was under a wedge of the previously failed coil about 3/8 of an inch from
the previous failure, as shown in Figure 5.

Because of the three failures, there was a concern about the integrity of the remaining “C” phase
coils. Coil XC17 was removed from the circuit. The remaining windings in “C” phase was energized to
14.2 KV, and successfully passed the test. Out of curiosity, DC step voltage testing was performed to 19
KV. The results up to 17 KV are displayed in Figure 6 along with test data from the old VPI’d winding.
The results are very similar to those obtained from the original VPI’d winding. The leakage current for the
hard coils at 19 KV was 2.5 uA.

4
FIGURE 1

Reactor Coolant Pump Motor


Winding Diagram

T1 T2 T3

XA1 YA1 XB1 YB1 XC1 YC1


XA2 YA2 XB2 YB2 XC2 YC2
XA3 YA3 XB3 YB3 XC3 YC3
XA4 YA4 XB4 YB4 XC4 YC4
XA5 YA5 XB5 YB5 XC5 YC5
XA6 YA6 XB6 YB6 XC6 YC6

XA7 YA7 XB7 YB7 XC7 YC7


XA8 YA8 XB8 YB8 XC8 YC8
XA9 YA9 XB9 YB9 XC9 YC9
XA10 YA10 XB10 YB10 XC10 YC10
XA11 YA11 XB11 YB11 XC11 YC11
XA12 YA12 XB12 YB12 XC1 2 YC12

XA13 YA13 XB13 YB13 XC13 YC13


XA14 YA14 XB14 YB14 XC14 YC14
XA15 YA15 XB15 YB15 XC15 YC15
XA16 YA16 XB16 YB16 XC16 YC16
XA17 YA17 XB17 YB17 XC17 YC17
XA18 YA18 XB18 YB18 XC18 YC18

T4 T5 T6

Note: winding is a 2-circuit wye-connection with


108 slots, and18 groups consisting of 6 coils per
group.

5
FIGURE 6

2D Reactor Coolant Pump Motor


Test Data

Date Date Date


10/29/98 1/11/99 3/6/99
Step Voltage Leakage Leakage Leakage
Current Current Current
KV uA uA uA
1 5 0.2 1 0.4
2 7 0.6 0.8 0.5
3 9 1 0.9 0.6
4 11 1 1 0.9
5 13 1.2 1.1 1.3
6 15 1.7 1.2 1.5
7 17 1.7 1.3 2
Test data
for 1st
rewind

Step Voltage Testing

2.5

2
Current (uA)

1.5 Series2
Series3
1 Series4
0.5

0
1 2 3 4 5 6 7
Applied Voltage (KV)

The second repair on coil XC17 was made using the same compound as the earlier repairs;
however, the coil failed again when energized to 14.2 KV. Three repair options were available. One was to
remove the straight section of coil XC17 from slot 45, and to splice another section in its place. A second
option was to electrically remove the coil from the circuit. The other option was to burn out all the coils,
and to install new ones. Even though the problem appeared to be isolated to one area of coil XC17, it could
not be removed without disturbing the other coils. Because of the critical application of the motor, Duke
Power engineers requested that the stator be rewound. Follow-up PD testing on this winding was not
performed.

6
TESTING THE SECOND WINDING

One of the reactor coolant pump motors in Catawba Unit 1 had a slightly bent shaft, but vibration
levels were well within the range of acceptance. The engineers at Duke Power decided not to expedite a
second winding, but to allow the vendor adequate time to determine the cause of failure and implement
appropriate corrective action. The reactor coolant pump motor with the bent shaft could operate safely for
an indefinite period, and would not be replaced during Catawba’s Unit 1’s refueling outage.

The vendor determined that the stator had failed because of inadequate temperatures of the hot
press used to press the hard coils. This problem was specific to the vendor’s facility in the United States,
and the new set of coils would come from Europe. Duke Power engineers were satisfied with the measures
taken. As an added assurance, the new coils would receive voltage endurance testing per IEEE1043.

PD testing was performed on the second winding, on July 26, 1999, after the coils were inserted
into the slots and connections made. The method for testing the second rewind was the same as for the first
set of coils. As seen in Tables IV and V, PD activity was very low. Two and three-dimensional plots for PD
are shown in Figure 7.

TABLE IV: PD Data, All Three Phases Energized

7/26/99 Line
Phase NQN+ NQN- Qm+ Qm-
A 4 5 5 4
B 6 7 5 5
C 44 40 18 16

TABLE V: PD Data, Each Phase Energized Separately

7/26/99 Line Neutral


Phase NQN+ NQN- Qm+ Qm- NQN+ NQN- Qm+ Qm-
A 22 25 18 16 33 30 15 14
B 13 11 7 5 7 9 6 5
C 59 56 28 32 71 72 38 34

After being VPI’d, DC step voltage testing was performed on the stator for comparative purposes,
and the results are shown in Figure 8. The maximum leakage current with an applied voltage of 17 KV was
1.7 mico-amps, which compares favorably with the values shown in Figure 6.

7
F IG U R E 8

R e w o u n d R e a c to r C o o la n t P u m p M o to r S ta to r
T e s t D a ta

D a te
8 /1 7 /9 9
S te p V o lta g e Leakage
C u rre n t
KV uA
1 5 0 .2
2 7 0 .6
3 9 1
4 11 1
5 13 1 .2
6 15 1 .7
7 17 1 .7
S e rie s 1

S te p V o lta g e T e s tin g

1 .8
1 .6
1 .4
Current (uA)

1 .2 S e rie s 2
1
S e rie s 3
0 .8
0 .6 S e rie s 4
0 .4
0 .2
0
1 2 3 4 5 6 7
A p p lie d V o lta g e (K V )

After VPI, each phase of the stator successfully passed the 14.2 KV AC high potential test. PD
testing was performed on August 17, 1999, and the results are shown in Tables VI and VII. It is interesting
to note the reduction in PD activity in the end-windings on “A” phase and overall in “C” phase. Some
reduction in PD would be expected because the resins would fill any voids in the flexible end-windings.

Another interesting note was that activity having high pulse count rate and low magnitude was
observed throughout the power frequency waveform. Activity for the current phase is shown on the two
and three-dimensional PD plots of Figures 9. A bus coupler had failed, and only one bus coupler, at the
motor terminals, was used during the test. Therefore, the analyzer could not distinguish between PD and
noise.

8
TABLE VI: PD Data, All Three Phases Energized

8/17/99 Line
Phase NQN+ NQN- Qm+ Qm-
A 9 9 5 4
B 13 7 7 3
C 9 9 5 3

TABLE VII: PD Data, Each Phase Energized Separately

8/17/99 Line Neutral


Phase NQN+ NQN- Qm+ Qm- NQN+ NQN- Qm+ Qm-
A 9 7 5 3 5 3 3 3
B 13 7 7 3 12 9 6 6
C 13 10 6 4 8 7 5 5

Two spare coils were available for testing. One coil was VPI’d; the other was not. An attempt was
made to perform AC high potential testing of the spare VPI’d coil to failure. The test began by maintaining
an applied voltage of 14.2 KV for one minute, and then increasing the voltage by 3 KV increments for.
Leakage current was recorded at each voltage level after being held there for one minute. The results are
shown in Figure 26. The test set tripped at 29.2 KV, but the coil never did fail. The coil was again
energized to 14.2 KV, and the voltage was raised steadily where the test set tripped at 36 KV.

An attempt was made to surge test the non-VPI’d coil to failure. Voltages up to 43.5 KV were
applied to the coil, but the coil never failed. Duke Power engineers were pleased with the results from the
stator and spare coils, and approved the stator for shipment to Catawba Nuclear Station.

CONCLUSION

All four failures occurred at the edge of the core. Three of the failures occurred within inches of
each other. The remaining windings successfully passed high potential testing. This suggests that the
problems were isolated to the two coils. The cause for failure has not yet been determined.

However, PD data obtained on the hard coils, prior to VPI, indicates that any voids present would
have been much smaller and less numerous than on the original VPI’d windings. This suggests that Duke
Engineers made the correct decision in selecting hard coils in an attempt to lower PD to acceptable values.
The results also show how important it is to examine a motor’s insulation from all aspects. The windings
need to constructed adequately to withstand substantial voltages as well as having few voids as possible.
The absence of either can shorten the life of a motor.

9
FIGURE 35

AC High Potential Test of Spare Coil


Test Data

Date
10/29/98
Step Voltage Leakage
Current
KV mA
1 14.2 5
2 17.2 8
3 20.2 11
4 23.2 13
5 26.2 15
6 29.2 19

AC High Potential Test


Spare Coil

20
Current (mA)

15
S
10 S
S
5

0
1 2 3 4 5 6 7
Applied Voltage (KV)

10

You might also like