7/22/2019 RM-2 Assess H2O Gen Simmonds
1/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
CONDITION ASSESSMENT OF A 103 MVA
HYDRO-GENERATOR DURING A FORCED OUTAGE
By I W Simmonds
Doble PowerTest Ltd, UK
ABSTRACT
This paper presents the case study of a Russian built (Electrosila) 103.5MVA, 72-pole hydro-generator built in 1953
with a rewound Epoxy/mica insulation system dating from 1979. The unit was tripped due to a stator earth fault
brought on by the breakdown of the insulation in one of the slots. Subsequent inspections (conducted by a 3rdparty)
indicated that several other bars were also needing to be replaced due to slot discharge damage.
The replacement of the affected bars was concluded before DPT arrived on-site but photographs provided by the
station show the repairs in progress. From these photos it would appear that some of the damaged bars were in the
same location as a previous failure (due to the presence of newer insulation at the series joins).
Figure 1 Figure 2
INTRODUCTION
The operators of a 103.5MVA hydro-generator contacted DPT to perform a condition assessment to one of their four
operating units after a forced outage. The operators were interested in knowing whether or not the unit would be
able to provide reliable full load operation over the next 5 years or more. All four units had condition assessments
planned for near future dates and the unit regarded as being in the best condition was chosen to be first.
From information provided by the station it was apparent that slot discharge was a well established problem with
this unit. Forced outages had occurred on two separate occasions within a 3 year time-span, before these current
events, for very similar reasons.
The condition assessment was conducted over several days and included High and Low voltage diagnostic tests on
the stator winding, stator core and rotor winding/poles. The rotor remained in place with the removal of 12 poles (2
pairs of sixpoles 180 apart) to facilitate the ElCID test and aid in the visual inspections.
7/22/2019 RM-2 Assess H2O Gen Simmonds
2/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
2
Figure 3 Figure 4
Electrical Tests
Rotor winding pole drop measurements
The results for the rotor pole drop tests indicated that eleven poles in Table 1 returned values 10% or more below the
average potential difference across each pole. This indicated a reduction in the inductive impedance of the circuit
and signified the existence of inter-turn faults within these poles.
Poles 65 & 66 returned values of 40% or more, below the average potential difference across each pole and
indicated that several coils were involved within the inter-turn fault, for each of these poles.
It should be noted that, with these tests conducted at stationary, any fault indications could change significantly with
the rotor running at normal operating speed. It is, therefore, possible that other inter-turn faults may exist, but only
become apparent under centrifuge and that existing faults may expand.
Table 1
Expected Volt Drop3V/pole 10% - Rotor poles connected to rotor
Measured Current1.5A
Pole No. Volt drop (V) Pole No. Volt drop (V)
16 2.53 40 2.49
22 2.57 64 2.53
23 2.04 65 1.82
24 2.57 66 1.32
35 2.62
Expected Volt Drop4.15V/pole 10% - Rotor poles removed from rotorMeasured Current6A
Pole No. Volt drop (V) Pole No. Volt drop (V)
12 3.50 60 3.73
39 3.65
7/22/2019 RM-2 Assess H2O Gen Simmonds
3/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
3
Stator winding IR/PI and winding resistance measurements
The stator winding resistance measurements were carried out between phases using a DLRO tester. The results for
the stator winding resistance tests were normalized to 75C and should be within 10% of the factory test
specifications. The results indicated that the resistance of the winding was balanced, between phases. Previous
results were not available during these tests, therefore, the results should be compared with factory test
specifications once they become available.
Table 2
Phase
Measured Resistance
at 23.50C (m)
Normalised Resistance
to 750C (m)
Factory Test Resistance
at 750C (m)
A - B 22.7 27.2 Not available
B - C 22.6 27.1 N/A
C - A 22.5 27.0 N/A
IR and PI measurements were carried out on each phase of the stator winding using a 5 kV insulation tester. The IR
and PI were measured both before and after the high voltage tests. The IR and PI measurements were considered
acceptable on all three phases, indicating that the winding was dry and free of any serious contamination.
Table 3
Insulation resistance and polarization index measurements performed at 5kV
Time after start of
measurements
(minutes)
Insulation Resistance (M)
Before HV Tests
A Phase B Phase C Phase
1 486 346 352
10 2040 1430 1430
Polarisation Index 4.19 4.13 4.06
Ambient temperature23.2C Relative humidity27.5%
Stator winding partial discharge measurements
The discharge analyzer used for the partial discharge measurements presents the PD pulses as a pattern showing an
amplitude-phase-height distribution in the 100-800MHz range i.e. each pixel within the pattern represents one
captured PD pulse with its position referring to amplitude and phase angle of occurrence, while the colour of each
pixel refers to the frequency of occurrence (number of pulses at the same amplitude and phase position).
The partial discharge values recorded at the maximum operating voltage to earth (phase voltage) were 48000, 45000
and 35000 pC for the A, B and C phases, respectively. The discharge magnitudes, on the A, B and C phases, were
considered to be high, for the type of insulation.
The phase resolved partial discharge (PRPD) patterns were reasonably similar between phases. The PD magnitudesindicated a very large step increase between inception voltage and 4kV and indicated that a significant number of
discharge sites would be active during service. The PRPD patterns, on all phases, suggested that significant arcing
through air was taking place. The PRPD patterns show several separate events of constant amplitude in both half
cycles. The dominant event, separated from the lower patterns, suggested that discharge was occurring from some
floating object. The other lower magnitude events were more indicative of discharge to the core, either from the
slot portion or near the slot exits. As the voltage was increased external surface discharge, probably due to
deterioration of the stress relief section, also became apparent.
7/22/2019 RM-2 Assess H2O Gen Simmonds
4/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
4
The discharge inception voltages (DIV) and discharge extinction voltages (DEV) were considered to be low and
were, on average, only 0.25x the working voltage. It is generally recommended that DIVs and DEVs are 0.5x the
working voltage or higher. Low DIVs and DEVs could indicate the onset of slot discharge, brought on by the
deterioration of stress grading paint and corona shields. Overall, the partial discharge measurements indicated a
thermally aged insulation system in a deteriorated condition.
Table 4
Partial discharge magnitudes for each phase of the stator winding.
Test Voltage (kV)Discharge Magnitude (pC)
A Phase B Phase C Phase
+ve
cycle
-ve
cycle
+ve
cycle
-ve
cycle
+ve
cycle
-ve
cycle
2.0 2500 2700 3000 3000 1000 1000
4.0 24000 25000 10000 10000 17000 15000
6.0 32000 34000 28000 29000 25000 28000
8.0 43000 48000 38000 45000 28000 35000
Discharge Inception Voltage 1.7 kV 1.7 kV 1.7 kV
Discharge Extinction Voltage 1.6 kV 1.6 kV 1.6 kV
A Phase at 4kV and 8kV
Figure 5
Possible arcing to the
core at the slot exits
Bar to bar or bar
to core sparking in
the endwinding
7/22/2019 RM-2 Assess H2O Gen Simmonds
5/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
5
B Phase at 4kV and 8kV
Figure 6
C Phase at 4kV and 8kV
Figure 7
Surface discharges from
stress relief section
Internal voids within
the insulation
7/22/2019 RM-2 Assess H2O Gen Simmonds
6/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
6
100
1000
10000
100000
0 2 4 6 8 10
Test voltage (kV)
Dischargemagnitude(pC)
Red (+ve cycle) Yellow (+ve cycle) Blue (+ve cycle)
Red (-ve cycle) Yellow (-ve cycle) Blue (-ve cycle)
Phase
voltage
Partial Discharge Magnitude (pC) v Applied Voltage (kV)
Figure 8
Stator winding dielectric loss angle (Tan Delta) and capacitance measurements
The Tan Delta/Capacitance test is a check of the bulk insulation. The Tan Delta and Capacitance were measured
while the applied voltage was increased in 0.1Un steps up to 8.0 kV. Each phase was tested separately, with the
other two phases earthed. Measurements were conducted on the complete phase and each phase end-winding
(Tables 5&6). Table 7 gives a comparison of the measured complete phase Tan Delta and Capacitance values.
The Tan Delta against voltage curves, shown in Figure 9, indicated significant tip-up as the test voltage was
increased above 4kV. The results were similar, between phases, for both Capacitance and Tan Delta measurements.
The Tan Delta measurements at 0.2Un indicated that the inherent dielectric losses of the insulation were high,
although previous test results were not available for comparison. The values derived from the Tan Delta tip-upindicated that the void content of the stator winding insulation was high and that a significant number of voids
would be active during service. The change in Capacitance measurements also increased to a higher degree than
that expected of epoxy-mica insulation in good condition (typically
7/22/2019 RM-2 Assess H2O Gen Simmonds
7/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
7
Overall, the Capacitance and Tan Delta measurements indicated a thermally aged insulation system in a deteriorated
condition.
The main advantage in measuring end-winding values is for comparison purposes against previous results. Since no
previous results exist for this machine these results should be used as a finger-print against future tests.
The negative Tan Delta measurements generated in the end-winding test are an effect created by the stress control
coating causing a non-linear path to earth. These measurements are repeatable and give valid insight to the
condition of the end-winding when trending is available.
Table 5
Tan Delta and Capacitance Measurements for the Complete Phase
Test Voltage Capacitance (nF) Measured Tan
nU
U(pu)
U
(kV)
A
Phase
B
Phase
C
Phase
A
Phase
B
Phase
C
Phase
0.1 1.38 786.8 798.8 795.2 0.03094 0.03026 0.03084
0.2 2.76 788.8 800.4 797.3 0.03135 0.03037 0.03096
0.3 4.14 790.5 802.0 799.2 0.03153 0.03075 0.03130
0.4 5.52 793.8 805.0 802.4 0.03450 0.03344 0.03353
0.5 6.90 802.1 811.8 809.8 0.04284 0.04030 0.04059
0.6 8.00 806.0 819.6 818.0 0.04795 0.04668 0.04724
Table 6
Tan Delta and Capacitance Measurements for the End-winding
Test Voltage Capacitance (nF) Measured Tan
nU
U(pu)
U
(kV)
A
Phase
B
Phase
C
Phase
A
Phase
B
Phase
C
Phase
0.1 1.38 7.37684 7.50329 6.53957 -0.0014 -0.0025 -0.0017
0.2 2.76 7.37649 7.50068 6.54095 -0.0029 -0.0027 -0.0047
0.3 4.14 7.37552 7.49909 6.53848 -0.0028 -0.0029 -0.0050
0.4 5.52 7.37511 7.49832 6.53608 -0.0028 -0.0029 -0.0052
0.5 6.90 7.37477 7.49790 6.53404 -0.0029 -0.0031 -0.0054
0.6 8.00 7.37426 7.49773 6.53333 -0.0031 -0.0031 -0.0055
7/22/2019 RM-2 Assess H2O Gen Simmonds
8/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
8
Table 7
Comparison of Measured Complete Phase Tan Delta and Capacitance Values
Phase Tan at
0.2 Un2
2.06.0nn
UU TanTan
Maximum
Tan
per 0.2Unstep
Increase in
capacitance at
phase voltage
(% of C at 0.2Un)
A 31 x 10-3 8.3 x 10-3 13.4 x 10-3 2.2
B 30 x 10-3 8.2 x 10-3 13.2 x 10-3 2.4
C 31 x 10-3 8.1 x 10-3 13.7 x 10-3 2.6
nb. Un is the rated line voltage.
0.02
0.025
0.03
0.035
0.04
0.045
0.05
0 2 4 6 8 10
Test voltage (kV)
Tan
Del
ta
A Phase B Phase C Phase
Phase
oltage
780
790
800
810
820
830
840
0 2 4 6 8 10
Test voltage (kV)
Capacitance(nF)
A Phase B Phase C Phase
Phase
oltage
Stator Complete Phase Tan Delta and Capacitance (nF) versus Applied Voltage (kV)
Figure 9
-0.02
-0.015
-0.01
-0.005
0
0 2 4 6 8 10
Test voltage (kV)
Tan
Delta
A Phase B Phase C Phase
Phase
oltage
5
5.5
6
6.5
7
7.5
8
0 2 4 6 8 10
Test voltage (kV)
Capacitance(nF)
A Phase B Phase C Phase
Phase
oltage
Stator Endwinding Tan Delta and Capacitance (nF) versus Applied Voltage (kV)
Figure 10
7/22/2019 RM-2 Assess H2O Gen Simmonds
9/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
9
Electro-magnetic core imperfection detector (ElCID) tests
The stator core of this unit had not previously undergone an ElCID test, therefore, there were no base line results
with which to compare. The traces contained within this report should be used as a finger-print against future
tests.
The rotor remained in place with the removal of 2 pairs of three poles 180 apart to facilitate the test and aid in thevisual inspections.
Figure 11
The hydro-generator stator core was 1.8m long, ~20m in circumference and had 486 slots. With two bars per slot
there were 972 bars housed within the core. The core was constructed in six segments, with the splits in the core at
the centers of the teeth on slots 40, 121, 202, 283, 364 and 445.
For this test the core was excited with a winding consisting of three groups of 15 turns, connected in series to give a
total of 45 turns. Each group was wound loosely around the core, and centered between the core splits as best as
possible. This type of excitation winding is often used to test hydro-generators; it is fairly easy to install, fits
through the air-gap and gives a reasonably uniform flux distribution around the core.
The stator core was made up of six segments. The joints between each segment tend to give high inter-laminationcurrent readings, during an ElCID test. Special consideration must be given in the vicinity of a core joint when
conducting an ElCID test on large hydro-generators. Inevitably when the sections of core are joined a non-uniform
air-gap exists between sections. Although the air-gap is usually small (0.2-0.5mm) it is significant in terms of mmf,
and a relatively large portion of the total mpd applied to the core is absorbed at such joints. Unfortunately, due to a
malfunction with the motor of the ElCID tractor near the end of the tests, the core joint measurements were not
repeated with the phase reference reset.
Even after factoring in the extra current required at the core joints it was noted that the current required to excite the
core, to 4% rated flux, was surprisingly high and could not be reached. The test was performed at 2.7% rated flux
and scaled up using the instrument software. At this reduced rating the excitation current was measured at 660At, so
where was it all going?
The phase values for all of the slots was extracted from the ElCID software and plotted on a graph. Averaged phasevalues between 0.1m and 1.6m were used to avoid end of core effects. From the graph it is clear that the phase
current at the core joints was very large. Since the joint mmf for a typical test at 4% rated flux is ~31A/mm gap,
therefore, at 2.7% this gives ~21At/mm, which sums to 126At at the core joints. The total joint mmf was directly
measured from Figure 12 as the average Phase mmf at the joints less the slot excitation i.e. 38+25+38+28+30+29-6
= 182At. These higher mmfs implies that the joints were quite wide and indicated either a very generous design
tolerance or structural movement in the machine. On inspection each core joint had a gap between sections of
~1.5mm.
7/22/2019 RM-2 Assess H2O Gen Simmonds
10/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
10
Figure 12
Also from Figure 8 we can see that the circulating currents are quite substantial and repetitive across section pairs.
The ElCID test was performed after the HV tests, therefore, the phase windings remained at earth potential.
Therefore, with three parallel paths per phase, there would have been 9 parallel circuits in total that would have
carried induced current from the excitation. At the core joints some of the circumferential flux escapes from the
main body of the core and crosses the gap by flowing up and down the teeth either side of the joint. This flux
flowing up and down the teeth links the stator winding inducing significant currents within it. Further currents
would be induced into the stator, by the same process, from the close proximity of the excitation cable to the core
(Fig.7). And finally, since the rotor was also in place with the removal of 12 of the poles to allow access, there
would have been an asymmetric flux leakage occurring to the remaining poles, inducing more current into the
parallel circuits. However, these are secondary as the rotor joints would be the dominant driver for the circulating
current.
The circulating current in the stator winding can only come from the excitation but since the magnetic coupling is
weak and both the mmf source and burden are in the winding their contribution to the excitation requirements
cannot be readily determined. However, from a simple sum of the absolute slot circulating currents (~330A across
all slots) and baring in mind that the currents will be varying during the test due to the movement of the rotor then
approximately 50-100At would be needed to drive these currents. The excitation current, 660At, is the sum of the
total mmf both for the main core and to drive the flux across the joints. The total joint mmf is directly measured as
the average Phase mmf at the joints less the slot excitation i.e. 38+25+38+28+30+29-6 = 182At. Therefore, the
current required to drive the core iron induction is approximately 660-182-70 = 400At. Normally the core induction
is estimated from the core geometry and based on a 20m circumference would be 300At at 4% rated flux (~15At/m).
This would leave 200At unaccounted for at the 2.7% rated test voltage, so could it be due to core damage?
The core induction current is in phase with the excitation current, whereas the core loss/damage is essentially
resistive and in quadrature to the induction current. The phase angle between the phase and quad components at thesix joints was plotted along the main slot length. Fig 8 shows a fairly consistent 17-20 angle between the core flux
and excitation current and is typical of hydro-cores. This indicated that globally there were no significant issues
with the core, however, small local faults and even large ones will be masked by the majority of the unfaulted core.
7/22/2019 RM-2 Assess H2O Gen Simmonds
11/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
11
Figure 13
It was suspected, due to the age of the core, that the increased current requirements were partly caused by the
reduced permeability of the core material or significant remenance e.g. if the permeability of the electrical steel was
as low as 1000 then the core circumference alone would require up to 620At to induce 4% of the rated flux. Typical
relative permeabilities of good electrical steels are 2000 or better. A combination of this and the presence of parallel
windings and wide core joints, plus induction of eddy currents in the main core frame and support structure, have
produced the high excitation currents. From this evidence it would not appear that the high currents were due to any
massive or extensive core faults. Approximately one hour into the test the 50A variac began to overheat.
Figure 14
Fault in centre
of tooth tip
Deep seated
fault
7/22/2019 RM-2 Assess H2O Gen Simmonds
12/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
12
Overall, the ElCID test indicated that up to 70% of the slots (352 out of 486) gave inter-laminar current readings
above 100mA. The majority of the high readings were recorded at the end core packets, where there were visible
signs of damaged and bent laminations. Many of the other high peak current readings coincided with impact
damage on the surface of the core packets. There were also several deep seated faults, the highest of which was
~300mA on slot 57. It should be remembered that the attenuation of the signal for a deep seated fault is
approximately 30%, so the actual fault level would have been considerably higher and able to generate a significant
amount of heat. Correlations between the low flux ElCID test and the high flux Ring tests indicates that a 100mA
fault current represents a 5-10C increase in local temperature around the fault. Generally, the base-line for the
average quadrature (core loss) current for each slot was high and would represent fairly high heat losses throughout
the core and also more localized hot-spots around the larger faults. Due to the high number of slots involved a
detailed analysis of each slot should be performed, with the rotor removed, to determine the peak quadrature
currents that are a real threat to the continued operation of the core. Repairs to the core teeth e.g. by acid etching
and grinding, should be performed and the fault current reduced below 100mA wherever possible. A ring flux test
should also be performed to assess the significance of the deep-seated faults contained within the core.
It was thought that the general deteriorated condition of the inter-laminar insulation had caused an increase in the
joules heating of the core, affirmed by the strong smell of burning coming from the core and core packets on the
face and back of the core. If this was the case then the additional heating will affect and contribute to the continued
thermal deterioration of the stator winding insulation.
Visual inspections
The visual inspections carried out at the top of the stator and end winding highlighted several defects brought on by
general looseness of the windings. Most noticeable was the high number of slot wedge fillers and slot wedges that
had migrated out of the top of the slots. The slot wedge filler was found to be protruding from the majority of the
slot exits around the top of the core. The migration of the under wedge packers was prevalent throughout the core.
The majority of the end wedges, around the top of the core, had migrated out of the slot, some by over 50mm. The
outer layer of tape can also be seen to be unraveling on one of the bars.
Figure 15 Figure 16
Damage to the outer tape of the winding insulation, at the slot exits, was noted in many of the locations where the
filler or wedge had vibrated out of the slot. Ridges on the bars, near the slot exits, show where the protruding edge
of the filler had chaffed against the insulation. On several outer bars, near the slot exits, the outer insulation
appeared to be flaking away and felt loose and brittle to the touch.
7/22/2019 RM-2 Assess H2O Gen Simmonds
13/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
13
Figure 17 Figure 18
Loose and broken bindings as well as loose blocking, at the end-windings and the phase ring connections at the top
of the core, was in clear evidence. Thermal discoloration of the insulation of the phase rings was also in evidence.
The movement of the loose binding had caused abrasions to the outer insulation at several locations. The insulation
also had hollow feeling areas and could be an indication that insulation delamination had occurred to some extent.
A small puncture in the insulation to one of the phase rings can be seen in Figure 20, probably caused by a
metallic/conductive material stuck to the insulation surface. Figure 21 shows a foreign object wedged between the
tape layers at a series connection at the bottom of the winding.
Figure 19 Figure 20
7/22/2019 RM-2 Assess H2O Gen Simmonds
14/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
14
Figure 21 Figure 22
Generally, there were high contamination levels around the whole generator. Debris from previous repairs e.g.shards of epoxy, old wedges and fillers, welding rods and in one instance a bastard file were scattered around the top
of the core and rotor poles, inside the stator frame and all across the bottom. Much similar debris was also observed,
and pulled out where possible, between the end-windings at the top and bottom of the core but predominantly from
the top. One particular piece of recovered debris shows clear indications of overheating possibly brought on by
intense discharge activity.
Figure 23 Figure 24
As a result of the stator earth fault, caused by the breakdown to earth between one of the bars in the slot to the core,18 new/reconditioned bars had been fitted, prior to the HV tests, to slots 80-97. Previous replacement bars also
includes those in slots 13 to 42 and 138-154. The stator bars in the slots had been found to be suffering from
excessive slot discharge and can be seen very clearly from bars that had recently been removed from the core.
7/22/2019 RM-2 Assess H2O Gen Simmonds
15/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
15
Figure 25 Figure 26
Figure 27 Figure 28
The visual inspections carried out on the stator core highlighted numerous core packets with bent or damaged
laminations, particularly at the top and bottom end packets of the core. Some of the core packets have several
laminations broken off at the corners. Impact damage, in the form of bent laminations, was noted on a number of
core packets and was also noticeable on several of the slot wedges.
7/22/2019 RM-2 Assess H2O Gen Simmonds
16/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
16
Figure 29 Figure 30
In general, the wedges were in poor condition, having frayed edges and damage to the dove tails as the wedge hadbeen forced into the slot, during re-wedging. This action had also caused the laminations in the wedge groove to be
bent together. Impact damage was also visible on the surface of many of the wedges. Where accessible, the wedges
were tapped to ascertain their tightness. A large proportion of the wedges were found to be loose. The looseness of
the wedges, which had allowed the wedge and slot filler to migrate out of the top and bottom of the slots, had
created gaps between the wedges allowing access to the under-wedge filler. The majority of the slots contained gaps
between wedges somewhere within the core, with gap sizes varying from a few millimetres up to 30mm.
Figure 31 Figure 32
Also very evident was a strong smell of oil and burning coming off of the core packets all the way around the bore.
It was not possible to determine whether the smell was coming from the core packets or from the stator bar
insulation but more likely it was a combination of both. There were fairly high levels of dirt and grease covering the
face of the core. Damage was observed on several of the step-down end packets where some of the teeth have
broken off.
Similar to the front of the core, there was a strong smell of burning coming off of the core packets, from the back of
the core. Many of the air ducts were found to be blocked with dirt accumulation and bits of wood. Some sort of
probe had been embedded into the back of a core packet. The areas around it had blackened as if from overheating.
7/22/2019 RM-2 Assess H2O Gen Simmonds
17/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
17
Resin had, at some point, flowed over the core packets at 3 or 4 locations, the cause of which has remained
undetermined. In general, the back of the core was full of loose debris, mainly epoxy chards and remains from
previous repairs.
Figure 33 Figure 34
Figure 35 Figure 36
The stator support structure was found to be in a deteriorated condition. Re-enforced concrete supports were found
to be crumbling away and one of the re-enforced beams was visibly buckled. The core packets in that area and in
others were buckled and sagging in places.
7/22/2019 RM-2 Assess H2O Gen Simmonds
18/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
18
Figure 37 Figure 38
Conclusions from the Condition Assessment
It seems clear from the results of the electrical tests and the visual inspections that this generator was in a
significantly deteriorated condition. But perhaps this shouldnt come as too much of a surprise since the core was
~50years old and the insulation system nearly 30years old.
The rotor pole drop test indicated that 15% of the 72 poles contained one or more shorted turns. Two of the poles
required immediate repair/replacement after indicating that several coils, in each pole, were involved.
The stator winding tests indicated that significant discharge/sparking events were occurring either between bars at
the end winding or from an electrically floating object. From the amount of debris found between the top bars it is
quite possible that electrically conductive material could be forming an intermittent bridge between HV conductors.
The PRPD patterns also indicated consistent discharges to the core which could have been emanating from/near the
slot exits or pressure fingers. The measured losses of the winding indicated that the insulation was in a thermally
deteriorated condition. Initial low voltage measurements, which are usually unaffected by PD, were found to be
relatively high and indicated that the void content of the insulation was high. The results could also be an indication
of poor contact between the semi-conductive coating and the core, which, considering the history of slot discharge
with this machine, seems a reasonable assumption. The Tan Delta tip-up was also high indicating a large number
of void becoming active at ~4kV and above. The capacitance measurements, for each complete winding, indicated
that the Blue phase had a noticeably lower value, compared to the other 2 phases. This could also be a pointer to
indicate that the insulation had suffered from long-term thermal deterioration. Again, this is the point where
previous results would be invaluable.
This also brings up the issue of contamination. Much of the HV tests could have been influenced by the high levels
of debris and contamination by conductive particles. It is, therefore, very important that levels of cleanliness within
the machine be improved and maintained and that all foreign objects are removed from the environment.
Problems regarding slot discharge appeared to be well established with this machine as was evidenced by the
number of new/repaired bars that had been installed in recent years. The mechanism responsible for this problem is,
of course, vibration. After conducting the visual inspections it became clear that the winding support system was
not doing its job. The majority of the wedges were loose and in poor condition, several of which had migrated out
of the core along with there under wedge fillers. Much of the end-winding blocking and binding showed evidence
of movement and chafing to the phase ring insulation had occurred against support bindings.
7/22/2019 RM-2 Assess H2O Gen Simmonds
19/20
2010 Doble Engineering Company -77 thAnnual International Doble Client ConferenceAll Rights Reserved
19
Fully re-wedging the core, using a re-designed arrangement incorporating resilient packing e.g. ripple springs, may
arrest the development of slot PD for a time, but many of the original slot bars will already have poorly deteriorated
slot coating/insulation. Consideration should have been given to the core before any bars were replaced, particularly
if it had suffered a ground fault. The energy involved in the ground fault could easily melt and fuse together several
of the core laminations creating a hot-spot. If the core teeth are not repaired in that area then the heat generated by
the inter-laminar fault current could damage the insulation on the replacement bar and cause another fault in the
same location.
If re-wedging takes place then wedge retaining blocks should also be fitted to prevent wedges and spacers migrating
out. The anti-corona shield should be repaired as necessary after re-wedging.
Due to the advanced age of the insulation it is likely that the occurrence of slot discharge will accelerate and that an
increasing number of bars will need replacing.
The ElCID test and the visual inspection of the core indicated that the inter-laminar insulation was in a seriously
deteriorated condition. The sagging of some of the core packets, the large air-gap between core sections and general
smell of burnt insulation, amongst other reasons, indicated that the core had loosened, probably due to long-term
overheating and shrinkage. This would also explain the numerous broken tooth-tip laminations, which would have
occurred due to chatter. However, the results did not indicate any massive or extensive damage to the core.
Of great concern was the integrity of the stator core support structure. To prevent the risk of collapse the core
supports should be tested for fatigue. Attempts should also be made to tighten the core to within manufacturers
guidelines.
Due to the high PD magnitudes and the advanced stages of slot discharge and thermal deterioration taking place
within the insulation, it was strongly recommended that consideration be given to completely rewinding this unit.
The present stator winding insulation is ~30years old.
Considering the age and condition of the generator core and insulation system it is possible for a catastrophic failure
to occur in service, at any time. There is a high risk that a stator ground fault will occur somewhere within the
insulation system due to advanced stages of slot discharge and thermal deterioration. There is also a high risk of
core failure due to hot-spots caused by damaged core packets or the general deterioration of the inter-laminar
insulation. Unfortunately, there is no time limit that can be attached to any of these failure mechanisms only that
there is a significant risk that one or more of them will occur sometime in the near future.
Therefore, the conclusions of this condition assessment can only support the complete renewal of this generator, if
continued reliable service is to be maintained. Remedial actions can be taken to minimise the effects from some of
the failure mechanism but the deterioration is considered to be too widespread for any action to have a long-term
effect.
7/22/2019 RM-2 Assess H2O Gen Simmonds
20/20
2010 Doble Engineering Company -77 thAnnual International Doble Client Conference 20
REFERENCES
Special thanks to David Bertenshaw for his aid in analysisng the ElCID data.
[1] Simmonds I.W, Unit 1 generator condition assessment report, DPT 2008.
[2] D. R. Bertenshaw and J. Sutton, "Application of the EL CID Test with Circulating Currents in Stator
Windings," Inductica 2004 Berlin, Germany: Coil Winding, Insulation & Electrical Manufacturing
International Conference and Exhibition, 2004, pp. 128-134..
[3] Jackson RJ, and Wilson A, Slot Discharge Activity In Air Cooled Motors And Generators Proc IEE, B 1982,
129(3) pp 154-167
[4] Stone, G. et al, Electrical Insulation for Rotating Machines, John Wiley & Sons 2004.
[5] D. R. Bertenshaw, "Analysis of stator core faults - a fresh look at the EL CID vector diagram," Hydro 2006
Porto Carras, Greece: The International Journal on Hydropower & Dams, 2006, pp. 15.02 1-10.
[6] IEC Duarte E.M, Update Of The Power Factor Database For Generator Stator Insulation,
Doble Engineering Company, 2004.
[7] IEEE Standard 492-1999 Guide for Operation and Maintenance of Hydro-Generators.
[8] IEEE Standard P62.2, Guide for diagnostic field testing of electrical power apparatus.
[9] IEEE Standard 56, Guide for insulation maintenance of large Alternating-current rotating machines.
BIOGRAPHY
I Simmonds has been employed at Doble PowerTest since 2002, and currently works as Team Leader
in the Rotating Machines Department. Mr. Simmonds received an Honors Degree in Electrical
Engineering from the University of Southampton, England.