You are on page 1of 67

Generator Testing

Power Users Conference


Nov. 3-4, 2015
Bert Milano, consultant
umbertomilano@hotmail.com
Littleton, Colorado
303-973-8479

1
AC and DC Test Procedures

Refer to IEEE standards and


guides for procedures, test
voltages etc.

2
AC Tests
• AC High Potential Test
• Corona Probe Test
• Power Factor
• Partial Discharge…. To be covered by
others

3
DC Tests
• Insulation Resistance (IR)
• Polarization Index (PI)
• DC High Potential Voltage Withstand
• IEEE Timed Voltage Step
• DC Ramp
• Reverse Energization

4
DC Testing Currents
• Leakage…
• Conduction…
• Polarization …Capacitive…
• Polarization …Absorption ***
molecular orientation very slow!
• Decays, hours, asphalt vs. epoxy
• Degree of cross linking

5
DC Insulation Resistance
Comments:
• Resistance corrected for temperature
• Sensitive to moisture, contamination
• Temperature correction issues
• Instrumentation
• 1 minute…Absorption current

6
Polarization Index Test
• PI testing stator windings is fairly common
• If PI decreases Insulation quality concern
• leakage, conduction, water, moisture,
contamination or deterioration
• For applied step voltage V:
PI = I1 minute / I10 minute
• Where I is the total insulation current:
leakage ,conduction, absorption (cap.≈0)

7
Polarization Index Test

PI = I1 (leakage + conduction + absorption)


I10 (leakage + conduction + absorption)

Leakage and Cross linking

8
Polarization Index Test

• Reading obtained at 0.5 minute ***


• Every minute to 10 minutes
• Ratio, no units, size and temperature fairly
independent
• Numbers greater than 2 acceptable
• High PI number comments

9
Polarization Index Test
• Significant Digits
• 0.5/0.07 = 7.14…. Wrong

• 7 … Correct
• Can raise concerns over the years
• Commercial test sets…. excellent

10
Polarization Index Test
• Below 2 winding wet, dirty, contaminated
• Instrumentation accuracy
• Thermal/PD damage/aging may affect IA
• 0.5 minute reading
• Plot the data
• See IEEE standards for further analysis
• My paper

11
Absorption Current Response
IA=ΔV•C•K•T-n
Where ΔV is the applied voltage step
C is the capacitance
T is time
K is the insulation constant
n is the absorption exponent
K and n can change with aging/damage

12
n ≠ Log10(PI)

I
I≠IA
IA (per IEEE)

1 10
Time (minutes)
13
PI Test

Why the 0.5 minute


reading?
&
Why plot?
14
PI ≠ I1/I10

I=IA

0.1 1 10
Time (minutes)
15
Crystal Bar Log-Log Plot
• New bar,11.5 kV (rated13.8 kV), 35 MVA
• Epoxy mica insulation, hydro unit
• Excellent insulation (new)
• Bar never placed in service
• 5 kV PI, inflection at 1.5 minutes
• Error without the plot… trending issue

16
Figure 2 Crystal Bar
0.1

0.017 uA

PI = 27.8
K = 1.43E-3 PI = 64
0.01 n = 1.44 K = 1.46E-3
Current (uA)

0.0155 uA n = 1.81

0.001

0.0001
0.1 1 10
Time (minutes)

17
Crystal Bar Ramp Test Plot

Normal – Grading Treatment Not Altered Grading Treatment Shorted

Figure 3: Ramp DC test of a single stator (Crystal Power Plant) bar with and without grading treatment
shorted
18
Insulation Ageing
• n & K, related to insulation characteristics
• IA=ΔV•C•K•T-n
• Damage mainly near line end & front coils
• Thermal - chemical changes
• Mechanical - radial forces/damage
• Voltage - Partial Discharge … conductive
byproducts

19
PI Differences Between Phases
New 6.9 kV winding passed:
• Acceptable 5 kV PI, passed 16 kV dc and
14 kV ac tests
• PI test (2.00, 3.21 and 4.51)
• Failed the Ramp DC test with distinction!
• Winding rejected per DC ramp test

20
PI Differences Between Phases
• Extreme conduction current
• 2 phases failed
• Traced to water in 3 coils… brazing
• Epoxy is not waterproof, once wet,
delaminates, appears as insufficient
impregnation, and stays delaminated
• PI after repairs: 3.25, 2.94, and 3.97

21
Measured Current (µA)

Scale Change

Voltage (kV)

FIGURE : Unacceptable ramped high-voltage DC test results

22
1 Minute High Potential
Withstand Test

Why one minute?

23
High Potential Test
• Answer:

That is about how long a test


engineer can hold his breath!

24
AC and DC High Potential
Withstand Testing
• Go – no go tests
• No information as to quality
• Insulation either passes or fails
• Insulation good for 3 to 5 years ! If tested
to the appropriate voltage

25
High Potential Test
Main objection to DC Testing:

• It stresses the end winding insulation

And AC high-pot doesn’t stress or Test


the end windings

26
Why DC Test
• Response:

• End winding has same voltage rating!


• Why not test the end winding?
• Contamination, cracks
• Grading Treatment

27
High Potential Test
• End winding AC voltage during operation
• Ac voltage distribution
• Coil insulation capacitance ..high
• Air space capacitance … low
• Thickness and Dielectric constant
• Voltage appears across the air gap,
not the insulation

28
29
High Potential Test
• End winding DC voltage during operation
• 1 minute test
• Absorption current decay….
• IR very high
• Ia determines the voltage distribution
• Voltage 15-30% of test value better (more
searching) than the AC test

30
DC High Potential Test
Secondary objection:
• DC testing is hazardous to insulation 1959

Response: GE promoted 0.1 Hz test set!!


• Absorption current to the rescue
• “1 minute test: conduction current is much
too low to affect the internal stress”, BB

31
When Contaminated End Turns
• Both tests stress the end windings
End Winding Condition:
• Aging minimal
• Mechanical minimal
• Thermal minimal
• Voltage minimal

32
Selecting the Test Voltage
Warning:

“Hi-Pot Testing at a very low voltage is an


unconscious decision to run the unit to
failure.”

Testing is Insurance

33
AC / DC Test Voltage Ratio
• Reclamation (2 or 3 efforts)
• EPRI and others
• No significant differences between ac & dc
• The Best study…
• Brown Boveri Review, April/May 1968,
Vol. 55, pages 208-214, Direct-Voltage
Testing of the Insulation of Windings for
Rotating Machines
34
BB: AC/DC Test V Ratio 1.56

35
AC / DC Test Voltage Ratio
• VDC maintenance test Voltage:
• 1.7 times 85% = 1.44
• Still lower than the 1.56!!

36
AC / DC Test Voltage Ratio
K factors for new insulation are larger, and
as the insulation ages the K factor drops.
Brown Boveri states “This may be taken as
evidence that long-term changes (fatigue)
are revealed earlier by a DC test.”

37
AC / DC Test Voltage Ratio
So:
• Decide if you want your end of life winding
to fail in service or under test
• There are other more valuable tests to
assess without the risk of failure
• DC Ramp
• Corona Probe

38
DC Ramp Testing

39
Ramp DC Testing Synopsis
• Valuable non-destructive test
• Does not damage insulation
• Early warning and early termination
• Automated 16.67 V/sec, or 1kV/sec
• It is a very low-frequency test
• Current plotted against voltage (time)

40
Ramp DC Testing Synopsis
• Trend over time
• Comparative analysis also of value
– Phase-to-phase
– Sister units
• Excellent repeatability

41
Ramp DC Test Description
• Ramp DC Test results provide information on the
following current components (see next slide example):
– Capacitive
– Absorption
– Leakage/Conduction
• The ramp linearizes absorption component
• Capacitive component response to ramp is a constant
current
• Provides ability to analyze individual insulation current
components
• Thus any non-linear response provides diagnostic value
– Leakage/Conduction

42
Example Ramp Test
Icon. or IL ~ 0.5 μA

IA ~ 2.6 μA

IC ~ 18.6 μA

43
Capacitive Component
• I = C times ∆V

• C = I / ∆V

• C = 18.6 micro amps / 16.67 v/sec =


1.12 micro farads

44
Capacitance

• PD activity eventually destroys several


layers of insulation
• Significantly deteriorated or wet insulation
will show an increase in capacitance
versus applied voltage
• Due to intense PD/ionization and
conductive byproducts of PD
• Due to dielectric constant of water

45
Capacitance

• Green powder resistance per 20 mill


thickness per square is about 100 to 200

• Good insulation resistance… several GΩ
• Effectively shorts/bypasses several layers
of insulation as voltage increases
• Capacitance increases

46
47
Ramp DC Test
Capacitance:
• Results in increased capacitance identified
as snaking
• This transition usually occurs between 8
and 12 kV dc for 13.8 kV insulation
• Thus the standard PF test cannot detect
• The Alternate Ramp test used by some is
less than adequate

48
Ramp DC & Power Factor Tests
Capacitance
– Capacitance also increases with
voltage due to grading treatment V-I
response
• Effect below 6 kV dc for 13.8 kV insulation
• Grading treatment properties related

49
Crystal Bar ramp Test Plot

Normal – Grading Treatment Not Altered Grading Treatment Shorted

Figure 3: Ramp DC test of a single stator (Crystal Power Plant) bar with and without grading treatment
shorted
50
Dielectric Properties (cont.)
Breakdown:
• Ramp DC: the rate of voltage increase is
slow enough to detect early avalanche
conditions so the test can be terminated
before tracking/breakdown occurs
• Lack of understand or expertise issue

51
Reclamation Ramp DC Test
History
• 30 years
• 240 stator windings
• 3 phases per winding
• Testing 3 to 5 years
• Results in approximately 5400 tests
• Four (4) sudden failures: excellent record

52
Winding Failures During Test
End Winding Failures:
• Failure due to mechanical impact in
grading paint area
• Flooding, water washed over the
winding, entered a crack in an end turn
and tracked to the slot paint
• Grading treatment protection

53
Winding Failures During Test
Failures in the Slot Section:
• 14 year old winding vibrating strands cut
through 50 t0 75% of the groundwall
insulation
• Crack in the ground wall insulation in the
slot

54
Winding Failures During Test

“I do not believe anyone would want to


return a unit to service if they knew the
unit had a cracked coil or if the there was
a coil in the unit with the turn insulation
and 50% of the ground wall insulation cut
through and destroyed.”

55
High Potential Testing Survey
2008 EPRI survey
• Some survey respondents perform
maintenance tests as low as 15.6 kV dc or
13.8 kV ac on a 13.8 kV winding
• This is of little to no value
• These low test levels can result in in-
service failures
• The Ramp DC test “inappropriately” limited
to 15.6 kV would be a much better value
56
High Potential Testing Survey
Experience:
• Subsequent test-to-failure voltages for
13.8 kV machines that failed in-service
failed under test in the range of 15 to 20
kV ac and 17 to 25 kV dc
• Thus testing at greatly reduced voltages is
an “unrecognized decision to run units to
failure”

57
Glen Canyon 1964 -1966
Vintage Asphalt-Mica Windings
• 8 units, 13.8 kV, ramp tested in mid 1970’s
• Excessive non-linear current, tests
terminated between 9 and 15 kV dc
• Rewind program initiated
• In-service failures occurred as new
windings were arriving on site
• The 5th unit to be rewound did not fail in
service, failed under test 18, 18, 20 kV ac
58
Corona Probe test
• Energize at 95% or 100% rated voltage
• Probe each slot phase by phase
• Measurements are not in milliamperes
• Very sensitive microvolts radiated
– New coils typically read 2 to 7
– Seriously deteriorated coils have
readings around 150 and higher.
– Highest ever… 280
59
1980 Corona Probe Data
Yellowtail Failure
• Corona Probed after failure, 2 coils with
readings of 260 and 280 were dissected
• Same condition as failed coil
• Turn and 50% of ground-wall insulation
gone
• 2 coils with readings of about 150 were
also dissected, voids, severely
deteriorated binder, and overheating
60
61
Comparative Corona Probe, PDA, and
Ramp DC Test Results
Glen Canyon 13.8 kV, multi turn coils
• 1984-5 epoxy mica-splitting rewinds
• 2006 Ramp DC tests very minor snaking,
and conduction at the higher test voltage
• Dec, 2005 PDA on Unit 6: Qm 54 to 76, no
predominance
• July, 2007: Qm 58 to 96, no predominance
• July, 2008, turn-to-turn failure
62
Comparative Corona Probe, PDA, and
Ramp DC Test Results (cont.)
Glen Canyon
• Sister Unit 5 had the rotor out
• Corona probe readings: line end 95, and a
few others around 50 - 80
• Unit 5 June, 2007 PDA: Qm 38 to 75, no
predominance (bus couplers)

63
Comparative Corona Probe, PDA, and
Ramp DC Test Results
Yellowtail 13.8 kV, multi turn coils
• 4 rewinds 1983-6, epoxy-mica insulation
• Unit 1: 1987 tests, slight snaking, corona
probe readings all below 10, highest
numbers were line end coils
• 1995 significant snaking
• 1999 PDA differential couplers installed
• Qm 421-1163, no predominance
64
Comparative Corona Probe, PDA, and
Ramp DC Test Results
Yellowtail
• “…Qm quite high for a machine of this age and
size, …voids within the bulk of the ground wall
insulation…”
• PD tested 2 years later, Qm 267 to 1075 and no
predominance
• Ramp DC testing shows snaking and
conduction/leakage at the higher voltage
• 2015 still in service: serious concern
• Subsequent corona probe tests will be
performed in 2016

65
Reverse Energization Testing
• End winding investigation
• Ground the winding and apply voltage to
insulation surface
• Aluminum foil
• Electrostatic gun, very low current
• Detects cracks, punctures, damage

66
Thank you

Bert Milano
Consultant
umbertomilano@hotmail.com
303-973-8479

67

You might also like