2011 Electrical Insulation Conference, Annapolis, Maryland, 5 to 8 June 2011
Insulation Resistance Measurements for Machine
Insulation
Andrew Brown
Kinetrics Mounir Essalihi
Toronto, Canada Hydro-Québec
[email protected] Montréal, Canada
[email protected] Eric David
École de Technologie Supérieure
Montréal, Canada
[email protected] Abstract— In order to gain assurance that a machine winding objective way to use IR values that might be more useful than
insulation can safely withstand the service voltage and assess its the general guidelines provided in IEEE Std-43.
reliability in operating conditions after a given number of years
in service, dc tests have been used for decades. Some specific
quantities, such as the 1-minute insulation resistance (IR) or the TABLE I. RECOMMENDED MINIMUM INSULATION RESISTANCE [1]
polarization index (PI), can be used to assess the condition of the
insulation system. The minimum IR value recommended by the Minimum insulation
Test specimen
IEEE Std-43 for most ac windings built after about 1970 is 100 resistance in MΩ
MΩ (corrected to 40oC). Whether this arbitrarily value is too For most windings made before about 1970,
high or too low is still under debate. This paper presents data IR1 min = kV + 1 all field windings, and
from windings of various machines and shows how insulation others not described below
resistance depends not only on the health of the insulation system For most ac windings built after about 1970
but also on machine size and on insulation technology. A IR1 min = 100 (form wound coils)
normalization technique using the winding capacity and
considering the insulation technology is suggested in order to For most machines with random-wound
compare values between different machines or with individual IR1 min = 5 stator coils and form-
bars or coils. wound coils rated below 1 kV and dc
armatures
Keywords-component; machine insulation, insulation
resistance, polarixation index II. THEORITICAL BACKGROUND
The theoretical background related to IR measurements is
I. INTRODUCTION summarized in [1] and can also be found in the general
Discussions on the minimum value require for the armature literature [2]. It has also been recently reviewed [3, 4] for the
winding of a large rotating machine operating under high particular case of machine insulation so we will merely restate
voltage at the phase exit have been going on since years. Table the basic equations. When a step voltage is applied between
1 below is a reproduction of Table 4 from the latest revision of the copper of a winding and ground, the current measured by
IEEE Std-43 (2000) [1]. A change from the 1974 version that an external meter will be given by
was made in this latest version is for the ac windings built after
1970, what could be qualified as modern thermosetting U o U o −τ
t
insulation system. Now the minimum required value for these I (t ) = + e + U oCo f s (t ) (1)
RL Rs
windings is 100 MΩ, instead of “kV + 1” that was used for any
winding in the 1974 version. Obviously, this change lead to a
higher minimum insulation resistance than previously, but still The physical meaning of the three terms in equation (1) are
there is some discussion on whether or not this value is high respectively, in order of appearance, the direct conduction
enough. In order to contribute to this discussion, this paper current (including surface leakage and bulk conductivity [1]),
presents data from windings of various machines and shows the capacitive current and the absorption or dielectric
how insulation resistance depends not only on the health of the relaxation current. RL is the leakage resistance related to the
insulation system but also on machine size and on insulation direct conduction current (not equivalent to the insulation
technology. A normalization technique using the winding resistance as defined by the IEEE Std-43), Uo is the applied
capacity and considering the insulation technology is suggested voltage, Rs is the instrumental resistance in series with the
in order to compare values between different machines or with specimen under test, Co is the geometric or vacuum
individual bars or coils. The suggested procedure provides an capacitance, τ is the instrumental time constant and fs(t) is the
978-1-4577-0279-2/11/$26.00
978-1-4577-0276-1/11/$26.00 ©2011 IEEE 261
Authorized licensed use limited to: Marathwada Mitra Mandal's College of Engg. Downloaded on August 07,2025 at 14:43:45 UTC from IEEE Xplore. Restrictions apply.
material dielectric response in the relevant time scale. For TABLE II. INSULATION RESISTANCE FROM FIGURE 1
almost every type of machine insulation, this latest term can IR (MΩ) C (μF) RC (s)
be reasonably fit by a power law function that can be written
A 650 3.64 2370
as [3]:
B 2910 0.90 2620
f s (t ) = At − n (2) Spare bar 1.34 x 106 0.0025 3350
III. INFLUENCE OF MACHINE SIZE
IV. INFLUENCE OF TECHNOLOGY
Unless a very high instrumental resistance is used, the
second term of (1) decays fast enough that its contribution is Except for the cases of moisture absorption and/or
negligible at 1 minute. Both conduction and absorption currents conductive contamination on the end windings, the absorption
are proportional to the number of bars or coils connected current (third term of (1)) normally dominates at 1 minute. The
together during the test if every bar/coil is electrically identical. absorption current is technology-dependent since it is related to
A very convenient way to take into account the winding (or the material dielectric losses. For instance, asphaltic insulation
machine) size is to use the winding capacitance as a systems have higher dielectric loss than epoxy-mica type
normalization factor since this value is roughly proportional to insulation systems mainly due to the chemical structure of the
the number of bars/coils and is routinely measured. Figure 1 bonding resin. A good review of the different material used for
shows the current as a function of time after a 1 kV step stator winding insulations with their historical development can
voltage for two large epoxy-mica insulated hydro-generators be found in [5]. Figure 2 shows the measurement of the
and an epoxy-mica spare bar. The measurement on machine A polarization current for three coils (polyester-1, polyester-2 and
was conducted on the three phases of the stator winding while asphalt) and two bars (epoxy-1 and epoxy-2). The currents are
the measurement on machine B is for phase A only. All three divided by the capacitance and accordingly are size-
measurements were taken around room temperature and no independent. Table III gives the normalized resistance at 1
temperature correction was applied. Table II gives the value of minute for each sample and the polarization index.
the 1-minute resistance and the capacitance for each sample. 10000
The last column gives the value of the product between the Epoxy-1
resistance and the capacitance, which could be referred as the Epoxy-2
normalized resistance. In all three cases, the specimen is in 1000 Asphalt
Current per unit of capacity (A/F)
Polyester-1
good condition. As it can be seen, the values of the 1 minute Polyester-2
resistance change tremendously from on sample to the other 100
due the size of the sample. For a relatively small machine or for
a measurement conducted on a single phase, when the 10
insulation is in good health, the resistance value will normally
be much higher than the minimum value given in Table 1.
1
When normalized by the capacitance, the 1 minute resistance 0.1 1 10 100 1000 10000
yields roughly similar values for the three samples.
Accordingly, for comparison purpose, the use of the 0.1
normalized resistance seems more appropriate than the absolute
value. 0.01
Time (s)
1
0.1 1 10 100 1000 10000
0.1
Machine A Figure 2. Normalized currents for different winding insulation systems after
0.01
Machine B
a voltage step of 5 kV (unguarded electrode configuration). Adapted from [4].
0.001 Spare bar
0.0001
TABLE III. NORMALIZED IR AND PI FROM FIGURE 2
Current (A)
0.00001
0.000001 PI RC (s)
0.0000001
Epoxy-1 10 4970
Epoxy-2 6.5 2160
1E-08
Polyester-1 7.4 1080
1E-09 Polyester-2 4.9 149
1E-10 Asphalt 5.1 462
1E-11
Time (s)
One can see in Table 3 that more than an order of
magnitude in the value of the normalized resistance can be
Figure 1. Polarization current at 1 kV for three diffent specimens. observed for the different technologies. Experience has shown
that for modern epoxy-mica insulated windings in good health,
the normalized resistance should be roughly above 1500 s
262
Authorized licensed use limited to: Marathwada Mitra Mandal's College of Engg. Downloaded on August 07,2025 at 14:43:45 UTC from IEEE Xplore. Restrictions apply.
when measured at room temperature while it is expected to be polyester based insulation system could have low values of
significantly lower for older insulation technologies. normalized IR, particularly early generation of polyester-based
insulation. For example, the last line of Table VII indicates a
A complication that arises for the insulating system labeled very low value of both IR and normalized IR which was
epoxy-1 in Figure 2 is the occurrence of an interfacial acceptable for this technology (same as “polyester-2 in Table
relaxation peak between the SiC-based stress grading coating III).
and the insulation wedge. Further discussion on this subject can
be found elsewhere [6, 7]. This leads to a deformation of the
polarization curve that can possibly affect the resistance TABLE V. IR DATA FROM CLASS F – EPOXY MICA WINDINGS
reading, particularly if the test is conducted at low voltage. IR @ PI C (μF) RC (s) @ 60 s
Figure 3 shows the polarization curves at 1 kV for the stator 60s
winding of two generators, machine C and machine D. (GΩ)
Machine C is a turbo-generator with SiC-based stress grading 4.82 11.3 0.349 1682
coatings and machine D is a very large hydro-generator with 13.8kV -
140MW Air
ferrous oxide-based stress grading coatings. One can see the Cooled Turbo 6.51 12.4 0.349 2272
strong contribution of the stress grading coating for the Generator 5.21 10.9 0.349 1818
measurement on machine C. This resulted in a low normalized
resistance value and an abnormally high value of the 18kV -270MW 2.23 3.25 0.381 850
polarization index as shown in Table 3. The interfacial H2 Cooled
Turbo 2.19 3.38 0.381 834
polarization mechanism yielding the relaxation peak is Generator
considerably smaller in the case of ferrous oxide based stress 2.19 3.4 0.381 834
grading coating or when measurements are conducted at higher
13.8 kV –
voltage. The normalized values of the 1 minute resistances are 32MW Air
40.4 5.62 0.08 3232
given in Table 3. Since the 1 minute value for machine C does Cooled Turbo
not reflect the groundwall insulation, the value at 10 minutes in Generator 40.3 5.64 0.08 3224
this case is more relevant. 40.1 5.83 0.08 3208
20 kV – 1.41 1.65 0.17 240
100
400MW Water
Cooled Turbo 0.797 1.19 0.17 135
Machine C Generator
Normalized current (A/F)
10 Machine D
2.55 3.04 0.17 433
13.8kV –
2.06 5.06 0.4 824
1
145MW Air
0.1 1 10 100 1000 10000 Cooled Turbo
Generator 1.98 5.95 0.4 792
0.1 1.99 5.91 0.4 796
13.8kV - 15 7.9 0.11 1650
34MW Air
0.01 Cooled Turbo 13 8.7 0.11 1430
Time (s)
Generator
12.5 8.5 0.11 1375
Figure 3. Normalized currents at 1 kV for the windings of a turbo and an
13.8kV -
hydrogenerator. Adapted from [8]. 15.5 7.53 0.16 2480
33MW Air
Cooled Turbo
Generator 19.6 8.08 0.16 3136
TABLE IV. NORMALIZED IR AND PI FROM FIGURE 2
20 8.36 0.16 3200
C (μF) PI R1min (MΩ) RC1min (s) RC10min (s)
Machine C 0.32 26 3350 1080 28 080 2.23 7.16 1.2 2676
13.8kV - 170
Machine D 4.0 4.6 520 2080 9 568 MW Hydro 2.19 7.27 1.2 2628
Generator
2.2 7.31 1.2 2640
Table V and Table VI show data from each phase of
respectively class F and class B hydro and turbo generators. 3.99 6.36 0.36 1436.4
13.2kV -
The last column gives the normalized IR value. As it can be 57MW Hydro 3.97 6.39 0.36 1429.2
seen, most of the data are above or in the vicinity of 1500 s. In Generator
the case of the 20 kV – 400 MW water cooled turbo, even if the 3.99 6.32 0.36 1436.4
IR values were well above the recommended minimum value
in Table 1, the normalized IR was much lower than the normal
values for class F epoxy-mica insulation which indicates a
problem that was also showed by low PI values. Normalized IR
values of around 250 s or more are adequate for asphaltic class
B insulation as shown in Table VI. Depending upon chemistry,
263
Authorized licensed use limited to: Marathwada Mitra Mandal's College of Engg. Downloaded on August 07,2025 at 14:43:45 UTC from IEEE Xplore. Restrictions apply.
TABLE VI. IR DATA FROM CLASS B –WINDINGS
Another debate related to IEEE-43 is whether the
IR @ 60s PI C (μF) RC (s) @ 60 s recommended values should also apply for measurements on
(GΩ) individual phases instead of the three phases together. Table
13.8KV - 1.9 2.78 0.19 361 VIII presents data from a 13.8 kV – 53 MVA asphaltic
150 MW H2 insulated winding. The normalized IR is about the same for
Cooled 1.6 3.04 0.19 304
Turbo phase A, B or for the three phase together. In this particular
Generator 1.59 3.07 0.19 302 case, the phase to phase leakage had a negligible contribution
on the value of IR, although the normalized values are
18.5kV -
300MW
0.167 1.95 0.23 38 somewhat low. The normalized IR value applies for individual
Water phases as well as for three phases measurements.
Cooled 0.554 1.14 0.23 127
Turbo
0.525 2.01 0.23 121
Generator
20kV - 3.19 2.66 0.18 574 CONCLUSION
400MW
Water 3.13 2.67 0.18 563 IR data from spare bars and coils and from complete
Cooled windings are presented in this paper. It is shown that the
Turbo 3.65 3.26 0.18 657 minimum recommended value does not take into account the
Generator size of the machine and technology. A better evaluation of the
22kV -
500MW H2 /
0.046 1.01 0.19 8.74 insulation condition can be performed using normalized IR
water cooled values and taking into account the type of technology.
Turbo 0.86 1.53 0.19 163
Generator 0.078 1.2 0.19 14.8 REFERENCES
[1] IEEE Std 43-2000, “IEEE Recommended Practice for Testing Insulation
Resistance of Rotating Machinery”, 2000.
TABLE VII. IR DATA FROM POLYESTER INSULATED WINDINGS
[2] A.K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielectrics
IR @ 60s PI C (μF) RC (s) @ 60 s Press, 1984.
(GΩ) [3] E. David, L. Lamarre, “Progress in DC Testing of Generator Stator
24kV Windings: Theoretical Considerations and Laboratory Tests”, IEEE
1.9 2.93 0.205 390 Trans. on Energy Conversion, Vol. 25, p. 49-58, 2010.
600MW
water cooled [4] E. David, R. Soltani, L. Lamarre, “PDC Measurements to Assess
Turbo 1.61 2.51 0.205 330 Machine Insulation”, IEEE Transaction on Dielectrics and Electrical
Generator Insulation, Vol. 17, p. 1461-1469, 2010.
1.78 2.93 0.205 365
[5] E.A. Boulter, G.C. Stone, “Historical Development of Rotor and Stator
13.2kV 0.25 Winding Insulation Materials and Systems”, IEEE Electrical Insulation
150MW 2.04 4.75 510 Magazine, Vol. 20, No. 3, pp. 25-39, 2004.
Hydro 0.25
2.02 4.6 505 [6] N. Taylor, H. Edin, “Stator End-winding Currents in Frequency-domain
Generator Dielectric Response Measurements”, IEEE Transaction on Dielectrics
0.25
2.05 4.69 513 and Electrical Insulation, Vol. 17, p. 1489-1498, 2010.
13.8 kV – 50 [7] E. David, L. Lamarre, “Low-Frequency Dielectric response of Epoxy-
MW Hydro 0.20 3.70 0.51 100 Mica Insulated Generator Bars During Multi-Stress Aging”, IEEE Trans.
Generator on Dielectr. and Electr. Insul., Vol. 14, pp. 212-226, 2007.
[8] E. David, L. Lamarre, D.N. Nguyen, “Measurements of
Polarization/Depolarization Currents for Modern Epoxy-Mica Bars in
Different Conditions”, Proc. Of the 28th Electrical Insulation Conference,
TABLE VIII. IR DATA FROM INDIVIDUAL PHASES AND THE THREE PHASES pp. 202-206, 2007.
TOGETHER
IR (MΩ) C (μF) RC (s)
A 412.1 0.33 136
B 443.4 0.33 146
ABC 153.8 0.97 149
264
Authorized licensed use limited to: Marathwada Mitra Mandal's College of Engg. Downloaded on August 07,2025 at 14:43:45 UTC from IEEE Xplore. Restrictions apply.
