Diagnostic Testing of Hydro and Turbo Generator
Diagnostic Testing of Hydro and Turbo Generator
Diagnostic Testing of Hydro and Turbo Generators and Large AC Motors in Service
Majority of the generating sets operating in our country are turbo generators contributing nearly 70
% of the total electrical power generated. Many of these generators are about 30–40 years old and
have already come to the end of their notional design life. The important components that have direct
bearing on the operational reliability and life of the machine are: (1) Stator winding, (2) Stator core,
and (3) Rotor winding. The stator winding is the most important and expensive part of the generator,
where full power is generated. The life of the stator is affected by one or more combinations of various
stresses, like electrical, thermal, mechanical and environmental, at any point of time during operation
of the machine. Since the mechanical stresses are relatively higher in turbo generators, many of the
stator winding failures are mechanically induced electrical failures. Some of the degradation processes
include loosening of wedges and coils, slot and end-winding discharges, erosion of stress grading and
corona shielding coatings, de-lamination of insulation, de-bonding of copper from insulation, etc.
Keywords: Degradation processes, Insulation resistance and polarisation index, DC Leakage current
measurement, Dielectric loss angle, Partial discharge, Surge comparison, Wedge tightness
detection, Electromagnetic core imperfection detection.
*Diagnostic, Cables and Capacitors Division, Central Power Research Institute, Bangalore, India. E-mail: [email protected]
172 The Journal of CPRI, Vol. 8, No. 3, September 2012
The stator winding is the most important and monitoring/diagnostic testing programme would
expensive part of the generator, where full go a long way in enhancing the service life of
power is generated. The life of the stator is the machines.
affected by one or more combinations of various
stresses, like electrical, thermal, mechanical
and environmental, at any point of time during 4.0 CONDITION ASSESSMENT
operation of the machine. In addition to these PROGRAMME
stresses, the stator insulation is also subjected
to unforeseen stresses during transient over- Condition assessment programme generally
voltage conditions. Steep fronted over voltages consists of following steps:
generated during switching actions and system y Collecting O and M history of the
faults propagate through the winding and have machine
deleterious effects on the insulation. Since the
mechanical stresses are relatively higher in y Visual inspection and examination
turbo generators, many of the stator winding y Conducting diagnostic tests on the
failures are mechanically induced electrical machine
failures. Some of the degradation processes
y Analysis of the data
include loosening of wedges and coils, slot and
end-winding discharges, erosion of stress grading y Identification of deteriorating factors
and corona shielding coatings, de-lamination of y Recommendation of appropriate
insulation, de-bonding of copper from insulation, remedial measures to enhance reliability
etc. and availability
In addition to the operating stresses, the stator Collection of historical data is a critical step to
winding is subjected to unforeseen higher stresses select the kind of tests and inspections which
during transient over voltage conditions. Steep are sensitive to the likely failure mechanisms.
fronted over voltages generated during switching Careful analysis of the data can warn of the
actions, system disturbances or direct on-line problems, which are of generic as well as develop
starting propagate through the winding and have over the years due to aging.
deleterious effects on the insulation. Some of
the degradation processes associated with these The historical data include:
stresses are coil looseness, vibration, erosion of
stress grading and corona shielding coatings, y Age of the equipment
slot discharges, end winding discharges, de- y Running hours
lamination, embrittlement, de-bonding of copper
from the insulation etc. Therefore, there is a strong y Number of starts and stops
incentive from the point of view of preventive y Load levels
maintenance to carry out certain non-destructive
diagnostic tests on the stator winding to detect y Over loading
any significant changes in the state and condition y Major electrical disturbances and faults
of the insulation. As condition assessment is
based on trend analysis, the diagnostic tests y Vibration and temperature abnormality
need to be conducted periodically and generate y Record of repair and replacement of
data. The data obtained over the years would components, etc.
help to initiate appropriate remedial measures to
y O and M practices
avoid forced outages and increase reliability and
availability of the machine. A systematic condition y Design documents
The Journal of CPRI, Vol. 8, No. 3, September 2012 173
4.2 Visual Inspection and Examination 4.3.2 Insulation Resistance and Polarisation
Index Measurement
Before carrying out the electrical tests, the
equipment must be thoroughly inspected and It is a routine method checking the stator
examined for visible symptoms of deterioration. insulation. The test provides an indication of
Symptoms of electrical tracking, corona, presence of cracks, contamination and moisture
surface discharges, surface erosion, abrasion, in the insulation. The PI is defined as the ratio
time temperature effects and deformation can of IR after ten minutes to the IR after one
be easily detected through visual inspection, minute of voltage application. It is regarded
tapping, touching and feeling by hand. as the index of dryness and cleanliness of the
insulation.
In most countries, the AC voltage test is the Machine manufacturers use tan δ test as a
preferred method for finding localized weak quality control test for new stator bars and coils
points. Because of the reduced capacity and [2–4]. A general weakness in the bulk insulation
size of the apparatus, DC testing at 1.6 times normally caused by incorrect composition or
the AC test voltage level has gained acceptance insulation that is not fully cured is indicated
alternative to power frequency testing. by an abnormally high dissipation factor.
Excessive voids occluded in the insulation
If there is a weakness in the ground wall involve in discharge activity at the operating
insulation, a sudden nonlinear increase in voltage resulting in a higher than normal
current will precede a breakdown as the voltage increase in the tan δ when the voltage is
is increased. An experienced operator can increased. The parameters that can be derived
interrupt the test when the first indication of from this test are tan δ at low voltage (usually
warning occurs. 0.2 V, where V is the rated line voltage), tan δ
tip-up (average change in tan δ between 0.2 V
and rated phase voltage) and capacitance
4.3.4 Dielectric Loss Angle Test tip up (percentage change in capacitance while
raising the voltage from 0.2 V to the phase
voltage) [5].
Dissipation factor, also called tan δ, is a measure
of dielectric losses in the insulation. It is the
property of the insulating material used. For 4.3.5 Partial Discharge test
a given insulating system, the tan δ shall be
as small as possible. An AC bridge such as a Partial discharge test is another important
Schering bridge or transformer ratio arm bridge diagnostic test for HV machines, as it is capable
is used to measure the tan δ and capacitance of of revealing incipient faults in the stator
the stator winding. winding structure. In HV electrical rotating
machines, three types of discharges can be
For conducting the tan δ test, the stator winding identified.
needs to be disconnected from the cables on
(a) Internal discharges that occur in voids
both neutral sides. When the measurement is
occluded in the bulk volume of the winding
conducted on one phase section, the other two
insulation.
phase sections, are shorted and grounded to the
stator frame. The tan δ is measured in steps up to (b) Slot discharges that occur in the air gaps
a maximum of rated phase to ground voltage. The between the core laminations and adjacent
tan δ at low voltage (at 20 % of the line voltage) coil sides in the slots.
is generally below the ionization threshold level
of the voids in the stator winding insulation, and (c) End winding discharges that occur at
tan δ is only dependent on the kind, temperature, the extremity of the conducting coating
humidity, degree of polymerization, aging, outside the end of the slot where there is an
contamination, etc. of the insulation. As the test interface on the coil surface between ground
voltage increases, tan delta increases due to partial and high voltages.
discharges in voids occluded in the insulation.
Therefore, the change in tan δ with voltage is The slot and end winding discharges are known
a measure of the gaseous losses in the winding to be more detrimental to the insulation than
insulation. The slope of the tan δ-voltage curve internal discharges. The internal discharges cause
is proportional to the volume of air or gas voids slow but gradual deterioration of the insulation in
short circuited by discharges at the test voltage. the course of service. The slot and end winding
Thus, tan δ measurements provide a good discharges are severe and can cause deterioration
indication of the average condition of insulation and eventual breakdown of the insulation within
of the stator. the span of few months.
The Journal of CPRI, Vol. 8, No. 3, September 2012 175
The PD test involves energizing the individual waves are superimposed on an oscilloscope.
phase winding to phase to earth voltage from an The two waveforms will be identical if both the
external source. The blocking capacitor Cb blocks phase windings are electrically identical and
the power frequency high voltage and allows the free from faults. Any discrepancy in the two
high-frequency current impulses of PD to be waveforms indicates inter-turn fault in one of
coupled to the discharge detector. The magnitudes the windings.
of PD are calibrated in pico coulombs.
The AC test voltage is raised gradually until PD 4.3.7 Wedge Tightness Detection
pulses are observed on the detector. The voltage
The stator bars are tightly fixed/held in the stator
at which PD starts occurring is called discharge
slots of the core by wedging system. Excessive
inception voltage (DIV). The test voltage is
vibration in the machine may cause loosening
increased up to the maximum of phase to earth
of the wedges in the slots. The loose stator
voltage and magnitude of the PD pulses is noted
wedges lead to excessive vibrations and erosion
down. As the test voltage is decreased, the voltage
of stator insulation and stress control coating in
at which the PD pulses disappear is recorded. This
the generators. In the extreme cases, the loose
voltage is called discharge extinction voltage
wedges cause failure of the stator winding.
(DEV) and is usually lower than the DIV.
Therefore, it is vital to detect tightness of the
wedges in the slots.
Analysis and interpretation of partial discharges are
still a subject of intense research. The PD is highly The stator wedges are traditionally tested for
stochastic in nature. Their magnitude, repetition tightness by tapping them with a hammer and
rate and phase angle of occurrence on supply listening to the hallow sound produced. However,
waveform change continuously depending on the it is a crude method and the assessment is highly
local conditions such as temperature, pressure subjective depending on the experience of the
and chemical composition that exist in the voids. operator.
In recent times, with the advent of computers
and data acquisition systems, sophisticated PD Electronic stator wedge tightness detection is
detectors and analysers have been developed. faster, more accurate, provides consistent results
With the available techniques, it is possible than traditional hand tapping method and the
to detect the presence of slot and end winding test procedure is repeatable. The tightness data
discharges. However, there is no general can be stored electrically for trend monitoring
agreement on the acceptable levels of PD at future date. The tightness of each wedge
magnitude, DIV and DEV. As the PD is known is measured electronically and compared to
to cause chemical and mechanical destruction all other wedges in the winding, the wedges
of the surrounding insulation, it is desirable that of another winding or an absolute external
magnitude of PD shall be as small as possible and reference. The data can be stored for trend
the DIV and DEV shall be as high as possible. The analysis to determine when maintenance will be
most useful method of interpreting the PD test required.
results is performing the test at regular intervals
and monitoring the trends. The hand-held probe consisting of a magnetic
hammer is small, light and can fit most slot
4.3.6 Surge Comparison Test widths and depths. The probe is moved from
wedge to wedge and each wedge is tapped 29
Surge comparison test is used to determine the times in three seconds. The accelerometer gathers
condition of inter-turn insulation of the stator the data and transmits this information to the
winding. An impulse voltage of appropriate signal processing circuitry in the electronic case.
magnitude is applied synchronously to the two These processed signals are sent to the computer
winding sections. The resultant damped oscillatory for final analysis and presentation in a map.
176 The Journal of CPRI, Vol. 8, No. 3, September 2012
The status and position of problem stator wedges The ELCID equipment tests a core for faults
can be read off on this map immediately. by exciting the core using a toroidal winding to
produce a ring flux similar to the conventional
method, but only to 4% of its normal working
4.3.8 Electromagnetic Core Imperfection level of excitation.
Detection
The stator core of a typical generator is built from 4.3.9 Electrical Tests on Generator Rotor
thousands of thin steel sheets (laminations) and
the winding is formed from electrically insulated The following tests are carried out on the
copper conductor bars which are embedded in generator rotor.
slots between teeth around the bore. The core is
held together by steel building bars, usually on (i) Insulation Resistance/Polarisation Index
its outside. (IR/PI): IR/PI test between slip rings
and the body is carried out to check the
The laminations are coated with a thin layer of integrity of the pole sleeve insulation of the
electrical insulation to prevent eddy currents rotor.
being induced between them by rotating magnetic
flux produced by the spinning rotor. (ii) DC Resistance test: The DC resistance of
the field winding is measured to identify
Defects in the inter-laminar insulation cause defective clamps/brazings of conductors.
fault currents to flow locally in the core. These (iii) Field Winding Impedance Measurements:
currents can produce dangerous local overheating The impedance of the field winding is
or hot spots in the damaged areas and the damage determined by measuring the AC voltage drop
to the core may become progressively worse. across winding on passing a known current.
In extreme cases, sufficient heat is generated to The inter turn short in any pole is reflected
melt small parts of the core and even modest by the change in the measured impedance.
rises in core temperature adjacent to the winding Normally, up to 5 % variation in the
can result in the premature failure of the winding measured impedance value is acceptable.
insulation. Clearly, hot spots should be detected Any variation over 5 % reflects shorted turns
and repaired during routine machine overhauls. in the pole winding.
Early test methods known as full ring flux testing
required the core to be executed to near its (iv) Recurring Surge Oscillogram Test: Rotor
normal working level for a period of time. The winding insulation inter-turn and earth
temperature rise of hot spots due to the core fault faults are detected using the recurrent surge
was then measured. oscillograph (RSO) method. The test is very
sensitive and will give an indication of the
An alternative method of detection of faults in core early stages of an interturn fault.
inter lamination insulation by electromagnetic
means was developed by the Central Electrical A DC voltage step is applied to each end of the
Research Laboratory of the UK C. E. G. B. (now rotor winding in turn. Each reflected wave at the
National Power plc). Instead of the previous full input end of the winding is monitored and the
flux working level, the ELCID method uses only two waveforms are superimposed automatically
a small fraction of rated excitation to generate and monitored on a single channel oscilloscope.
fault currents within the core body which are As the half windings in a rotor are identical, the
sensed by a pick-up coil. This avoids the testing two waveforms monitored at each end of the
problems usually found with high excitation, yet rotor will also be identical for a healthy winding.
gives an accurate indication of damaged area A winding with a fault will cause different
along tooth tips and walls, as well as possible voltages to be monitored at the two ends for a
sub-surface damage. winding with an inter-turn fault.
The Journal of CPRI, Vol. 8, No. 3, September 2012 177
TABLE 1
IR AND PI VALUES
DC leakage current Conductor Resistance
Phase IR (GΩ) PI
at 9 KvDC (μA) (mΩ) at 30° C
R 4.19 4.32 1.2 1.901
Y 4.41 4.26 1.0 1.902
B 3.71 3.99 1.2 1.902
y No white/brown powder formation on the are satisfactory. The leakage currents increase
surface of the stator. gradually with test voltage and no sudden
change in the slope has been observed.
y No symptoms of arcing or surface
These results indicate that there are no weak
discharges.
points in the overhang portions of the stator
y No symptoms of mechanical damage to the winding.
stator winding.
y No migration of wedges. The stator conductor resistance values are
comparable with the factory value with due
y No loose lashes/space bars were found. consideration to the temperature at the time
of measurement. From these results, it can be
6.0 ELECTRICAL TESTS observed that the stator conductor joints are in
good condition.
The results of IR, PI, DC leakage current
measurements and conductor resistance The results of tan δ and capacitance tests obtained
obtained on the Stator winding are presented on the three phases of the stator winding are
in Table 1. presented in Table 2.
Referring to Table 1, IR and PI values are quite The tan δ test parameters lie in the normal
high. Since PI is regarded as the index of dryness acceptable range for an in-service machine.
of the insulation, it can be inferred that the stator From these results, it can be inferred that
winding is clean and dry. For a clean and dry the dielectric losses in the stator winding
insulation, the PI shall be higher 2.0. insulation are quite low. Tan δ tip up (ΔT),
and capacitance tip up (ΔC), indicate that the
The variations in DC leakage current against void content in the stator insulation system is
DC voltage obtained on the stator winding low.
TABLE 4
IR, PI, CONDUCTOR RESISTANCE, AC FIELD
6.1 Stator Core IMPEDANCE AND SURGE COMPARISON
TESTS
The ELCID test has been carried out on the IR (GΩ) 2.74
Stator winding to detect imperfections/faults PI 3.5
and hotspots in the Stator core. The ELCID test Conductor Measured value
has been carried out by passing an AC current 214.5
resistance (mΩ) at 28ºC
(Excitation Current) of 6.35 A. through 07 turns
Field imped-
of loop around the Stator frame. The traces show Measured value 6.99
ance (Ω)
the variations in the fault current as a function of
Surge test No abnormality
the core length. The core laminations are coated
with a thin layer of electrical insulation to prevent
eddy currents being induced between them by
the rotating magnetic field. Any defect in the The Conductor Resistance of the Rotor winding
inter-laminar insulation causes fault currents is in the permissible range with due consideration
to flow locally between the laminations. These to the temperature of the conductor. The Surge
currents can produce dangerous local overheating Comparison test indicate that there are no
at the damaged areas and the damage to the core inter-turn faults in the Rotor winding.
may become progressively worse. In extreme
cases, sufficient heat is generated to melt small On the basis of the data, it can be inferred that
parts of the core which will ultimately lead to the insulation conditions of the Stator winding,
failure of the winding insulation. Generally, for Rotor winding and Stator core are healthy.
a good inservice machine, the currents induced However, it is recommended to repeat the
in the core are flat to within 50 mA. The condition monitoring tests after three years with
maximum limiting value of this fault current is a view to monitor state and condition of the
100 mA. Results of the ELCID test indicate that Generator.
there are no imperfections or hot spots in the
stator core.
6.3 13.8 kV, 120MW TURBO GENERATOR
6.2 Rotor Winding
Class B insulation, Year of commission: 1974
6.2.1 Visual inspection and examination
6.3.1 Stator winding
Low contamination on the rotor winding,
No deformation of the retaining rings.
Visual inspection and examination of the stator
winding revealed damage to two bars (bar nos.
6.2.2 Electrical Tests 48 and 49) on their overhang portions at the
exciter end. The damage has caused a big hole
Results of the IR, PI, Conductor Resistance, in the insulation on each stator bar.
The Journal of CPRI, Vol. 8, No. 3, September 2012 181
TABLE 5
RESULTS OF THE IR, PI, DC LEAKAGE CURRENT AND CONDUCTOR RESISTANCE
MEASUREMENT
IR-GΩ DC Leakage current Conductor Resistance Previous value
Phase PI
60 sec. at 1 (kVDC) (mΩ) at 32°C (mΩ)
R 1.16 4.16 5.2 1.290
1.296
Y 1.07 4.08 6.0 1.288
Results of the IR, PI, DC leakage current and insulation up to the phase voltage (7.97 kV)
conductor resistance measurement obtained on of the machine. These results supplement the
the three stator winding sections of the generator tan δ test results indicating low void content in
are presented in Table 5. the stator winding insulation.
TABLE 8
RESULTS OF THE IR/PI, WINDING IMPEDANCE, CONDUCTOR RESISTANCE AND SURGE
COMPARISON TESTS
PD magnitude
Phase IR(GΩ) PI Id (μA) Tan δ (%) ∆T(%) ∆C(%) Vi (kV)
(pC)
R 0.166 1.23 28.0 4.71 0.28 4.51 2.20 46,000
Y 1.06 1.64 4.2 4.37 0.05 1.60 3.40 3680
B 1.80 2.41 3.4 5.26 0.086 4.16 3.75 5980
The Journal of CPRI, Vol. 8, No. 3, September 2012 183
insulation. While R-phase section of the winding content in the stator insulation systems of the
is exhibiting higher level of deterioration of the motors is also low. There are no inter-turn faults
insulation, the Y-phase section appears to be in in the stator windings as revealed by the Surge
good condition. The PD test also revealed higher Comparison test.
level of deterioration in the R-phase section. As
the PD was asymmetric and their magnitudes The insulation conditions of the stator windings
were extremely high, it was suspected that the of both the motors are healthy. However, it is
motor was suffering from slot and end winding recommended to repeat the measurements after
discharges, especially in the R-phase section 3 years.
of the winding. On the basis of these results,
it was recommended to withdraw the machine
and thoroughly inspect the stator winding for 6.6 6.6 kV, 570 kW, BALL MILL MOTORS
symptoms of slot/end winding discharges, loose (2 No’s)
or damaged wedges, erosion of stress grading
coating and presence of heavy contamination. Motor A:Cls F insulation; Date of commission:
On physical inspection, it was found that the 1992; Motor B: Class B insulation.
stator winding was heavily contaminated, 2 nos.
of stator wedges were missing and most of the The results of PI, DC leakage current (Id), and
wedges in the slots were loose and deteriorated. tan δ tests obtained on the stator windings of the
Subsequently, the user company initiated action motors are presented in Table 10.
for re-wedging of the stator winding.
The ball motor B is exhibiting higher tan δ values
6.5 6.6 kV, 3500 kW, BFP Motors (2 no.s) due to the presence of heavy contamination.
However, the tan δ tip up and capacitance tip up
Motor A: Class B, year: 1975; Motor B: Class values are quite low, indicating low void content
B; year 1975. in the stator winding insulation. The stator
conductor resistance values are comparable and
in the normal range.
The results of PI, DC leakage current (Id) and
tan δ tests obtained on the stator winding of the
motor are presented in Table 9. Insulation Condition of the stator windings
of both the motors is healthy. However, it is
The IR and PI values of all the motors are in the recommended to thoroughly clean and dry the
normal acceptable range. The tan δ tests indicate stator windings and varnish be applied to prevent
low dielectric losses in the insulation. The void any possible degradation of the structural
TABLE 9
RESULTS OF PI, DC LEAKAGE CURRENT (ID) AND TAN δ TESTS
Motor A Motor B
Test parameters
All the phases are shorted All the phases are shorted
IR (GΩ) 4.14 5.69
PI 3.48 3.50
Id(μA) 0.8 0.8
Tan δ(%) 2.245 2.250
ΔT (%) 0.093 0.086
ΔC (%) 0.211 0.160
Surge test Stable smooth and identical pattern Stable smooth and identical pattern
184 The Journal of CPRI, Vol. 8, No. 3, September 2012
integrity of the motors due to heavy deposits of of the insulation, it is preferable to monitor as
contamination. many relevant parameters as possible to enable
to arrive at a realistic assessment. The data of the
diagnostic tests obtained over the years enable to
initiate appropriate corrective measures to avoid
TABLE 10 forced outages and hence to enhance remaining
RESULTS OF PI, DC LEAKAGE CURRENT (ID) useful service life of the machines.
AND TAN δ TESTS
Test parameter Motor A Motor B
REFERENCES
IR (MΩ) 1150 1810
[1] “Dielectric diagnosis of electrical equipment
PI 1.62 3.18 for AC applications and its effects on
Id at 5 kV DC (μA) 3.62 2.45 insulation coordination”. State-of-the-art
report presented by the working Group
Tan δ (%) 4.747 9.608
33/15.08, CIGRE 1990 Session.
ΔT (%) 0.01 0.07
[2] Stone G C and Sedding H G. “The ability
ΔC (%) 0.141 0.26 of diagnostic tests to estimate the remaining
759.0 1102 life of stator insulation”, IEEE, Trans. On
Stator Conductor Energy Conversion, Vol. 3, 1988.
758.8 1105
resistance (mΩ)
758.4 1105
[3] Dakin T W. Trans. on PAS AIEE, Vol. 78,
October 1959.
[4] Mallikarjunappa K, Moorching S N and
7.0 CONCLUSIONS Vishwanath C. “An overview on the
estimation of life of high voltage machine
insulation”, CBI&P 59th R&D Session,
Progressive deterioration of the stator insulation
Calcutta, February 1994.
system of inservice HV machines can be
assessed by monitoring certain non-destructive [5] Simmons J S. “Diagnostic testing of HV
test parameters. As each test parameter can be machine insulation”, IEE Proc.B, Vol. 127,
related to certain information regarding condition No. 3, 1980.