DIAGNOSTIC TESTING AND

CONDITION ASSESSMENT OF
LARGE GENERATORS

Introduction
This presentation briefly describes some of the failure mechanisms
on large generators, and then describes some of the testing we
perform, in our Service Facilities or on-site during an outage, to
more accurately assess the condition of a machine.

FAILURE MECHANISMS
40% of failures on generators occur due to problems with the windings.
The following is a list of possible causes of winding failure and stator
winding failure mechanisms.
Inadequate bonding of the winding insulation. Results in voids in the
groundwall insulation, either at the copper surface or within the
groundwall insulation itself. Depending on the size of the void, the
voltage of the conductor and the pressure of the cooling gas, a spark or
partial discharge may occur within the voids, attacking the organic resin
holding the insulation together. This results in the voids growing, to the
point where the insulation fails and an earth fault occurs.

Introduction
Slot discharges. If the surface corona or semiconductive coating on the
coils in the slot section is incorrectly applied or eroded over time due to
movement of the coils, discharges will take place between the coil and
the core. These too will attack the organic resin, and eventually result in
the failure of the winding insulation.
Stress Grading failure. This results in a sharp voltage gradient over the
surface of the coil in the endwinding, and consequently corona
discharges in air-cooled machines. These are prevalent at the interface
between the stress grading and the slot semiconductive coating and will
start breaking down the interface area. Eventually they will attack the
slot semiconductive coating until it is eroded back into the slot section,
resulting in slot discharges and eventual winding insulation failure. Both
the above two failure mechanisms are slow, but irreversible once
started.

Introduction
Loose windings. If the windings can move in the slots (due to loose wedges,
loose packing, ingress of oil, shrinking of resin, etc.), the semiconductive coating
can be eroded against the core, resulting in sparking as described above. In
addition, the groundwall insulation can then be mechanically eroded, and the
combination of the electrical and mechanical erosion mechanisms can result in
failure of the winding insulation in a relatively short period (two years or even
less).
Inadequate spacing. Inadequate spacing between the coils in the overhang can
result in a high voltage stress between bars of different phases, and
consequently corona discharges. These will gradually erode the groundwall
insulation until failure occurs, although the failure mechanism is a slow one.
Endwinding failure. Endwindings of high voltage machines are subject to
strong twice frequency (i.e. 100Hz in a 50Hz machine) mechanical (from the
core/rotor magnetic interaction) and electro-magnetic forces, which can cause
vibrations in poorly blocked or inadequately designed machines. These
vibrations can abrade the windings against the blocking or support structure,
eventually causing the copper to be exposed and a possible phase-to-phase or
earth fault. Also, the vibrations can cause the coils to loosen in the slots.

Introduction
Surges. These result in voltage spikes, which stress the interturn
insulation in multi-turn coils. Eventually the interturn insulation may fail,
resulting in very high level circulating currents flowing in the loop caused
by the shorted turn and failure of the groundwall insulation. The main
problem here is that it is almost impossible to predict failure in the
majority of cases, as the groundwall insulation will appear in good
condition until failure occurs the failure is taking place from within the
winding.
Thermal Deterioration. When an insulation system is exposed to
overheating, the bonding material loses its mechanical strength, and the
insulation layers delaminate. This eventually results in voids and long
term failure, depending on the degree of overheating. With asphaltic
systems, the black asphalt material sometimes oozes out of the winding.

Introduction
Load Cycling. As copper expands due to increase in temperature under
loading of a machine, the insulation bonded to the copper is placed
under stress, as it has a different coeffecient of expansion. This can lead
to delamination, and can also cause damage to interturn and strand
insulation. In older asphaltic systems, a phenomenon called girth
cracking can occur, where the coils exit the core. This is because when
the copper expands, it drags the insulation out of the slot, and when the
copper cools and contracts again, the insulation doesnt contract back
into the slot but collects just outside the core, eventually leading to
cracking and sometimes splitting of the insulation.
Endwinding Contamination. Any conductive contamination from
moisture or oil mixed with dust or dirt can result in electrical tracking and
partial discharges on the surface of the insulation. Phase-to-phase and
phase-to-ground deterioration may occur, and eventual failure of the
groundwall insulation may result.

Introduction
In the core, the following are the main failure mechanisms which occur.
Overheating or hot spots. Failure of the interlaminar insulation can
result in circulating currents in the core and localised overheating. This
then can degrade more of the insulation, resulting in bigger cicrculating
currents and more heat. Eventually, the hot spot will cause the winding
insulation near the spot to overheat and fail, and an earth fault will occur.
Loose Laminations. Particularly near the ends of the core, the press
plates and press fingers can relax and result in the laminations
becoming loose and vibrating in service. These can then abrade the
winding insulation, or even fatigue, crack and break off resulting in iron
particle contamination of the machine.

Introduction
In the rotor, the main concern is mechanical stresses on the shaft, body and
Coil Retaining Rings resulting in possible cracking, problems on the bearings
due to contaminated or insufficient oil, out of tolerance clearances or electrical
pitting on the surface. The two main electrical failure mechanisms are as follows.
Interturn Faults. Failure of the interturn insulation on one or more of the poles
results in a thermal and possibly magnetic unbalance of the rotor and a possible
increase in vibrations. In addition, the fault can deteriorate to the point where an
earth fault occurs. Interturn faults result in a reduction in the net ampere-turns of
the rotor, so more excitation current than prior to the fault may be necessary to
reach a certain loading of the machine.
Earth Fault. An earth fault is a failure of the insulation between the rotor winding
and the body of the rotor, either in the slot section or the overhang (to the Coil
Retaining Ring). A single earth fault may not necessarily negatively impact the
operation of the rotor. However, if a double earth fault occurs, there is a path for
circular currents to flow, which will result in catastrophic damage to the rotor and
potentially cause it to disintegrate at speed. The main reasons for failure of
either interturn or groundwall insulation on generator rotors are mechanical
abrasion of the insulation, or overheating.

Introduction

There are a variety of tests,

Some general to all machines, some specific to certain types
(e.g. pneumatic/vacuum, water flow, etc only to stators with
direct water cooled windings)
We will only be looking at the common tests - IR/PI,HV , RSO,
volt drop, PD, tan delta, ELCID and wedge tightness
Mostly harmless, however, HV testing may cause damage
Condition assessment of generators is difficult, therefore it is best
to do as many tests as possible when the opportunity arises

Insulation Assessment Testing

IR / PI

AC / DC / VLF
PD/Tan Delta

Insulation Resistance (IR) Test

Ratio of DC voltage applied

between winding and earth for 1
minute and the resultant current
(R=V/I).
Applicable to stators and to rotors
( or field and armature windings)

Insulation Resistance (IR) Test - Stators

From IEEE 43-2013, the applied DC Voltage for the
majority of generators is 5000 V

Winding Rated Voltage

<1000V
1000 2500V
2501 5000V
5001 12000V
>12000V

Insulation Resistance Test Voltage

500V
500 1000V
1000 2500V
2500 5000V
5000 10000V

Insulation Resistance (IR) Test -Stators

The acceptance criteria for stators is usually 100 M

Description
Most windings made before 1970, all rotor
windings and others not described below
For most DC armature and AC stator
windings after 1970 (form-wound coils)
Most machines with random-wound coils
and form-wound coils rated below 1kV

Minimum IR Value
(kV + 1) M
100 M
5 M

Insulation Resistance (IR) Test

Results very sensitive to humidity and

temperature
- they are usually corrected to 40 C

Difficult to trend results

Generally a pass / fail test,
But very simple and useful

Creepage Paths in the rotor

Rotors rely on creepage paths for
their IR value, so the acceptance
limits are much lower

As can be seen in this photo, the

creepage path is the gap between
the copper winding and the body
If it gets contaminated, by moisture
or dirt, the IR value drops
considerably

Insulation Resistance (IR) Test - Rotors

The test voltage for rotors is typically 500 VDC

Winding Rated Voltage

<1000V
1000 2500V
2501 5000V
5001 12000V
>12000V

Insulation Resistance Test Voltage

500V
500 1000V
1000 2500V
2500 5000V
5000 10000V

Insulation Resistance (IR) Test - Rotors

Acceptance criteria of (kV+1)M e.g. for a 400V rotor,
acceptance is 1.4M

Description
Most windings made before 1970, all rotor
windings and others not described below
For most DC armature and AC stator
windings after 1970 (form-wound coils)
Most machines with random-wound coils
and form-wound coils rated below 1kV

Minimum IR Value
(kV + 1) M
100 M
5 M

Insulation Resistance (IR) Test - Rotors

So for most rotors, were looking at an IEEE acceptable IR value
of 1.2 to 1.6 M

However, we use 100 M as an alert value from experience

most test results are in the G range, so we want to see whats
happening.

Some OEMs use 50 M for rotor acceptance

Some utilities minimum of 20 M

Polarisation Index (PI) Test

With a constant DC voltage applied with a Megger, PI is the ratio of

the 10 minute reading to the 1 minute reading
Provides a criteria to ascertain the dryness and cleanliness of the
winding
Ratio, hence relatively insensitive to winding temperature
With wet/dirty windings, the surface leakage current will dominate, so
the total current will not vary much with time

Polarization Index (PI) Test

IEEE: PI > 2 means the winding is clean and dry
Older machines, 1.5 is acceptable
General Guideline:

<1: dangerous
1-2: questionable
2-4: good
>4: excellent

BUT if the insulation resistance is very high (5000 M or more), the

PI can be considered meaningless:
- some modern insulation systems polarize quickly (within the 1
minute limit for IR);
- also very low leakage current makes the readings unreliable

High Voltage AC Test

Final proving test - pass / fail
Often called a pressure test - applies pressure to the winding insulation
Apply 2Un + 1 for new winding, between 1 - 1.5Un (or less phase
voltage in some cases) for old winding
Requires large power supply, up to 100kVA or greater, to charge the
capacitance of the winding

Ensure that CTs, VTs, thermocouples and RTDs are disconnected or

shorted out
Safety precautions are paramount

High Voltage AC Test Set

High Voltage AC Test

Apply test voltage for 1 minute - the only aging that will occur is if
there is excessive PD, but this will not take much off the life of the
machine. DC has no aging effects.
Leakage current must be monitored and trended.
The winding must be properly grounded (through a resistor) after the
test, and an IR repeated.

Alternate tests - VLF (0.1Hz) or DC (recommended value 1.7 x AC

value). DC places more stress on the overhangs.

High Voltage DC Test

Major advantage small test set!
Plotting leakage current against voltage
can give some useful information.
Stresses slot section almost the same as
AC.

Doesnt stress the overhang the same

stress grading not designed for DC
(higher stress)!
Cant do PD or Tan Delta testing.

High Voltage DC Test

High Voltage VLF Test

Major advantage small test set, but produces a
sinusoidal wave form
Less stress on the overhang in stator testing than DC
Stresses slot section the same as AC.
Can see on the waveform where any breakdown occurs
Can do PD and Tan Delta testing with specialized
equipment
Often used mainly for rotors

The decisions that need to be made when HV testing:

Test or not?
AC / DC / VLF?
What voltage?

A properly conducted high voltage test:

1) Will not damage a winding that is not already severely deteriorated.
2) Will give good assurance against service failure for a stator winding that
is not experiencing aggressive local or general deterioration.

Partial Discharges

Partial discharges are tiny sparks (a flow of electrons and ions) occurring
within or around a high voltage insulation system.
Voltage buildup occurs across gaps or voids in the insulation system as the
voltage waveform builds up, and depending on the gap length or cavity
diameter, gas pressure and nature of the surfaces involved a breakdown
will occur resulting in the spark.
Partial discharges are also produced in the overhang in air-cooled
machines in the form of corona (discharges to air adjacent to high voltage
conductors).

Partial Discharges

PD has a very fast rise time and is generated at high frequencies.

Bus couplers are 80pF capacitors that detect the high frequency signals
and block signals of lower frequencies (specifically power frequency!).
Lower frequencies (<100 MHz) see deeper into the winding, but can be
affected by circuit parameters.
Higher frequencies (>100 MHz) attenuate more rapidly, but are more
repeatable.
PD is then grouped and plotted according to number of pulses,
magnitude and phase angle.
The higher the magnitude the bigger the void that the spark is jumping
across
The more pulses, the more voids

Linear Plot of PD

400

300
200

100
0
-100
-200
-300
-400
0

45

90

135

180

[ ]

Bipolar Machine PD

225

270

315

360

Magnitude Plot

Partial Discharge Testing

Interpretation
No standardization
Readings > 5000pC indicative of winding deterioration
IEEE for new stator windings max PD = 2000 pC
Only reliable method of interpretation (opinion!)
TRENDING!

TVA Test (RF probe or Peak Pulse Meter)

Technical background
PD radiates radio frequency energy
Uses a modified AM radio loop antenna
(tuned coil) at the end of a well insulated rod
Test Setup
Rotor removed safety paramount!
Procedure
Operator stands inside bore of machine,
voltage is raised, and each slot is scanned
Discharges induce a current in the coil
proportional to the magnitude of the
discharge, which is measured on a peak
pulse meter

TVA Test (RF probe or Peak Pulse Meter)

Interpretation
On a new class F epoxy mica based winding readings should be less than 5
mA
For older machines, the limits depend on the insulation system
Included in IEEE standard 1434 (PD testing)

Tan Delta (Power Factor) Testing

Ic

Perfect Insulation

Ir

Real Insulation

Stator insulation system is a capacitor

In a perfect capacitor, no losses, I and V exactly 90 out of phase, ie PF = 0
In the real world, voids and losses in the resin affect the power factor, causing
resistive (real) losses in phase with voltage
Resultant current vector out of phase with perfect current vector by angle losses proportional to tan
Measured with a Schering / Transformer Ratio bridge

Tan Delta Testing

Practically the Cos (power factor) = tan at very small angles
In a very good capacitor, losses would increase linearly with voltage

In a poor insulation system (poor capacitor) as voids start to discharge with

increased voltage the losses increase, so no longer linear relationship
The more voids, the greater the losses, hence the larger the angle

Good insulation power factors (tan deltas) of 0.2 to 1.5 %

Deteriorating systems PF/tan delta 5 10 %
Tip up (deviation from linearity) a good indicator as well

Tan Delta Testing

Tan Delta Testing

Rotor Tests

RSO
Recurrent Surge Generator/Oscilloscope
(RSG/RSO) Testing

Used to check for inter-turn insulation faults in

rotor windings
Fast rise time, low voltage pulses / surges (5 10 V) injected into both ends of the winding
Inter-turn faults, earth faults or breaks in the
winding produce discontinuities in the surge
impedance.
These cause part of the pulse to reflect back to
the injection point, changing the voltage
waveform

RSO
The waveforms from both ends of the winding viewed
simultaneously
on an
oscilloscope
should theoretically be(RSG/RSO)
identical
Recurrent
Surge
Generator/Oscillograph
if no problems
Testingare present
A problem shows as a difference in the two
traces
Slip-rings

Ground wire

RSO
Oscilloscope

220 Vac

Good trace

Trace showing fault

Volt Drop Test

DC or AC Voltage applied (DC more stable, but large current source
required. AC more practical)
Typical voltage 80 110 V (to get at least 2 5 V drop across each coil)

Can measure across poles, coils or turns 2% pole balance, 10% coils
Looking for shorted turns - less current will flow and voltage will be less
(little or no change turn-to-turn)

STATOR
TESTS

Stator Wedge Testing - Two Methods

Tap Test:

Simple done by tapping each wedge with a ball

peen hammer or spanner and listening (and feeling) for

its response.
No recorded data manual check sheet filled out.

High level of experience required.

Highly Subjective two different operators can get
completely different results!

Digital Wedge Tightness Detector

Digital Testing :

Hand held probe with magnetic adjustable side units.

Hammer strikes the wedge approximately 10 times

per second.

Typical scan rate of approximately 3 seconds per wedge

Accelerometer detects frequency response of wedge

after it is struck

WTD-501 Wedge Tightness Detector

Probe

System Configuration

Signal Processing Unit

(SPU)

Probe

Operation

Operator chooses
first slot/wedge.
Initiates test with
switch on probe.
Results displayed on
PC and stored.
Large PC display
allows remote
reading.

WTD Results

Results may be displayed in colour coding up to 5 categories.

Alternate calibrations may be used after test to re-qualify data.
Multiple tests may be compared to establish trends.
Results are repeatable and not subjective two different operators
will get the same results

ELCID Testing Stator Cores

ELectromagnetic Core Imperfection Detection: a technique used to
assess the integrity of inter-laminar insulation for stator cores of large
generators and motors
Only requires a low voltage supply to excite the core to 4% of rated flux
Use a Chattock Coil to look for axial currents (ie in quadrature to the
main currents)
These represent fault currents. A reference coil tells us the direction
and magnitude of the main currents, and any current induced in the
Chattock coil 90out of phase to this is plotted on the computer against
the distance from the core end.
A relatively flat line indicates a healthy core - acceptance usually
100mA or less.

Fault Current

Positioning the Chattock

Test Trolley
Chattock

Adjustments for width (4)

Clamp screws for Chattock (2)

Screws for cable clamp plate (2)

Adjustments for bore radius (2)

Clamp nuts for side guidance rails (4)

Typical Turbogenerator EL CID Excitation System

Typical Hydrogenerator EL CID Excitation System

Electromagnetic Core Imperfection

Detection System Configuration

Digital ELCID - Model 601

Digital ELCID Evolution (Latest Model)

Scanning
End of slot N

Start of Slot N+1

Sample Fault Indication (Turbo)

ELCID Signal and Thermal Response to Faults

Thermal
Imaging
Camera

EL CID
Trace

Trending

When things are changing inside a machine, without an obvious reason for
the change such as increased load or a very hot day, then there is cause
for concern.

And the only way to know if something is changing is to trend machine

operating or test data.

Conclusions
Generators big, but relatively sensitive and fragile
They contain and direct huge amounts of power, often sufficient to light up a town.
When that power gets out of control within the generator, it can do severe
damage in a very short space of time.
Hence is it vital to know in advance when something is starting to go wrong regular inspection and testing of the machine, routine maintenance, and good
trending of all available information is vital.
Test and inspection on a generator is relatively inexpensive and short duration
However, a full rewind or core restack, is extremely expensive and can take from
a few months to over a year to complete, depending on the size of the machine
and the availability of spares.
A plant owner can save the cost of testing over the years by not doing the
necessary maintenance and information gathering on his machine. But when the
machine eventually fails, the cost of a major repair and outage will far outweigh
the saved cost of testing.

Field Case: Repair of a GT Generator Rotor

CUSTOMER:
REMACO
LOCATION :
JOHOR, MALAYSIA
JOB NO.
:
P.1621 Aug 2014
Work Scope:
Rewind coil No.1 pole A
and Pole B, replace broken and damaged leads.

PASIR GUDANG GT 6A - ROTOR

RATING: 143400KVA
GE FRAME 9 GAS TURBINE GENERATOR

Field Case: Repair of a GT Generator Rotor

The slot liners have some fretting damage caused by coil
movement. It was also noticed that there is considerable radial
clearance in the slot. This is apparent by looking at the
clearance between the coils and the bottom of the slot. The
calculated clearance is 4.5 mm.

Field Case: Repair of a GT Generator Rotor

The radial connection pins show evidence of movement and
wear between the pin insulation and the retaining rings in the
shaft.

The leads also show wear marks caused by looseness in the

bores and in the sealing plugs..

Field Case: Repair of a GT Generator Rotor

There was red iron oxide dust on the bottom of the radial
connection studs, in the stud holes and on the connection
leads. The up-shaft leads were removed and found to have
considerable clearance in the shaft.
Suspect water leakage from the stator.
HV Testing would have detected the problem.

The leads were re-installed with additional insulation to achieve

a tight fit in the shaft bores. The new radial connections studs
were shaped to add flexibility to the field leads.

Field Case: Repair of a GT Generator Rotor

A new flexible jumper connection was shaped and brazed to
coils 1 and 2 of pole A.

HV test was done to confirm the validity of the new insulation.

The hydrogen seal plugs were pressure tested to 4 bar for 4

hours.

Field Case: Repair of a GT Generator Rotor

Installing Collector End (Excitation End) Retaining Ring and
Heaters.

Set up rotor for installation of turbine end retaining ring.

Installing the turbine end retaining ring.

Final electrical tests include rotor winding tests, HV tests IR, PI,
RSO.

END