Professional Documents
Culture Documents
GENERATOR DESIGN
AND CONSTRUCTION
The focus of this chapter is on the design and construction of the generator and its
major individual components. Although not a design book, this chapter will go into
enough detail on how the components are designed and fabricated, to assist the
reader in maintaining them. In addition, issues that significantly influence the
design of the various generator components are discussed. If the reader wishes
to learn more in depth about a specific component, it is recommended that the orig-
inal equipment manufacturer (OEM) be consulted.
The class of generators under consideration is water-driven generators com-
monly called hydro or waterwheel generators. They range from relatively small
machines of a few megawatts (MW) to very large generators with ratings up to
944.5 MW. These generators typically have speed ratings of 72–900 RPM and
are installed all over the world.
The basic function of the generator is to convert mechanical power, delivered
from the shaft of the turbine, into electrical power. There are many different types
of turbines such as Francis, Kaplan, Pelton, and Deriaz which will impact the gen-
erator design. The discussion of each type of turbine is out of the scope of this
book, and the reader is referred to Ref. [1] for a comprehensive discussion on tur-
bines. The mechanical energy from the turbine is converted by means of a rotating
magnetic field produced by direct current in the copper winding of the rotor or
field, into three-phase alternating currents and voltages in the copper winding of
the stator (armature). The stator winding is connected to terminals, which are in
turn connected to the power system for delivery of the output power to the system.
Generators are made up of two basic members, the stator and the rotor, but
the stator and rotor are themselves constructed from numerous parts. Rotors are the
rotating member of the two, and they undergo severe dynamic mechanical loads as
well as the magnetic and thermal loads. The stator is stationary, as the term
35
36 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
The core is built up from these thin laminations to limit eddy current losses in
the core. Some manufacturers bond 10 or 20 laminations together at one time and
install as a group to increase core stiffness. Each lamination is insulated from other
laminations; laminations when installed may be grounded at the back of the core via
the keybar, the insulation ensures that circulating currents between laminations as a
result of the induced magnetic flux cannot occur. If the insulation between the lami-
nations is damaged, and the laminations are grounded at the back of the core, loca-
lized current flow and overheating of the laminations would result. Figure 2.1-1
shows an exaggerated yoke for a hydro machine for the purposes of illustrating
the circulating current concept. Normally, the yoke on a hydro machine versus
the slot depth would be more in line with Figure 2.1-4.
The stator core is designed to carry the electromagnetic fluxes and must be
capable of handling the magnetic flux density in the stator teeth and core back areas
which can on average be between 1.7 and 1.45 T, respectively, depending on the
design and manufacturer. These flux levels can vary from these stated values, but
they are in the ballpark of what can be expected.
Figure 2.1-2 demonstrates, in a simplified manner, how breaches in the inter-
laminar insulation result in larger than normal eddy current and losses. Larger than
normal means currents and losses substantially higher than those found when the
machine operates within its designed parameters. Figure 2.1-3 carries the analysis
in Figure 2.1-2 further. It can be seen therein that with 100 or so divisions, the eddy
current losses are reduced to about 2% of the original losses. Further divisions
make this type of loss very small when compared to that of an equivalent single
Insulation
Conductor damage
bars
Groundwall
insulation
Flux
L L
I L/4
Steel
L lamination
material
- Induced volts ∝ L2
- Loop resistance ∝ 4L
- Current ∝ L/4
- Losses ∝ L3/4
This means that by dividing the original metal sheet into four insulated
laminations, the total eddy-current loss was reduced 2.5 times
Figure 2.1-2 Inter-laminar insulation reduces eddy current losses in the steel.
body core. From the content of this paragraph, it becomes obvious how by dam-
aging the insulation between the laminations in a few spots, the eddy current losses
can significantly increase, with consequential temperature rises and additional
damage to the insulation, further increasing the amount of short circuited core. This
has the potential to becoming a runaway situation, leading to melted core material
and a catastrophic failure. The issue of short-circuited laminations and consequen-
tial core damage due to increased eddy current loss is broached in other places in
this book when discussing foreign metallic objects left inadvertently on the core
2.1 STATOR CORE 39
120
100
Loss in % of unlaminated block
80
60
40
20
0
1 10 100 1000
Number of laminations
Figure 2.1-3 Laminated-core eddy-current loss as percentage of full-block loss. This
graph shows the cumulative effect of decreasing eddy current losses by laminating the core steel.
laminations in the bore, or metal parts (e.g. bolts and washers) becoming loose dur-
ing operation and landing on the core in such a way that laminations are shorted.
Stator yolk
sections that fit into the slots are rectangular in shaped and consequently the core
slots for them are also rectangular in shape. The slots for this type of winding have
grooves (one on each side of the slot) near the inner diameter of the lamination to
retain the nonmetallic wedges that hold the winding coils or bars tightly in the sta-
tor core slots. Magnetic wedges are not common but are usually used with small
airgaps. There are also machines out there, although rare that have a semi-closed
slot instead of an open rectangular slot. These machines have “hair pin or
U-shaped” windings pressed in one end of the machine and then joined at the other.
Early days of asphalt insulation had these types of windings, but modern insulation
is also used in the present day when replicating this type of winding. Also, this has
been done with larger machines where the coil cross section is more than one
square inch and connections are then installed at either end, see Figure 2.1-5.
The portion of the core behind the bottom of the winding slots is called the
yoke or core-back area and has lower flux density than the tooth area. For example
(these are average values), if the teeth are loaded at 1.5 T, then the core yolk area
would be loaded at 1.2 T for a typical hydro machine. The size of the yolk is due to
several factors such as the amount of heat dissipation required, reactance of the
machine, core stiffness, and whether it is a high or low flux machine.
Core Keybar
bolt Gusset
Piling
pin
Vent
lamination
Figure 2.1-6 Lamination stacking in the stator, also showing the conventional keybars
(rectangular) welded to gussets at the back of the core as well as the round core clamping
studs. Piling pins are used in the slots during construction to align laminations. Vent
laminations have a red colored protective coating applied from the factory, Figure 2.1-7.
Radial duct
Radial duct
curled to
direct air flow
into core from
the air gap
Figure 2.1-7 Heavy lamination segment with I-shaped blocks (radial duct) – red protective
coating applied.
42 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Section A-B
A B
48
1.5
Stator frame
Clamping strap
46.5
36
5.5
43
Figure 2.1-8 Shows double dovetail design that prevents core buckling. Source: Courtesy
of Voith.
(which are in the thousandths of an inch) with respect to the keybar placement are
not maintained, the core will not pile correctly and will not compress correctly.
Thus, ensure that the OEM is following their internal or industry-accepted standard
for keybar circularity, concentricity, axial trueness, keybar flatness, and keybar
stepping and core compressing (normally every 304–457 mm (12–18 ) of core
height). In fact, when placing laminations onto the keybars, the laminations will
not fit smoothly and may have to be forced into position if some of the previously
mentioned tolerances are not in check. One method of checking the keybar place-
ment is to take a few laminations, stack them together on a piece of plywood, screw
the yolk to the plywood so that the keybar interface portion is hanging out in the air,
and run the laminations up and down the keybar assembly all the way around the
machine. This should be able to be done with ease without any shaking or jiggling
of the lamination stack. It will be evident very quickly whether or not the keybars
are in the right position or not. When doing a core restack, it is highly advisable that
the keybar arrangement be verified against the original construction tolerances. If
the keybars are out of tolerance, resetting of the keybars is paramount.
The core can be stacked in sections onto the frame that are shipped to site (if
convenient for shipping) or continuously piled onto the frame which is done at site.
In either case, the core must be stacked as level as possible from front to back and as
well as circumferentially (minimal wave at the top of the core when measuring core
height). As a result of the lamination punching, the tooth assembly at the front of
the core will tend to pile higher than the back of the core due to distortion of the
2.1 STATOR CORE 43
Lamination
shim
Tangential
clearance
Radial
Core to frame clearance
clearance
(some designs)
Figure 2.1-9 Core-to-keybar mounting arrangement in the stator frame. Shim located at
outer core diameter to level the stacking.
lamination when punched. Shims are added to the back of the core to level the stack
from front to back (see Figure 2.1-9). The distortion problem is largely eliminated
with laser-cutting of the laminations; however, it is not economical to laser cut
large amounts.
Figure 2.1-6 shows the stator core bolts attached to the frame assembly and
not going through the core laminations and separate keybars for the core attach-
ment. Another variation of the core attachment and bolting arrangement is shown
in Figures 2.1-10–2.1-12. In this design, the core bolts go through the core itself in
order to provide the clamping pressure, and the keybars are again separate for core
attachment. The clamping is achieved by using a threaded core bolt at both ends
with an insulating sleeve. A bolt with a washer arrangement as shown in
Figure 2.1-10 can be used to secure the clamping plate. The insulating sleeve pre-
vents the core bolt from contacting the core causing circulating currents as shown
in Figure 2.1-10. The sleeve then passes through an insulated bushing that is
housed inside the vent lamination structure as shown in Figure 2.1-11. In this fig-
ure, for demonstration purposes, the bushing is placed inside the lamination hole
where the core bolt goes through. The core bolts may have a powder coating
applied as an additional barrier of protection. This bushing will prevent the core
bolt and sleeve from rattling inside the core during service. Again, for demonstra-
tion purposes, the bushing is removed from Figure 2.1-11, the red colored venti-
lation layer is put in place as shown in Figure 2.1-12 and then bushings would be
placed in every hole in this layer. This layer accommodates this insulating bushing
along with the I-shaped blocks used for airflow. The I-shaped blocks will be
44 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Keybar
welded to
Clamping frame
plate
Corebolt
insulating
sleeve
Insulating bushing
used for the
core through bolts
Figure 2.1-11 For demonstration purposes this figure shows an insulating bushing for core
through bolt. The bushing does not sit on this lamination but on top of the red vent
laminations as shown in Figure 2.1-12.
discussed shortly. Some OEMs do not paint the ventilation layer with insulating
varnish, thus this layer will look like the rest of the core in terms of color.
OEMs have different nut and washer arrangements that can include a com-
pression washer assembly as shown in Figure 2.1-13.
2.1 STATOR CORE 45
Insulating bushings
will be here
I-shaped
blocks
Figure 2.1-12 Showing the vent lamination with the hole to accommodate the bushing in
Figure 2.1-11.
Figure 2.1-14 Shows keybar and core clamping bolt as one piece with threaded end.
split the teeth into smaller sections in the radial direction (also called “slitting”), or
use step-back punchings, see Figures 2.1-15 and 2.1-16 respectively. These special
laminations are punched or laser cut at the factory along with the rest of the regular
laminations. This reduces the eddy current effect in the teeth and hence the losses
and core-end heating effect.
In order to cool the core, nonmagnetic radial ventilation ducts, “space blocks
or I-shaped spacers” are spot welded and secured to a thicker core lamination (typ-
ically called the “heavy” and is typically 0.711 mm (0.028 ) thick installed at stra-
tegic and set locations in the core stack to form a radial air path from the stator bore
Core bolt
Clamping finger nut and
not welded to washer assembly
clamping plate
Figure 2.1-15 76 MVA newly piled stator core showing the clamping finger assembly at
the top of the core, I-spacer vent assembly, first two packets with a slit tooth to reduce eddy
currents due to fringing fluxes.
2.1 STATOR CORE 47
Clamping
finger
to the back of core area, and vice versa, see Figure 2.1-7 for the heavy lamination
itself and Figure 2.1-6 for the locations along the stack height.
Once the stator core is properly stacked, it must be held together tightly under
pressure in the axial direction to ensure long-term performance and a fretting free
environment. In order to achieve this performance, tightness of the core should be
checked periodically and may need retightening after decades of operation. If the
core is properly stacked and clamped, and no buckling or waves develop, issues of
widespread or local area core looseness should not develop over time. For larger
hydro machines, the clamping mechanism typically consists of heavy steel plate
(length varies by OEM and design) that has heavy finger assemblies welded to
it. This assembly is then placed on top of the core clamping bolts and torque into
their final position. Another method is to have the heavy fingers as part of the core
assembly (welded to a heavy lamination just like the I-shaped assemblies are) and
then use the heavy plate on top to secure the entire assembly. Each OEM has their
own proprietary method of using these finger plate assemblies, some are more
effective than others over time. See Figures 2.1-15 and 2.1-16 for a couple of var-
iations of this assembly. The assembly in Figure 2.1-15 has the fingers welded to
the heavy lamination in the core assembly and the plate sitting on top separately,
whereas Figure 2.1-16 has the plate and fingers welded together as one piece sep-
arate from the laminations. There are various combinations on how fingers and
plates fit together at the top and bottom of the stator core. The general concept
in most designs is that the frame is used as a fulcrum and the fingers have lever
action onto the stator core. In a typical design, the outer edge of the clamping plate
48 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Adjustable
fulcrum
points
Figure 2.1-17 Shows fin-
ger plate installation with
adjustable fulcrum points.
sits on the frame and the bolts when tightened exert typically more axial force on
the teeth than they do on the back iron. Again, providing more force on the ends of
the teeth than the back of the core (see Figure 2.1-17).
For smaller hydro machines, the piling and clamping mechanisms may be the
same as for a large hydro or would be done as described in the following. In cores
with single-piece laminations, the core can either (i) be built into the stator frame
and clamped by tooth support fingers and steel rings (rings that are one piece equal
to the circumference of the stator core at each end and then welded to the core sup-
port bars), or (ii) built as a separate assembly and then fitted into a stator frame with
support bars that have been machined with a profile and dimensions that provide a
tight fit between the two assemblies. In either case vent ducts are installed during
the core building process and the core has to be placed under a high axial pressure
before the end support structures, consisting of end fingers and clamping rings, are
fixed in place. Segmented cores use either (i) through bolts that are installed
through holes punched in the core laminations, or (ii) keybars or dove tails at
the core back to which the laminations are assembled to. In either case, an even
axial force is applied over the surface of the laminations by the use of ring flanges
or plates.
The core design, and thereby the tightness, must be able to accommodate the
steady load machine torque as well as the transient torques experienced during fault
conditions. Such torques are transmitted through the laminations to the stator
frame, via the keybars mounted on the stator frame.
The gussets that are welded onto the keybar must be capable of taking these
tangential loads without failing, particularly during phase-to-phase faults and
faulty synchronization (see Figure 2.1-6). Ensure the welds on these gussets are
not cracked, if they are, they must be repaired.
Overtightening of the stator core can result in damage to the stator frame,
laminations, and ventilation duct “I-shaped blocks or beams” (“I,” “Square,” or
“U” shaped assemblies), resulting in a weakening or even cracking of the beams
and thin steel laminations, which in time will result in slackening of the core.
Adversely, not tightening the core sufficiently will result in lamination vibration,
producing a low-pitch hum and resulting in fretting of the lamination insulation
followed by potential burning, cracking, and breaking of the lamination steel
and also a high potential for damage to the stator bar insulation installed in these
2.1 STATOR CORE 49
slots. Also, a loose stator core will not be able to withstand the additional forces
during fault conditions and may result in premature failure of the stator. Further-
more, a low clamping force will decrease the core capacity to resist the buckling
phenomenon. Core pressures once the core bolts have been tightened on modern
new cores are typically in the range of 150–200 PSI. Once a core buckle or wave
develops, it is nearly impossible to remove it. Buckling can occur for a variety of
reasons such as
• overheating of the core
• frame not accommodating thermal expansion of the core
• tolerances used up when the core was assembled due to stator frame
manufacturing
• lamination being out of tolerance when manufactured
The core is designed to thermally expand by a certain amount radially, circumferen-
tially, and axially when at rated load with rated ventilation and cooling medium flow-
ing. There are tolerances between the keybar and laminations to accommodate
thermal expansion of the laminations with respect to the stator keybar and frame
assembly. When the machine heats up, the stator frame expands radially and circum-
ferentially as does the core. If a stator core is overheated, the allowed clearance
between the laminations and the keybars may be exhausted and the core will then
push up against the stator frame assembly, radially and tangentially, and begin to
buckle. Refer to Figure 2.1-9 and pay particular attention to the clearance or interface
gap (tangential and radial) between the laminations and the keybar and the lamina-
tion and the stator frame. If the stator frame is not able to expand freely on the sole-
plates for whatever reason, this may cause the core in that section to expand hard up
against the stator frame and core buckling may result. There are many stator designs
that have no initial clearance between the lamination and the stator frame. The core in
these designs will thermally expand in operation more than the frame and is normally
able to safely elastically expand the frame to accommodate the expansion difference.
When these generators operate at excessive temperatures, have lost core clamping
compression, or in some designs the frames are prevented from radially expanding,
the cores can buckle from excessive compressive stresses.
Cooling of the core is accomplished in large generators with the use of air to
water heat exchangers and modern-day designers use Computational Fluid
Dynamics software to optimize airflow and analyze temperature rises for different
components of the machine. Radial “ducts” which are formed by the vent lamina-
tions allow airflow in the core for this purpose. The losses generated in the core are
dissipated to the cooling air at the surface of the radial ducts. The width of the ducts
and the thickness of the core packages are chosen as required by the ventilation
needs of the machine and the temperature permitted in the core. The ventilation
scheme of the generator, that is, the amount of airflow and where the air is flowing
should never be adjusted without consulting the manufacturer first. In particular,
the rotor fan arrangements should be left as designed unless the manufacturer has
sanctioned a change.
50 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Stator frame
Surface air
Figure 2.2-1 Shows outside
cooler
cutout part of stator frame showing
the wrapper plate and frame
rings for the two shelves. The
surface air cooler opening
Wrapper cutout is covered by cardboard
plate for protection during
construction.
Stator frame
Leveling key
Stator soleplate
one manufacturer to another and some assemblies are very complex. The soleplates
are required to carry the weight of the generator, rotational torque and transient
tangential and axial loads due to system disturbances, and if required, provide
means of radial expansion. Referencing Figure 2.2-2 as an example, this particular
soleplate assembly consists of four major components as outlined beneath.
for example) is only possible if the stator frame can be lifted in the air exposing
the soleplate surface. The surface should be checked for corrosion and unifor-
mity of contact between the mating surfaces. The radial keys or dowel fit can also
be checked and all surfaces can be cleaned and lubricated before the stator is
lowered.
“J or T” hook
Leveling bolts
poured to the correct height with respect to the soleplate elevation. The grout serves
as a permanent encasement of the soleplate assembly.
The frame structure must also be capable of withstanding abnormal events
from the power system and generator faults, which cause high transient stresses in
the frame. Since the frame provides the basic support for the stator core, it must also
be able to move with the core expansion and contraction from heating and the mag-
netic pulsating forces associated with the rotating flux patterns in the core. To
accommodate all this, the core-to-frame mechanical coupling is usually done with
some flexibility installed. This is typically done by providing keybar (frame) to
core clearance when assembled, as well as using the soleplates which allow the
frame to expand and contract radially.
Frame stiffness and natural frequencies of vibration are important parameters
due to the (120 or 100 Hz) mechanical and electromagnetic forces developed in the
generators in conjunction with the stimulus from the power system frequency.
Therefore, great care is taken to ensure that the natural frequencies of the core
and the frame together are not near 120 (100) Hz or any multiple thereof. It is
suggested that these natural frequencies differ at least 20% from all multiples
and modes of 120 Hz to allow for safe operation of the machine.
To provide stiffness for the outer shell of the frame or casing, there are
frame rings or shelves welded to the wrapper at spaced axial intervals over the
height of the stator as shown in Figure 2.2-1. These are designed to give the stator
frame the strength it needs for its intended purpose of supporting the core. The
entire frame structure is dimensioned to ensure the correct strength and to avoid
the natural frequencies of the once-and twice-per-revolution characteristics of the
generator. The type of material used in the frame is generally mild steel which is
easy to weld with good strength and low-temperature ductility. A ventilation path
must be provided to direct cooling air from the exit of the core (stator hot air) to
the surface air coolers. The stator hot air is sent through stator frame ventilation
path to the surface air coolers to become stator cold air after passing through the
coolers. The air is then sent back through the generator to the various components
such as the rotor field poles, stator core, circuit rings, and main and neutral leads
to remove heat and become stator hot air again. Of course, the sizing of the cut-
outs and cooling passages is determined by the amount of cooling required in
each part of the generator.
Stator frames are also designed with lifting and handling in mind. Once a
machine is built, it must be delivered to a site, and to do this, requires transportation
by any number of means such as a large truck for smaller machines and by rail and
ship for larger machines. The method of lifting is generally by craning and to
achieve this, lifting beams are designed to get the stator lifted and set into the
pit area. The machine may be erected in sections or as one piece depending on
the size. It is, in fact, the transportation mode that governs the maximum size that
a component can be manufactured. There is no point in building a machine so big
that it cannot be transported to the generation site. Therefore, such things as overall
weight of the stator and the transport system must be accounted for, as well as the
overall dimensions. Some of the things to consider are time of year, rural road
54 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
2.3 ELECTROMAGNETICS
For simplicity, cross-sectional view presented in Figure 2.3-1 shows an airgap
separating the slotted outer surfaces of both the rotor and the stator. The major ele-
ments of the magnetic circuit, as shown, are the rotor (including the rotor winding,
pole bodies and the rim), the airgap (which constitutes the principal reluctance in
the circuit), and the laminated steel stator core (including the stator teeth/slots and
stator yoke below the slots).
The airgap is the annular region between the rotor body and the stator core
and probably has the largest influence on the electromagnetic design of the gener-
ator. Although the airgap is large to accommodate insertion of the rotor, it is small
in relative terms to the rest of the magnetic circuit of the generator. It has a major
influence with regard to the reluctance of the total magnetic circuit and, hence, the
overall stability of the generator. The airgap greatly affects the steady-state stability
of the generator when connected to the power system by simple variation of the
length of the space between the stator and rotor outer surfaces. The length of this
1 Stator yoke
2 Stator teeth
3 Stator winding in
slots
4 Salient pole rotor Figure 2.3-1 Four pole
5 Excitation winding generator flux pattern.
6 Magnetic field lines Source: Courtesy of Voith.
2.3 ELECTROMAGNETICS 55
airgap is used to determine the short circuit ratio (SCR), which is calculated as
described elsewhere in this book.
In practical terms, this means that the longer the airgap, the higher is the mag-
netic circuit reluctance, and, therefore, the higher the short circuit ratio. Further-
more, the generator will tend to be more stable, producing higher ampere-turns
(A-T) to achieve the required level of magnetic flux across the airgap. In real terms,
this means more field current is required.
A reasonable rule of thumb for the ampere-turns of the generator as a whole
is that the airgap generally accounts for up to 90% of the total ampere-turns pro-
duced by the rotor. The remainder of the iron in the total magnetic circuit uses the
other 10% or more and yet accounts for the majority of the electromagnetic flux
path. This is because of the high permeability of the iron and high reluctance of
air in the airgap. Therefore, a larger generator is required for higher apparent power
output if the SCR ratio is to remain constant. This is because a larger rotor is
required to handle the extra field current for the higher output and the airgap would
be required to be about the same size to maintain the constant SCR. The airgap
always needs to be large enough to permit insertion of the rotor through the stator
bore with sufficient clearance for safe handling. This and stability requirements
limit the minimum possible SCR in generators.
Electromagnetic finite element analysis (FEA) is the preferred method to
determine the actual magnetic field and its distribution in the machine and serves
as a good visual representation to better understand what is happening within the
machine at various loads. An example of a salient pole generator analysis on open
circuit is shown in Figure 2.3-2, at full load in Figure 2.3-3, and during a sudden
short circuit in Figure 2.3-4.
In the open circuit example of Figure 2.3-2, the flux pattern is completely
symmetrical about the pole axis of the rotor. Although the flux path includes
the stator, the stator winding is on open circuit, and no current is flowing. There-
fore, there is no back electromotive force (EMF) from the stator winding, and no
electromagnetic torque coupling between the stator and rotor windings.
In the case where the generator is connected to the system, there is current in
the stator winding that is leading or lagging the voltage and significant torque is
developed (see Figure 2.3-3). This shows the increase in pole flux density with
increased load to compensate for the stator back EMF. As the turbine drives the
rotor (counter clockwise direction in this example), the electromagnetic coupling
between the stator and rotor windings tries to pull the rotor back in line with the
axis of the stator poles. This difference in position of the stator and rotor pole axis
creates a load angle that can be varied by changing the power output from the tur-
bine and the field current for magnetic coupling between the stator and rotor.
Increased field current pulls the rotor back toward the direct axis in the clockwise
direction. In the case of the short circuit, the flux pattern symmetry is lost on the
stator side and is now different on the rotor side.
A few more interesting examples of FEA when the generator is in the over-
excited and under excited modes of operation are shown in Figure 2.3-5–2.3-8.
Again, these examples are presented to give the reader a visual idea of what is
occurring magnetically inside the generator during these modes of operation.
56 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
A(Wb/m)
1. 2240e–001
1. 0490e–001
8. 7394e–002
6. 9891e–002
5. 2389e–002
3. 4886e–002
1. 7384e–002
–1. 1900e–004
–1. 7622e–002
–3. 5124e–002
–5. 2627e–002
–7. 0129e–002
–8. 7632e–002
–1. 0513e–001
–1. 2264e–001
B(T)
2. 4000e+000
2. 2286e+000
2. 0571e+000
1. 8857e+000
1. 7143e+000
1. 5429e+000
1. 3714e+000
1. 2000e+000
1. 0286e+000
8.5714e–001
Y 6.8571e–001
5.1429e–001
3.4286e–001
X 1.7143e–001
Time = 0.633 506 486 209 975 s Z 0.0000e+000
Speed = 94.7368 RPM
Position = 360.098°
Figure 2.3-2 Flux distribution and flux density at no load obtained by FE simulation
(65 MVA, 76 poles, 13.8 kV, 396 slots). Source: Courtesy of Dr. Arezki Merkhouf.
A(Wb/m)
1. 5913e–001
1. 3282e–001
1. 0650e–001
8. 0191e–002
5. 3879e–002
2. 7566e–002
1. 2540e–003
–2.5058e–002
–5.1371e–002
–7. 7683e–002
–1. 0400e–001
–1. 3031e–001
–1. 5662e–001
B(T)
2. 4000e+000
2. 2286e+000
2. 0571e+000
1. 8857e+000
1. 7143e+000
1. 5429e+000
1. 3714e+000
1. 2000e+000
1. 0286e+000
8. 5714e–001
Y 6. 8571e–001
5. 1429e–001
Y
3. 4286e–001
X 1. 7143e–001
Time = 0.633 506 486 209 975 s Z 0. 0000e–000X
Speed = 94.736 800 RPM
Position = 360.098264° Z
Figure 2.3-3 Flux distribution and flux density at rated load obtained by FE simulation
(65 MVA, 76 poles, 13.8 kV, 396 slots). Source: Courtesy of Dr. Arezki Merkhouf.
2.3 ELECTROMAGNETICS 57
B(T)
2. 4000e+000
2. 2286e+000
2. 0571e+000
1. 8857e+000
1. 7143e+000
1. 5429e+000
1. 3714e+000
1. 2000e+000
1. 0286e+000
8. 5714e–001
6.8571e–001
5.1429e–001
3. 4286e–001
1. 7143e–001
0.0000e+000
A(Wb/m)
5. 4090e–002
4. 5075e–002
3. 6060e–002
2. 7045e–002
1. 8030e–002
9. 0147e–003
–3. 4785e–007
–9. 0154e–003
–1. 8030e–002
Y –2. 7046e–002
–3. 6061e–002
X –4. 5076e–002
Y
Time = 0.633 506 486 209 975 s Z –5. 4092e–002
Speed = 94.736 800 RPM
Position = 360.098 264°
Figure 2.3-4 Flux distribution and flux density during sudden short circuit obtained by
FE simulation (65 MVA, 76 poles, 13.8 kV, 396 slots). Source: Courtesy of Dr. Arezki
Merkhouf.
Simplified:
V LL ∗k
Machine flux φ = Wb (2.2)
7 7∗f ∗kw ∗N ph
where,
VLL = line-to-line stator terminal voltage in volts
k = number of stator winding parallel paths per phase
f = frequency
kw = stator winding factor (includes pitch and distribution)
Nph = number of stator winding turns-per-phase
58 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
A(Wb/m)
2. 8369E–01
2. 4427E–01
2. 0485E–01
1. 6543E–01
1. 2601E–01
8. 6588E–02
4. 7168E–02
7. 7475E–03
–3. 1673E–02
–7. 1093E–02
–1. 1051E–01
–1. 4993E–01
–1. 8935E–01
–2. 2877E–01
–2. 6819E–01
–3. 0761E–01
X
Time = 0.505 013 299 449 292 s Z
Speed = 128.600 000 RPM
Position = 37.420 362°
0 200 400 (mm)
Figure 2.3-5 Shows the flux distribution in the over-excited mode with field current at
1800 A (310 MVA, 56 Poles, 128.6 RPM). Source: Courtesy of Dr. Arezki Merkhouf.
The winding factor of the machine is largely concerned with reducing harmonic
effects and wave shaping. It is comprised of the pitch and distribution factors.
The pitch factor is determined from a winding diagram and depends on the
number of slots separating the distance (the coil span) between connection from
top and bottom coil legs or bars in series, that is, a top leg in slot 1 connected
to a bottom leg in slot 7 gives a span of 6 and for 195 slots, 26 poles, gives slots
per pole of 7.5. The machine therefore would have a stator winding pitch of 6/7.5 or
0.8. The distribution factor deals with the fact that the EMF induced in different
slots are not in phase, therefore, their vector sum must be less than their arithmetic
sum. The distribution factor therefore is the ratio of the vector sum divided by the
arithmetic sum of the stator coil EMFs for this distribution.
To work out the winding factor (kw) from the pitch (kp) and distribution (kd)
factors see Equation (2.3):
kw = k d ∗kp = sin β 2 η × sin γ 2 × sin ρπ 2 (2.3)
2.3 ELECTROMAGNETICS 59
B (tesla)
3. 1367E+00
2. 9276E+00
2. 7185E+00
2. 5094E+00
2. 3002E+00
2. 0911E+00
1. 8820E+00
1. 6729E+00
1. 4638E+00
1. 2547E+00
1. 0456E+00
8. 3645E–01
6. 2734E–01
4. 1823E–01
2. 0911E–01 Y
6. 3159E–07
X
Z
Time = 0.505 013 299 449 292 s
Speed = 128.600 000 RPM
Position = 37.420 362°
0 200 400 (mm)
Figure 2.3-6 Shows the flux density in the over-excited mode with field current at 1800 A
(310 MVA, 56 Poles, 128.6 RPM). Source: Courtesy of Dr. Arezki Merkhouf.
where,
β = π/number of phases = π/3
η = number of slots/number of poles/number of phases
γ = π/(number of slots/number of poles)
ρ = stator winding pitch (from winding diagram)
Equation (2.2) provides the basic level of machine flux required to achieve rated
line-to-line terminal voltage in a generator, given a specific winding configuration.
This formula will be elaborated on in Chapter 11 and an example provided for
determining excitation levels in a flux test.
Generators are made with different power factor ratings. The most common
are 0.90 and 0.85 lagging. Two machines of the same MVA rating will have dif-
ferent capability design parameters for the two different power factors. The 0.85
60 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
A (Wb/m)
2. 2736E–01
1. 9625E–01
1. 6514E–01
1. 3402E–01
1. 0291E–01
7. 1796E–02
4. 0683E–02
9. 5699E–03
–2. 1543E–02
–5. 2657E–02
–8. 3770E–02
–1. 1488E–01
–1. 4600E–01
–1. 7711E–01
–2. 0822E–01
–2. 3934E–01
X
Time = 0.505 013 299 449 292 s Z
Speed = 128.600 000 RPM
Position = 37.420 362°
200 400 (mm)
Figure 2.3-7 Shows the flux distribution in the under excited mode with field current at
1190 A (310 MVA, 56 Poles, 128.6 RPM). Source: Courtesy of Dr. Arezki Merkhouf.
power factor machine will require more field current to achieve the same power at
the 0.85 power factor. Hence, the machine is somewhat larger to accommodate a
rotor that can handle more field current and cooling capacity and is more costly to
build. It is easy to see that design optimization to make the best utilization of the
magnetic materials is a design priority.
The flux density becomes the driving factor for the amount of stator core
material that is required. As can be seen from Figures 2.3-2 and 2.3-3, the flux den-
sities are different between open circuit and full load, but only marginally higher on
load. However, there is considerable redistribution of the flux when the machine is
on load, due to the stator currents. On open circuit, the stator core does not
approach the electromagnetic loss limits of the iron, which are typically in the
2.3 ELECTROMAGNETICS 61
B (tesla)
2.9979E+00
2. 7980E+00
2. 5981E+00
2. 3983E+00
2. 1984E+00
1. 9986E+00
1. 7987E+00
1. 5989E+00
1. 3990E+00
1. 1991E+00
9. 9929E–01
7. 9943E–01
5. 9957E–01
3. 9972E–01
Y
1. 9986E–01
0. 0000E+00
X
Time = 0.505 013 299 449 292 s Z
Speed = 128.600 000 RPM
Position = 37.420 362°
200 400 (mm)
Figure 2.3-8 Shows the flux density in the under excited mode with field current at 1190 A
(310 MVA, 56 Poles, 128.6 RPM). Source: Courtesy of Dr. Arezki Merkhouf.
1.7 T range in the stator teeth and under 1.45 T in the stator core back. Lower flux
densities will typically be found in the rotor rim, but they are induced by the DC
current in the field winding, and so do not cause losses. That is to say, they are
unidirectional as far as the rotor is concerned and so there are no eddy current losses
in the rotor body due to the main flux. It is the alternating effect in the stator that
designers are concerned with, in this instance. Heating of the rotor components is a
concern, but more so because of the I2R losses in the field winding as opposed to
the effects of magnetic interaction with the stator slots causing heating due to
induced stator slot ripple effects (i.e. the variation of the main field due to the slot-
ting of the stator).
62 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
where,
MVA = rated apparent power
K = a proportionality constant
Ma = armature reaction
Φ = magnetic flux per pole at rated voltage in Webers
P = number of poles
F = frequency
This is really the same as the product of the stator current and the stator terminal
voltage. The stator or armature current is proportional to the armature reaction. The
stator voltage is proportional to the flux. The field winding ampere-turns or field
current at rated load is directly related to the level of armature reaction. Calculation
of the flux per pole is described in Section 2.3 and the calculation of armature reac-
tion is as shown in Equation (2.5):
Nst
Nph
Nph k
Ma = × ∗Ia A − T (2.5)
2P Kp Kd
2.5 FLUX AND ARMATURE REACTION 63
where,
Nph = number of phases
P = number of poles
Nst = number of full turns
k = number of parallel paths
Kp = winding pitch factor
Kd = winding distribution factor
Ia = stator current in amperes
One other basic relationship that governs the rating of a generator is the output
coefficient. Simply put the output of the generator increases with the square of
the diameter of the rotor or stator bore, and with the height of the machine, based
on the following relationship as shown in Equation (2.6)
MVA
Output coefficient = MVA min m3 (2.6)
D2b LS
where,
MVA = rated apparent power
Db = diameter of the stator bore in meters
L = height of the active iron in the stator in meters
S = speed of the rotor in RPM
Specific generator ratings are accommodated in machine design by trading off the
levels of magnetic flux against the level of armature reaction. The actual compo-
nent dimensions as described above also play a role in optimizing designs of large
generators. Therefore, a specific rating can be achieved by a relatively high value
of flux and a low level of armature reaction, and vice versa, or some combination in
between. Increasing the generator output at a specific combination of flux and
armature reaction can also be done by making the machine taller or longer. Using
all these factors, one can design a machine to fit any output rating desired.
However,when one parameter changes, it affects all the other parameters, some
marginally, but some others significantly.
Two additional formulas that help to describe the output of the generator in
relative terms are specific magnetic loading and specific electric loading. These
two formulas as shown in Equations (2.7) and (2.8) in their more basic form
can be multiplied together to produce the output coefficient above:
PΦ
Magnetic loading = Wb m2 (2.7)
LDb
I st N c
Electric loading = A m (2.8)
Db k
64 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
where,
P = number of poles
Φ = flux per pole in Webers
Db = stator bore diameter in meters
L = core iron effective length in meters
Ist = stator, single-phase current in amperes
Nc = total number of stator conductors
k = number of parallel paths in stator circuit
Using the above formulas, one can compare basic machine design outputs to deter-
mine which is more highly loaded in specific terms. For instance, if a machine is
prone to high core-end heating, the specific electric loading of the generator is
likely to be high relative to other machines, indicating that high stray losses are
present. High stray losses can directly affect core-end flux penetration and, subse-
quently, the level of core-end heating.
Machines with a high level of flux require a relatively large volume of iron to
carry the flux and a relatively small amount of copper to carry the stator and field
currents. Such machines tend to be larger and more costly to build. Machines with a
low level of flux require a relatively small volume of iron to carry the flux but a
relatively large volume of copper in their windings. Such machines are termed
“copper rich,” and they increase the problem of heat removal from the windings.
These machines tend to be smaller and less costly to build.
The per-unit transient and subtransient reactances, which play a significant
role in the electrical performance of the generator connected to the power system,
tend to be low with high-flux levels. The higher-flux generator will, therefore, tend
to have a somewhat better inherent transient stability. It will also tend to have
higher per-unit transient currents during severe disturbances and, therefore, higher
winding forces and torques, than a lower-flux machine. To limit fault currents in
the generator and, hence, the forces and torques, minimum values of subtransient
reactance are usually specified. The subtransient reactance is a function of the sta-
tor leakage reactance and the effects of the rotor amortisseur or damper winding.
As discussed earlier, the principle function of the stator core is to carry electromag-
netic flux. The core must handle magnetic field flux densities in the stator teeth and
in the core-back or yoke area. The magnetic field is revolving, so it creates an alter-
nating voltage and current effect in the generator components, which is a source of
high losses and heating. This alternating effect also causes vibration of the core at
the rotational frequency and with harmonics due to the nature of the flux patterns.
Because of the inherent vibration and the large mechanical and thermal
forces involved, the core must be held solidly together so that there are no natural
2.7 STATOR WINDINGS 65
frequencies near the once and twice per revolution forcing frequencies. Some cores
are installed as continuous pile, while others have core splits. Care must be taken
with core splits as the packing material between the splits installed in the original
day may be dried out, brittle, coming out of the split at the back of the core, and no
longer able to perform its designed function. The function of the packing material
is to consolidate the split sections of core so that the core behaves as a solid mass
when excited by the field current. If this packing is deteriorated or missing, the core
may not behave has designed once excited. A sure sign the packing material is
missing is fretting at the split of the core. Designers take great care to ensure that
the natural frequencies of the core or core and frame are not near 120 (100) Hz or
equivalent to other induced forcing frequencies. It is desirable to keep the natural
frequencies of the core and frame at least 20% away from forcing frequencies. An
example of a forcing frequency above 120 (100) Hz is typically a tooth-pass fre-
quency which is equal to the stator slots per pole rounded to the closest integer
multiplied by 120 (100) Hz.
There also is a large rotational torque created by the electromagnetic cou-
pling of the rotor and stator across the airgap. This is in the direction of rotor rota-
tion. The torque due to the magnetic field in the stator core iron is transmitted to the
core frame via the keybar structure at the core back. Therefore, the stator frame and
foundation must be capable of withstanding this torque, as well as large changes in
torque when there are transient upsets in the system or the machine.
The natural vibration inherent in the core must also be accounted for in the
core to frame coupling. Heating and cooling effects in the core and frame materials
will also affect this coupling and vibration, due to differences and rates of thermal
expansion and contraction in the core and frame components.
The stator winding is made up of insulated copper conductor bars or coils that are
distributed around the inside diameter of the stator core, commonly called the stator
bore, see Figure 2.7-1 for typical multi-turn coil configurations and Figure 2.7-2 for
a typical single turn Roebel bar configuration. The winding is installed in equally
spaced slots in the core to ensure symmetrical flux linkage with the field produced
by the rotor. Each slot contains two Roebel bars or coils, one on top of the other
(see Figures 2.7-3 and 2.7-4). These are generally referred to as top and bottom bars
or top and bottom legs. Top bars or legs are the ones nearest the slot opening (just
under the wedge) and the bottom bars or legs are the ones at the slot bottom. The
core area adjacent to the slot is generally called the core teeth as shown in
Figure 2.7-5. The stator winding is then divided into three phases, which are almost
always wye-connected. Wye connection is done to allow a neutral grounding point
and for relay protection of the winding. The three phases are connected to create
symmetry between them in the 360 arc of the stator bore. The distribution of the
winding is done in such a way as to produce a 120 difference in voltage peaks
from one phase to the other, hence the term “three-phase voltage.” Each of the three
66 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Copper strands
Groundwall insulation
phases may have one or more parallel circuits within the phase. Multiple groups of
coils can be connected in series to form the entire phase circuit. The parallels in all
of the phases are equal, on average, in their performance in the machine. Therefore,
they each “see” equal voltage and current, and magnitudes and phase angles, when
averaged over one alternating cycle.
2.7 STATOR WINDINGS 67
Strand insulation
Semiconducting layer
Strand insulation
Semiconducting layer
Figure 2.7-4 Cross section of a stator multi-turn coil in the stator slot.
68 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Slot wedge
Under wedge
ripple spring
Conductive slot
Laminated
side spring
stator core
Copper strands
(20 per turn)
RTD three
wire lead
Conductive bottom
slot strip
Turn to turn Conductive middle
insulation slot strip
The stator winding in any particular phase group are arranged such that
there are parallel paths that overlap between top and bottom bars or coils, see
Figures 2.7-6 and 2.7-7, for examples of reversing group jumpers. The overlap
is staggered between the top and bottom bars or coils. The top bars or legs in
the first pole group are connected to the bottom bars or legs in the next pole group
in one direction, whereas the bottom bars or legs in the first pole group are con-
nected in the other direction on the opposite side pole group. This connection with
the bars or legs on progressive pole groups around the stator creates a “reach” or
“pitch” of a certain number of slots. The pitch is, therefore, the number of slots that
the stator bars or coils have to reach in the stator bore arc, separating the two bars or
coils to be connected. This is almost always less than one pole pitch and it is done to
assist in reducing the harmonics induced in the stator winding.
Once locally connected, bars or coils form a group. A group may be a parallel
circuit, or a full phase. Parallel circuits may be connected in series with other par-
allel circuits to form a full phase. The total width of the overlapping parallels is
called the “breadth.” The combination of pitch and breadth create a “winding or
distribution factor.” The distribution factor is used to minimize the harmonic con-
tent of the generated voltage. In the case of a two parallel path or more winding,
2.7 STATOR WINDINGS 69
Figure 2.7-6 Stator endwinding showing reversing group jumper connection on a bar
winding.
Circuit rings
Back legs
Front legs
Figure 2.7-7 Stator endwinding showing reversing group jumper connections on a multi
turn winding.
these are connected in series or parallel via the circuit rings outside the stator bore,
see Figures 2.7-6 and 2.7-7. The connection type will depend on a number of other
design issues regarding current-carrying ability of the copper in the winding.
70 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.7-8 Showing reduced eddy current losses with individual strands of copper
instead of one large piece. Source: Courtesy of Dr. Michael Znidarich & Engineers
Australia [4].
2.7 STATOR WINDINGS 71
Inverted turn
Coil loop
Figure 2.7-10 Looping of a coil with inverted turn. Source: Courtesy of Dr. Michael
Znidarich & Engineers Australia [4].
inverted turn is done in the endwinding of the multi-turn coil, and depending on the
design, it can be done on any turn of the coil, wherever the most reduction in eddy
currents is calculated during the design stage, see Figure 2.7-10 during the looping
process of coil manufacturing.
72 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Similarly, to reduce the effect of the eddy currents within each individual stator
bar, the conductors are made up of numerous copper “strands” (refer back to
Figure 2.7-2). This is analogous to the reasoning behind the stator core being made
up of very thin insulated laminations rather than a solid mass of steel. However,
although the strands are insulated from one another in the bar or coil, they are even-
tually connected at each end of the stator bar or coil. Therefore, additional circulating
current could flow from the top to the bottom strands in a single bar or coil. This is due
to the difference in the magnetic field from the top to bottom of the slot. To reduce the
effect of the circulating currents, the strands are “Roebel transposed” in each bar (see
Figures 2.7-11 and 2.7-12). Roebel transposition of the copper strands refers to the
repositioning of each strand in the stator bar stack such that it occupies each position
in the stack at least once over the full length of the stator bar.
Roebel transpositions are mainly 360 and 540 . A 360 transposition means
that each strand occupies each position once over the length of the bar, and a 540
transposition means that each strand occupies each position one-and-a half times.
The 360 transposition is generally done in the slot only and the 540 transposition
includes the very ends of the stator bars, and in the curved endwinding portion as well.
5 1
6 2
7 3
6 5 8 4
7 1
8 2
7 6 4 3
8 5
4 1
8 7 3 2
4 6
3 5
4 8 2 1
3 7
2 6
3 4 1 5
2 8
1 7
2 3 5 6
1 4
5 8
1 2 6 7
5 3
6 4
5 1 7 8
6 2
7 3
8 4
1′ 2′ 3′ 4′ 1′ 2′ 3′ 4′
1′ 2′ 3′ 4′ 1′ 2′ 3′ 4′
1′ 2′ 3′ 4′
1′ 2′ 3′ 4′
1′ 2′ 3′ 4′
1′ 2′ 3′ 4′
Figure 2.7-12 Roebel bar principle. Source: Courtesy of Dr. Michael Znidarich &
Engineers Australia [4].
40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
1
5
9
13
17
21
25
29
33
37
41
45
48
53
57
61
65
69
73
77
81
Side A Side B Side A Side B
Strands
Figure 2.7-13 Temperature profile of a double-stack stator bar with separate Roebel
transposed stacks. The average temperature difference between side A and side
B of the bar example shown is about 10 C. Source: Courtesy of Alstom Power Inc.
Figure 2.7-15 Cross Roebel transposition – temperature profile of a double stack stator
bar. All cooling strands are temperature equalized due to the elimination of stack-to-stack
circulating currents. Source: Courtesy of Alstom Power Inc.
2.7 STATOR WINDINGS 75
There are many ways of designing stator conductor bars, depending on the
size and cooling method required for the machine. Cooling is particularly critical in
designing machines for higher outputs. In this regard, direct cooling is the most
desirable type of cooling because it increases the generator stators current carrying
capability considerably.
The advantage of this is to reduce flux levels and, hence, the physical size
and weight of the generator. The basic limit for conventionally cooled generators
(i.e. indirect cooling with air) is now in the 944.5 MVA range. Hydro generators up
to 855 MW have been built with direct conductor cooling.
In indirectly cooled machines, the strands within the conductor bars are all
solid and the heat generated in the conductors is removed by conduction through
the groundwall insulation to the stator core. The size of the generator is signifi-
cantly limited by the temperature conduction through the groundwall insulation
to the stator core.
In direct water cooled windings, the copper strands are made hollow, to carry
liquid coolant. The stands are generally rectangular in shape to allow stacking and
they are each individually insulated from one another and Roebel transposed.
Figure 2.7-16 shows a three-dimensional representation of what a typical sta-
tor bar looks like when inserted into the stator core as well as the strand arrange-
ment for this particular design. In this mixed strand arrangement, the hollow
Figure 2.7-16 Shows a 3D representation of a typical water cooled stator bar in the slot
section of the stator core. Source: Courtesy Voith.
76 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
strands are evenly interspersed among solid strands. The strands can be arranged in
various combinations to produce more efficient winding designs.
In directly cooled stators, it is possible to increase the current density in the
copper winding of the stator to achieve higher ratings. Trade-offs are also made
between slot sizes and winding configurations to find the optimum terminal volt-
age level versus the current flowing in the stator winding, all in consideration with
keeping magnetic flux densities in the stator iron at manageable levels.
Because the stator current densities in directly cooled windings are so much
higher than in indirectly cooled windings, designers must also consider the effect of
transients and temperature rise. Considerations of reactance and stability also come
into play and, therefore, so do short circuit ratio and excitation performance.
Some modern generator designs mix solid copper stands for conduction of
the electrical current and hollow stainless steel strands for carrying the coolant.
Figure 2.7-17 shows an arrangement for a direct cooled winding where the headers
and hoses carry the coolant to the winding. This design has been in service for the
last 30 years and has been successful. The use of stainless steel strands for cooling
has eliminated certain industry problems of copper erosion and corrosion in the
stator bars. The mixed steel and copper stator bars also tend to be more rigid than
fully copper bars and allow higher wedging pressures in the slot.
In direct water cooled machines, the cooling method dictates the need for an
external system to remove the heat picked up by the stator cooling water after it
passes through the stator winding. Therefore, an external system is attached to
the generator that employs heat exchangers to accomplish this function. To circu-
late the water, pumps and a piping system are provided. In addition a filtering sys-
tem is provided to remove any large particles suspended in the stator cooling water
Figure 2.7-17 Direct cooled stator winding (rotor poles removed). Source: Courtesy
of Voith.
2.7 STATOR WINDINGS 77
that can cause blockage within the stator windings inside the generator. Since the
water is in contact with current-carrying copper conductors, which are also oper-
ating at voltage levels from ground potential up to 23 kV, the water must be kept
absolutely as pure as possible to avoid flashovers by conduction through the water.
To maintain pure water, a de-ionizing system is provided. See Chapter 3 for a
description of the stator cooling water system.
The basic functions of electrical insulation in the stator winding are to main-
tain ground insulation between the conductors and the stator core and other
grounded objects, and to maintain insulation between turns of multi-turn coils
and between the strands within a turn.
The groundwall insulation must be designed to withstand line-to-line AC
voltages over the entire life of the generator. In addition, it must be capable of with-
standing overvoltages from system faults. The turn insulation must withstand nor-
mal coil voltage over its lifetime, with substantial short time overvoltages in the
event of a steep-front voltage surge such as system faults or lightning strikes local
to the generating station. Strand insulation is exposed to only a few volts with brief
overvoltages during occasional high-current transients.
A high resistance coating or “semiconducting” system is applied on top of
the groundwall insulation in the slot to control the voltage distribution over the
length of the slot for machines with terminal voltages in excess of 6 kV depending
on the designer’s preference (see Figure 2.7-18). In addition, a special “grading”
system is applied to the bars or coils over a short distance starting a few inches from
the bar or coil exit from the slot to part way into the endwinding area. This grading
system is typically a Silicon Carbide or Iron Oxide coating. The grading system
allows for a gradual voltage drop in the endwinding to the stator core. The end-
winding is at line potential and surface current flows from the endwinding through
the grading system to the stator core. Some manufacturers apply the grading mate-
rial over the full length of the end turn.
Figure 2.7-18 Shows the black semiconducting material and the grey Silicon Carbide
grading material.
78 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
To ensure good contact between the stator winding and the core in the slots, a
side packing filler is inserted between the coil and core section inside the along the
entire length. The side filler is impregnated with semiconducting material to assist
with the electrical contact to the stator core. The base material is usually made up of
strong resin-filled woven glass material. It may be a flat piece but a ripple-spring
filler is now commonly used to ensure continual pressure and contact over the life
of the winding (see Figures 2.7-19 and 2.7-20).
Another popular method used by some manufacturers to side pack the wind-
ing is to use a sheet of soft semiconducting material with an adhesive applied to it,
and wrap the slot section of the coil or bar forming a blanket over the winding. The
adhesive is applied to one side of the blanket only. To illustrate the application of
the adhesive, imagine a piece of cloth that is as long as the coil or bar and wide
enough to wrap three sides of the coil or bar. Then, place a bead of adhesive in
an “S” shape covering the cloth. The cloth is then wrapped around three sides
of the coil or bar, the bare side (without the blanket) faces the airgap of the machine.
The coil or bar with this new “blanket” is then inserted into the slot section and the
adhesive conforms to the slot section securing it into place. The adhesive is
between the winding and the blanket only, the core does not come into contact with
the adhesive. Only the dry side (where no adhesive has been applied) of the blanket
contacts the core. This type of semiconductive installation requires the installer to
apply just the right amount of adhesive. Too little will allow for a loose winding,
and too much will be wasteful and create a much larger cleanup than necessary.
Further, if the adhesive material smears into the wedge groove, wedging will be
difficult. Some manufacturers apply the bead of conducting or nonconducting
Figure 2.7-19 Flat side-packing (top) and ripple spring (bottom) with semiconducting
impregnation.
2.8 STATOR WINDING WEDGES 79
Figure 2.7-20 Shows stand up view of Figure 2.7-19 to illustrate the ripple spring.
adhesive on the blanket, fold it in half, and then wrap this strip in a spiral fashion
around the coil or bar.
Due to the current flowing in the stator bars or coils, there is a reaction force
in each slot, which varies according to the level of current and direction of flow at
any instant. This creates forces between bars or coils that are both repulsive and
attractive at any given time in the alternating cycle. Therefore, the slot
section of a stator conductor bar or coil “sees” significant and constant (mainly
radial) vibration forces at the twice-per-revolution frequency. The stator bars or
coils tend to vibrate in the slot, a phenomenon called “bar or coil bouncing”
(see Chapter 5). Therefore, the stator bars or coils must be tightly wedged in
the slot to eliminate the relative motion and avoid fretting damage from contact
against themselves and the stator core and bar or coil packing systems. Coil side
packing typically should be on the trialing side referred to the direction of rotation
of the unit. Stator windings have been known to fail quickly once they become
loose in the slot.
Slab
wedges
Packing
materials
Stationary and
drive wedges
packing material depends on how much space exists to be filled. Sufficient packing
is installed to tighten the wedge assembly to the OEM installation procedures.
These procedures are normally supplied when wedges are purchased for a stator
re-wedge (ask for the procedure) or when the machine is new and wedges are being
installed. If there is insufficient packing material, the wedge system will be loose. If
there is too much packing material, the wedge may crack under the extreme pres-
sure that is placed on it, this is particularly true for systems with drive wedges. To
the right of the packing material in Figure 2.8-1 is a typical “slab” wedge made of
some sort of insulating material from the early days such as Micarta™ or even
wood such as Maple. Split Maple wedges are split axially with opposing tapers
and were very popular in the early days of hydro generators. The slab wedge will
have different methods used to tighten it against the top coil or bar in the slot.
For example, a stationary or drive wedge between it and the packing material can
be used or the slab can be driven over flat filler. The final wedge assembly to the
right of the slab wedge in Figure 2.8-1 is a three-part wedge system consisting of
the main wedge itself, a drive wedge, and a stationary wedge. The drive and station-
ary wedge are fixed in length and taper. All materials in modern systems typically
consist of some sort of epoxy mixed with fiberglass and processed to form the shapes
as shown. This system typically allows for the main wedge itself to deflect slightly
when in its final position providing the “spring” action pushing against the coil sur-
face to keep everything tight. The amount of “spring” provided by this design is min-
imal due to the small amount of deflection of the wedge.
In this system, it is very easy to put too much pressure on the main wedge by
putting too much packing material and overdriving the “drive” wedge into its final
position. Cracking of the main wedge over time will result if too much deflection is
allowed particularly since this system has a very thin wedge. When designing slab
type wedges over flat sliding fillers, it is important to know the flexural strength of
2.8 STATOR WINDING WEDGES 81
the wedge material to ensure it can provide the necessary retention force for the
coils in the slot. For this reason one has to carefully choose the “fiberglass lami-
nate” that is used. Not all laminates are the same.
During installation, it is important to utilize the gauge that is supplied by the
manufacturer during the winding process in order to get the optimal deflection of
the wedge. The gauge is manufactured on site by using a dial indicator placed on
the wedge surface and driving the “drive wedge” into position. Depending on how
much packing is in place, the drive wedge will need to be driven a longer length or a
shorter length to get the dial indicator to the desired reading as the main wedge
deflects outwards. It is important to observe that the dial indicator reading is actu-
ally measuring the deflection of the wedge outwards and not the movement of the
wedge outwards because it is not seated properly in the groove. A gauge can then
be made out of aluminum with various numbers of lines marked on it that can be
inserted into the drive wedge position to determine if enough or not enough pack-
ing has been installed to get the proper wedge deflection. It is important to recog-
nize that when this main wedge deflects (crowns) into the airgap, the amount of
wedge left “holding on” in the wedge groove is reduced and thus the coil retention
effectiveness may also be reduced. To summarize, the amount of packing material
is critical in ensuring the wedge remains tight for as long as possible for the life of
the winding. It is quite likely that a re-wedge may be necessary more than once in
the life of the winding using this system.
Figure 2.8-2 has four different styles or components of wedge assemblies
which will now be discussed. The top wedge assembly is a typical fiberglass wedge
body with a ripple spring and drive wedge assembly to compress the spring once
installed. As with the other systems discussed, the depth packing is critical in
ensuring the right amount of spring compression occurs during in installation.
Most wedge spring assemblies should be compressed to 80% of their original value
to ensure proper pressure over the life of the winding. Of course, consult the spring
manufacturer or generator OEM for the exact compression amount for the specific
installation at hand. Checking the amount of spring compression is done in two
ways. The first way is to insert a feeler gauge into the air vent notches in the wedge
to see how much compression the spring has. Again, the spring manufacturer or
generator OEM will have the feeler gauge thickness to use as a “go-no-go” gauge.
Another way to measure spring compression is to use a wedge with holes
drilled into the wedge assembly as shown in the last wedge body in Figure 2.8-2.
A gauge with a small needle head is used to measure the spring compression in each
successive hole. These numbers should be within a tolerance set by the spring man-
ufacturer or generator OEM. A variation of these holes drilled into the wedge body is
to machine a groove the same length and width as all of the holes so the same gauge
can now slide along the groove length to measure the spring compression. The slot
will contain at least three wedges of this type (top, middle, bottom) or more depend-
ing on the stack length. The wedge itself, having a groove or holes machined into it, is
obviously weaker than a wedge that is solid. However, if designed correctly, it
should be sufficiently strong to provide its intended function along with the rest
of the solid wedges in the slots. Consultation with the generator OEM is recom-
mended if there are any doubts about the wedge integrity. The small wedge at the
82 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
bottom of Figure 2.8-2 has an interesting design feature; it has Kevlar® wrapped
around it. The Kevlar® provides a less abrasive and softer surface against the stator
iron so if the wedge were to become loose and vibrate in the groove, damage to the
core would be minimized (see Figures 2.8-3 and 2.8-4). The typical fiberglass and
epoxy wedge is quite abrasive when it moves relative to the core and can cause abra-
sion of the iron if left unattended. The Kevlar® wrap is applied in a mold to the wedge
(the wedge is made in a mold as well, it is not a machined wedge) and is a very expen-
sive, not very common, and extremely difficult to source. It has been used success-
fully in machines with a very small wedge groove where machining a fiberglass
wedge proves difficult with the tight tolerances required for a proper fit.
The second and third wedge assemblies are typical ripple spring and driver
types. There are two important fundamental differences between the two wedge
assemblies. The first is that the ripple spring in the third wedge sits between the
depth packing and the wedge body with the driver in the wedge groove. The third
wedge assembly has the ripple spring in the wedge groove with the tapered driver
against the depth packing surface. Either assembly will perform well over the life if
the depth packing is correctly installed and the wedge assembly remains tight. The
2.8 STATOR WINDING WEDGES 83
Kevlar wrapped
wedge
Fretting dust
Core packet
Figure 2.8-3 Shows core packet and fretting at the interface of wedge groove.
Wedge
groove
after
cleaning
second difference is that the third wedge assembly is Kevlar® wrapped, while the
first, second, and third assembly are simply machined epoxy glass.
Stator bar or coil looseness is one of the main reasons that tight stator wedges
in the slots is so important. The resulting vibrations of the bars in the slot due to
looseness can quickly wear the groundwall insulation on the bar or coil right
through to the copper and cause a stator ground failure.
Maximum instantaneous bar/coil bounce force per unit length of stator
winding in the slot occurs when the top and bottom bars or coils in the same slot
are in phase and carrying maximum stator current as shown in Equation (2.9):
Total bar or coil leg bounce force,
(2.9)
F total = F bottom + F top
where,
3μo I 2
F top = N m
ws k2
μo I 2
F bottom = N m
ws k 2
Therefore,
4μo I 2
F total = N m
ws k 2
μo = 4π × 10−7 H/m
I = stator phase current in amperes
ws = stator slot width in meters
k = number of parallel stator circuits per phase
This force is toward the bottom of the stator slot and is also sinusoidal in nature, due
to the fact that it is proportional to the current squared. This means that the force
is associated with the pole-pass forcing function and produces vibration at
120 (100) Hz, similar to the vibration forces on the stator core and frame. Since
the magnetic field in the slot is highest near the top of the slot and diminishes
toward the bottom of the slot, it can be shown that the resulting difference between
the forces on the top and bottom bars or coil legs is substantial. In fact, the top bar or
coil leg forces can be up to three times that of the bottom bar forces when both bars
are in the same phase. The net effect for maximum bar bounce forces is
described above.
Wedging of the stator bars or coils, however, is not strictly concerned with
just the bar or coil leg bouncing effects. Since there is considerable heat generated
in a stator bar or coil, there are also thermal expansion, contraction, and insulation
shrinkage issues to consider. Thermal expansion and contraction can easily loosen
2.9 ENDWINDING SUPPORT SYSTEMS 85
bars in the slot if they are not wedged properly, and the heat impact on the insu-
lation systems can also be a factor if the insulation is not preshrunk (winding can
lose a very small amount of mass as it continues to cure in service) prior to wed-
ging. To elaborate on the preshrunk condition, when a winding is new, there is a
possibility of some very minute amount of shrinkage of the insulation system due
to the continued curing of the resins at specific temperatures depending on the man-
ufacturer. It is important to realize this shrinkage is extremely small, so if the
wedges are installed on the lower end of being tight, then this extra shrinkage could
put the wedge into the loose category. It is more probable, however, that that the bar
or coil has not been properly bottomed in the slot when wedged and the vibrations
have assisted in this task and now the wedges are loose.
keep the MVA rating constant. This relationship has significant consequences for
generator design. For example the number of coils in series for a given core length
will increase with voltage and the number of parallel circuits will increase for
higher line current leading to the requirement for addition number of slots. These
factors of core length and the number of slots is the basis for machine design. The
product of these two is proportional to the MVA capability of the machine (see
Figures 2.10-1 and 2.10-2).
A B C
A B C
A B C
If the connection is parallel, the terminal voltage tends to be lower and the
stator current higher. For the same MVA rating, if the connection of the stator
winding is in series, the terminal voltage will be higher and the current lower.
The physical consequence of this is that the higher voltage machine requires a
thicker groundwall insulation to withstand the higher voltage. For parallel con-
nected winding, there would need to be a large amount of copper and increased
cooling to accommodate the higher stator current.
In another example, there are four parallel paths in the stator winding as
shown in Figure 2.10-2. Figure 2.10-3 shows a Y-connection comprised of two
of the parallels connected in series for a 720 slot machine. These configurations
described above (and there are many more) allow flexibility in design to achieve
a machine with a smaller overall size, lower cost, and lowest losses for best
efficiency.
All generators require a means to deliver the power produced inside the machine,
out to the main transformer, via an isolated phase bus (IPB) system, copper bus or
cables.
Since there are three phases in the generator, three-phase lead connections
are required, commonly called stator terminal connections. These are used to make
the connection from the stator winding inside the generator, out through the gen-
erator frame and casing, to the system. Each stator terminal carries the same current
as the sum of the currents of all the parallels in a single phase. Since the terminals
are at the rated voltage of the generator they need to be insulated, and generally, the
same type of materials used for the stator winding insulation are used for the term-
inals as well (see Figure 2.11-1). In this photograph, the main leads are insulated
with mica tape and epoxy resin, then painted with a beige protective paint which
allows for easy cleaning, is essentially cosmetic in nature, and has almost no insu-
lating capability. The high voltage terminals may also have a series of split phase
2.11 STATOR TERMINAL CONNECTIONS 89
Figure 2.11-1 Generator main leads – new installation with tags to show testing
of the CT’s completed.
current transformers in addition to the conventional CTs used for protection of the
stator winding from turn to turn faults associated with multi-turn windings as
shown in Figure 2.11-2. In this photo, the winding is a two parallel or 2Y connec-
tion and is painted with red protective paint similar to the beige paint. Each leg of
the phase goes through the CT, thus the term “split phase,” and the CT monitors the
current in each phase as they should be equal if the machine has no coils cut out and
the airgap is perfectly balanced. In reality, the airgap is never perfect, and there will
90 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
always be some split phase reading on the winding. These numbers should be
documented when the machine is new or the winding is new and monitored there-
after and trended.
In addition to the high-voltage terminals, there are also three neutral term-
inals that make up the common connection point at the zero voltage or wye/star
connection of the stator winding. Although these are essentially at zero or ground
potential, they do carry the full stator current that the high-voltage connections
carry and so must be given the same cooling as the high-voltage terminals. They
are also insulated from ground, except at the actual connection or “star” point, to
ensure no circulating currents or faults occur anywhere else in the winding system.
This end of the stator winding (neutral end) will also have conventional CTs used
for protection of the stator winding as shown in Figure 2.11-3. This arrangement
can contain a number of redundant CTs for a “B” series protection if the “A” series
were to fail. In protection systems nowadays, there are normally two redundant
duplicate protection schemes in case one fails, these are “A” and “B.”
Rim
Laminated
end
rim
plates
Spider/drum
assembly
Figure 2.12-1 Shows a modern day rotor. Source: Courtesy of Dr. Michael Znidarich.
92 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.12-2 Shows rotor from 1925 without the field poles installed.
5. Torque blocks (not all designs have this) (Figure 2.12-3) – these transmit the
torque from the spider/drum assembly to the rim but do not provide a shrink
interface and are partial length of the rim stack.
Rotor rims have evolved over time from the late 1890s when the rim was
actually an integral part of the cast spider assembly (see Figure 2.12-2) to being
a separate laminated structure interfacing with the spider or drum via rim keys
on a modern day machine as shown in Figures 2.12-1 and 2.12-9. Of course,
depending on how large the machine is in diameter and the speed, integral type
rims are still used today as they do prove economical in the right design
circumstance.
A rim and spider assembly that is integral behaves differently in service than
a rim that has a laminated assembly constructed from steel segments (see
Figure 2.12-8) and separate from the spider or drum assembly. When the integral
rim is constructed the shape is set by the cast process, so the rim itself is circular and
concentric when rotating providing the cast process was done accurately.
The circularity or concentricity does not change when the machine is in serv-
ice since it is part of the spider assembly and really has nowhere to move. The rim
cannot expand independently from the spider on this type of arrangement. Suppos-
ing that the stator is circular as well, then the airgap for the machine remains stable
and consistent all the way around the machine. The reason a more stable and con-
sistent airgap is desirable will be discussed later in the book.
The rim is normally constructed in a separate area of the powerhouse that has
ample room for scaffolding and measuring devices to ensure the circularity, ver-
ticality, and concentricity is maintained in accordance with the design standard
from the manufacturer.
The construction procedure for assembling a rotor rim is proprietary to the
manufacturer but a typical sequence is something like this. To begin, the bottom
rim end plate segments as shown in Figure 2.12-9, or simply steel segments as
shown in Figure 2.12-3, are installed onto the spider or drum assembly all the
2.12 ROTOR RIM 93
Rim support
Torque
shelves
block
Butting steel
segments Torque block
with keys
installed
Large rim
keys go here
Piling
pin
Overlapping
steel segments
way around the circumference and are supported by stationary stands that can be
adjusted for height. The height adjustment is critical in making sure the rim is
erected as level as possible. The rim end plates are typically a very heavy construc-
tion that is much thicker than the steel segment. This is because in some designs,
the rim end plates will be acting as the pressing plates when the bolts are tightened
at the end of the assembly process instead of just the steel segments. Either design
serves the same purpose and is equally effective. Once the initial circle is made,
piling pins and adjustable thickness rim piling keys are installed, and the steel
94 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
segments are piled in a circle in a specific sequence, with overlap, see Figure 2.12-4.
The sequence of overlap is a proprietary feature of the OEM and this is what will
determine the operating behavior of the rim while in service.
The rim piling keys are temporary and are typically a tapered set that are in
place of the permanent rim keys. These temporary keys during construction pro-
vide a couple of important functions. First, they provide circular shape control for
the steel segments when being piled and secondly provide concentric control of the
piled rim stack.
Both of these quantities are extremely important when the final assembly is
complete so that the rim rotates as close to on center with the spider or drum assem-
bly as possible. Further, both of these quantities can be adjusted as the piling is
ongoing (by adjusting the tapered keys) and thus is checked after so many inches
of piled rim is achieved, usually coinciding with the intermittent pressing opera-
tion. Once the rim reaches a specified height, the piling pins (same diameter as
the rim bolts and shown in Figure 2.12-4), are removed and replaced with the
rim bolts as shown in Figure 2.12-5.
Depending on the size of the steel segments, more than one person is
required to place the steel segments over the rim bolts. The tolerance of the rim
bolts to the holes in the steel segments is very tight, thus, sand-filled mallets are
used to strike and move the steel segment down the rim bolts to its final position.
One person in the erection process will be going around the piled rim as each steel
segment is installed and marking it with a grease pen to ensure the next segment
goes in the proper location. After piling one steel segment after another, it can
become confusing where the next segment should be installed in the sequence,
so the person marking with the grease pen is a great way to check the process.
At set points in the piling process, when a certain amount of rim height has been
Figure 2.12-5 Rim being piled showing the stacks of steel laminations and rim bolts finally
in place.
2.12 ROTOR RIM 95
piled, pressing of the stack is done to encourage proper settling and compression
and also serves as a checkpoint for various dimensional controls such as level, ver-
ticality, and stack thickness as shown in Figure 2.12-6. If the level or stack verti-
cality is not within the required tolerances, now is the time to fix it. The more steel
segments that are stacked, the more difficult it will be to adjust the verticality or
level to within specified tolerances. At some critical point, it is no longer to pos-
sible to adjust these quantities and the rim will forever remain with these charac-
teristics. The piling continues until the proper stack height is reached, then, the final
pressing is completed. The proper amount of stack height is critical to ensure there
is sufficient stack height to accommodate the tightening of the core bolts since they
are of fixed length and require so many threads to be engaged once the nut is put on
and torqued to the final value to give the desired rim stack pressure.
For example, if one too few layers are installed, when it comes time to tighten
the nuts on the rim bolts, the nuts will bottom out onto the non threaded portion of
the rim bolt and no more compression can be achieved without adding more steel
segments (if any spares are available) or some other arrangement. Conversely, if
one too many layers are stacked, then there may not be sufficient thread engage-
ment of the rim bolt when the nuts are torqued to the specified value.
What is presented in the erection procedure is only one way of erecting a rim,
there are many different ways this can be done, so consult the OEM for the pro-
cedure that was used for the rim in question.
Now that the rim has been erected, it is appropriate to discuss in detail the
main components of an assembled rim:
• Individual steel segments that together form the laminated rim
• Rim bolts and nuts
96 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
The rim bolts and nuts vary in diameter depending on the size and speed of
the machine but are typically made from a higher strength steel since they are ten-
sioned when the rim is completed. The bolts are machined to a very precise toler-
ance and the steel segment holes are punched or laser cut to a similar tight
tolerance. The typical clearance for the bolt and the hole in the steel segment is
approximately 0.076–0.127 mm (0.003–0.005 ). Piling these steel segments that
can weight 100 lb or more in a special sequence onto the rim bolts to make a lami-
nated structure is no easy task.
As previously mentioned, not all rim designs have a rim end plate as shown
in Figure 2.12-8; the steel segments are used instead. The rim end plates, if so
equipped (Figure 2.12-9), are typically made from medium carbon steel and are
much thicker than the steel segments, but are the same pattern. They vary in thick-
ness depending on the machine diameter, speed, and compression required in the
finished rim assembly. A typical pressure for a rim when compressed is 500 PSI but
can vary depending on the size and speed of the generator.
The rim keys are the critical interface between the rim itself and the rotor
spider or drum assembly. Manufactures will use different grades of steel for rim
keys depending on their rim to spider fit design. The key sits in the keyway that
is punched or laser cut for the laminated steel segment and the keyway that is
machined for the spider arm or drum assembly as shown in Figures 2.12-10 and
2.12-11.
There are many different rim key designs, and we will touch on a few exam-
ples here. In Figure 2.12-10, the rim key is a solid rectangular piece that is driven
Rim bolts
and nuts
No end plate
just steel
segments
Figure 2.12-8 Fully piled rim with no end plate – just steel segments on top.
98 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Rim key
Spider post
Steel segments
Spider hub
Spider post
Rim key
Steel segments
into the keyway when the rim (steel segment) are heated sufficiently to allow the
key to be inserted. When the rim is cooled down to ambient temperature, an inter-
ference fit is established, known as a “shrink” fit. This design also allows for a shim,
a thin piece of steel stock, to make up the difference in interference fit if needed.
2.12 ROTOR RIM 99
Black
reference
line
Figure 2.12-11 Tapered rim keys set in final position by hydraulic jack.
Shims are used when the key itself has not been machined to the exact radial dimen-
sion needed for the final interference fit. It is desirable to have a single piece key to
keep things very simple unless the shim is being used as a liner or shrink dimension
on the rim (steel segment) side in order to provide a smooth surface for the key to
ride on when being inserted. As previously discussed, this interference fit, depend-
ing on the OEM, is maintained even during load rejection speeds.
In Figure 2.12-11, a tapered key system is being used to make up the inter-
ference fit required for the designed shrink to be applied. These keys (one station-
ary and one drive key) are machined with graduated slope in mils/in or μm/m, so
when the key is driven down or pulled up, a radial displacement in terms of key
thickness can be calculated. In this scenario, there are two tapered keys, the station-
ary key on the rim side (just the tip visible in the picture) and the drive key which
has the hydraulic jack underneath to pull the key outwards a little bit to achieve the
correct fit. The drive key is normally installed using a plastic sledge hammer to
push the key inwards for a proper fit. When the rim is sufficiently heated, the drive
key is inserted until the black reference line is the same level as the stationary key.
If the key is driven too far down, since a mallet and human force is being used, a
hydraulic jack can be used to reverse the key insertion slightly. This system is
advantageous and simple as one can calculate how much the drive key must be
adjusted to have the interference fit required based on the taper.
Referring back to Figures 2.12-3 and 2.12-4, for this particular rotor design,
there are rim keys which we have discussed but also a torque keys as well. The
torque keys on this particular design are what transmit the torque from the shaft
to the rest of the drum assembly. These torque keys are not full axial length like
the rim keys; they are smaller in length and do not have an interference fit. The keys
100 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
are driven in with a hammer until refusal and locked into position by welding.
A torque key installation in progress with the final key to be hammered into posi-
tion is shown in Figure 2.12-12.
In some designs, the rim keys take the torque from the spider or drum and
transmit it to the rim, or a combination thereof. The rim keys as previously men-
tioned control the circularity and concentricity of the rim as well as take the shrink
forces (if the rim is a shrunk design) from the rim and transfer it to the spider or
drum assembly. If the rim is a shrunk design, the spider/drum, rim key, and rim are
always in contact even while the machine is at rated speed or even higher such as
load rejection speeds.
If the rim is a floating design or nonshrunk, the rim keys or torque keys are in
place to transmit the torque from the spider/drum to the rim assembly and to maintain
rim-to-spider concentricity. These keys are typically driven to refusal during the
installation process at ambient temperatures. The radial and tangential tolerances
on the key assembly are very tight, in the neighbourhood of (0.050 mm) 0.002 .
The concentricity and circularity is now a function of the rims sole ability to maintain
rigidity and shape with respect to the airgap while in service. Since, in this design, the
spider/drum, rim keys and torque keys, and rim are not in contact when the machine
is at speeds other than stand still, airgap uniformity is of paramount importance. Non-
shrunk rims are typically used when the normal airgap magnetic forces are not suf-
ficiently high to change the circular shape of a sufficiently mechanically stiff
laminated rim. Normal assumes a uniform airgap and, of course, some deviation
from perfection as no airgap is 100% uniform. Should this airgap become compro-
mised to the point where the magnetic forces exceed the rim mechanical stiffness, the
magnetic forces can distort the shape of the rim where it is in line with the critically
small airgap on a once per revolution frequency. The nonshrunk rim will slide on the
spider ledge as it is pulled by the narrow airgap’s high magnetic force and then will
2.12 ROTOR RIM 101
slide back into its original position when the airgap is sufficiently large again. This
type of activity will cause potentially severe fretting on rim support and key compo-
nents leading to mechanical vibration of the stator and rotor assemblies as the rim
loses circularity and concentricity. The fretting of the rim support contact surface
can cause failure of the support structure which represents a serious risk to the safe
operation of the generator. On a shrunk rim design, compromised airgaps can also
cause the rotor assembly as a whole to migrate toward the smallest airgap and can
also cause vibrations depending on the circularity and concentricity of the rim. Since
the rim is shrunk onto the spider/drum assembly during normal operating speeds
(may float during an overspeed event), it will take more force to pull the entire rotor
assembly, which is restrained by the guide bearings, toward the stator as opposed
to the unshrunk design where the rim can move independently of the spider/drum
assembly.
As mentioned previously, there are many different designs for the rim-to-spider/
drum interface, too many to discuss in this book. The important thing to recognize
is if the machine has a floating or shrunk rim design, consultation with the OEM
may be required if there is any uncertainty.
When the rotor rim is first assembled and compressed, the frictional forces
between the steel “lamination” segments in some rim designs is not enough to pre-
vent segmental movement in the radial direction for the first time the machine
achieves an overspeed condition such as a load rejection. This depends on the
machine diameter, speed, lamination thickness, and mass of the rim. The design
of the rim can take into account that the steel segments are allowed to move
and become tight up against the rim bolts, this is also known as “rim slip.” In this
condition, the steel segments move in the radial direction outwards toward the air-
gap and the clearance between the steel bolts and the steel segment holes is taken
up. In other words, the steel segments are butt up against the steel bolts. The cir-
cumference of the rim is now slightly larger than the original thus making the air-
gap slightly smaller. The formula to calculate the amount of radial airgap reduction
once slip occurs is shown in Equation (2.10):
AGAS = AGBFS − LCTRB × RSPC 2π (2.10)
where,
AGAS = Airgap after slip – which is the calculated airgap after slip
AGBFS = Airgap before first spin of the machine – this is the airgap after the
rim has been piled and before the rotor has been spun for the first time –
this can be measured at site during construction or on an OEM drawing
LCTRB = Lamination clearance to rim bolts – the clearance on the rim lam-
ination can be found on the rim segment drawing and the bolt size can also
be found on a drawing or both dimensions can be acquired from the OEM
RSPC = is the number of rim segments per circle – which is the number of
rim segments it takes to complete one circle of the rim. This information
should be on the rim segment drawing or can be counted at site or can be
acquired from the OEM.
102 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Let us take an example of a 140 MVA machine with an airgap of 0.796 (20.21
mm), 12 rim segments per circle with a clearance of 0.005 (0.127 mm) between
the lamination and the rim bolts. Using Equation (2.10) yields the following result:
AGAS = 0 796 − 0 005 × 12 6 283 = 0 786 19 96 mm
Here, we can see that the radial airgap reduction with the machine at standstill
would be 0.010 (0.25 mm). This may seem like a small amount for a machine that
has this large of an airgap, but the smaller the airgap at the beginning, the more this
slip reduction will impact the final airgap. This exercise is simply to make the
reader aware that if rim slip occurs, this is the magnitude of the reduction given
the parameters listed above. Notice that the more the clearance between the rim
lamination and the rim bolts and the more segments per circle, the more rim slip
will affect the final value.
Once this “slip” has occurred, it is not reversible, and the speed at which this
occurs is design-specific. Not all rims slip as described, some OEMs claim their
designs are not subjected to this event, thus a conversation with the OEM may
be in order.
This brings up another question about the airgap when the machine is
assembled versus after the machine has experienced this “rim slip.”
It is now appropriate to discuss the airgap and the different types of airgaps
that may be listed on a drawing, operating manual, or in a proposal from the OEM.
From experience, there are at least three designations for airgap:
Erected airgap – this is the airgap that exists when the machine is first
assembled before it spins for the first time. This value may be present
on an assembly drawing from the OEM as construction is ongoing at site.
Design airgap – this is the airgap that exists after the machine has been spun
for the first time having experienced a load rejection. A load rejection is at
some speed higher, typically between 25 and 40% above rated speed.
This allows the rim steel segments (if they are going to move) to settle
into a final position radially, the concept of rim slip. It is important to rec-
ognize that if the machine reached a runaway speed condition, that is typ-
ically twice rated speed, the rim segments may settle a bit more if the load
rejection speed did not butt the steel laminations against the steel bolts.
The design airgap value is likely on a drawing or operating manual that
is issued to the customer. For the laminated rim designs that do not suffer
rim slip (friction rim) because they have sufficient friction between lami-
nations to remain stable for all operating modes including overspeed, the
erected airgap and design airgap may be the same.
Running airgap – this is the theoretical airgap that is present dynamically
while the machine is in service running at rated load and temperature. The
only way to measure this is with a dynamic airgap monitoring system. This
theoretical calculated value is not normally given to the customer as it is an
internal number for the OEM for design purposes. This airgap will be the
result of centrifugal forces expanding the rim, temperature of the machine,
and the expansion difference between the rotor and stator structures.
2.13 ROTOR SPIDER/DRUM 103
These three airgap types are invaluable when trying to understand drawings,
dynamic operation, and discussions about the machine with the OEM. Discuss
with the OEM which airgap is being referenced on the drawing as each OEM
may have a different interpretation of the definitions as given here.
These components can be some of the most complex and tricky to understand
depending on the design and manufacturer. There are many types of spiders and
drums in service today, and it is not possible to touch on each and every unique
design, instead a more general description of a rotor spider/drum design and con-
struction and its intended function will be presented. For a more in depth under-
standing of a specific design, the reader is encouraged to consult with the
drawings that came with the machine (if any) and the OEM.
The main purpose of the spider/drum is to transfer the torque from the main
shaft to the rotor rim assembly. Many spider/drums made in the early 1900s made
use of cast technology as shown in Figure 2.12-2, and as technology improved they
were made from fabricated steel components that were welded and/or bolted together
as shown in Figure 2.12-1. Some spiders even incorporated both a cast and fabricated
steel design together as shown in Figure 2.13-1. The spider is composed of a hub and
arms that act as a single unit to transmit the mechanical energy to the rim. Spider
arms take on many shapes and sizes depending on how many arms there are, rim
weight that needs to be supported, speed, diameter of the spider, and whether or
not the rim is floating or a shrunk design and the amount of shrink that is applied.
Cast
Fabricated
Machines with cast spiders have the advantage that the spider and hub assembly and
sometimes even the rim (cast as well, no steel segments as shown in Figure 2.12-2)
are all one integral piece. There are no welds to worry about, and if the original cast-
ing was done properly with no inherent defects, there is little that can go wrong if the
machine is operated and maintained within the design nameplate rating.
A fabricated spider, similar to the one shown in Figure 2.13-1 (except it is all
fabricated steel with no cast pieces), have many more considerations in the way
they are designed and constructed. Starting with the hub it can be constructed
as a single cylinder section or many sections welded together to make one cylinder.
The hub, when finally completed, will ultimately have to carry the entire weight of
the spider arms, rotor rim, field poles, and field winding. The hub will have to
endure the high compressive forces of a shrunk rim without collapsing at standstill
as well as the torsional forces when the machine is stopped and started. The welds
that make up the hub assembly will also have to endure a portion if not all of these
forces, depending on the hub design.
The spider arms are attached to the hub by heavy welds. Anytime a weld is
used to attach one piece of metal to another there are inherently going to be stresses
developed in the pieces that are being joined. These stresses must be managed by
stress relieving techniques or by the welding procedure so that stress cracks do not
develop while the machine is in service and undergoing cyclical loading. The spi-
der arms must support the weight of the rotor rim, field poles, and field winding on
a small shelf at the bottom of each arm and transfer this weight to the hub assembly.
The spider arms will also have to endure the high compressive forces from a shrunk
rim design without buckling at standstill. It is very important not to overcompress
the arms during the shrink process as this yielding is irreversible. This activity is
better left to the OEM if there is any uncertainty of the shrink value or procedure to
apply the shrink to the machine.
A drum assembly is typically made up of an upper and lower steel disk separated
and held together with contoured steel webbings or vanes that are welded in place.
Ultimately, the vanes form part of the powerful fan assembly for the rotor which will
circulate air inside the generator. The lower disk couples with the generator shaft and
transfers all or a portion of the torque to the rim via the torque blocks and/or rim keys in
cooperation with the top disk if so designed as shown in Figure 2.13-2. In this design,
the rim keys transfer the torque and absorb the shrink forces.
Depending on the manufacturer, the lower disk may transfer 80% of the tor-
que and the upper disk may transfer 20% of the torque or some other percentage
variation, depending on the design. The lower disk also has the rim shelf that will
accommodate the steel segments of the assembled rim. This rim support shelf area
in contact with the steal segments or bottom end plate varies greatly from machine
to machine. For example, on the drum design as shown in Figure 2.13-3, the area
seems larger than the rim support shelf area at the end of the spider arm as shown in
Figure 2.13-4. On the other hand, looking at the support shelf on the drum design in
Figure 2.12-3, the area seems closer to the spider arm design. It all depends on the
diameter of the rotor, weight of the rim, speed, and if the rim is a shrunk or floating
design. As with the spider design, if the rim is shrunk, the drum will have to endure
2.13 ROTOR SPIDER/DRUM 105
Figure 2.13-2 Shows spider drum design. Source: Courtesy of Dr. Michael Znidarich.
Steel segments
Support shelves
high compressive forces to accommodate the rim shrink. Depending on how large
these forces are, it will drive the designer to re-enforce the drum components to
accommodate the additional loading. In all cases, it is more economical from a spi-
der and drum point of view to have a floating rim since the components do not have
106 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Bottom rim
end plate
Spider arm
support shelf
to be as robust in order to accommodate the additional forces from the rim shrink.
The rim keys on a full floating rim may be driven cold by a sledge or pneumatic
hammer, so there is little compression between the rim and spider/drum assembly
at standstill, and there is no heating of the rim during installation of keys on this
style. Other floating designs may incorporate some rim heating during key instal-
lation to give a rim float at some speed lower than synchronous. It is important to
keep in mind that later in the operating life of the machine, the airgap as previously
mentioned will no longer be as uniform as when first assembled (assuming the
OEM has erected the machine properly) and a floating rim will be more susceptible
to the unbalanced airgap and vibrational issues may arise. Keep in mind that a float-
ing rim is more sensitive to eccentricities in the airgap than a shrunk design. In
Chapter 9 of the book, the consequences of a loose rim on the support shelf struc-
ture will be discussed in detail.
Through bolts
and nuts
Amortisseur
bar (winding)
Punchings
Amortisseur
End plate shorting plate
Figure 2.14-2 Shows parts of field pole being assembled. Source: Courtesy
of Dr. Michael Znidarich.
These components are illustrated in Figure 2.14-2. The pole punching is the main
component of the pole body assembly. It is the piece that holds the copper field
winding from moving out in the radial direction while in operation and houses
the amortisseur bars, ground insulation, and the through bolts which hold the entire
108 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.14-3 Shows finished rotor pole end plate with the “L” shaped portion. Source:
Courtesy of RPR Hydro.
pole body assembly together. The pole punching is typically made of medium car-
bon steel (0.3–0.8% carbon content) that has no insulation except oxidation on
either side of the punching and is typically 1.5 mm (0.060 ) thick.
In some cases, since there are a large number of pieces required to make up
the pole body assemblies for one generator, the pieces are made using a die and
punch press. In other cases, if the quantity required is sufficiently low, laser cutting
may be used to produce the pole lamination, but this is generally a more expensive
process than die and punch, however, in recent years, the price has come down
significantly.
The end plate, as appropriately named, is installed on both ends of the pole
body assembly. The end plate is typically made from forged or cast steel depending
on the speed of the machine and the size of the rotor pole body. The “L” shaped part
of the end plate is under quite a bit of stress from centrifugal forces since it is at the
outer radius of the rotor assembly, see Figure 2.14-3.
Its purpose is to provide a pressing surface for the copper at each end in con-
junction with the through bolts and nuts. The end plates help hold the copper field
winding in place in the radial direction to prevent distortion during operation.
Higher speed machines may have an interpole wedge style brace Vee blocks
to prevent the copper winding from distorting into the space between poles during
operation. Depending on the design and size of the field poles and the
speed the machine achieves during normal operation, load rejection, or even run-
away, the copper winding wants to occupy the interpolar space which is more tan-
gential in direction than radial. The wedge prevents this from happening (see
Figures 2.14-4 and 2.14-5). The wedge design typically has minimal centrifugal
loading by utilizing high-strength aluminum alloys to make them as light as
2.14 ROTOR POLE BODY 109
Cast aluminum
alloy bracket
Holding down
bolt
Epoxy glass
laminate Rectangular slot
insulation punched in rim plates
Figure 2.14-4 Concept of interpolar wedge. Source: Courtesy of Dr. Michael Znidarich.
Figure 2.14-5 Example of an interpolar wedge. Source: Courtesy of Dr. Michael Znidarich.
The through bolts as previously mentioned hold the entire pole body assem-
bly together. They are typically made of a higher strength, more durable steel than
the pole punching since these bolts are under tension when the pole body is com-
pleted. There are a many different ways the through bolts are secured to the pole
body. One modern and common method is to apply nuts on each end of the through
bolt and then tighten and weld the nuts to the end plate.
The construction of the pole body assembly is proprietary to the manufac-
turer but a typical example follows. One end plate and pole punchings are stacked
on a special horizontal table, positioned with a jig to get the proper alignment for
the through bolts and amortisseur bars, and pressed every so many feet of stack
length to ensure proper compression.
Once the final stack length is reached, the stack is compressed to a specified
value to properly seat all the pole punchings before the amortisseur bars are
installed and brazed to the shorting plate. A typical pressure the stack is com-
pressed to is 580 PSI (4 MPA) or higher depending on the design of the pole.
The other end plate is installed and the through bolts are secured either by swaging
or with nuts that are tightened and welded. The pole body is now ready to be insu-
lated and accept the copper field winding.
Laminar insulation is not required since the pole body is experiencing a DC
field while in synchronous operation and not an AC field. The only location where
the pole body will experience an AC field (cross slot leakage flux or tooth ripple
flux) during synchronous operation is at the pole-face since typical hydro generator
airgap dimensions are small relative to the stator slot pitch. With regards to the
cross slot leakage flux, one of the issues a designer must account for when selecting
the size of the generator airgap is the width of the stator slots. During operation,
there is slot-to-slot leakage flux in the stator and it is important that this leakage
flux linkage with surface of the rotor field poles be kept to a minimum so as
not to cause additional and potentially excessive pole-face losses (this will all
depend how close the pole-face gets to the cross slot leakage flux). The previous
statement goes back to ensuring the airgap on the machine remains as close as pos-
sible to design parameters. In older and smaller MVA machines where the design
airgap is very small, the stator will have a closed slot or nearly closed slot to keep
this interaction to a minimum.
The amortisseur winding and shorting plate are described later in this chapter
and the reader is referred to Section 2.16.
coils are made with large rectangular continuous copper strap which is edge bent as
shown in Figure 2.15-1. In this manufacturing process, long pieces of copper strap
are coiled using a special machine. When the machine gets to the end and needs to
make a turn, the copper is bent around the edges thus giving the profile in the figure
(thicker at the small radius against the pole body and thinner at the outer radius).
Brazed joints in the copper coil are made along the straight portion if needed when
the long piece of copper strap runs out. The coil shown in Figure 2.15-1 may only
have two or three brazed joints in the entire assembly along the straight portion.
The only real consequence to edge bent is the thinning of the turns at the ends
of each coil along the outer edges as shown in Figure 2.15-2. The effect shown
in this figure is an extreme case of what it can look like, most edge bending results
in some variation less that what is presented. The other method is to braze pieces of
copper strap (butt or interlocking joint) to form the coil as shown in Figures 2.15-3
and 2.15-4. In this example, the interlocking joint is first subjected to a very high
clamping pressure to put the pieces together and then brazed. The difference is the
fabricated joint is more uniform in coil thickness at the ends as shown but more
brazed joints are required. A butt braze (copper segments are brazed perpendicular
to each other to form the coil) has similar features of uniformity of coil thickness
and more brazed joints like the interlocking. Experience has shown that all meth-
ods are equally effective, and there should be no discernable difference in reliabil-
ity provided the manufacturing has been done correctly. During a reinsulation is a
good time to check the copper sections for any type of cracking or deformity. The
brazing process can be done using a torch or automated process using special
machinery. In either case, a common brazing material called Silfos® which is a
Thicker
copper
here
Thinner
Edge
copper
bent portion of
here
copper strap
Figure 2.15-1 Copper strap edge bent coils.
112 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Jigsaw
joint
Finned turns
for cooling
Jigsaw
joint
Figure 2.15-3 Assembled copper pieces at the jigsaw (interlocking) joint similar to a
puzzle. Source: Courtesy of Dr. Michael Znidarich.
copper alloy containing copper, silver, and phosphorous can be used to fuse the
copper segments together under high heat.
Strap wound coils need turn insulation inserted between the copper segments
prior to the consolidation of the coils. Some older machines, pre-1970s had turn
insulation consisting of an asbestos paper in sheet or tape form bonded with shellac
2.15 ROTOR WINDING AND INSULATION 113
Uniform copper
thickness
Brazed jigsaw
(interlocking)
joint
Figure 2.15-4 Brazed jig saw (interlocking) sections and uniform copper. Source:
Courtesy of RPR Hydro.
and heat press cured. The asbestos paper was thicker (more like a sponge consist-
ency) and provided a nice filler for the edge bent uniformity problem, hence the
gaps as shown in Figure 2.15-2.
The insulation shown in this figure is a thin Nomex®, so the copper thinning
is much more evident. If desired, these gaps are easily mitigated with an insulating
filler to prevent ingress of moisture and contaminants as well as for esthetic pur-
poses. Adding more insulation between turns is not economical nor will it likely
completely eliminate the gaps at the edges of the coils.
As mentioned, modern strap wound coil designs use Nomex® or similar
materials since their thermal, mechanical, and electrical properties are more than
adequate for this application. Thin strips of this insulating material are placed
between turns and epoxy is used to consolidate everything in place under a heat
cure process. Since there are large rotational mechanical forces acting on the coils,
114 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
particularly during overspeed and runaway conditions, it is essential that they are
well bonded into a rigid structure.
The ground insulation is between the copper coil and the pole body assembly
and can be composed of many materials depending on the vintage of the machine.
Some such materials are asbestos, epoxy-bonded glass fibers, micafolium, and
Nomex® [8]. This insulation can be applied in several different ways depending
on the manufacturer and vintage of the machine. In the early days of field pole man-
ufacturing, the ground insulation was applied to the copper winding along the inside
of the coil and then placed over the pole body assembly which may or may not have
had a collar around the pole body. The other more traditional way is to wrap the pole
body material with ground insulation such as an appropriate grade of Nomex® or
equivalent and installing an insulating collar made from synthetic resin-bonded
glass or Daglas® backed mica to insulate between the copper winding and the pole
body tips as well as increase leakage distance from the copper to the pole body tips
[8]. Figure 2.15-5 shows the bottom collar installed during the field pole refurbish-
ment. These collars are preferably made in once piece, but for larger poles, a pinned
lapped joint may be necessary. This insulating collar is normally sealed to the pole
body insulation with silicone to isolate the copper winding from the grounded pole
and prevent contamination ingress.
Once the coil assembly is placed onto the pole body, it is desirable that the
coil does not move while the machine is in service as this would cause abrasion of
the pole body insulation. To prevent this, the manufacturer would secure the coil
onto the pole body using wedges and a gluing compound to secure the wedges in
place as shown in Figure 2.15-6. It is important to realize here that bonding of the
Pole body
Nomex
groundwall
insulation
Silicone
sealant
Bottom
insulating
collar
coil to the pole body assembly is done to varying degrees depending on the coil-to-
coil connections used on the field winding. If the coil-to-coil connection is a bolted
and soldered connection (consolidated with no flexibility), then the bonding to the
pole body must not be so rigid, allowing for a finite amount of flexibility as the
connection is stressed in operation. If the connection is the flexible type, then the
manufacturer may choose to more solidly bond the copper to the pole body by using
more wedges and even fill the void between the copper and pole body with epoxy.
Consult the manufacturer whenever a reinsulation of the pole body assembly is
required to determine which system best suits the machine. Finally, a top collar
(the one closest to the rim) may also be used to ensure adequate creepage distance
to the pole body/rim assembly, but not all designs incorporate the top collar. Silicone
is used to seal the gap between the coil collar and pole body assembly to prevent
contamination ingress, see completed pole in Figure 2.15-7.
The outside surfaces of the copper coil may be left bare to provide the most
efficient cooling. Sometimes the copper is painted with an insulating paint for pro-
tection and ease of cleaning. Lastly, some designs of copper strap coil utilize a
high-low approach or finned turns in order to improve heat transfer into the venti-
lating air (see Figure 2.15-3).
116 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.15-7 Finished field pole with red insulating varnish applied.
Some very large machine designs have springs inside the rim assembly that
push against the inner pole collar to keep it radially in place.
On some generators, the rotor coils are designed to carry the current in one
direction on one pole, and in the other direction on the next pole. These are called
open or crossed coils. On other generators, all the rotor coils are wound in the same
direction and special connections are installed on top or bottom of the rotor rim to
carry the current in one direction in one pole and in the other direction in the
next pole.
Most rotor field poles employ a damper (also called amortisseur or damping)
winding to dampen torsional oscillations and provide a path for induced currents
to flow. The amortisseur winding is essentially a separate winding installed under
the face of the pole body that is connected in a way similar to the squirrel-cage of an
induction motor. The winding is typically made from a tough pitch copper, brass,
Everdur (copper 95%, silicon 4%, manganese 1%), aluminum, or iron. The amor-
tisseur winding is typically buried in the pole body steel and is not always visible
over its length (see Figures 2.16-1 and 2.16-2). It produces an opposing torque
when currents flow in it and this helps dampen torsional oscillations and add to
the stability of the rotor during system excursions from normal operating condi-
tions. Negative sequence currents in the stator winding will also cause the amor-
tisseur bar to be active. The limit for negative sequence currents while in operation
2.16 AMORTISSEUR WINDING 117
Figure 2.16-2 Pole under construction with the amortisseur shorting bar shown –
amortisseur bar (not installed yet) will not be visible under pole punching when installed,
see also Figure 2.14-2 for where the amortisseur bars reside.
is 5% of the rated current for nonconnected and 10% amortisseur windings with
interpole connections, as outlined in the Ref. [5].
On most machines, the amortisseur winding segments are only connected to
each other in one pole in a noninterconnected fashion. In special machines, such as
118 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
a pump generator, the amortisseur winding segments are interconnected with the
adjacent pole, forming a “squirrel cage.” This is due to the large starting currents
the amortisseur winding will be exposed to when the rotor is started from a stand-
still by applying system voltage to the stator, when a starting motor is not used.
Some manufacturers have a design philosophy where the amortisseur is intercon-
nected in all machines that are designed due to better damping performance (sub-
transient reactance X d ) for little cost difference.
During manufacture, the amortisseur bars are installed into the pole body
such that there is intermittent contact between the pole body and the amortisseur
bar itself through the entire length. This helps ensure that the bar does not come
loose while in service.
The bars are then joined at the ends of each pole by brazing them together
using a single copper plate or multiple layers of copper plate as shown in
Figure 2.16-3. These plate assemblies offer some flexibility since when the bars
are active in service, the current flowing through each bar may not be equal and
thus axial thermal expansion may be different from bar to bar. If the brazed con-
nection in combination with the copper plate(s) does not offer the correct amount of
flexibility, cracking of the brazed connection or plate(s) will result affecting the
performance of the amortisseur circuit [7]. It is very important that proper periodic
inspections are done on these connections to ensure their integrity. Figure 2.16-1 is
an example of a less flexible arrangement for axial expansion for the amortisseur
connections.
There are many different connections that can be used to connect one set of
amortisseur bars to the adjacent set on the next pole. A flexible connection is nor-
mally used in order to accommodate expansion and contraction between shorting
bars from operating stresses. Typical connections may include leaf copper, solid
omega shaped, and flexible braid style as shown in Figure 2.16-3. These flexible
connections may be solidly bolted, bolted and brazed, or just brazed depending
on the manufacturer. Other solid designs as shown in Figure 2.16-4 can be solidly
bolted, bolted and brazed or just brazed as well. Inspection should include ensuring
the bolts have not worked themselves loose and that there is no cracking or fraying
anywhere on the connector. Signs of overheating on the connector could be an indi-
cator of a poor or a high resistance connection. It is generally understood that the
flexible braided connections can be more prone to filament breakage due to thermal
and stop/start cycling and centrifugal forces at operating speeds. Make sure that this
type of connection has been authorized by the OEM and that the OEM has done an
analysis of the conditions in which this connection must function.
EXC
3~
Figure 2.17-2 Traditional rotating brushless exciter with diodes. Source: Courtesy
of Voith.
collector rings as shown in Figures 2.17-3 and 2.17-4. For the slip/collector ring
type of current delivery system, the rings are supported by an insulating block
which is generally made of epoxy glass or some other insulating system. The rings
are typically made of mild steel, but other materials such as brass and copper alloys
were also used. Each ring is opposite in polarity to the other as one conducts current
into the rotor winding and the other collector ring brings it back out.
The current transfer to the rings takes place using a sliding contact surface by
carbon-loaded brushes that slide along the rotating surface of the rings as the rotor
spins. The brushes in more modern systems utilize a constant pressure spring to
maintain a consistent pressure against the ring surface during operation as shown
2.17 SLIP/COLLECTOR RINGS AND BRUSH GEAR 121
Brush and
brush
holder Constant
pressure spring
Slip ring
Adjustable pressure
spring using box
rachet
Figure 2.17-4 Slipring assembly with old style adjustable spring.
in Figure 2.17-3. In older designs, the spring pressure must be adjusted every so
often since the spring pressure is dependent on the setting on the brush box as
shown in Figure 2.17-4. This type of system requires much more observation
and adjustment than the constant pressure type. Good contact is difficult to achieve
if the surface of the rings and brushes is not properly prepared when installed.
122 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
The friction between the ring and brush surfaces and the I2R across the brush-
collector contact resistance generates heat. To theoretically maintain reasonable
current sharing between multiple brushes on some rings there are helical grooves
cut into the ring surface to wipe the brush surface in operation. The rings them-
selves may be machined with a slight radial offset or the rings themselves may
be mounted slightly off-center so as to move the brush in and out slightly to ensure
the dust from the brush does not jam the brush in the holder. It is very important that
the brushes move freely and unobstructed inside the brush box assembly. Failure to
maintain this freedom of movement can lead to many problems with the sliprings
including unequal loading of brushes and excessive sparking and eventually a ring/
brush gear failure. More discussion on the brush boxes, brushes, clearances to the
ring, and so forth, later in the book.
Collector
enclosure
filter
Stator winding
and connections
Upper bracket
Station
heating
Rotor duct
coil
Air baffle
Rotor Rotor Stator
spider rim frame
Collector
support Rotor
pole
Stator
core
Lower air
deflector
Lower bearing
Generator bracket EL. 283 FT. 0.00 IN.
shaft
Pit diaphragm
Shaft Bracket
seal soleplate
(CMM) of air needs to be circulating in order to meet the temperature rise guaran-
tee. It comes down to how close to the maximum total temperature of the machine
the manufacturer wants to operate to. The way in which the airflow is directed
through the machine should never be adjusted or modified unless the manufacturer
is consulted. Changing the airflow in one area may give rise to an airflow defi-
ciency in another. Conversely, if it is evident that the machine is not being cooled
uniformly, preferably during the heat run tests during commissioning (while the
machine is still under warranty), the manufacturer should be consulted as soon
as possible.
Both systems are efficient provided the air that is being circulated within the
machine is kept clean and free from insects and other airborne contaminants
124 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
The purpose of the rotor fan or blower is to circulate or draw air into a main central
cavity and then distribute that air throughout the machine. There are two main
types of rotor fans in a vertical hydro generator, a rotor centrifugal fan or a rotor
axial fan. Pictures of each type of rotor fan are shown in Figures 2.19-1 and 2.19-2.
The ventilation diagram for the rotor centrifugal fan is shown in Figure 2.18-1.
The rotor centrifugal fan is situated in the center of the rotor assembly with
large vanes that pump air from the center of the generator in through the rim, into
the airgap and through the core. The ends of the airgap on the rotor are sealed with
covers preventing air escape. In this design, the rim has airflowing through it and is
designed accordingly and is known as a ventilated rim.
The rotor axial fan is situated on the top and bottom end of the rotor where
the airgap is. The fan forces air into the airgap from the top and bottom of the rotor
pressuring this area and thus forcing air through the core. In this case, the rim itself
does not have the ability to allow air passage, so it is known as a nonventilated rim.
The air pumping loss component in the generator affecting overall windage
loss is directly proportion to the volume of air pumped. In more recent hydro gen-
erator designs, the amount of air being pumped with modern ventilation shrouds/
baffling is in the range of 55–75 CFM/kW (1.55–2.1 m3/min/kW) loss within the
Axial fans
generator ventilation circuit, that is, not including bearing friction losses that are
extracted from the unit by other means. Older open ventilated units, with less
air baffling and large core ventilation ducts, could have windage loss of around
120 CFM/kW (3.39 m3/min/kW).
Extra laminations
The torque on the rotor as seen at the rotor surface in the airgap is as follows
in Equation (2.12):
Airgap torque (at rotor surface),
BAπDr L
T= N−m (2.12)
2
where,
Flux density,
RPϕ
B= Wb m2
πDr L
Electric loading,
N ph N c I a
A= A m
πDr
where
P = number of poles
R = ratio of the pole-face width to the pole pitch
Φ = flux per pole in Webers
Dr = rotor diameter in meters
L = core iron effective length in meters
Nph = number of phases
Nc = number stator conductors per phase
Ia = stator phase current in amperes.
The above formula is an elegant result derived from basic physical principles.
However, the typical user of a hydro generator can make use of the following very
simple expression for the shaft torque as shown in Equation (2.13):
MW output
Torque~ (2.13)
Speed × Efficiency
The efficiency of a typical hydro generator is above 98%; thus, an approximate and
conservative simplification of the equation yields:
MW output
Torque~
Speed
Using SI units:
kW output × 60 000
Torque = , Nm
2πRPM
This torque on the rotor shaft can be significant and creates torsional forging stres-
ses all along its length. Torsional stresses are basically shear stresses in the rotor
shaft due to the twist in the shaft that is created by the action of the rotor’s magnetic
coupling to the stator magnetic field, as opposed to the opposite force imposed by
the water flow to the turbine. Increasing the water flow to the turbine causes the
rotor load angle to increase and, hence, MW load to increase, and produces
increased mechanical torque in the shaft. The magnetic coupling between the rotor
and stator is what inhibits the rotor from running away and keeps the turbine and
generator system in synchronous equilibrium. Increasing or decreasing the rotor
magnetic field causes the load angle to increase or decrease, but does not actually
change the torque applied, only the angle of the torque, or, in electrical terms, the
power factor and reactive power output of the machine.
Under some electrical fault conditions, the airgap torque can be significantly
higher than the rated torque. It is not unusual to have faulty synchronization of the
unit onto the grid (phase angle difference between the generator and the system
grid) being 10 times the rated torque. During this event, the rotor shaft can be
exposed to additional torque loading that is dependent on the ratio of the generator
rotational inertia to that of the rotational inertia of the turbine runner.
2.21.1 Introduction
All generators require bearings to operate with minimal friction and vibration. For
a vertical hydro generator, there are a several bearings that are part of the design
and variations of these bearings can be found depending on the manufacturer. The
types of bearings are, the thrust bearing, thrust bearing/guide bearing combination,
and additional guide bearings along the shaft length if required particularly for
long shaft lengths where extra support is required, both styles are shown in
Figures 2.21-1 and 2.21-2.
The upper and lower bracket, and the thrust and guide bearings, make up the
generator load bearing structure. Bearing structures support axial and radial loads.
For vertical generators, the axial load consists of the weight of the rotating com-
ponents and of the hydraulic thrust. The hydraulic thrust is exerted by the water
flow through the water passage and the turbine. The hydraulic thrust is transferred
to the headcover, spiral casing, and along the shaft to the generator thrust bearing.
The generator and turbine guide bearings manage the radial forces exerted from
water transients and flow disturbances occurring in rough zones of operation, stops
and starts, along with load rejections [9]. As well, any mass and magnetic imbal-
ances that exist on the rotor are transferred to the powerhouse foundation through
the guide bearings and their support structures [8].
2.21 THRUST AND GUIDE BEARINGS 129
Overspeed device
(or P.M.G. for governor control
not supplied with generator)
Pilot exciter
Main exciter
Exciter inspection platform
Upper oil reservoir
Air housing
Cooler
section
Field coil
Rotor spider
Laminated
rotor rim
Stator sole
plate
Foundation
bolt
Combined
brakes and
jacks
Figure 2.21-1 Shows a cross section of a conventional two guide bearing generator.
Source: Courtesy of Voith.
Overspeed device
(or P.M.G. for governor control
not supplied with generator)
Air housing
Pilot exciter
Stator Upper
Main exciter Stator coil laminations bracket
Cooler
section
Rotor
spider
Field coil
Laminated
rotor rim
Stator sole
plate
Combined guide and thrust Foundation
bearing assembly bolt
Combined brakes
and jacks
Oil reservoir Lower bracket
sole plate
Lower bracket
Bearing cooling coil
Shaft and coupling flange Compression tube
jack screws
Figure 2.21-2 Shows a cross section of an umbrella style generator with one guide
bearing. Source: Courtesy of Voith.
130 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
There are a couple of different materials that can be used for thrust bearings,
Babbitt or Teflon™. Babbitt was invented back in 1839 by Isaac Babbitt who was a
goldsmith at the time and contains multiple alloys [10]. Polytetrafluoroethylene
(PTFE) is a synthetic fluoropolymer of tetrafluoroethylene that has numerous
applications. The best known brand name of PTFE-based formulas is Teflon™
by The Chemours Company. The Chemours Company is a 2015 spin-off of
DuPont Co., which discovered the compound in 1938 [11]. Because of the relative
softness of pure Teflon™ many PTFE-based bearing liners used for higher speed
and highly loaded thrust bearings today have a mixture of Teflon™ and carbon
(carbon-filled PTFE) to provide better dimensional shape stability under high
temperature exposure [8].
Radial loads on vertical generators could be the result of poor static balance
of the generator rotor, incorrect shaft alignment, an ambiguous hydraulic force due
to transient behavior, or a magnetic unbalance of the rotor caused by a compro-
mised airgap in the generator. Compromised airgaps can be the result of poor
circularity or concentricity of the rotor or stator or both. All of these loads are
supported by radial guide bearings housed within their respective brackets and
transfer the load to the generator foundation [9].
Theoretically, in the case of radial forces, the radial loads on the guide bear-
ings are minimal if not negligible. Within the normal operational sequence, guide
bearings are indeed exposed to constant fluctuating radial loads that are minimal in
nature. However, in runaway mode or in the case of a load rejection, these loads are
transferred to the two (two bearings is common) generator guide bearings. The
third guide bearing (turbine guide) in this case is not taken into consideration when
calculating the load support. In the case that generator has only one guide bearing,
the turbine guide bearing is then taken into consideration for the load support cal-
culation [9]. In modern hydro-generator design, the ultimate design load is that
which would result from a short circuit of the rotor poles [8].
Grooves and
hole for oil
Figure 2.21-3 Shows Babbitt bearing pie pieces or “shoes.” Source: Courtesy of Ryan
Gillespie of Ontario Power Generation.
Shoes
Load cells
Figure 2.21-5 Shows load cells supporting the thrust bearing shoes.
Hydrodynamic oil
wedge
Pivot Stationary
point bearing shoe
Centerline
of shoe
Resultant
Oil
pressure
Weight
Figure 2.21-7 Sketch showing oil wedge principle for a pivoting shoe on jack screw
(pivot). Source: Courtesy of Voith.
Load
Pressure
Rotating surface
Oil film
Segment
Spring support
Figure 2.21-8 Oil pressure distribution between the runner plate for a spring supported
thrust bearing design.
Figure 2.21-9 Shows runner plate that sits on top of thrust bearing shoes. Source: Courtesy
of Ryan Gillespie of Ontario Power Generation.
136 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
using Babbitt may also employ an oil lift system used during routine start-ups and
shut-downs as well as after prolonged shutdown periods. The oil lift system works
by pumping high pressure oil between the mating surfaces introducing a thin film
of oil which will then produce the “hydro dynamic wedge” and properly lubricate
the surfaces. This is achieved by having circular grooves and holes strategically
placed on each shoe as shown in Figure 2.21-3. When the machine is slowing down
or shut down, the oil between the mating surfaces is slowly being squeezed out.
There will be a point in time during and/or after shutdown when enough oil is
squeezed out that there will not be enough oil to initially prelubricate the bearing
surfaces during start-up rotation. In this case, insufficient oil is in between the sur-
faces and will quickly overheat and the bearing will wipe. The manufacturer will be
able to determine what timing after shutdown will require an oil lift system to be
activated.
Another thrust bearing design that is more simplified employs special spring
assemblies to form a bed for each shoe as shown in Figures 2.21-10 and 2.21-11.
Using compressible springs as the segment support eliminates the need for high
tolerances of the spring assemblies and for perfect alignment of the shaft system
components as the springs act as a pressure relief should the loading or the part
condition not be as initially assembled [8]. This allows the bed of springs to support
each shoe equally around the ring. If Babbitt is used an oil lift system would be
used in this design as well.
Cooling of the thrust bearing oil can be done in several ways. The most tra-
ditional method is to have a cooling coil inside the oil reservoir as shown in Fig-
ures 2.21-12 and 2.21-13.
In the case of Figure 2.21-13, the thrust bearing is on the inside of the cooler
tubes shown. There are many variations of cooling tubes depending on the water
conditions present such as the degree of silt in the water, micro-organisms that
attach themselves to the cooler tubes, etc. Stainless steel or other alloys may also
be used to make the tubes and may be combined with the nickel.
Figure 2.21-12 Shows the copper cooling coil at the bottom of the oil reservoir.
Water is circulated inside the cooling coil and the heat is exchanged through
the tubes. The oil is self-pumping in these enclosed cooling designs so the oil mixes
inside the oil pot to avoid stratification of oil at different temperatures. Fins may be
added to increase the efficiency of the cooling coil. Another method is to pump the
oil to an external cooler and cycle it back into the oil pot using external electric
pumps. The oil reservoir size depends on the machine design and the specifications
required by the customer. Considerations such as available water temperature,
speed of the machine (bearing losses), available space for the thrust bearing (size
restraints), and sustained time for runaway speed conditions all play a role in the
amount of oil in the reservoir and cooling capacity the bearing must have. The seal
for the oil reservoir is discussed in Chapter 3.
138 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.21-13 Shows thrust bearing cooler (extruded aluminum fin CuNi) tubes inside
the oil reservoir.
Journal bearing
Rotation
Journal
Oil
Adhesion
Oil wedge
Figure 2.21-14 Demonstrates the principle of the hydrodynamic wedge in the guide
bearing. Source: Courtesy of Kingsbury Inc.
as shown in Figure 2.21-14. This action sets up the pressure in the oil film which
thereby supports the load. This wedge-shaped film was shown by Reynolds to be
the absolutely essential feature of effective journal lubrication [13].
The guide bearings are there to make sure the entire shaft and rotor assembly
once aligned, remains in that position. The guide bearings are typically made from
segmented Babbitt metal as shown in Figure 2.21-15 and are lubricated and cooled
using oil.
Some guide bearings are part of the thrust bearing assembly and share the
same oil reservoir as shown in Figure 2.21-16.
Figure 2.21-16 Guide bearing shoe in the thrust/guide bearing combination assembly
(segmental type).
where,
D = guide bearing internal diameter
d = shaft bearing journal diameter
The absolute gap can be defined as the need for handling small variations of the
shaft journal positioning within the bearing.
2.21 THRUST AND GUIDE BEARINGS 141
Bearings with hydrodynamic oil film lubrication have values of the Relative
Gap defined as follows:
The quoted minimum and maximum respective values are for power transmission
shafts [9].
D−d
φ= re − arranging yields
d
Z = φd (2.16)
Substituting values of φ for the minimum and maximum (0.0003
and 0.0005):
Z = 0 0003 × 1295 4 = 0 388 mm 0 0153 − Minimum absolute gap calculated
for 51 diameter hydrodynamic lubricated bearing
Figure 2.21-17 Shows a Babbitt bearing wipe. Source: Courtesy of Ryan Gillespie of
Ontario Power Generation.
2.21 THRUST AND GUIDE BEARINGS 143
Figure 2.21-18 Shows another bearing wipe. Source: Courtesy of Ryan Gillespie of
Ontario Power Generation.
Figure 2.21-19 Shows what is left of a PTFE bearing when shoe removed and spring bed
exposed. Source: Courtesy of Ryan Gillespie of Ontario Power Generation.
2.21.6.6 Abrasion
A bearing surface exhibiting circumferential scratches is the result of abrasion
damage. Abrasion is caused by hard debris, which is larger than the film thickness,
passing through the oil film. The debris may embed itself in the soft Babbitt, exhi-
biting a short arc on the shoe surface, ending at the point where the debris becomes
embedded as shown in Figure 2.21-20. Depending on the debris size, the scratch
may continue across the entire shoe surface. Abrasion damage becomes worse with
Figure 2.21-20 Bearing shoe with surface abrasion. Source: Courtesy of Kingsbury Inc.
146 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
time. Surface scratches allow an escape for lubricating oil in the oil wedge, decreas-
ing the film thickness. This will eventually lead to bearing wipe.
Another source of abrasion damage is a rough journal, collar, or runner sur-
face. Roughness may be due to previous abrasion damage. It may also be from rust
formed after extended periods of down time. New bearings should not be installed
when the rotating component is visibly damaged.
In order to eliminate abrasion damage, the lubricating oil must be filtered. If
the oil cannot be filtered or has degraded, it should be replaced. It is important to
evaluate the filtering system, since the problem may be an incorrectly sized filter.
The filter should only pass debris smaller in size than the predicted bearing min-
imum film thickness. In addition to filtering/replacing the oil, the entire bearing
assembly, oil reservoir and piping should be flushed and cleaned [13]. It should
be noted that even oil filtering cannot extract some of the “heavier” debris that
may rest on the floor of the oil pot. For this reason, if significant Babbitt abrasion
is encountered, the oil should be drained from the pot and a manual clean-up of the
reservoir and the bearing parts should be completed [8].
Figure 2.21-21 Bearing shoe with tin oxide damage. Source: Courtesy of Kingsbury Inc.
2.21 THRUST AND GUIDE BEARINGS 147
A reduction in oil temperature may also discourage the formation of tin oxide. In
addition to replacing the oil, the entire bearing assembly, oil reservoir, and piping
should be flushed and cleaned with mineral spirits [13].
2.21.6.8 Overheating
Overheating damage may represent itself in many ways, such as Babbitt discolor-
ation as shown in Figure 2.21-22, cracking, wiping, or deformation. Repeated
cycles of heating may produce thermal ratcheting as shown in Figure 2.21-23,
a type of surface deformation that occurs in anisotropic materials.
Figure 2.21-22 Bearing that exhibits overheating. Source: Courtesy of Kingsbury Inc.
Figure 2.21-25 Show exhibiting outer edge Babbitt erosion. Source: Courtesy of
Kingsbury Inc.
no oil coming in at the leading edge of the shoe, this area typically shuts down,
resulting in the Babbitt being eliminated in the corner near the outer diameter as
shown in Figure 2.21-25 [13].
2.21.6.11 Fatigue
Fatigue damage, as shown in Figures 2.21-27–2.21-29, may represent itself as
intergranular or hairline cracks in the Babbitt. The cracks may appear to open
in the direction of rotation. Pieces of Babbitt may spall out or appear to be pulled
away in the direction of rotation. The cracks extend toward the Babbitt bond line,
150 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
and may reveal the shoe backing. A combination of causes may contribute to
fatigue damage, but concentrated cyclic loading is usually involved. The fatigue
mechanism involves repeated bending or flexing of the bearing, and damage
occurs more rapidly with poor bonding. As well, fatigue is more prevalent in bear-
ing designs where the Babbitt thickness is larger [8]. It is important to note that
fatigue damage will occur without poor bonding. Fatigue can occur when condi-
tions produce concentrated cyclic loads, such as
• Misalignment
• Journal eccentricity
• Imbalance
• Vibration
• Thermal Cycling
• Bent Shaft [13]
2.21.6.12 Cavitation
Cavitation damage appears as discreet irregularly shaped Babbitt voids which may
or may not extend to the bond line as shown in Figure 2.21-30. It may also appear
as localized Babbitt erosion. The location of the damage is important in
2.21 THRUST AND GUIDE BEARINGS 151
Figure 2.21-27 Edge load pivoted show showing Babbitt mechanical fatigue.
Source: Courtesy of Kingsbury Inc.
Figure 2.21-28 Edge load journal shell with mechanical fatigue. Source: Courtesy of
Kingsbury Inc.
152 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
Figure 2.21-29 Shoe segment showing fatigue. Source: Courtesy of Kingsbury Inc.
determining the trouble source. Often called cavitation erosion, cavitation damage
is caused by the formation and implosion of vapor bubbles in areas of rapid pres-
sure change. Damage often occurs at the outside diameter of thrust bearings due to
the existence of higher velocities. This type of damage can also affect stationary
machine components in close proximity to the rotor. Based on its source, cavitation
can be eliminated in a number of ways:
2.21 THRUST AND GUIDE BEARINGS 153
2.21.6.13 Oil
A quick visual examination of the oil or oil filter may be all that is required to deter-
mine that a problem exists and that further investigation is necessary. Cloudy or
discolored oil indicates that a problem exists. A thorough oil analysis can provide
very useful data to assist in diagnosing bearing or machine distress. Be aware that
the usefulness of the analysis is directly related to the information you request. As a
minimum, the following should be supplied:
• Particulate density
• Particulate breakdown
• Viscosity
• Water contamination
• Chemical breakdown
The amount of particulate, as well as its content, can identify potential trouble
spots. Oil viscosity will decrease in time, and whether or not distress is suspected,
it should be periodically evaluated. Water contamination is extremely unwanted,
since it can cause rust and oil foaming, and if it is drawn into the oil film, bearing
failure. A chemical breakdown of the oil will help to determine the integrity of
additive packages and the presence of unwanted contaminants [13].
154 CHAPTER 2 GENERATOR DESIGN AND CONSTRUCTION
2.22 REFERENCES
Culbert, I.M., Dhirani, H. and Stone, G.C. (1988). Handbook to Assess Rotating Machine
Insulation Condition, Vol. 16, EPRI.