IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 33, NO.

2, MARCH/APRIL 1997

577

System Electrical Parameters and

Their Effects on Bearing Currents
Doyle Busse, Jay Erdman, Member, IEEE, Russel J. Kerkman, Senior Member, IEEE,
Dave Schlegel, Student Member, IEEE, and Gary Skibinski, Member, IEEE

Abstract This paper examines ac motor shaft voltages and

resulting bearing currents when operated under pulse width modulation (PWM) voltage source inverters. The paper reviews the
electrical characteristics of bearings and motors that cause shaft
voltages and bearing currents. A brief review of previous work
is presented, including a system model for electrical analysis of
bearing currents. Relying on the work of a companion paper, the
propensity for electric discharge machining (EDM) is determined
by a design equation that is a function of system components.
Pertinent machine parameters and their formulas are presented
and values calculated for machines from 5 to 1000 hp. The effects
of system elements on shaft voltages and bearing currents are
evaluated experimentally and the results compared to theory.
Finally, the paper will present quantitative results for one solution
to the shaft voltage and bearing current problem.
Index TermsBearing capacitance (Cb ), bearing current (Ib ),
bearing threshold voltage (Vth ), bearing voltage ratio (BVR), electric discharge machining (EDM), electrostatic shielded induction
motor (ESIM), rotor shaft to ground voltage (Vrg ), stator to rotor
capacitance (Csr ).

I. INTRODUCTION

RIVE systems engineers typically concern themselves

with the distribution of developed motor torque. An
analysis of mechanical components (e.g., motor bearings)
seldom is of interest. However, the presence of insulated
gate bipolar transistors (IGBT) and higher carrier frequencies
require the design engineer to be aware of the effects of
pulsewidth modulation (PWM) waveforms on the system
mechanical components.
Recently, investigators observed the existence of significant
shaft voltages induced by PWM voltage source inverters.
The values exceed those associated with magnetic dissymetries reported on by Alger and others over three quarters of
a century ago [1]. The effect these voltages can have on
the bearing race surfaces is shown in Fig. 1 [2]. With the
continuing increase in bearing life through improvements in
mechanical design and lubrication, the fluting of Fig. 1 is
troubling because recent bearing failures have shown to be
Paper IPCSD 9654, approved by the Industrial Drives Committee of the
IEEE Industry Applications Society for presentation at the 1996 IEEE Applied
Power Electronics Conference and Exposition-APEC96, San Jose, CA, March
37. Manuscript released for publication October 7, 1996.
D. Busse, R. J. Kerkman, D. Schlegel, and G. Skibinski are with Rockwell
AutomationAllen Bradley Company, Inc., Mequon, WI 53092 USA.
J. Erdman, retired, was with Rockwell AutomationAllen Bradley Company, Inc., Mequon, WI 53092 USA. He is now at W. 175 S. 7121 Lake
Drive, Muskego, WI 53150 USA.
Publisher Item Identifier S 0093-9994(97)01541-7.

Fig. 1. Surface roughness of a ball bearing race due to electrical fluting [2].

the result of electric discharge machining (EDM); voltage

breakdown of the lubricant with coincident gap discharge.
More recent investigators include Costello and Lawson
[3], [4]. They reported on shaft voltage and bearing current
problems, but were primarily concerned with magnetically
induced bearing currents. Possible mechanisms for bearing
damage when operating on variable frequency drives (VFD)
or electrostatically induced currents, oil film dielecare
tric breakdown causing EDM currents, and current causing
chemical changes within the lubricant. A recent investigation
was conducted by Chen et al. on this EDM phenomenon [5],
[6]. Recently, the authors presented their findings on EDM
and its relationship to PWM inverter operation [7], [8]. The
authors suggested the sources for rotor shaft to ground voltage ( ) include electrostatic charge buildup and capacitive
coupling. These studies resulted in an electrical model of the
inverter, motor, and bearing system, and the development of
an electrostatic shielded induction motor (ESIM), a solution
to the electrostatically induced bearing damage.
and bearing
The electrical model accurately predicted
currents measured when operating with PWM voltage source
inverters (VSI). The electrical system model consists of a
balanced three-phase source with a common mode or zero
sequence source from neutral to ground and two sets of
balanced three-phase impedances coupled by an equivalent
network of machine capacitances. The zero sequence or
common mode equivalent circuit is shown in Fig. 2. The
bearing model combines a bearing resistance in series with
the parallel combination of the bearing capacitance ( ) and a
nonlinear device; the device accounts for the random charging
and discharging of the rotor shaft.

00939994/97$10.00 1997 IEEE

578

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 33, NO. 2, MARCH/APRIL 1997

(a)
Fig. 2. Common mode equivalent model.

This paper further examines the zero sequence model and

explains the electrical factors driving the shaft voltage coupling mechanism. Motor capacitance formulas are presented
and values calculated for a range of horsepower ratings.
Effects of machine parameters and interface components (e.g.,
common mode chokes, cables) are examined analytically and
compared with experimental results. The paper shows secondand third-order reduced models accurately predict the frequency response and damping factor of the
and system
current. Experimental results suggest bearing current densities
with PWM VSI drives can exceed bearing life thresholds.
Finally, results employing an ESIM and identical system interface components show the efficacy of the ESIM in reducing
rotor voltage buildup.
II. THE COMMON MODE EQUIVALENT CIRCUIT
For purposes of investigating
buildup, capacitively
coupled
current, and EDM discharge, the common mode
or zero sequence equivalent circuit of Fig. 2 provides accurate
results without the complexity of the distributed system. The
common mode models for the ac machine, cable, common
mode chokes, transformers, and line reactors are included in
the figure. Although greatly simplified, the equivalent circuit
provides a useful tool for the analysis of system parameters
and their effect on
and bearing current.
From Fig. 2, it is clear the existence of capacitively coupled
and EDM bearing currents with PWM VSI drives
depends on the following three conditions: 1) a source of
excitation ( ), which is transferred by the zero sequence or
common mode components to the stator neutral to ground volt); 2) a capacitive coupling mechanism, accomplished
age (
by the stator to rotor capacitance ( ); and 3) sufficient
buildup, a random occurrence depending on the existence of
. All three of these conditions must simultaneously exist for
capacitively coupled EDM currents to occur.
This section of the paper will explore the system factors
contributing to the development of
buildup. Section IIA develops the machine components of Fig. 2, with
calculations based on results by researchers in Tribology.
Following the presentation of relevant mechanical properties,
machine capacitance formulas are derived for the components
in Fig. 2. Section II-B examines experimental evaluations of
the model parameters and compares the values to the design
calculations.

(b)

(c)

Fig. 3. Capacitance system models: (a) stator to rotor and rotor to frame
capacitance, (b) stator to frame capacitance, and (c) bearing capacitance.

A. Capacitance Calculations for the Shaft

Voltage and Bearing Current Model
1) Mechanical Components : The occurrence of
and bearing currents depends on the existence of . Furthermore, the bearing impedance becomes capacitive only when a
lubricant film occurs in the contact regions between the balls
or rollers and the raceways [9]. The minimum film thickness
is given by
(1)
is a function of the fluid velocity and viscosity,
where
a function of the pressure coefficient of viscosity and
modulus of elasticity, and
the force or load acting on the
ball or roller [10]. Other factors influencing
include the
temperature ( ), viscosity ( ), additives ( ), lubricant film
thickness relationship to the rms value of the contact surface
( ), and dielectric strength of the lubricant ( ) [9].
The dielectric strength of lubricants is determined by static
tests [11]. Data provided by lubricant vendors indicates dielectric strengths range from 130 kV/mm. Typical bearing loads
together with (1) and technical literature indicate lubricant
film thickness ranges from 0.22.0 m. These values are
significantly lower than those employed by the static tests.
Based on tests, the authors conclude that 15
m dielectric
strength is reasonable. This suggests shaft voltages from 330
V can produce EDM currents [7]. Furthermore, tests performed
on the 15-hp induction motor of [7] showed a maximum
withstand voltage of 30 V peak at pulse durations of 10 s.
Thus,
becomes a complicated function of all the above
variables [
] [9].
2) Electrical Components
: Although
a distributed parameter system, lumped parameters adequately
model the system as shown in Fig. 2. This system consists
of the stator winding zero sequence impedance ( and
),
the stator winding to frame capacitance ( ),
, the rotor
to frame capacitance ( ), and
. A formula for each
capacitance follows, together with calculations for machines
from 5 to 1000 hp. The formulas assume the geometrical
shapes depicted in Fig. 3. A comparison with experimental
values for the 15-hp machine of [7] is presented in Section
II-B.
a) Calculation of
: The
model consisted of
parallel capacitors, where
is the number of stator slots.
Each slot consisted of a conductor
meters long,
meters

BUSSE et al.: SYSTEM ELECTRICAL PARAMETERS AND THEIR EFFECTS ON BEARING CURRENTS

579

TABLE I
CAPACITANCE VALUES FOR 15 HP MACHINE OF [7]

Csf
Crf
Csr
Cb

Fig. 4. Calculated motor and bearing capacitance values.

deep, and
meters wide centered within a rectangular
conduit with all sides at the same potential. A dielectric
material separates the conductor and conduit by meters with
a relative permittivity of (slot paper). Equation (2) provides
the
for
slots [12]. Fig. 4 shows calculated values of
for induction machines from 5 to 1000 hp:

15-hp Machine [7]

Calculated 15-hp Machine

11 nF
1.1 nF
100 pF
200 pF

7.7 nF
1.0 nF
123 pF
225 pF

increases with increasing eccentricity and length/diameter ratio

[13]. The capacitance of all bearings depends on the load angle
and relative permittivity of the lubricant.
The model selected for ball bearings, shown in Fig. 3,
assumes a set of
pairs of concentric spheres, where
is the number of balls. Each capacitor pair includes an inner
sphere (modeling the balls) within an outer sphere (modeling
the raceways). Equation (5) provides the mathematical formula
for this capacitance [12]. The radius of the inner sphere
( ) corresponds to the radius of the ball; the radius of
the equivalent outer sphere equals the radius of the inner
sphere plus the radial clearance (
), the distance to
the outer raceway. The bearing capacitance varies with the
shaft diameter and radial clearance and is plotted in Fig. 4:
(5)

(2)
b) Calculation of
: The stator to rotor coupling capacitance, shown in Fig. 3, consists of
sets of parallel
conducting plates. The area of each plate equals the product of
the length of the rotor ( ) and the width of the rotor conductor
near the rotor surface ( ). This capacitance is given by (3);
where the distance between the parallel plates ( ) is the air gap
of the machine [12]. Fig. 4 shows calculated
for induction
machines from 5 to 1000 hp:
(3)
c) Calculation of
: The capacitive coupling between
the rotor and frame, shown in Fig. 3, is determined as the
capacitance of two concentric cylinders or a coaxial capacitor.
In this case, the effective gap between the cylinders must
compensate for the effect of the stator slot widths. If the
inside radius of the outer cylinder (stator) is
and the outer
radius of the inner cylinder (rotor)
, then the capacitance is
given by (4) [12]. Fig. 4 shows calculated
for induction
machines from 5 to 1000 hp:
(4)

d) Calculation of : The bearing capacitance depends

on the geometrical configuration of the bearing, load, speed,
temperature, and characteristics of the lubricant. Each bearing
typeball, roller, journal, etc.yields a capacitance model,
with the capacitance value a function of physical and operating parameters. For example, a journal bearings capacitance

decreases;
Fig. 4 shows with increasing machine size
the machine capacitances, however, increase with increasing
horsepower [8]. These calculations are based on design data for
four-pole, 460 Vac induction machines and associated bearing
dimensions.
B. Experimentally Determined System Capacitances
The machine zero sequence inductance and parasitic capacitances were measured on the induction machine of [7].
Measurement results and methodology for each element of the
system model follow. Table I lists measured and calculated
capacitance values for the machine of [7]. The measured
capacitance values were made with the rotor externally driven
at controlled speeds when appropriate.
1)
and : The common mode or zero sequence
impedance of the machine equals one third of the stator
resistance in series with one third of the stator leakage inductance. They were obtained by connecting the three stator lines
and measuring the impedance line-to-neutral with a HewlettPackard 4284A LCR meter. A value of 300 H and 59.8
was measured at 100 kHz.
2)
: For the 15-hp machine of [7], the
obtained
by LCR measurement with the rotor removed was 11.1 nF.
, and
are
By removing the rotor, the effects of
eliminated. The 11.1 nF compares well with the calculated
value 7.7 nF in Fig. 4, which is based on a different stack
length than the motor of [7].
3)
: Measurement of
was achieved by shorting
the rotor shaft to frame and connecting a LCR meter to the
three commonly connected stator terminals and the machine

580

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 33, NO. 2, MARCH/APRIL 1997

frame. To obtain
, the value of
is subtracted from the
capacitance reading of the LCR meter. For the 15-hp machine
of [7], the measured value was 100 pF; Fig. 4 shows a value
of 123 pF. Fig. 4 suggests an increasing
with increasing
horsepower, which is consistent with the increasing machine
length and number of slots of higher power machines.
4)
: The bearing capacitance is a function of dielectric
characteristics, resistivity, and temperature of the lubricant,
geometrical construction, dynamics of the asperity contact
of the balls with the race, and speed of the rotor. The
, therefore, is dynamic and dependent on the operating
conditions of the machine. Tests were performed with a
segmented bearing and a pressure contact between the race,
film, a known insulator, and the ball. For the 15-hp machine
of [7], a
of 200 pF was measured. This compares favorably
with the calculated value of 225 pF of Fig. 4, predicted by the
bearing model.
5)
: An indirect measurement of
is possible
once
, and
are known. Measureing the impedance
from rotor to frame, the dominance of
can be reduced.
The value obtained with a LCR meter for the 15-hp induction
machine of [7] was 1.1 nF; Fig. 4 indicates 1.0 nF for a 15-hp
machine, which compares favorably with the measurement.
III. EFFECT OF DRIVE VARIABLES ON MOTOR
SHAFT VOLTAGE AND BEARING CURRENT
This section examines drive variablescommon mode
chokes, line reactors, long cablesand their effect on
and bearing current. These passive elements often provide
the impedance necessary for proper functioning of ac
drive systems. For example, common mode chokes reduce
conducted noise and series line reactors control voltage
reflection at a motors terminals. Therefore, the effects these
elements have on
and bearing currents are important to
quantify. To accomplish this, first a design equationthe
bearing voltage ratio (BVR)establishes a machine design
criterion for evaluating the potential for
and bearing
current. Next, the common mode circuit above is reduced in
complexity and a simple analysis tool is presented.
A. System Model and Analysis
With the common mode model for the drive established,
an analysis of the effects of system parameters on
and
bearing currents is possible. Fig. 2 allows for the investigation
of common mode chokes or transformers, line reactors, and
long cables through the modification of the series and parallel
impedance elements; it provides a model capable of examining
PWM modulation techniques and power device rise times;
and it allows for an investigation of source to ground voltage
levels.
1) Steady-State Shaft Voltage Level: With PWM frequencies much less than the natural frequency of the system zero
sequence network impedance, the capacitors divide
and
yield the following algebraic relationship for the BVR:
BVR

(6)

Fig. 5. Bearing voltage ratio.

This relationship, although simple, provides substantial information about bearing charge and discharge phenomena
and potential improvements. For example, a value of
,
the bearing threshold voltage ( ), exists for each value
of film thickness below which dielectric breakdown EDM
does not occur. This threshold depends on pulse duration
and characteristics of the lubricant. However, (6) provides
an estimate of
. This estimate when compared to
determines the likelihood of EDM discharge. For example,
with a dielectric strength of 15
m and lubricant film
thickness varying between 0.2 and 2 m,
ranges from 330
. With a BVR of 0.1 (Fig. 5),
is in the neighborhood of
35
for a 460 Vac system having a
equal to one half
bus voltage or 350 Vdc. A
of this magnitude is sufficient
to cause EDM discharge.
Equation (6) also suggests that a large
reduces the
bearing voltage; thus, to maintain bearing or shaft voltage
below
(the maximum sustainable voltage without dielectric
breakdown EDM), increase the relative permittivity of the
lubricant. This expression also shows how the ESIM eliminates
the potential for bearing or shaft static voltage buildup: for
an ESIM, the
in (6) is zero. In addition, the capacitive
voltage divider indicates inserting an insulating sleeve or
barrier may exacerbate the bearing charging since this reduces
the effective .
Using (6) and combining it with results of the capacitance
curves of the previous section, the BVR as a function of
horsepower was derived with the results shown in Fig. 5. From
Fig. 5, the machine of [7] has a predicted BVR of 0.074.
, bearing current, and
Fig. 6 shows a typical sequence of
traces. It shows three different shaft voltage phenomena
occurring in the bearing. Region A depicts the shaft and
bearing charging according to the capacitor divider action of
(6) followed by an EDM discharge. Region B represents a
charging and discharging of the bearing without EDM current.
Finally, Region C shows the rotor and bearing charging, but
to a much lower voltage level before EDM discharge [8].
The BVR is obtained by dividing
by the
at a point
where the machines rotor rides the lubricant (Region A).
The experimental value (0.064) is in good agreement with
the theoretical calculation of 0.074.
2) A Second-Order Model Approximation: The common
mode model of Fig. 2 adequately describes most of the
observed phenomena associated with shaft voltages and
common mode currents. However, the complexity of this

BUSSE et al.: SYSTEM ELECTRICAL PARAMETERS AND THEIR EFFECTS ON BEARING CURRENTS

Fig. 6. Examples of bearing breakdown mechanisms due to film breakdown,

dv=dt currents, and asperity contacts with an IGBT drive.

Fig. 7. Second-order model.

model often obscures the cause and effect of PWM voltage

source inverters on shaft voltages and bearing currents. A
reduced order model, if applied correctly, would have a
distinct advantage to the circuit of Fig. 2. Common mode
chokes, line reactors, and output filters, for example, often
are employed to reduce electromagnetic interference (EMI)
from PWM voltage source inverters. Also, many applications
require long cable lengths between the inverter and load. The
reduced order model of Fig. 7, therefore, provides a simple
model retaining the important effects of these elements on the
of the machine [12], [14].
The second-order system of Fig. 7 has the following general
solution for a step input:

(7)
(8)
where

581

Fig. 8. Inverter time constant and damped natural frequency as a function

of common mode inductance.

and
is the undamped natural frequency, is the dampis the characteristic impedance, and
is the
ing ratio,
. The equivalent capacitance, (
), equals
phase angle of
]
in parallel with the series
[
and the parallel combination of the
combination of the
and
.
This formulation of the system equations also allows for
an easy analysis of the rise time of the forcing function
, the effect of the PWM frequency, and influence of
the system parameters on damping, natural frequency, and
is longer than
overshoot. If the rise time of the stepped
one half of the oscillation period, the zero sequence current
current
is reduced substantially; thus reducing the
through the bearing and frame. Furthermore, increasing the
common mode inductancewith common mode chokes and
line reactorswithout considering the effect on the damping
factor can raise the of the circuit. The higher and lower
natural frequency may result in a near resonance condition
with the stepped waveform of the forcing functions PWM
carrier.
) and damped
Fig. 8 shows system time constant (
) as functions of common mode
natural frequency (
) for the 15-hp induction motor of [7]. Both
inductance (
quantities have been converted into hertz or 1/s, for easy comparison with typical carrier frequencies employed by IGBT
inverters. IGBT VSIs often incorporate common mode chokes
current. Fig. 8 indicates the damped
to reduce the
natural frequency and time constant decrease with increasing
common mode inductance. For typical common mode inductances, the damping in the system decreases and the damped
natural frequency is well within the dominant frequencies of
the common mode voltage source of IGBT inverters, setting
up a potential resonance condition.
3) A Third-Order ModelThe EDM Discharge Current: The
second-order system, very useful for voltage and common
mode analysis, fails to describe the EDM discharge phenomena. The full-order model of Fig. 2 is too complex. However,
the third-order system of Fig. 9 is manageable and accurately

582

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 33, NO. 2, MARCH/APRIL 1997

Fig. 9. Third-order system model.

describes the common mode and EDM discharge. In this

figure, the
of Fig. 7 is resolved into
in parallel with an
equivalent circuit (
) for the
. In series with the parallel
combination of the
and
. This can also be expressed
as:
.
An eigenvalue analysis of this third-order system with
parameters corresponding to the conditions of Fig. 6 showed
a pair of complex poles at 95.7 kHz with a time constant of
8.57 s. The third pole, associated with the bearing voltage
and current, is located on the negative real axis with a time
constant of 0.01 ps, accurately modeling the response observed
following an EDM discharge (Fig. 6, Region A).

Fig. 10. Common mode choke response.

B. Model Evaluation and Component Analysis

Evaluation of the second-order model requires experimental
results that allow a comparison of the natural frequency and
damping factors with the predicted values based on Fig. 7.
The response of the stator neutral voltage, rotor shaft voltage,
and bearing current to a PWM VSI with various system
components inserted between the inverter and motor provides
data for model evaluation and demonstrates the effect of
system components on bearing currents.
1) Effects of Common Mode Components, Line Reactors, and
Cable Lengths: With the appearance of IGBT inverter drives,
common mode noise presents a significant challenge to drive
design. Common mode chokes and transformers, inserted
between the inverter output and load motor, provide additional
impedance to common mode current without affecting the
fundamental component. Another approach inserts a threephase series reactor, but at the price of reduced fundamental
voltage at the terminals of the machine.
Fig. 10 shows the response of
, and bearing current
with a common mode choke of 270 H and 2.6
inserted
between inverter output and load motor. The
oscillates
at 60 kHz with a damping ratio of 0.12. Using the model of
Fig. 7, the calculated values are 62.7 kHz and a damping factor
of 0.12. Adding the common mode choke to reduce
current also affects the response of
and
. The reduced
to overshoot the nominal
damping causes the machines
steady-state value for each switching instant. The decreased
damping also provides the rotor the opportunity to charge once
the bearing rides the lubricant film.
To examine the effects of reduced damping in more detail,
a three-phase series reactor with a common mode reactance
of 600 H was inserted between the inverter output and load

Fig. 11. Series reactor response.

motor. The theoretical frequency and damping factor were 50.3

kHz and 0.0158, respectively. Experimental results for a 15hp induction machine (Fig. 11) show a lightly damped 50-kHz
oscillation. The decrease in damping increases the probability
charging. This is because the system capacitance never
of
achieves the steady-state charge associated with the forcing
function. Each time the bearing rides the film, the presence of
alters the system topology and the voltage distribution must
change to reflect the change in impedance. Thus, with relatively
is excited and rings to an excessively large
light damping,
exceeds 590
, which is
value. In the case of Fig. 11,
larger than one half
.
280
current, shaft voltage
A cables length also affects
buildup, and bearing current discharge. Fig. 12 shows the
, and bearing current with a 600-ft cable. At the
frequencies of interest, the cable presented an equivalent series
and 80 H, and a parallel resistance of
impedance of 3.2
in series with 22 nF of capacitance. The Thevenin
3.0
in series with 129
equivalent equals a resistance of 10.9
H. The calculated damped natural frequency and damping
ratio for the model of Fig. 7 are 71.7 kHz and 0.18. These

BUSSE et al.: SYSTEM ELECTRICAL PARAMETERS AND THEIR EFFECTS ON BEARING CURRENTS

Fig. 12.

Long-cable-length response.

Fig. 13.

Common mode choke response with a standard motor and an ESIM.

Fig. 14. Series reactor response with a standard motor and an ESIM.

compare well with the experimental values of 76.0 kHz and

0.19, respectively.
The transient response of the long-cable system shows the
rings up to over 600
, with a nominal 630 Vdc bus.
The bearing rides the lubricant film and charges to 25
just
before the next ring up of
. Once the stator begins to ring
up to the 600
level,
responds with a slight delay and
achieves almost 65
before an EDM of 3.2
occurs.
Experimental results similar to these confirm excessive
and
are possible with long cable lengths. The resulting
current densities2.48 to 5.16
are in the region
to reduce bearing life.
IV. SYSTEM PERFORMANCE

583

OF AN

ESIM

The three conditions necessary for the existence of bearing

current outlined in Section II provide the basis for investigations into solutions to the problem. One solution proposed,
prototyped, and tested by the authors is the Electrostatic
Shielded Induction Motor (ESIM). The ESIM essentially decouples the stator and rotor by inserting a Faraday shield

Fig. 15. Long-cable-length response with a standard motor and an ESIM.

between the stator and rotor. The prototype reported on in

[7], [8], and [15] proved effective in eliminating EDM current
and in reducing
current to acceptable levels.
To examine the effectiveness of the ESIM, tests were performed using typical system components reported in Section
III. Figs. 1315 show experimental results of a four-pole, 460Vac, 15-hp ESIM with a common mode choke, series reactor,
and long cable, respectively. Each figure shows traces of the
rotor voltage with and without the Faraday shield of the ESIM
active. As discussed earlier, the magnitude of rotor voltage is
a measurement of the potential for EDM discharge.
In each case, the ESIM reduces the rotor voltage; the rotor
voltage ranges from approximately 10% to 25% of the value
without the Faraday shield. This demonstrates the universality
of the ESIM as a solution to the shaft voltage and bearing
current problem. Furthermore, the results without the Faraday
shield are consistent with those reported in Section III and [7],
[8], and [15] for a standard induction motor. Note the reduced
damping for the case of the series reactor; this correlates well
with the generalized damping and frequency results of Fig. 8.
In addition, point of Fig. 14 corresponds to a partial EDM

584

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 33, NO. 2, MARCH/APRIL 1997

discharge (note the abrupt discharge and lack of oscillation).

In contrast, the ESIM revealed no EDMs.
V. CONCLUSION
This paper reviewed recently reported bearing failures and
examined the important system parameters and their relationship to EDM and
bearing current. Models and formulas
were presented for the major system elements influencing rotor
shaft voltage and bearing current. Parameters were calculated
for machines from 5 to 1000 hp based on machine design
data and correlated with tests on a 15-hp machine. The effects
of system components on bearings were evaluated through
reduced-order models and experimental results. Finally, test
results for an ESIM demonstrated its ability to attenuate
current and eliminate EDM current for all system components
tested.

[9] D. Busse, J. Erdman, R. J. Kerkman, D. Schlegel, and G. Skibinski, The

effects of PWM voltage source inverters on the mechanical performance
of rolling bearings, in APEC 96, Eleventh Ann. Applied Power Elec.
Conf. and Expo., Mar. 37, 1996, pp. 561569.
[10] T. Harris, Rolling Bearing Analysis, 3rd ed. New York: Wiley, 1991.
[11] L. Alston, High Voltage Technology. New York: Oxford, 1968.
[12] W. H. Hayt, Engineering Electromagnetics, 5th ed. New York:
McGraw-Hill, 1989.
[13] H. Prashad, Theoretical evaluation of capacitance, capacitive reactance,
resistance and their effects on performance of hydrodynamic journal
bearings, Trans. ASME, vol. 113, pp. 762767, Oct. 1991.
[14] J. L. Melsa and D. G. Schultz, Linear Control Systems. New York:
McGraw-Hill, 1969.
[15] D. Busse, J. Erdman, R. J. Kerkman, D. Schlegel, and G. Skibinski, An
evaluation of the electrostatic shielded induction motor: A solution for
rotor shaft voltage buildup and bearing currents, in 1996 IEEE Industry
Application Society Conf. Rec., vol. 1, pp. 610617, October 6, 1996,
San Diego, CA.

Doyle F. Busse, for a photograph and biography, see this issue, p. 576.

REFERENCES
[1] P. Alger and H. Samson, Shaft currents in electric machines, in AIRE
Conf., Philadelphia, PA, Feb. 1924.
[2] T. Tallian, G. Baile, H. Dalal, and O. Gustafsson, Rolling bearing
damageA morphological atlas, SKF Industries, Inc., Technology
Center, King of Prussia, PA.
[3] M. Costello, Shaft voltage and rotating machinery, IEEE Trans. Ind.
Applicat., Mar. 1993, pp. 419425.
[4] J. Lawson, Motor bearing fluting, CH3331-6/93/0000-0032, in 1993IEEE Pulp and Paper technical conference, pp. 3235, Hyannis, MA,
June, 1993.
[5] S. Chen, T. A. Lipo, and D. Fitzgerald, Measurement and analysis of
induction motor bearing currents in PWM inverter drives, in IEEE PES
Summer Meet., July 2427, 1995.
[6] S. Chen, T. A. Lipo, and D. Fitzgerald, Modeling of motor bearing
currents in PWM inverter drives, in IEEE IAS Ann. Conf. Rec., Oct.
812, 1995, vol. 1, pp. 388393.
[7] J. Erdman, R. J. Kerkman, D. Schlegel, and G. Skibinski, Effect of
PWM inverters on ac motor bearing currents and shaft voltages, in
APEC 95, Tenth Ann. Applied Power Elec. Conf. and Expo., Mar. 59,
1995, vol. 1, pp. 2433.
[8] D. Busse, J. Erdman, R. J. Kerkman, D. Schlegel, and G. Skibinski,
Bearing currents and their relationship to PWM drives, in IECON 95,
IEEE 21st Ann. Ind. Elec. Conf., Nov. 610, 1995, vol. 1, pp. 698705.

Jay M. Erdman, (M58) for a photograph and biography, see this issue, p.
576.

Russel J. Kerkman (S67M76SM87), for a photograph and biography,

see this issue, p. 576.

David W. Schlegel (S89), for a photograph and biography, see this issue,
p. 576.

Gary L. Skibinski (M96), for a photograph and biography, see this issue,
p. 576.