J.C. Gomez, M.M. Morcos, C. Reineri, G.

Campetelli

Induction Motor Behavior

Under Short Interruptions
and Voltage Sags
O f all power quality issues, voltage sags and short interrup-
tions are considered to be the main cause of more than
80% of the problems experienced by sensitive equipment. The
part of an extensive project with the goal of developing a simple
tool for the study of induction motor behavior and system effects
under short interruption and voltage sag conditions.
consequences of a power quality problem are sensitive equip-
ment dropout and possible full-process or industrial-line disrup- Short Interruptions and Voltage Sags
tion, with the obvious customer economic losses and A short interruption is defined as the complete loss of voltage
complaints. This type of problem occurs frequently due to the (< 0.1 pu) on one or more phases for a time period between 0.5
increasing widespread of highly sensitive control equipment, cycles and 3 s. This phenomenon can be due to the supply in-
such as programmable logic controllers, adjustable speed terruption or due to the trip and subsequent reconnection by the
drives, and personal computers. motor undervoltage protection. The present situation is more
Customers normally suffer from the effect of the induc- severe than the normal motor start due to several reasons, such
tion-motor and supply-system interaction, and utilities can expe- as the motor generated voltage that is out of phase, heavily
rience significant loss of load [1]. The motor undervoltage loaded machinery, and a rigorous hot-load pickup. The indus-
protection could trip the motor contactor if the supply voltage trial plant should have a reacceleration scheme in order to al-
stays too low for a long time [2]. New power quality requirements low its production process restart without interfering with its
have an important effect on the motor system interaction, for ex- own sensitive equipment and with other customers connected
ample, the increasingly popular motor fast reconnection to the to the same supply system.
same source or to an alternative source. The load characteristics Without considering the induction motor effects, voltage
during the reconnection instant are also critical for the motor be- sags are normally represented by a square waveform [4]. Most
havior, since it is possible that
the motor would stall and not
start when the supply voltage is
restored [3].
Several reports related to
the modeling of induction mo-
tor behavior under voltage sag
conditions have been pub-
lished, but only a few on
short-interruption behavior [2].
To the authors’ knowledge, no
extensive experimental study
on these two phenomena has
been available in the literature.
This article documents an ex-
perimental study that is the first

J.C. Gomez, C. Reineri, and G.

Campetelli are with the Electric
Power System Protection Institute,
Rio Cuarto National University, Rio
Cuarto, Cordoba, Argentina. M.M.
Morcos is with the Department of
Electrical and Computer Engi-
neering, Kansas State University,
Manhattan, KS, United States.
Digital Vision Ltd.
IEEE Power Engineering Review, February 2001 0272-1724/01/$10.00©2001 IEEE 11
of the voltage sags lasted 10 cycles or less and were 20-30% in oscillation with initial amplitude of 12%. The observed
magnitude. Transmission faults are usually cleared in less than 6 oscillation seems to have constant frequency and exponential
cycles, while distribution faults last between 10 and 20 cycles. amplitude-attenuation. The maximum values and oscillation
Voltage sag results in the initial reduction of the motor speed, amplitudes are similar for both no-load and 85% rated load
keeping for a while a higher voltage supplied by its internal, or starts, with start durations of approximately 0.18 s and 0.24 s,
back, electromotive force (emf). When the voltage sag ends, the respectively. The effect of the phase angle, at the instant of con-
motor speed increases, demanding more energy from the supply necting the motor to the supply, on the maximum current peaks
until the steady-state speed is reached, hence, extending the is noticeable - ranging from 125 A to 148 A. It should be noted
voltage sag duration. The load torque in this case shows very dif- that the voltage recovery is slower in the 85% rated load start
ferent characteristics as compared to normal startup conditions than in the no-load start.
(presence of compressor unloading valves, counter pressure, fan
dampers, etc.). The motor current is now a function of two phe- No-Load and 85% Motor Rated Load Decelerations
nomena, mechanical and electrical, each having its own time Until Zero Speed due to Supply Interruption
constant. The presence of the induction motor causes a voltage The voltage drop follows a double-exponential variation due to
sag distortion, smoothing and prolonging the voltage variation. the speed reduction in addition to magnetic decays. The initial
The result is that some of the sensitive equipment that was able value is the back emf that differs from the supply voltage by
to withstand the original voltage sag would drop out during the 13% for no-load and 17% for 85% rated load, respectively. Be-
post-sag period due to the induction motor effects. This indi- sides, the speed reduction is governed by the mechanical time
cates that the addition of motor loads to a system known to be constant, which is proportional to the kinetic energy at the shaft
operating without harmful voltage sags can be critical to the sen- power. The magnetic decay is governed by the rotor-circuit time
sitive equipment operation. It would be very convenient to deter- constant. As the rotor resistance is a function of speed, the time
mine the motor load limit for each particular system based on its constant will change with the speed. Since the circuit is open,
sensitive equipment. the current goes immediately to zero.
The single line-to-ground fault is the most probable type of The voltage measured values are easily fitted by the applica-
fault and, through a delta-wye transformer, is transferred as a tion of two time constants, 0.3 s for the magnetic phenomenon
two-phase voltage sag, in which case voltage sags should be and 27 s for no-load (or 0.44 s for 85% rated load) for the speed
considered as a case of unbalanced transient supply. change, which is very small for the time period considered in the
Three-phase voltage sags (to the same level of the imbalance no-load case. Analytical and experimental values show an ex-
sags) represent the worst stability condition [3]. Therefore, only cellent agreement up to at least 0.2 s after the disconnection has
balanced phenomena were experimentally studied, leaving the taken place, Figure 2.
unbalanced behavior for future investigation.

Experimental Setup 150

The tested induction motor is a standard three-phase, squir-
100
rel-cage machine of the following ratings: 5.5 kW, 380 V, 50 Hz,
and 1,450 rpm. The load was based on an eddy-current brake, 50
Current (A)

having torque characteristics nearly proportional to the square

of speed. Voltage and current were measured and recorded 0
through a digital oscilloscope and a standard power data ana-
−50
lyzer, both sampling 32 bits. The investigated voltage sags and
short interruptions were always balanced with a duration of ap- −100
proximately 5 cycles.
−150
0 0.05 0.1 0.15 0.2
Test Cases and Results Time (s)
In order to get a clear and step-by-step idea about the induction
motor behavior, the following tests were carried out. Figure 1. Three-phase no-load direct start currents

No-Load and 85% Motor Rated Load Direct Starts 600

Steady-State Voltage: This test shows a small distortion (< 3%),
and very small voltage imbalance (< 1%) and a phase-to-phase 500
open circuit rms voltage of 390 V (slightly higher than rated).
Steady-State Current: This test shows distortion without low 400
Voltage (V)

frequency swings or oscillations, current imbalance higher than

voltage imbalance (approximately 3%), and steady-state no-load 300
and 85% load currents of 5.56 A and 9.38 A, respectively.
From the steady-state voltage and current oscillograms, no 200
important constructive asymmetries were detected.
Start Voltage: The voltage waveforms for both conditions 100
show a smooth increase caused by a normal start phenomenon
plus a slight oscillation. These slight voltage variations are due 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
to voltage drops caused by power oscillations. Time (s)
Start Current: Figure 1 shows the characteristic shape (start
current approximately 8 times the rating value) and an important Figure 2. Back emf decay for no-load motor and analytical exponential line

12 IEEE Power Engineering Review, February 2001

No-Load and 85% Motor Rated Load Decelerations the rated current. There are also slight power oscillations with a
due to Supply Interruption During 5 Cycles total transient duration shorter than the direct start phenomenon.
and then Reconnection to Supply The situation is noticeably different for the loaded case,
The voltage drop follows the previous double-exponential where the reconnection is completely out of phase, showing a
variation. The same principle was applied to determine the large voltage shift [5]. The measured intensity values were
voltage difference (magnitude and phase angle) for the higher than the direct start currents, but their time duration was
reconnection instant (off time 96.7 ms and 93.5 ms for no-load shorter, showing 60% amplitude oscillations (Figure 3). The
and 85% load, respectively), with a very good agreement be- voltage oscillations due to voltage drops on the circuit imped-
tween experimental values (0.375 pu for no-load and 1.35 pu ances are, therefore, more noticeable in the out-of-phase
for 85% rated load) and analytical values (0.378 pu for no-load reconnection than in the no-load case. The thermal effect pro-
and 1.46 pu for 85% rated load). The voltage difference mea- duced by the reconnection current is only 53% of that corre-
sured for the loaded case is very close to the reclosing maxi- sponding to the 85% load start current.
mum allowed value of 1.33 pu. At the reconnection instant,
three transient phenomena take place that are due to three dif- No-Load and 85% Motor Rated Load Decelerations
ferent processes: due to Supply Short-Circuiting During 5 Cycles
● Magnetic inrush current and then Reconnection to the Supply
● Mechanical inrush current As soon as the motor terminals are short circuited, the voltage
● Power oscillation. falls sharply to zero. The three-phase currents follow the classi-
The magnetic inrush is caused by the discrepancy between cal two-component short-circuit time variation. The dc compo-
the supply-established magnetic field and air-gap residual flux nent is attenuated by the stator time constant, and the ac
(in spatial position and value). The mechanical inrush is due to component is also attenuated due to the emf decay. Besides, the
the difference between the actual and steady-state speeds. The energy dissipation process produces a new speed-variation time
power oscillation is caused by the induction motor response to constant, where the power value should now represent the
the applied power step, generating several power interchanges no-load losses, shaft load, and energy dissipation.
with negative and positive torque until passive loads smooth the The analytical and experimental results have similar
phenomenon down. waveshapes and values, clearly showing that the variation is
For the no-load case, the speed drop is very small, then the rather complex. For the motor under study, the current reaches
reconnection takes place with small magnetic field and speed dif- zero in approximately 80 to 100 ms, with half of the maximum
ferences, showing overcurrent values not higher than 2.5 times value in nearly one-third of the time. The Joule’s heat (or Joule’s

200 600

400
100
200
Voltage (V)
Current (A)

0 0

−200
−100
−400

−200 −600
0 0.04 0.08 0.12 0.16 0 0.04 0.08 0.12 0.16
Time (s) Time (s)

Figure 3. Reconnection current for 85% rated load motor Figure 5. Voltage recovery after a 5.5-cycle short circuit with no-load motor

100 400

50 200
Voltage (V)
Current (A)

0 0

−200
−50

−400
−100 0 0.04 0.08 0.12 0.16
0 0.05 0.1 0.15 0.2 0.25 0.3 Time (s)
Time (s)
Figure 6. Voltage recovery after a 5.5-cycle short circuit with 85% rated loaded
Figure 4. Short-circuit and reconnection currents for no-load motor motor

IEEE Power Engineering Review, February 2001 13

integral) of short-circuit currents is much smaller than that cor- reaction. Figures 5 and 6 show the recovery voltages for no-load
responding to the loaded-motor start current. The experimental and loaded cases, respectively. It can be seen that, in the no-load
magnetic-decay time constant was 0.07 s, and the speed-reduc- case, the steady state is reached without great difficulties. How-
tion time constant was 0.09 to 0.1 s. This means that the genera- ever, in the loaded situation, the voltage is nearly stabilized at
tion process does not affect the short interruption waveform. In 65% of the presag value (torque is about 42% of the rated
this case, the first part of the short interruption (or voltage sag) torque), showing the difficulties that the induction motor is ex-
shows a stepwise waveshape. periencing for the load reacceleration. The Joule’s heat for the
The comparison between short-circuit currents with and loaded reconnection at 0.3 s is already 2.3 times the total amount
without mechanical load shows small differences. The no-load for the no-load case.
motor short-circuit current is approximately 10-15% higher
than the loaded current case, and the attenuation of the loaded No-Load and 85% Motor Rated Load
case is slightly higher than the no-load situation. Voltage Sags to 45% During 5 Cycles
At reconnection, the motor emf is practically zero, since the The voltage sag is an intermediate situation between the
magnetically stored energy has been dissipated in the rotor and open-circuit and short-circuit cases described previously. The
stator resistances. The voltage difference is virtually the supply no-load and loaded voltage waveforms present slight differ-
voltage, thus the current will follow the variation explained in ences during the low-voltage period. Besides, it can be seen that
the previous case. The main difference is that the voltage recov- the variation from 100% to 45% and back are not stepwise
ery is rather slow now, lasting approximately 0.15 s and 0.65 s changes. After the reconnection to 100% voltage, the slow volt-
for the no-load and 85 % rated load, respectively, and producing age recovery for the loaded motor is noticeable, taking nearly
smaller maximum current values, as shown in Figure 4. It 0.12 s to reach the steady-state value as opposed to only 60 ms
should be pointed out that the transient that lasted nearly 0.65 s for the no-load case, as shown in Figure 7. During the on-sag pe-
was caused by a short circuit present in the circuit for less than 6 riod, the motor slows down, and a higher and more reactive cur-
cycles. The slow voltage recovery is due to the test circuit rent can be detected. At the moment of voltage recovery, a large
hot-load pickup, which represents a typical industrial system. In inrush current is present, which slows down the recovery pro-
the 85% rated load case, the situation is drastically different be- cess. The current magnitudes and durations were higher in the
cause of the long reacceleration with a high Joule’s heat (current loaded case than in the no-load case, and the oscillations were
is kept approximately constant at 60 A), which can cause ther- more noticeable in the first case, as shown in Figure 8. From the
mal problems to the induction motor and a possible protection comparison with the direct start, it can be concluded that the

600 600

400 400

200 200
Voltage (V)

Voltage (V)

0 0

−200 −200

−400 −400

−600 −600
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s) Time (s)
Figure 7. Voltage sag to 43% during 5.5 cycles with 85% motor rated load Figure 9. Voltage sag to 30% during 5.4 cycles

80 80

40 40
Current (A)
Current (A)

0 0

−40 −40

−80 −80
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s) Time (s)
Figure 8. On-sag and post-sag currents for voltage sag to 43% during 5.5 cy- Figure 10. On-sag and post-sag currents for voltage sag to 30% for 85%
cles and 85% loaded motor loaded motor

14 IEEE Power Engineering Review, February 2001

whole process was faster and that the reconnection maximum IEE Proc. Generation, Transmission and Distribution, vol. 143,
current was nearly 50% of the direct-start maximum current. 1996, pp. 56-60.
[3] J.C. Das, “Effects of momentary voltage dips on the oper-
85% Motor Rated Load Voltage Sags to ation of induction and synchronous motors,” IEEE Trans. Ind.
30%, 56%, 71%, and 85% During 5 Cycles Applicat., vol. 26, 1990, pp. 711-718.
The observed behaviors were similar to the previous case. The [4] M.H.J. Bollen, “The influence of motor reacceleration on
mentioned effects are very noticeable with the 30% voltage sag voltage sags,” IEEE Trans. Ind. Applicat., vol. 31, 1995, pp.
being attenuated while the voltage sag increases from 30% to 667-674.
85%. The current phase shift is evident in both the sag start and [5] T.S. Key, “Predicting behavior of induction motors dur-
end instants (Figures 9 and 10). The transient durations of volt- ing service faults and interruptions,” IEEE Ind. Applicat. Mag.,
age decrease and increase are reduced from 0.1-0.13 s to January/February 1995, pp. 6-11.
0.03-0.025 s, as voltage sags change
from 30% to 85%. The on-sag and
reconnection peak-current to load-cur-
rent relationships move from 2.9-5 to 2002 IEEE Fellow Nominations
1.2-1.9, while the voltage sags change Deadline: 15 March 2001
from 30% to 85%.
Recognizing the achievements of its members is an important part of the mission of
the IEEE. The IEEE grade of Fellow is conferred upon a person of “outstanding and ex-
Conclusions traordinary qualifications and experience in IEEE designated fields, and who has made
From the experimental study related to important individual contributions to one or more of these fields.” The total number of
short interruptions and balanced volt- Fellows selected each year does not exceed 0.1% of the total IEEE membership.
age sags, the following conclusions can The number of IEEE Fellow nominations for PES members has declined in recent years.
be drawn. As a result, fewer of our colleagues were considered for the recognition that they deserve
● The induction motor greatly influ- through their contributions to power engineering. Many of our PES colleagues made sig-
ences the voltage sag waveform nificant contributions to the profession through their work in engineering, technical lead-
and duration. ership, and education. As a professional community, we need to be more proactive in
● There are situations where the sys- nominating our colleagues for this significant award.
tem recovery can be seriously af- Any person, including a nonmember, is eligible to serve as a nominator with the follow-
fected by the induction motor ing exceptions: members of the IEEE Board of Directors, members of the IEEE Fellow
presence. Committee, IEEE Technical Society/Council Fellow Evaluating Committee Chairs, mem-
● The motor-load characteristics bers of IEEE Technical Society/Council Evaluating Committees reviewing the nomination,
should be considered in voltage sag or IEEE staff. The deadline for nominations is 15 March 2001.
studies. The candidate must be an IEEE Senior Member at the time the nomination is submitted,
● The on-sag and post-sag currents and he/she must have completed 5 years of service in any grade of IEEE membership.
can reach levels higher than the di- All the necessary material to assist you in the nomination process is available on the IEEE
rect start values and the post-sag Web site: http://www.ieee.org/about/awards/fellows/fellows.htm. If you prefer a hard copy,
overcurrent duration can last more please send an e-mail to fellow-kit@ieee.org. Include your name, street address, city,
than twice the normal start time state/province, postal code, country, and telephone/fax numbers. For more information,
period. contact Chen-Ching Liu, PES Fellows Committee chair, liu@ee.washington.edu.
● The circuit hot-load pickup together

with the motor load can drastically

extend or delay the reacceleration
process and, in particular cases, pre-
vent the start completely.
● The worst case is related with the

motor size, system hot-load pickup,

and shaft load characteristics.
● Knowledge of the circuit hot-load

pickup characteristics is decisive in

order to get a reasonable accurate
circuit representation for short-in-
terruption and voltage-sag studies.

References
[1] J.W. Shaffer, “Air conditioner re-
sponse to transmission faults,” IEEE
Trans. Power Syst., vol. 12, 1997, pp.
614-621.
[2] M.H.J. Bollen, P.M.E. Dirix,
“Simple model for post-fault motor be-
havior for reliability/power quality as-
sessment of industrial power systems,”
IEEE Power Engineering Review, February 2001 15