of 9
7/29/2019 milanovic_2008E
1/9
IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 2, APRIL 2008 1063
The Influence of Induction Motors on Voltage SagPropagationPart I: Accounting for the Change in
Sag CharacteristicsJovica V. Milanovic, Senior Member, IEEE, Myo T. Aung, Member, IEEE, and Sarat C. Vegunta
AbstractThis paper presents an analytical approach to theanalysis of the interaction between induction motors (IMs) andvoltage sags. The presented methodology enables quick assess-ment of the influence of IM on sag characteristics and avoidstime-consuming transient simulations. The accuracy of the de-rived analytical formulae is verified through classical transientsimulations using the PSCAD/EMTDC package. This paper fur-ther proposes a simple, yet efficient transformation of resultingnonrectangular voltage sags, based on the loss of voltage index,into equivalent rectangular sags suitable for conventional voltagesag performance benchmarking.
Index TermsInduction motors (IMs), power quality (PQ),voltage sags.
I. INTRODUCTION
POWER-QUALITY (PQ) issues continue to receive signif-
icant attention due to their potentially huge economic im-
pact on the industry. Voltage sags, transients, and momentary
interruptions are among the most common PQ disturbances and,
by large, the most frequent cause of equipment malfunctions and
costly disruptions of industrial processes [1].
The IEC 61000-4-30 standard defines the voltage dip (sag)as a temporary reduction of the voltage at a point of the elec-
trical system below a threshold. This threshold is defined by
the IEEE 1159-1995 standard as 90% of the nominal voltage.
If the magnitude of the retained voltage drops below 10% of
the rated voltage, the disturbance is classified as an interrup-
tion. Most voltage sags and short interruptions recorded in ac-
tual power systems last from a few cycles to 1 min. Voltage
sag characteristics (i.e., magnitude, duration, phase-angle jump,
sag shape, etc.) change when sags propagate through power
system networks. The extent of this change depends on network
topology (mesh or radial), line and cable impedances, trans-
former winding connections, system protection and groundingpractices, load connections (star or delta), and load dynamics
(e.g., IMs) [2], [3]. The influence of all of these factors has been
Manuscript received October 18, 2006; revised June 5, 2007. This work wassupported in part by the EU Framework V Project Microgrids and in part by theEU Framework VI Project RISE under Contract FP6-INCO-CT-2004-509161.Paper no. TPWRD-00650-2006.
J. V. Milanovic and S. C. Vegunta are with the School of Electrical and Elec-tronic Engineering, The University of Manchester, Manchester, M601QD, U.K.(e-mail: [email protected]).
M. T. Aung is with the University of Bath, Bath, BA2 7AY, U.K. (e-mail:[email protected]).
Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRD.2007.915846
studied in the past either by considering an individual effect or
through more inclusive studies that dealt with several influences
at the same time (typically not all of them). A large number of
published and easily accessible reports describe the results of
those studies.
One of the major deficiencies of the past studies remains as
the accounting for dynamic effects in large system sag prop-
agation studies. Typically, sag calculation and propagation in
large systems are studied by applying static fault calculations
based on system impedance (or admittance) matrix and equiva-lent positive-, negative-, and zero-sequence networks. It is fur-
ther assumed that voltage sags originate from faults (cleared by
primary protection) on transmission and/or distribution systems
and that they have a rectangular shape. The assumption about
the shape is based on neglecting the change in fault impedance
during the fault and the effect of load dynamics.
The shape of voltage sags, however, ceases to be rectangular
if there is an appreciable number of IMs in the network and the
faults last longer (e.g., distribution system faults or faults cleared
by secondary protection). In such cases, the shape of the sag
changes due to the IMs dynamic responses. When a voltage
sag appears at the terminals of an induction motor, the torqueand the speed of the motor decrease to levels below their nom-
inal values. Once the voltage sag is removed, the IM attempts
to reaccelerate, and starts drawing an excessive amount of cur-
rent from the power supply. The flow of such current through
the supply impedance prevents fast recovery of the voltage and
causes a prolonged sag duration and change in the shape of the
sag. This new resulting sag may cause sensitive equipment to
trip even though it might have been able to ride through the orig-
inal voltage sag.
Effectively, all existing, standard benchmarking approaches
(e.g., SARFI, DISDIP, ESKOM, and UNIPED sag tables, 3-D
cross tabulation of the rms variation, magnitude-duration bar-
charts, etc.) register voltage sags as rectangular events [4]. (Forexample, for a typical voltage sag, only the information about
phase-to-neutral or line-to-line voltage magnitude and sag dura-
tion is provided.) These tabulated results are then used for com-
paring the sag performance of different buses/systems and even
for the assessment of equipment/process sensitivity to voltage
sags.
A few past works (e.g., [3] and [5]) dealt with the influence
of IMs on voltage sags. They typically modeled IMs using dy-
namic models and, therefore, are not suitable for large system
applications. The results of that analysis could not be presented
using standardized, widely accepted voltage sag tables due to
nonrectangularity of the resulting voltage sags.
0885-8977/$25.00 2008 IEEE
7/29/2019 milanovic_2008E
2/9
1064 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 2, APRIL 2008
Fig. 1. Simplified system impedance model.
To capture the IM behavior during the sag and its influence on
sag characteristics, a comprehensive analytical tool was devel-
oped and partially reported in [5]. The method presented there
described the general behavior of IMs when subjected to sym-
metrical and asymmetrical voltage sags using positive- and neg-
ative-sequence IM equivalent circuits. It did not discuss the in-
fluence of IM operating condition and parameters on the char-
acteristics of the resulting voltage sag.
This paper builds on and is an extension of the work presented
in [5]. It considers comprehensively the influence of IM on in-
dividual phase voltages and the conversion of resulting nonrect-
angular voltage sags into rectangular ones suitable for conven-
tional sag benchmarking purposes.
The results of developed analytical method (implemented in
Matlab [6] and based on static calculations) are verified using
classical transient simulations in PSCAD/EMTDC [7]. The
proposed methodology accounts for the influence of IM on
voltage sag characteristics and propagation, without performing
lengthy transient simulations, making it applicable for large,
practical system studies where hundreds of sags at numerous
bases have to be evaluated. It further enables studies of prop-
agation of voltage sags from the point of common coupling
(PCC) to the end-user equipment in an industrial plant with a
number of IMs or the propagation of voltage sags through the
electrical system of an industrial park to be built. The majoradvantage of the method is that it starts with conventionally
recorded/represented voltage sags and produces the output in
the same, universally adopted, format.
II. MODELLING OF AGGREGATE IM
FOR VOLTAGE SAG STUDIES
Voltage sagIM interaction is studied based on a group of
identical IMs connected at the PCC (as shown in Fig. 1) and
subjected to a typical, asymmetrical, three-phase, rectangular
voltage sag. IM and sag parameters are given in Appendix A.
The model is implemented in Matlab to facilitate static calcu-
lations. The integration step of 20 ms (to account for the delay
in rms calculation in PSCAD/EMTDC) is used in order to val-
idate the results by transient simulations in PSCAD/EMTDC.
The methodology is developed for the most general case of an
asymmetrical sag and a group of IMs, but it can be easily simpli-
fied and applied either to a single IM or a group of IMs subjected
to symmetrical voltage sags.
A. Sequence Components Voltages
For a given presag, during the sag and after sag voltages (see
Appendix C), corresponding sequence component voltages can
be calculated as follows:
(1)
Fig. 2. Simplified induction motor equivalent circuits. (a) Positive-sequencecircuit. (b) Negative-sequence circuit.
With , where is a com-
plex operator. It is assumed that at the end of the voltage re-covery, the voltages in all three phases reach their respective
nominal presag values, i.e.,
(2)
B. Aggregation of IMs
The aggregate model of a group of IMs retains the same form
as a model of an individual motor of the group. The aggregated
parameters of the equivalent IM representing a group ofniden-
tical (or different) IMs can be calculated as suggested in [8].
However, for simplicity, only identical IMs are chosen in thisstudy and could well extend to a group of nonidentical IMs.
For identical IMs, the aggregated parameters are obtained by
dividing electrical parameters of an individual motor by
(number of aggregated motors) and multiplying mechanical pa-
rameters by , e.g.,
(3)
(Note: The aggregated motor parameters are represented in
the paper with suffix agg.)
C. Electromagnetic Torque Variations
Fig. 2 shows the equivalent positive- and negative-sequence
circuits of the IM. Since the magnetizing impedance of the IM
is much larger than the stator and the rotor impedances, it is
neglected. The stator and the rotor currents are assumed equal.
The positive- and negative-sequence electromagnetic torques
are given in [9] as
(4)
(5)
7/29/2019 milanovic_2008E
3/9
MILANOVIC et al.: INFLUENCE OF INDUCTION MOTORS ON VOLTAGE SAG PROPAGATIONPART I 1065
Fig. 3. Speed-torque curves under normal and voltage sag conditions [4].
where and
Under balanced supply conditions, only positive-sequence
components exist and they contribute to the development of
positive-sequence torque only. During unbalanced supply con-
ditions, however, negative-sequence components appear, giving
rise to negative-sequence torque, which opposes the existing
positive-sequence torque. The resulting electromagnetic torque,
or net torque, in the machine under unbalanced operating con-
ditions, is a vector sum of positive- and negative-sequence
torques
(6)
D. Machine Speed Variation During and After Voltage Sag
The presence of either balanced or unbalanced voltage sags at
the IM terminals will cause electrical and mechanical transients
in the IM. The dynamic performance of the IM during the sag
can be described by the following electromechanical equation:
(7)
Due to small time constants (fast response of electrical cir-
cuits of the IM), the electrical transients in the IM are mostly
finished before the onset of the mechanical transients [3], [5].
The electrical torque in the machine therefore will reduce in
proportion to the square of the rms value of the voltage sag
magnitude, while the mechanical torque will remain largely un-
changed. Due to the reduction in electrical torque, the motor
operates at a new operating point given by intersection of the
motor-speed curve during the voltage sag and load torque char-
acteristic, as shown in Fig. 3. The speed of motor reduces from
the nominal value at time to that at the end of the
sag at time along the load torque curve. Depending
on the duration of the voltage sag, the during voltage sag speedwill be between and .
Under balanced or unbalanced voltage sags, (6) can be trans-
formed to a first-order differential (8) which, when solved, gives
the expression (9) for or motor deceleration speed
(8)
(9)
where
At the end of the sag, the balanced nominal voltage supply is
restored, resulting in positive-sequence torque only. This now
causesthe motorto reacceleratefrom at time toat time . During this condition, (6) can be transformed to (10).
The solution of (10) gives the expression (11), which represents
voltage recovery speed or motor acceleration speed
(10)
(11)
where and
.The reacceleration time from to , at the end of
the sag, is given by
(12)
where .
E. Machine Current Variations During and After Voltage Sag
Positive- and negative-sequence currents during the sag are
given by
(13)
(14)
where ; and
At the onset of the voltage sag, the positive-sequence current
falls almost instantly with a drop in terminal positive-sequence
voltage. As the voltage sag magnitude stays constant during the
sag (for a sag duration ), the positive-sequence currentincreases gradually followed by a gradual increase in motor slip
7/29/2019 milanovic_2008E
4/9
1066 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 2, APRIL 2008
during the sag; hence, the motor slows down. How-
ever, in the case of unbalanced operation, the negative-sequence
current rises instantly with the appearance of voltage unbalance
and as negative-sequence voltage stays constant during the sag,
the current remains almost unchanged with little or a negligible
change in slip (as ).
Similarly, positive- and negative-sequence currents after thesag are given by
(15)
(16)
where .
At the start of voltage recovery, the voltage rises to nominal
balanced value and the slip is large (depending on the speed
drop seen at the end of the sag). These contribute to a large
positive-sequence inrush current (to build up the flux in the
air gap) which gradually decays as the motor accelerates (de-
creasing motor slip) to nominal speed. As the nominal and bal-
anced voltage is restored, the negative-sequence current dropsclose to zero.
III. INDUCTION MOTORVOLTAGE SAGS INTERACTION
A. Basic Change in Sag Shape Due to IM Dynamics
As a result of this continuous variation in the motor cur-
rents, the motor terminal voltages will not remain constant and a
change in terminal voltages occurs in accordance with the vari-
ation in currents. To describe this phenomenon, a simple test
system consisting of an aggregated IM connected to the infi-
nite (system) bus through a system impedance , as shown
in Fig. 1, is considered.
During the voltage sag, positive- and negative-sequence volt-
ages at the system bus are calculated using (1) and positive- and
negative-sequencecurrents at the end of voltage sag are obtained
from (13) and (14). Once the initial conditions ( and
) are established, the variation in IM terminal volt-
ages (i.e., voltages at the PCC) are calculated from
(17)
(18)
(19)
(20)
Similarly, positive- and negative-sequence (21), (22) voltages
at the system bus ( and ) after the sag can
be obtained. Finally, the variation in IM terminal voltages is
given by (23) and (24)
(21)
(22)
(23)
(24)
It can be seen from (23) and (24) that the change in IM ter-
minal voltage depends on the motor current, system voltage, and
impedance. The zero-sequence voltage during the sag is not in-fluenced by the motor dynamic behavior and, hence, remains
Fig. 4. IM dynamic responses following the asymmetrical three-phase voltagesag (sag parameters given in Appendix B) obtained with the Matlab model.
constant. Finally, the phase components of IM terminal voltages
during and after the sag are ob-
tained from
(25)
(26)
Responses of the aggregate IM to an asymmetrical voltage
sag obtained using (1)(26) are shown in Fig. 4. The top dia-
gram in this figure illustrates the original voltage sag applied,
and the subsequent diagrams present speed, voltage, and cur-
rent responses of the IM.
B. Model Validation
The developed IM model has been partially validated in [5]
by comparing the IM speed response (following a three-phase
symmetrical sag) with a previously published speed response
obtained with the more detailed model of the IM. A good agree-ment between the two sets of results was found and discussed in
detail in [5]. In order to validate the model further, in particular
responses of IM to asymmetrical sags, the test system of Fig. 1
is modeled in PSCAD/EMTDC as shown in Fig. 5.
The model consists of an aggregated IM of three identical
squirrel-cage induction motors (with the parameters similar to
those in Appendix A) connected to the PCC bus (as shown in
Fig. 5). A voltage sag generator originally developed in [10]
is also connected to a PCC bus via a system impedance de-
fined in Appendix A. Classical transient simulations with full
dynamic models of IMs are performed in PSCAD/EMTDC to
observe the motor transient response to voltage sags. In order to
initiate the PSCAD/EMTDC model smoothly, all motors werebegan with 0.07914-p.u. torque, and then when they reached full
7/29/2019 milanovic_2008E
5/9
MILANOVIC et al.: INFLUENCE OF INDUCTION MOTORS ON VOLTAGE SAG PROPAGATIONPART I 1067
Fig. 5. IM connected to a voltage sag generator through system impedance.
Fig. 6. Comparison of Matlab (dashed lines) and PSCAD (solid lines) results.
speed, switched to a full load (0.7914 p.u.). Fig. 6 compares IM
terminal voltages following an asymmetrical three-phase sag,obtained with Matlab (dashed lines) and PSCAD (solid lines)
models. The top pair of curves represent Phase A voltages, the
middle pair is Phase B voltages, while the bottom pair repre-
sents Phase C voltages. It is evident from Fig. 6 that the re-
sults of Matlab and PSCAD/EMTDC agree well, not only qual-
itatively but also quantitatively, (both in terms of sag magni-
tude and duration) with a maximum difference occurring during
the sag voltage magnitudes at the beginning of the sag of less
than 10%. (Note: The sag magnitude that matters is the min-
imum value of the rms voltage during the sag, the value that is
usually recorded as the equivalent sag magnitude in commonly
used in sag representation tables. The difference in this valuebetween two sets of curves, as can be seen from Fig. 6, is far
less than 10%.) The observed difference can be attributed not
only to simplifications made in the development of the Matlab
model [based on (1)(26)] but also to the originally different IM
model used in PSCAD/EMTDC. Namely, the IM model avail-
able in the PSCAD/EMTDC library is based on double-cage IM.
A similar or better agreement between responses obtained with
Matlab and PSCAD/EMTDC models was observed for other
types of voltage sags. Therefore, it can be deduced that the sim-
plified IM model developed in Section II can be used to ade-
quately describe the IM behavior when subjected to any type of
single stage (constant sag magnitude for the whole duration of
the sag), rectangular voltage sag, and, consequently, the IM in-fluence on voltage sag shape.
Fig. 7. Comparison of the IM speed change with respect to symmetrical andasymmetrical sags (adopted from [4]).
Fig. 8. Comparison of the IM speed change with respect to asymmetrical sagswith a small and large amount of negative-sequence contents.
The results of the analysis obtained for different case studies
are presented in Sections III-C and D. All results are obtained
with the model developed in Section II.
C. Effects of Voltage Sags on IM
Fig. 7 (reproduced from [5] for the completeness of dis-
cussion) illustrates that the voltage sags of different types
and severity will result in different variations in motor speed.
For this specific case, a symmetrical voltage sag caused by
a single line-to-ground fault results in a slight drop in speed
compared to that caused by a symmetrical three-phase voltage
sag. According to the previous discussion, it can be concluded
that this particular asymmetrical voltage sag contributes only
to a small amount to the negative-sequence voltage component
while its positive-sequence component is much higher than thatdeveloped by the symmetrical sag. In the same cases, however,
unbalanced voltage sags might cause a significant drop in speed
as shown in Fig. 8. This is particularly the case when they are
associated with large negative-sequence voltage components.
Fig. 9 illustrates the influence of a combined inertia of motors
and loads on IM behavior during the sag. It can be seen that the
IM with a low inertia load decelerates rapidly during the sag
though it also restores the nominal speed soon after the voltage
sag is over. In contrast, a steady slow speed drop and a slow
speed recovery can be expected in the case of the motor with a
high inertia load.
As far as the influence of motor parameters on the change in
speed is concerned, Fig. 10 shows that different motor param-eters will cause different changes in motor speed. The greater
7/29/2019 milanovic_2008E
6/9
1068 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 2, APRIL 2008
Fig. 9. Comparison of the induction motor speed changes with respect to lowand high inertia loads.
Fig. 10. Comparison of the IM speed changes with respect to low and highrotor resistances.
Fig. 11. Effects of the dynamic response of 168 aggregated small IMs on sagswith different magnitude.
speed drop can be expected in case of motors with high rotor
resistance. Speed recovery time, however, is not affected signif-
icantly.
D. Effects of IM Dynamic Responses on Voltage Sags
It can be seen from Figs. 11 and 12 that the shape of voltage
sags is not rectangular when the effect of IMs is taken into ac-
count. Additionally, Fig. 13 shows that different groups of IMs
differently influence the shape of the resulting voltage sags.
(The effect of 168 aggregated small IMs is particularly pro-
nounced).
Table I shows that the output power produced by groups 3 and
5 is almost the same though group 5 needs a longer recovery
time (i.e., 1.38 s versus 0.48 s). Fig. 14, however, shows an al-most identical influence of individual (small and large) IMs on
Fig. 12. Effects of dynamic responses of 168 aggregated small IMs on sags
with different magnitude and duration.
Fig. 13. Different changes in the shape of voltage sags due to different groups
of IMs.
TABLE I
GROUPS OF IMS USED IN THE STUDY
Fig. 14. Effects of a single large or small induction motor on the change in theshape of the voltage sag.
voltage sag characteristics. The change in sag characteristics is
highly dependent on the number and type of motors involved in
the group.
Fig. 15 illustrates that different inertia of the group results in
a small change in sag shape even dough the difference in inertia
is significant. A motor driving a low inertia load will recover abit faster than the one driving a high inertia load.
7/29/2019 milanovic_2008E
7/9
MILANOVIC et al.: INFLUENCE OF INDUCTION MOTORS ON VOLTAGE SAG PROPAGATIONPART I 1069
Fig. 15. Effects of inertia on the change in the shape of voltage sag.
Fig. 16. Effects of dynamics responses of IMs on voltage sags with a differentmagnitude.
Fig. 17. Effects of dynamics responses of IMs on voltage sags with differentduration.
By applying this methodology, voltage sag performance at
any given bus, with or without IMs, can be explored. For this
purpose, voltage sags can be conveniently arranged in three dif-
ferent categories: severe, with a magnitude of less then 0.5 p.u.;
medium with a magnitude in the range of 0.50.7 p.u.; and
shallow with a magnitude greater than 0.7 p.u. Figs. 16 and
17 show that IM dynamics do not change the characteristics
of every voltage sag. (Note: the original rectangular sags (sags
at the bus without the IMs) corresponding to those shown in
Figs. 16 and 17 are illustrated in Figs. 18 and 19, respectively).
This is particularly true for very shallow voltage sags, with a
magnitude of p.u., or for those lasting less than 0.08 s.(Note: The change in duration of sags with magnitudes
Fig. 18. Original rectangular sags for nonrectangular sags of Fig. 16.
Fig. 19. Original rectangular sags for nonrectangular sags of Fig. 17.
Fig. 20. Nonrectangular sag (solid line) originated by a rectangular0.6.-p.u./0.3-s sag and its respective equivalent rectangular sag (dashedline).
p.u. is not relevant anyway since their magnitude is above the
sag threshold of 0.9 p.u.) However, when the sag magnitude is
p.u., or when the duration is s, the IM dynamicsalter sag characteristics as shown by the solid line in Fig. 20.
As a result of this, the rectangular voltage sag becomes nonrect-
angular, and the methodology proposed in Section IV should
be applied to convert it back into the equivalent rectangular sag
(shown by the dashed line in Fig. 20) for benchmarking pur-
poses.
IV. NONRECTANGULAR TO RECTANGULAR SAG CONVERSION
A major consequence of the IM dynamics on voltage sag
characteristics is the change in sag shape from rectangular to
nonrectangular as illustrated in the previous section. The non-
rectangular part of the sag has basically a time-varying magni-tude [as illustrated in Fig. 21(a)] and, as such, is not suitable for
7/29/2019 milanovic_2008E
8/9
1070 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 2, APRIL 2008
Fig. 21. IMs connected to a voltage sag generator through system impedance.(a) Voltsseconds lost due to IM nonrectangular sag (shaded area). (b)Voltsseconds lost due to new IM rectangular sag (shaded area).
inclusion in existing magnitude-duration tables used for bench-
marking purposes.
Any conversion of nonrectangular to rectangular shape is as-
sociated with a certain level of simplification and the equivalent
rectangular sag becomes an approximation of the actual distur-
bance. Even though it is an approximation, it still represents the
actual voltage sags in low-voltage networks (in particular) more
accurately then current practices. The existing approaches al-
most completely ignore the presence and, consequently, the ef-
fects of IMs on voltage sags in the network. One possible way
for converting nonrectangular to rectangular voltage sags is to
use the loss of voltage concept/index [11] and (27). In thisway, the original nonrectangular and corresponding rectangular
voltage sags have the same loss of voltage (or voltseconds)
(27)
Fig. 21 illustrates the methodology for voltage sag conversion
based on the loss of voltage concept. Consider an IM exposed
to rectangular sag represented by the dashed curve abcefijm in
Fig. 21(a). Due to the effect of IM dynamics, this sag is changed
to a nonrectangular sag represented by the solid line abcfihm.
Using (27) and considering only the area below 0.9-p.u. voltage
threshold, the nonrectangular sag represented by the shaded areain Fig. 21(a) is converted to a rectangular sag represented by the
Fig. 22. Study procedure for sag conversion.
shaded area in Fig. 21(b). A new rectangular sag, represented by
the dashed curve abcdghk, is equivalent to the original nonrect-
angular sag in terms of volt-seconds lost. (Note: The original
integration step of 20 ms for determining the nonrectangular
voltage sags by procedure described in Section II is reduced in
this case to 1 ms to increase the accuracy of the conversion.)
The original nonrectangular sag is shown in Fig. 21(a) by a
solid line (only the part below 0.9-p.u. voltage threshold is used
for conversion). The curve bcefih is defined by a set of points
and it encompasses the area
(28)
with a sag duration given by
(29)
The new rectangular sag will have the same duration as the
original sag while its magnitude will be reduced and given by
(30)
Using (29) and (30), the original nonrectangular sag (bcefih)
is transformed into an equivalent rectangular sag with magni-
tude and duration , as shown in Fig. 21(b).
The consequence of this conversion is that the resulting rectan-
gular sags have a slightly higher (and constant) magnitude than
the original nonrectangular sags and the same duration (only the
part below 0.9-p.u. magnitude is considered in accordance with
the definition of voltage sags). Since this conversion is applied
to all three phases independently, the resulting rectangular sags
may have different sag durations as illustrated and discussed in
detail in the adjoining paper (Part II).
A. Study Procedure
The study procedure is divided into two parts (see Fig. 22):
IM dynamic response analyzer and nonrectangular to rectan-
gular sag converter.
1) Part 1: Induction Motor Dynamic Response Analyzer:
Using the input data (IM equivalent circuit data, original
rectangular sag parameters, number of motors connected at
the PCC and system impedance) and the model developed in
Section II, this part calculates the change in sag parameters at
IM terminals using the following algorithm.
Parameters of the equivalent aggregate IM connected at
PCC are calculated based on parameters of individual IMs.
The original (without the effect of IMs) pre, during, andafter sag voltages are calculated using (1)(3).
7/29/2019 milanovic_2008E
9/9
MILANOVIC et al.: INFLUENCE OF INDUCTION MOTORS ON VOLTAGE SAG PROPAGATIONPART I 1071
The aggregate IM speed during the sag is calculated every
millisecond of the sag duration using (12) and aggregate
motor parameters.
Motor reacceleration time is computed using (15).
Motor speed after the sag is calculated every millisecond
during the reacceleration of the motor using (14).
The motor currents during and after the sag are determinedusing (1821).
Finally, the change in sag shape due to IM dynamics is
determined using (22), (23), and (26)(29)
2) Part 2: Nonrectangular to Rectangular Sag Converter:
This block uses the output from the part 1 (i.e., nonrectangular
voltage sag parameters) and converts them to parameters of
the equivalent rectangular voltage sag using the following
algorithm.
The voltseconds lost in each phase are calculated using
(30). Only the part of the voltage sag curve that falls below
the defined voltage sag threshold (90% of nominal voltage)
is considered in this calculation.
The sag duration in each phase is calculated using (34). The equivalent rectangular sag magnitude in each phase is
calculated using (35).
V. CONCLUSION
This paper presented an analytical tool to describe the influ-
ence of voltage sags on a group of IMs (aggregated model) and
the influence of IM dynamics on voltage sag characteristics.
The results show that the presence of IM(s) at a bus may
initially provide voltage support, however, during the voltage
recovery, the sag duration may increase due to the reaccelera-
tion of IM(s). The change in voltage sag shape due to IM dy-
namics depends on various parameters, such as number of mo-tors, motor electrical and mechanical parameters, number of
sagged phases, and severity of voltage sag.
To facilitate the assessment of the impact of IMs on voltage
sag performance of a bus using conventional benchmarking ta-
bles, an analytical nonrectangular to rectangular sag conversion
method is introduced. The method accounts for the effect of IMs
on sag characteristics and enables quick and convenient repre-
sentation of resulting voltage sags using conventional bench-
marking approaches. Further details about the sag conversion
method and practical application issues are discussed in detail
in the adjoining paper (Part II).
APPENDIX A
LARGE INDUCTION MOTOR AND SYSTEM PARAMETERS:
Eight poles V/ph, kW,
rad/s, rad/s, ,
, , ,
, , Nm,
, .
APPENDIX B
SMALL INDUCTION MOTOR PARAMETERS
Four poles V/ph, kW,
, , ,
, , ,Nm
APPENDIX C
THREE-PHASE TEST ASYMMETRICAL SAG PARAMETERS
Phase A presag p.u., ; during
the sag p.u., , ms.
Phase B presag p.u., ;
during the sag: p.u., ,
ms.Phase C presag p.u., ; during
the sag: p.u., , ms.
ACKNOWLEDGMENT
The authors would like to thank the Microgrids and RISE
partners for their contributions.
REFERENCES
[1] W. E. Brumsickle, R. S. Schneider, G. A. Kuckjiff, D. M. Divan, and
M. F. Mcgranaghan, Dynamic sag correctors: cost-effective industrialpower line conditioning, IEEE Trans. Ind. Appl., vol. 37, no. 1, pp.212217, Jan./Feb. 2001.
[2] R. Gnativ and J. V. Milanovic, Voltage sag propagation in systemswith embedded generation and induction motors, in Proc. IEEE Power
Eng. Soc. Summer Meeting, Jul. 2001, vol. 1, pp. 474479.[3] M. H. J. Bollen, Understanding Power Quality Problems: Voltage Sags
and Interruptions. New York, NY: IEEE, 1995, pp. 153248.[4] D. L. Brooks, E. W. Gunther, and A. Sundaram, Recommen-
dations for Tabulating RMS Variation Disturbances With Spe-cific Reference to Tility Power Contracts, Cigr 36.05/CIRED2 CC02Voltage Quality Working Group. [Online]. Available:http://www.ccu2.org/count-d2.pdf.
[5] M. T. Aung and J. V. Milanovic, Analytical assessment of the effectsof voltage sags on induction motor dynamic responses, in Proc. IEEESt Petersburg PowerTech, St. Petersburg, Russia, Jun. 2730, 2005.
[6] Matlab Help File. Natick, MA, Mathworks, 2004.[7] PSCAD 4.1.1 On-Line Help File. Winnipeg, MB, Canada, Manitoba
HVDC Research Centre, 2004.[8] D. C. Franklin and A. Morelato, Improving dynamic aggregation of
induction motor models, IEEE Trans. Power Syst., vol. 9, no. 4, pp.19341741, Nov. 1994.
[9] Y. J. Wang, An analytical study of steady-state performance of aninduction motor connected to unbalanced three-phase voltage, in Proc.
IEEE Power Eng. Soc. Winter Meeting, Jan. 2000, vol. 1, pp. 159164.[10] S. Z. Djokic, J. V. Milanovic, and K. A. Charalambous, Computer
simulation of voltage sag generator, in Proc.10th IEEE Int.Conf. Har-monics Quality Power, Rio de Janeiro, Brazil, Oct. 69, 2002.
[11] IEEE Recommended Practice for Monitoring Electric Power Quality,IEEE Std. 1159-1995, 1995.
Jovica V. Milanovic (M95SM98) received the Dipl.Ing. and M.Sc. degreesfrom the University of Belgrade, Belgrade, Yugoslavia, and the Ph.D. degree
from the University of Newcastle, Newcastle, Australia.Currently, he is a Professorof Electrical Power Engineering and DeputyHead
of School (Research) of Electrical and Electronic Engineering, The Universityof Manchester, Manchester, U.K.
Myo T. Aung (M05)receivedthe B.E.degree in electrical engineering from theYangon Technological University, Rangoon, Burma, the M.E. degree in electricpower system management from the Asian Institute of Technology, Bangkok,Thailand, and the Ph.D. degree from the University of Manchester (formerlyUMIST), Manchester, U.K.
Currently, he is a Lecturer with the Department of Electronic and ElectricalEngineering at the University of Bath, Bath, U.K.
Sarat C. Vegunta received the B.E. degree in power electronics from NagpurUniversity, Nagpur, India, and the M.Sc. degree from the University of Man-
chester Institute of Science and Technology (UMIST), Manchester, U.K.He is currently a Ph.D. student in the School of Electrical and Electronic
Engineering at The University of Manchester, Manchester.