1

Short Circuit Analysis of Induction Machines –

Wind Power Application
Dustin F. Howard, Graduate Student Member, IEEE, Travis M. Smith, Senior Member, IEEE,
Michael Starke, Member, IEEE, and Ronald G. Harley, Fellow, IEEE

 conditions to aid in the protection system design for wind

Abstract--The short circuit behavior of Type I (fixed speed) power plants.
wind turbine-generators is analyzed in this paper to aid in the The various WTG types are discussed in detail in [1].
protection coordination of wind plants of this type. A simple This paper focuses on the Type I WTG; a fixed speed, squirrel
network consisting of one wind turbine-generator is analyzed for
cage induction generator with power control through the
two network faults: a three phase short circuit and a phase A to
ground fault. Electromagnetic transient simulations and turbine blade pitch angle. Previous work analyzing the short
sequence network calculations are compared for the two fault circuit behavior of induction machines within power systems
scenarios. It is found that traditional sequence network has typically considered balanced three phase short circuits,
calculations give accurate results for the short circuit currents in and the analysis for this particular kind of fault is described in
the balanced fault case, but are inaccurate for the un-faulted detail in [2]-[4]. The analysis in [2]-[4] is useful for
phases in the unbalanced fault case. The time-current behavior
determining the maximum short circuit current expected to
of the fundamental frequency component of the short circuit
currents for both fault cases are described, and found to differ flow in the induction machine and the rate at which the current
significantly in the unbalanced and balanced fault cases. decays after the fault. Unbalanced faults, however, are more
common in the power system; thus detailed analysis for the
Index Terms--Generators, Induction generators, Power system unbalanced fault cases are also critical to ensure the wind
faults, Power system planning, Power system protection, Power turbine and interconnecting network are protected against
system transients, Smart grids, Wind energy, Wind farms, Wind these faults. Unbalanced operation of power systems is
power generation, Wind generators.
typically analyzed using symmetrical components [5][6],
which is a steady state analysis technique. Symmetrical
I. INTRODUCTION
component theory is applied to unbalanced power system

E nvironmental concerns, energy security, and higher

energy costs are driving a worldwide increase in the
production of electricity from wind energy. As wind power
operation through the assembly of sequence network circuits,
which are often used in fault studies and protection planning.
Analysis of induction machines under symmetrical and
capacity continues to increase as a part of the world’s energy unsymmetrical faults is provided in [7], but that analysis
portfolio, the reliability of the electric power system should assumes a machine with a high X/R ratio for the machine
not be compromised to accommodate for this growing wind parameters and assumes that the machine is connected to an
capacity. Fault protection, among many other factors, is infinitely “stiff” source, which is not always the case.
essential for maintaining a reliable and resilient power system. Electromagnetic transient (EMT) simulations are generally
Fault protection systems are intended to isolate the faulted accepted as sufficiently accurate for analyzing the short circuit
components of the system and minimize the impact on the rest behavior of WTGs. In [8], EMT simulations are compared
of the network. Further, improper fault protection design can with sequence network calculations, and found to have good
lead to equipment damage or personnel injury. Wind power agreement for a three phase fault and a phase A to ground
plants present a new challenge for protection system design, fault. However, only the faulted phase (phase A) was
since they behave differently than conventional power plants compared in that study with the EMT simulation.
under system faults. This paper describes the behavior of For protection purposes, the fundamental frequency
Type I wind turbine-generators (WTGs) under faulted (50/60 Hz) component of the short circuit current is of primary
interest. Therefore, important parameters for protection
This material is based upon work supported in part by the National studies are the fundamental frequency initial short circuit
Science Foundation under Grant No. ECS1028546. Any opinions, findings, current and its time-rate of decay. For the three phase
and conclusions or recommendations expressed in this material are those of
the author(s) and do not necessarily reflect the views of the National Science
balanced short circuit case, the sequence network calculation
Foundation. Additional support is acknowledged from the U.S. Department gives the initial short circuit current at the instant of the fault,
of Energy and Oak Ridge National Laboratory as a part of the Grid Innovation and the decay rate of the short circuit current is based on the
Leaders Fellowship Program, Contract No. DE-AC05-00OR22725 and
Subcontract No. 4000105351.
parameters of the machine. However, for unbalanced faults,
Dustin F. Howard and Ronald G. Harley are with the Georgia Institute of sequence network circuit calculations can provide the initial
Technology, School of Electrical and Computer Engineering, Atlanta, GA, short circuit current value, but give errors in the calculations
USA (e-mail: [email protected], [email protected]). for the un-faulted phases, as explained in this paper.
Travis Smith and Michael Starke are with Oak Ridge National Laboratory,
Oak Ridge, TN 37831 USA (e-mail: [email protected], [email protected]). Additionally, the rate of decay of the short circuit current for

978-1-4673-1935-5/12/$31.00 ©2012 IEEE

 2

unbalanced faults does not follow an exponential decay as in 1.5

the balanced fault case, and also depends on the parameters of 1

Phase A

Machine Voltages (pu)

Phase B
the interconnecting network, not only the machine parameters. 0.5 Phase C
A review of the sequence network circuit representations 0

of induction machines is provided in this paper, and -0.5

calculations are performed on a simple network for a single -1

line-to-ground (SLG) fault and three line-to-ground (TLG) -1.5

-0.05 0 0.05 0.1 0.15 0.2
fault. EMT simulations are run for the same network and the time (sec)
same faults as used for the sequence network calculations, and
the two analysis techniques are compared by using a curve Phase A

Machine Currents (pu)

10 Phase B
fitting algorithm on the simulation results and extracting the Phase C
fundamental frequency component of the waveforms. The 0
time-current behavior of the fundamental frequency
component of the short circuit currents is analyzed using EMT -10

simulation results and mathematical techniques.

-0.05 0 0.05 0.1 0.15 0.2
time (sec)
II. EMT SIMULATIONS (a)
EMT simulations can accurately represent the short circuit 1.5

dynamics of a wind turbine-generator, and are used as a 1

Phase A

Machine Voltages (pu)

Phase B
reference in this study to examine closed form analytical and 0.5 Phase C

“short-hand” calculation methods of short circuit currents. 0

Short circuit calculation methods involving electric machines -0.5

are usually concerned with the RMS value of the short circuit -1
current immediately after the fault occurrence. Obtaining this -1.5
-0.05 0 0.05 0.1 0.15 0.2
current value is not always straightforward with EMT time (sec)
simulations since the current waveforms contain DC offsets
and are decaying with time. A method for extracting the Phase A
Machine Currents (pu)

10 Phase B
initial RMS value and rate of decay of the short circuit current Phase C
from the EMT simulations is given in this section. 0
EMT simulations of the network shown in Fig. 1 are run
using PSCAD/EMTDC. The parameters of the machine are -10
given in the Appendix. The machine speed is set to 1.01 pu
and remains constant for the duration of the fault. The fault is -0.05 0 0.05 0.1 0.15 0.2
time (sec)
applied directly to the machine terminals at some arbitrary (b)
time. The wye-connected stator winding neutral is Fig. 2. Machine terminal voltage and current under a (a) three phase short
ungrounded, thus no zero sequence current flows in the circuit and (b) phase A to ground fault.
machine. The machine is connected through an impedance In short circuit studies, it is useful to have an analytical
(0.01 + j0.05 pu) to an ideal voltage source (600 V, 60 Hz). expression for the short circuit currents for quick calculations
Simulation results for the machine voltage and currents are such as the maximum fault current magnitude, initial RMS
shown in Fig. 2(a) for a three phase short circuit and Fig. 2(b) fault current, and rate of decay of the fault current. Obtaining
for a phase A to ground fault. The quantities used for per- these values using EMT simulations is cumbersome, since
unitization of the voltages and currents are given in the many of these parameters vary depending on the angle on the
Appendix. In each case, the machine currents sharply increase voltage waveform at which the fault occurs. In [7], an
after the fault is applied, and decrease in magnitude after a few approximate solution for the short circuit currents of a
cycles. For the three phase fault case, the currents decay to squirrel-cage induction machine under balanced or unbalanced
zero since there is no longer a source to magnetize the faults is of the form given by
machine. However, in the single line to ground fault case, the
i (t )  I t e  t / Tt sin r t     I dc e  t / Tdc  I ss sin s t    (1)
two healthy phases can maintain flux in the machine, thus a
new, unbalanced steady state operating point is reached after Equation (1) indicates that the short circuit current contains a
several cycles if the fault fails to clear. transient AC component (It) which decays to zero and a DC
component (Idc) which decays to zero. For an unbalanced
fault, a steady state current is present since the un-faulted
phases maintain flux in the machine. However, for a three
phase short circuit, the steady state component is zero.
Equation (1) provides the decay time constants of the AC and
DC components of the short circuit current, and the initial
fundamental frequency, RMS short circuit current phasor can
Fig. 1. Network used for EMT simulations of short circuit faults.
 3

be found by assuming the rotor speed is approximately is at 1.01 pu. For the SLG fault fitted waveform data shown in
synchronous speed (ωr ≈ ωs) and setting t = 0 in (1), giving Table II, the machine steady state and transient component (It
 I I and Iss) are all different since the fault is unbalanced.
I  t e j  ss e j (2) Additionally, the time constants of the current decay are
2 2
different than that of the TLG fault since the interconnecting
where the DC component has been neglected since the
network affects the current behavior in the SLG fault case.
fundamental frequency component is of interest. In Type I
The data given in Tables I and II for the simulated short
WTGs, the operating slip is typically 1% or less, thus
circuit waveforms is used in the rest of the paper as a
assuming the rotor speed is roughly equal synchronous speed
reference to compare calculation methods for the initial short
is a good approximation.
circuit currents and time rate of decay of the short circuit
To obtain the parameters of (1) for each phase current
currents. These fitted waveform results are used as the
displayed in Fig. 2, the EMT simulation data is imported into
reference since the fitted waveforms are close enough to the
Matlab’s Curve Fitting Toolbox. Given the general form of
actual simulated waveforms.
the time-domain waveform in (1), and some initial “first-try”
values for the amplitudes, phase angles, and time constants in TABLE I
(1), the Curve Fitting toolbox can fit the EMT simulation data PHASE CURRENT PARAMETERS IN (1) FOR THREE PHASE SHORT
to a waveform using a non-linear, least squares method. A CIRCUIT (CURRENTS IN PER-UNIT)
custom Matlab script is written to utilize this toolbox, which Phase It Tt ωr θ Idc Tdc Iss ωs φ
ia 8.36 0.0413 380.2 -80.6 8.87 0.1047 0 0 0
outputs the various magnitudes, phase angles, and time
ib 8.35 0.0414 380.2 159.4 -4.36 0.0935 0 0 0
constants of (1) which fit the given phase current simulation ic 8.37 0.0413 380.2 39.4 -4.53 0.1157 0 0 0
data. Knowing these parameters is useful for comparing the
calculation methods in the following sections with the EMT TABLE II
simulation results. Performing the fitting function in this way PHASE CURRENT PARAMETERS IN (1) FOR PHASE A TO GROUND
on the phase currents shown in Fig. 2 produces the waveforms SHORT CIRCUIT (CURRENTS IN PER-UNIT)
Phase It Tt ωr θ Idc Tdc Iss ωs φ
in Fig. 3(a) for a three phase short circuit and Fig. 3(b) for a ia 2.71 0.0495 379.7 -79.1 5.32 0.0630 2.31 377 -61.7
phase A to ground short circuit. Only the phase A currents are ib 2.38 0.0495 379.6 157.3 -2.85 0.0584 1.96 377 16.2
displayed in Fig. 3 due to limited space, but similar results ic 2.43 0.0487 379.8 45.9 -2.48 0.0680 3.33 377 153.3
were obtained for phases B and C. The fitted curve calculated
by Matlab is overlaid on the actual phase currents in Fig. 3, III. SEQUENCE NETWORK ANALYSIS
and match the EMT simulation results very closely. The Sequence network analysis is a steady-state, fundamental
parameters of (1) calculated for each phase current for the frequency analysis method often used to study unbalanced
TLG and SLG fault cases are given in Table I and Table II, systems. Typically, sequence network analysis is divided into
respectively (all currents are given in per-unit). For the TLG three different time segments: sub-transient, transient, and
fault case, the steady state components (Iss, ωs, and φ) are zero steady state. Since sequence network analysis is steady state
since the machine current decays to zero after the fault. The analysis, these different time segments are used to capture
transient components of the fault currents are roughly equal, “pieces” of the time-varying nature of the circuits. For fault
as expected since this is a balanced fault. Additionally, ωr is studies, the sub-transient and transient time segments are of
slightly above synchronous frequency since the machine speed interest. The induction machine has a different representation
in each of the above time segments. The sequence network
Phase A Machine Current (pu)

15 Current model of induction machines in the sub-transient period is

Fitted Curve
represented using a sub-transient reactance and voltage behind
10
sub-transient reactance [3]. Similarly, the induction machine
5 is modeled as a transient reactance and voltage behind
transient reactance in the transient time segment. Sub-
0
transient effects are present in squirrel-cage induction
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
time (sec)
machines with either a double-cage rotor or large induction
(a) machines with deep rotor bars [3], but will not be considered
10 in this study. The sub-transient effects cause an increase in
Phase A Machine Current (pu)

Current the fault current magnitude immediately after a fault, but

Fitted Curve
5
generally die away much faster than the transient effects.
The network in Fig. 1 is used again for sequence network
analysis and Fig. 4 is the per-phase, steady state equivalent
0
circuit of the network in Fig. 1. The sequence network
representation of the induction machine is calculated from the
0 0.02 0.04 0.06 0.08 0.1
time (sec)
0.12 0.14 0.16 0.18 0.2
circuit in Fig. 4. The steady state current flowing in the circuit
(b) in Fig. 4 can be calculated by
Fig. 3. Machine phase A current and fitted curve under a (a) three phase short
circuit and (b) phase A to ground short circuit.
 4


 Vs terminals. The sequence components of the currents are
Is    (3) calculated based on the sequence network circuit, and
Zs  Zm
transformed to phase quantities using (8). These calculations,
where as described in the following sub-sections, are compared with

Z s  RL  jX L (4) the simulation results given in the previous section.
and A. Three Phase Short Circuit
 jX m ( Rr / s  jX lr )
Z m  Rs  jX ls  (5) Using the V’ and X’ calculated in (6) and (7), the transient
Rr / s  j ( X m  X lr ) sequence network of the network shown in Fig. 1 for a three
A derivation of the transient reactance and voltage behind phase short circuit is shown in Fig. 5. Since the faulted
transient reactance representation of an induction machine is network in Fig. 1 maintains symmetry, only positive sequence
detailed in [2], and is not repeated here. The theory is based current flows. There is no steady state sequence network for
on the principle that the flux linkages in the machine cannot the three phase short circuit case because the currents go to
change instantaneously, thus the induced emf in the machine zero in steady state. At the instant of the fault, the switch in
remains constant for a short time immediately after a change Fig. 5 closes, thus short circuiting the machine terminals.
in the terminal voltage. The transient reactance X’ and voltage Thus, the fault currents flowing in the machine are given by

behind transient reactance V’ of an induction machine are  V'
calculated by [2][3] I  (10)
Rs  jX '
X m X lr Calculation of the fault current using (10), inserting this value
X '  X ls  (6)
X m  X lr into (8) for Ia+ (with Ia- and Ia0 set to 0) gives the results shown
and in Table III, along with the comparable quantities from the
   
V '  Vs  ( Rs  RL ) I s  j ( X ' X L ) I s (7) simulation parameters displayed in Table I applied to (2).
Comparing the magnitude of the short circuit current phasor
The resistances are often omitted in the calculation of V’, but
calculated using (10) and the magnitude of the short circuit
are included here for completeness.
current phasor from the data in Table I (as applied in (2))
In sequence network analysis, the voltages and currents are
gives the percent error as displayed in the right column of
found using circuit analysis, where the values are determined
Table III. The error is less than 1% for each phase current
as positive, negative, and zero sequence quantities. Once the
calculation, indicating that this method is quite accurate for a
particular currents or voltages of interest are determined, the
three phase short circuit.
actual phase quantities are found from the symmetrical TABLE III
components transformation [6], given by COMPARISON OF CALCULATED IN SIMULATED SHORT CIRCUIT
  CURRENTS FOR A TLG FAULT
 I a  1 1 1   I a0 
  
  
 Seq. Net. Calc. Sim. Results
 Ib   1 
2
   I a  (8) Mag. (pu) Ang. (deg) Mag. (pu) Ang. (deg) % Error

 I  1   2   I   Ia 5.95 -76.7 5.91 -80.6 0.7 %
 c   a  Ib 5.95 163.3 5.90 159.4 0.8 %
j120° Ic 5.95 43.3 5.92 39.4 0.6 %
where α = e . A similar relationship holds for voltages,
fluxes, etc. Similarly, the transforming from phase quantities B. Phase A to Ground Short Circuit
to sequence quantities is found using the inverse
The sequence network for the SLG fault case is more
transformation [6], given by
  complicated than the TLG fault case since positive, negative,
 I a0  1 1 1  I a  and zero sequence currents now flow in the circuit. The
  1 
 2  

 Ia   1     Ib  (9) steady-state sequence network of the diagram in Fig. 1 is
 I   3 1  2    I  shown in Fig. 6(a) for the SLG fault case. The positive
 a   c 
sequence network is the typical steady state equivalent circuit
Thus, calculating the sequence components of the current or of the induction machine. The negative sequence circuit is
voltage from a particular sequence network circuit gives the similar to the positive sequence circuit, except that the slip
actual phase quantities by using the symmetrical components term in the rotor circuit has changed. In negative sequence,
transformation matrix given in (8). the synchronous frequency is rotating in the opposite direction
Using the V’ and X’ representation of the induction machine as positive sequence, thus the slip is given by
given by (6) and (7), sequence network analysis of the s  r 
network shown in Fig. 1 is studied for a three phase short sn  1 r  2  s (11)

 s s
circuit and a phase A to ground short circuit at the machine
Xlr Xls Rs XL RL

Is
+
Rr Xm
s VS
-

Fig. 4. Per-phase, steady state equivalent circuit of network shown in Fig. 1. Fig. 5. Sequence network circuit for TLG fault.
 5

Xlr Xls Rs XL RL

+ +
Ia +
Rr Pos.
Xm Va+ Sequence VS
s Network
-
-
Xlr Xls Rs XL RL

- +
Ia
Rr Xm
Neg.
If /3
Va- Sequence
2-s Network

-
XL RL
+
Zero
Va0 Sequence
Network

(a) (b)
Fig. 6. (a) Steady state and (b) transient sequence network for single line to ground fault.
No zero sequence current flows in the machine since the stator
neutral is un-grounded. The steady state solution to the phase IV. PHASE CURRENT SENSITIVITY TO V’
currents under a SLG fault can be found by traditional circuit The results of the previous section indicate that the
analysis on the circuit shown in Fig. 6(a), and is equivalent to sequence network calculations may be less accurate for
the steady-state component of the short circuit current unsymmetrical short circuit calculations than for symmetrical
equation given in (1). short circuits. However, as displayed in Table IV, the results
The transient sequence network for the SLG fault case is calculated for the phase A current were quite accurate, yet the
shown in Fig. 6(b). The same V’ and X’ calculated in (6) and results for phases B and C produce much larger error. This is
(7) above are used in this case. The negative sequence particularly unusual since the same sequence components of
impedance is usually assumed to be the same as the positive the machine current calculated from the sequence network
sequence impedance in the transient period. However, there is circuit displayed in Fig. 6(b) are used to calculate all three
some variation in the values for X’ used in literature, where [2] phase currents. In other words, the Ia+ and Ia- inserted into (8)
defines it as given in (6), but [5] defines it as the sum of the give good results for phase A, but less accurate results for
stator and rotor leakage reactances. If the magnetizing phases B and C.
reactance is large compared to the leakage reactances, the two As a further study, the value of the voltage behind transient
different ways of calculating X’ are roughly the same. Solving reactance V’ is re-evaluated to find if there is a value to use in
the sequence network circuit for the currents Ia+ and Ia-, and the sequence network circuit shown in Fig. 6(b) which gives
substituting these into (8) gives the short circuit current better results for all three phase currents. The original V’
phasors immediately after the fault (Ia0 is set to 0). calculated using (7) is found to be 0.94ej11.8°. Therefore,
Calculation of the short circuit current phasors in this way is analysis of the sequence network circuit displayed in Fig. 6(b)
analogous to the calculation in (2) from the time-domain is performed for several different magnitudes and angles of V’
solution. The sequence network calculation results are around a neighborhood of the original calculated V’. The
displayed in Table IV, along with comparable quantities from results of this analysis appear in Fig. 7. The results are
the simulation data displayed in Table II applied to (2). As obtained by repeating the analysis method of the previous
displayed in Table IV, the calculation of the phase currents section for the SLG fault case, and using several combinations
using sequence network circuits produces some error, of magnitudes and angles for V’. The percent error is found
particularly in the phase B and phase C calculation, giving between the calculated phase current magnitude and the
roughly 14% and 7% error, respectively. These results reference phase current magnitude calculated using (2) with
indicate that using sequence networks to calculate short circuit the data in Table II. The results of Fig. 7 indicate that the
currents is not as accurate in the case of unsymmetrical faults phase A current calculation using sequence networks is less
as in the case of symmetrical faults. A discussion of the sensitive to the value used for V’, since the percent error
possible source of this error is given in the next section. displayed in Fig. 7(a) stays well below 10% for the entire
range of V’ phasors used. Phase B is very sensitive to the
TABLE IV
COMPARISON OF CALCULATED IN SIMULATED SHORT CIRCUIT
value used for V’, where the percent error reaches as high as
CURRENTS FOR A SLG FAULT 50% over the range of V’ studied, as displayed in Fig. 7(b).
Seq. Net. Calc. Sim. Results Similarly, the phase C current calculation shows higher
Mag. (pu) Ang. (deg) Mag. (pu) Ang. (deg) % Error sensitivity to the value used for V’ than the phase A current
Ia 3.54 -69.5 3.51 -71.1 0.8 % calculation. Upon examination of the three plots in Fig. 7, a
Ib 0.91 105.4 1.06 102.0 13.8 %
Ic 2.63 112.3 2.46 111.6 6.8 %
magnitude and angle for V’ can be found which gives nearly
0% error for all three phases, and is found to be approximately
 6

50
A. Three Phase Short Circuit
Based on (1), the short circuit current of an induction
50 40 machine after a TLG fault has a RMS magnitude given by
Phase A Current % Error

40  I
30
I (t )  t e  t / T (12) t

30
2
20
where the steady state component in (1) is zero and the DC
20
10 component in (1) is neglected. The decay of the fundamental
0 frequency component is determined by the time constant Tt in
0.98 10
0.96 16 (12). The value of this time constant can be calculated based
14
0.94 12 on the machine parameters [2][4][7], given by
0.92 10 0
Mag. V' (pu) 8 L'
Ang. V' (deg)
Tt  (13)
(a) Rr
50 where L’ = X’/ωs. Calculation of the time constant in this way
using the machine parameters given in the Appendix gives Tt
50 40 = 0.0415 sec. Note that this value matches closely with the
Phase B Current % Error

40 value of Tt estimated by the curve fitting as displayed in Table

30 30 I. Using the values of It and Tt from Table I in (12), the
20 fundamental frequency component of the phase currents from
10
20 the simulation results of Section II for a TLG fault gives the
0
curves shown in Fig. 8 (solid line). Additionally, the
0.98 10 magnitude of the phase current calculated using (10) and the
0.96 16
0.94
14 time constant calculated using (13) are applied to (12), and
12
0.92 10 0 plotted in Fig. 8 (dashed line). Note: the square root of two
Mag. V' (pu) 8
Ang. V' (deg)
term in (12) is not necessary from the sequence network
(b) results since this current is already an RMS value. Each phase
50 current overlaps in Fig. 8 since the fault is symmetrical, and
therefore the fundamental frequency components remain
50 40 balanced. Similarly, the calculated results overlap the
Phase C Current % Error

40 simulated results since the initial short circuit current

30 30 calculated using sequence networks and decay time constant
20 match well with simulations in the TLG fault case.
20
10
B. Single Line to Ground Short Circuit
0
0.98 10
In the SLG fault case, the total short circuit phase current is
0.96
14
16 the sum of two components: a transient and steady state
0.94
0.92 10
12
0
component, It and Iss, as described by (2). Both the steady
Mag. V' (pu) 8
Ang. V' (deg) state and transient components contribute significantly to the
(c) total short circuit current. The phasors for the steady state and
Fig. 7. Percent error of the (a) phase A, (b) phase B, and (c) phase C current transient components of the short circuit current from the data
calculations using sequence networks for varying magnitude and angle of V’. in Table II are displayed graphically in Fig. 9(a), with the
V’ = 0.94ej9.6°. Compared to the originally used V’ of transient components in blue and the steady state components
0.94ej11.8°, the correct V’ differs only in phase angle by in red (Note: the components are not drawn exactly to scale).
approximately 2.2°. This small difference in the value of V’ Notice that the phase A transient and steady state components
explains the high sensitivity of the phase B and C currents to have only a small phase shift between them, but the phase
the value used for V’, as this small difference produces a shift between the phase B and C components is large ( > 90°).
significant error in the current calculation, as described in the The phasors representing the total initial short circuit current
previous section. are found by adding the steady and transient components (as
1
10
RMS Machine Currents (pu)

V. TIME-CURRENT BEHAVIOR OF SHORT CIRCUIT CURRENT Phase A

Phase B
0
In the previous sections, the focus was on the calculation of 10 Phase C

the fundamental frequency, initial short circuit current after a

fault. In this section, the focus is on the behavior of the -1
10

current a few cycles after the fault occurrence to evaluate the

time-current behavior of the fault current waveforms. Again, -2
10
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
the TLG and SLG cases are discussed separately in the sub- time (sec)

sections to follow. Fig. 8. RMS machine current magnitude over time for a three phase short
circuit occurrence at t = 0 (solid = simulated, dashed = calculated).
 7

 I I
I (t )  t e  t / T e j  ss e j (14) t

2 2
Ict Calculating the RMS phase currents for the SLG fault case
Icss
using (14) and the data from the simulations given in Table II
Ibss
gives the (solid line) curves shown in Fig. 10. Phases B and C
Ibt
reach a minimum in the plot shown in Fig. 10, for reasons
explained above. Phase A decays steadily until a steady state
operating point is reached. Similarly, the fundamental
frequency, RMS magnitude of the phase currents calculated
Iass
using sequence network circuits is shown in Fig. 10 (dashed
Iat line). The calculated curves are found using (14), except the
square root of 2 term is not present since the currents
calculated using sequence networks are already RMS
(a)
quantities. The value of Iss is found by solving the sequence
network of Fig. 6(a) for the current Ia+ and Ia- and inserting
Ic these values into (8) to get the phase quantities. The value of
It is found by solving the sequence network in Fig. 6(b) for the
currents Ia+ and Ia-, inserting these quantities into (8) to obtain
Ib the phase quantities, and subtracting the steady state
component calculated previously, given by

I t e j  I  I ss e j (15)
where I is the total phase current calculated using the sequence
network of Fig. 6(b). The value of Tt used in the calculated
curves of Fig. 10 is calculated using (13). However, the time
constant with which the transient component in (14) decays is
not necessarily the same as calculated in (13) above, as
indicated by the value of Tt from the fitted data displayed in
Table II. In other words, the value of Tt is different in a TLG
Ia
fault than in a SLG fault. Hence, the discrepancy in the
(b)
Fig. 9. Phasor diagram of the (a) steady state and transient components of the calculated curves in Fig. 10 and the simulated curves is due to
short circuit current and (b) total short circuit current. both the error in the initial short circuit current calculated
in (2)), as displayed in Fig. 9(b). Because the phase A using sequence networks (as described in the previous section)
components have a small phase shift, adding the components and the difference between the calculated time constant and
makes the total phase A short circuit current larger than the the actual time constant of the system for a SLG fault. The
individual components. However, for the phase B and C reason for this difference is that in a SLG fault, the two
initial short circuit currents, the large phase shift between the healthy phases remain connected to the network, thus the
components results in relatively small difference between the network impedance affects the dynamic behavior of the phase
magnitude of the total short circuit currents and the individual currents. Whereas in the TLG fault case the machine is
components. Figure 9(b) also indicates that the initial short completely isolated from the network. However, as indicated
circuit currents in phases B and C are roughly in-phase with in [7], if the machine is connected to a very “stiff” grid, the
one another, and roughly 180° out of phase with phase A. The value of Tt is close to that calculated in (13).
diagram of Fig. 9(b) also gives insight into the time-varying It is important to distinguish between the time-current
behavior of the short circuit currents. From Section II, the behavior presented in this section to that which a relay would
transient component of the current dies away exponentially. “see” in real time. The behavior of the fundamental
Thus, the phase currents in Fig. 9(b) move from their initial frequency, RMS current over time as displayed in this section
position towards the steady state solution as the transient represents the true quantities. However, a relay is performing
component dies away. Therefore, the magnitudes of the
currents in phases B and C reach a minimum approximately
RMS Machine Currents (pu)

Phase A
when the transient component has decayed half-way, whereas Phase B
Phase C
for phase A, the current is a minimum when it reaches steady
state. Another interesting result displayed in Fig. 9(b) is that
the phase C current is larger in steady state than the phase A 0
10
current for a phase A to ground fault.
From (1), the RMS magnitude of the fundamental frequency 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
component of the induction machine phase current under a time (sec)

SLG fault is given by Fig. 10. RMS machine current magnitude over time for a phase A to ground
short circuit occurrence at t = 0 (solid = simulated, dashed = calculated).
 8

a similar functtion to extracct the fundam mental frequen ncy [5] P. M

M. Anderson, Anallysis of Faulted Poower Systems. IEE EE Press, 1995,
p. 2222-228.
coomponent, but is doing so by b measuring and filtering the [6] A. R R. Bergen and V V. Vittal, Power Systems Analysiss (2nd Edition).
cuurrent waveformms. An additioonal study wouuld be to comp
pare Prenntice Hall, 2000, p . 445-487.
thee results preseented in this section to the actual curreents [7] F. Sulla, J. Svens son, and O. S Samuelsson, “Sym mmetrical and
unsyymmetrical short--circuit current oof squirrel-cage aand doubly-fed
“seeen” by the rellay. induuction generators,”” Electric Power Systems Researcch, vol. 81, pp.
16100-1618, Jul. 2011.
VI. CON
NCLUSION [8] D. FF. Howard, J. Re strepo, T. Smith, M. Starke, J. Daang, and R. G.
Harlley, “Calculation of Fault Currentt Contribution off Type I Wind
Short circuit analysis
a of induuction machines is essential for Turbbine-Generators,” in Proc. 2011 IIEEE Power & E Energy Society
prooper rating an nd coordinatio on of protection equipmentt of Gen eral Meeting, Julyy 24-28, 2011, Dettroit, MI.
wiind power plaants. The an nalysis presentted in this paaper
inddicates that traaditional short circuit analysiis using sequennce IX. BIOGRAPH
HIES
neetworks is acccurate for sym mmetrical faultts. Additionaally,
seqquence networrk analysis provides good ressults for the ph hase D ustin F. Howard (S'09) is currentlyy a PhD student
inn electrical enginneering at Georggia Institute of
A short circuit current in a phase A to ground g fault, but Teechnology under the advisementt of Dr. Ron
preesents some error in the un-faulted (phasee B and C) sh hort H arley. He receiveed his B.S. and M M.S. degrees in
cirrcuit currents. The reason for f this error is due to the high
h Ellectrical Engineering in 20088 and 2010,
reespectively, from GGeorgia Institute of Technology,
sennsitivity of phaases B and C tot the value used for the voltage wh
where his primary focus was in elecctric power and
beehind transientt reactance in the sequence network circcuit. coontrol systems. H He is currently ppresident of the
Beecause the derrivation of th he conventionaally used voltage G eorgia Tech chaapter of the Pow wer & Energy
Soociety. His reseearch interests innclude the grid
beehind transientt reactance involves various assumptionss, a integrationn of renewable energy, smart ggrid technologiess, and electric
neew method forr calculating V’ V may be neecessary if hig gher transportaation systems.
levvels of accuraccy in all threee phase currentts for unbalanced
short circuit calculations are required. Travis M. Smith (SM’91) is an Electric Power
Syystems Research Engineer in thee Energy and
Trransportation Scieence Division att ORNL (Oak
VII. APPENDIX Ri dge National Labooratory). He has oover 19 years of
TAB BLE V expperience in poweer systems and ellectrical system
INDUCT TION MACHINE AND
A LINE PARAAMETERS moodeling. He has w worked for a nummber of utilities
Power - 1.816 MVVA 3 pu
p (Tennessee Valley Authority, Georggia Power, and
Soouthern Companyy) and has also worked as a
Voltage Vs 600 V 1.7321 pu
p
connsultant. He has a professional engiineering license
Stator Resistaance Rs 0.0008 oh
hm 0.0040 pu
p
in the state of Tennnessee with activee capabilities in
Stator Leakage Reeactance Xls 0.0173 oh
hm 0.0873 pu
p muultiple jurisdictionns as a NCEES record holder.
Stator Leakage Indductance Lls 4.589e-5 H 0.0873 pu
p Travis is aalso a NSPE, NC EES, and Senior IIEEE member. Reesearch focuses
Magnetizing Reaactance Xm 0.7783 oh
hm 3.9261 pu
p include ssimulation and modeling, substtation design, pprotection, and
Magnetizing Indu uctance Lm 0.002065 H 3.9261 pu
p identificattion of power systeem issues such as harmonics and flicker.
Rotor Resistaance Rr 0.002 oh
hm 0.0101 pu
p
Rotor Leakage Reeactance Xlr 0.0143 oh
hm 0.0721 pu
p Michael Starkee (M’06) is a P Power System
Rotor Leakage Ind ductance Llr 3.794e-5 H 0.0721 pu
p Reesearch Engineer at the Oak R Ridge National
Inertia J 560.05 *m2
kg* 4.87 sec
s Laaboratory. He receeived his B.S, M.S S. and Ph.D. in
eleectrical and coomputer engineeering at The
Poles - 6 - - -
Unniversity of Tennnessee in 2004, 2006, and 2009
Line Resistannce RL 0.002 oh
hm 0.01 pu
p resspectively. He is a member of IE EEE and of the
Line Inductaance LL 2.63e-5 H 0.05 pu
p Poower and Energgy Society withh number of
puublications in poweer systems and powwer electronics.
TABLE VI Reesearch focus is oon wind, solar, annd hydro power
BASE
E QUANTITIES FO OR PER-UNITIZA
ATION tecchnologies.
Q
Quantity Sym
mbol Base Vallue (Eq.) Base Value
V (Num.) Unit
U
Power Pb 1,816,000/3 605,333 VA
V
Voltage Vb 600/sqqrt(3) 346.41 V Ronald G Harlley (M’77-SM’86--F’92) received
Im
mpedance Zb Vb2/P
/ b 0.1982 Ohm
O thhe MScEng deggree (cum laude)) in electrical
Current Ib Vb/Z
/ b 1,747.4 A enngineering from tthe University of Pretoria, South
A
Africa in 1965, andd the Ph.D. degreee from London
U
University in 19699. He is a Regentts’ Professor at
VIII. REFERENCES
E thhe Georgia Instituute of Technology,, Atlanta, USA.
[1] IEEE PES WindW Plant Collecctor System Desiign Working Group, H
His research interrests include thee dynamics of
“Characteristics of Wind Turbinee Generators for Wind
W Power Plantss” in ellectric machines, power systems, w wind and solar
Proc. 2009 IEEEE Power & Energ gy Society Generall Meeting. ennergy, and the usee of intelligent conntrol algorithms.
[2] M. S. Sarma, Electric Machinees: Steady-State TheoryT and Dyna amic D
Dr. Harley is also a Fellow of the British IEE, a
Performance (2 2nd Edition). Thom
mson Learning, Inc., 1994, p. 488-4966. Fellow off the South Africann Institute of Elecctrical Engineers, a Fellow of the
[3] S. S. Kalsi, D. D. Stephen, and B.
B Adkins, “Calcullation of System Fault
F Royal Socciety in South Afrrica, and a Member of the Academ my of Science in
Currents Due to t Induction Moto ors,” Proc. IEE, vol. 118, No. 1, Jan. South Affrica. In 2005 hee received The C Cyril Veinott Eleectromechanical
1971. Energy C Conversion Awardd from the IEEE Power Engineeriing Society for
[4] J. Morren and d S. W. H. de Haan,
H “Short-Circcuit Current of Wind
W “Outstandding contributionns to the field of electromechhanical energy
Turbines with Doubly Fed Indu uction Generator,” IEEE Trans. Eneergy conversionn”, and in 2009 tthe IEEE Richardd H. Kaufmann field award with
Conversion, vool. 22, pp. 174-180, Mar. 2007. citation ““For contributionns to monitoringg, control and ooptimization of
electrical pprocesses includinng electrical machiines and power neetworks”.