See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261047827

Park's transformation analitycal approach of transient signal analysis for power

systems

Conference Paper · September 2012

DOI: 10.1109/NAPS.2012.6336412

CITATIONS READS

7 118

4 authors, including:

R.G. Ferraz Leonardo Iurinic

Universidade de Caxias do Sul (UCS) Universidade Federal do Rio Grande do Sul
14 PUBLICATIONS   38 CITATIONS    12 PUBLICATIONS   31 CITATIONS   

SEE PROFILE SEE PROFILE

Arturo Bretas
University of Florida
195 PUBLICATIONS   1,144 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Extended TLM-2D Numerical Technique: An Algorithm Considering Ionization of Non-Homogenous Media View project

Ph.D Project View project

All content following this page was uploaded by Arturo Bretas on 04 September 2014.

The user has requested enhancement of the downloaded file.

Park’s transformation analitycal approach of transient
signal analysis for power systems
R.G. Ferraz; L.U. Iurinic; A.D. Filomena; A.S. Bretas
Electrical Engineering Department
Federal University of Rio Grande do Sul
Porto Alegre, Brazil

Abstract— A rapid change in power system states leads to (vd) obtained through Park’s transformation (Tdq0). The fault
appearing of electromagnetic transients signals in system point is located knowing the time difference between the wave
variables. Such transients are in the form of DC offset or fronts arrival in each terminal and the known wave velocity.
frequencies above de fundamental and have information about The same approach is presented in [9] but adding a self-
its producer phenomenon. Thus, system signals can be recorded adaptive block to account system unbalance levels. These
using adequate equipment in order to diagnose the phenomenon. works highlight the vantages of Tdq0 to monitor the three
Within the previous contextualization, this paper present an phases with only one signal and eliminate the influence of
analytical study of Park’s transformation applied in the fundamental frequency, improving the transient detection.
framework of power system three-phase signals. Furthermore,
Nevertheless, a more rigorous analytical study of the
this paper highlights the potential use of such transformation as
a filter that could be useful to improve the transient signal
application of Tdq0 to power system signals has not been done.
analysis. By means of a numerical implementation of Park’s Thus, this paper intends to present an analytical study of
transformation taken on Matlab environment, four cases are Tdq0 contextualized on electric power system in order to assess
analyzed. Three of them consist on studying synthetics and transient signals. The potential application of this tool as a
controlled signals, whereas the last analyze the phase voltages of filter to eliminate the fundamental frequency is presented.
a faulted power system simulation performed in ATP-EMTP. Furthermore, the effects of amplitude unbalance and distortion
of transient’s characteristics are analyzed. A numerical
I. INTRODUCTION implementation of the Tdq0 is assessed with four cases. The
first three are taken on to validate the resulting equations and
It is well established that rapid changes in circuit states show the effects of unbalances and transients in a controlled
create electromagnetic transients seem in the system variables. manner. The fourth employ signals from a digital simulation
The term transient indicates the notion of an event that is of a faulted transmission line.
undesirable but momentary in nature, disappearing during the
transition from one steady-state to another. Common sources II. ANALYTICAL APPROACH
of electromagnetics transient in power systems are lightings
and commutations. Such creates impulsive or oscillatory Park’s transformation allows relating variables from a
transients that can affect the performance of equipment or static reference frame with a rotating reference frame. Its main
damage their electrical insulation [1]. and best known application is in analysis of rotating electric
machines, turning the variables inductances of the stator in
Transient waveforms contain oscillatory components constant inductances on a rotating reference with synchronous
above the fundamental frequency and the natural frequency of speed [10]. The given Park’s transformation name of was
these oscillations characterizes the phenomenon that produced adopted due to R. H. Park work’s consecration, that presents a
the transient. From all possible sources of transients, faults generalized theory to analyze rotating electric machines [11].
(short circuits) are common and worrisome due to the high
Consider the matrix Park’s transformation TP presented in
magnification produced in system currents. Therefore, a
(1) [10]:
reliable detection and location of faults is necessary to
minimize the period of interruptions and associated costs. ⎡cos (ωt + θ)
2π
cos (ωt −
2π
+ θ) cos (ωt + ⎤ + θ)
Faults create travelling waves and resonances phenomena that
2
⎢ 3 3 ⎥
2π 2π
produce transients used in many works to perform an event TP = ⎢-sin (ωt + θ) -sin (ωt − + θ) -sin (ωt + + θ) ⎥ ,
diagnostic [2-7]. 3⎢ 3 3 ⎥
Reference [8] presents a fault location approach for a two ⎢ 1 1 1 ⎥
⎣ 2 2 2 ⎦
terminal time-synchronized algorithm. The concept is to
monitor all three-phase by the direct axis voltage component (1)

The authors gratefully acknowledge Coordenação de Aperfeiçoamento

de Pessoal de Nível Superior (CAPES), Conselho Nacional de
Desenvolvimento Cientifico e Tecnologico (CNPq), and Companhia
Estadual de Geração e Transmissão de Energia Elétrica (CEEE-GT) for the
financial support of this study.

978-1-4673-2308-6/12/$31.00 ©2012 IEEE

where: (5)
ω is the angular power frequency; for the zero voltage component;
t is the time instant;
2 1
θ is the angle between the axis of phase a and the direct axis vd = { (Vma + Vmb + Vmc ) +
component (d-axis). 3 2
1
Park’s transformation allows the transformation from the [V cos (2ωt) −
2 ma
abc phase variables to the dq0 variables in the following
matrix form [10]: Vmb cos (2ωt − π / 3) −
Vmc cos (2ωt + π / 3)] +
fdq 0 = TP ⋅ fabc , (2)
f (t ) cos(ωt )},
where: T
fdq0 = [fd fq f0]T is the vector of signals dq0; (6)
fd, fq e f0 are the signals of direct axis, quadrature axis, and for the direct axis voltage component and
zero, respectively;
fabc = [fa fb fc]T is the vector of signals abc; 2 1
vq = − { [Vma sin (2ωt) +
fa, fb e fc are the signals of the phases a, b, and c, respectively. 3 2
Since the transformation (2) is orthogonal; i.e., the inverse of Vmb sin (2ωt − 4π / 3) +
the transformation matrix TP is equal to its transpose, can be Vmc sin (2ωt + 4π / 3)] +
the conversion of signals abc to components dq0 and vice-
versa [10]. f (t ) sin (ωt)},
T
(7)
A. Three-Phase System with Voltage abc Signals and for the quadrature axis voltage component.
Electromagnetic Transient
For simplicity, the angle θ of the transformation matrix TP was
Consider the following set of voltage signals from a three considered zero.
phase system with an electromagnetic transient in phase a as:
Consider the following trigonometric relationships by
⎡va ⎤ ⎡Vma cos (ωt) + fT (t ) ⎤ definition:
⎢vb ⎥ = ⎢Vmb cos (ωt − 2π / 3) ⎥ , (3) cos x cos y = 0.5 [cos( x − y ) + cos( x + y )] (8)
⎢ v ⎥ ⎢V cos (ωt + 2π / 3) ⎥
⎣ c ⎦ ⎣ mc ⎦ sin x cos y = 0.5 [sin ( x + y ) + sin ( x − y )] (9)
where:
fT (t) is the function that represent the electromagnetic cos( x + y ) = cos x cos y − sin x sin y (10)
transient;
ω is the angular power frequency; cos( x + y ) = cos x cos y + sin x sin y (11)
Vma, Vmb e Vmc are the maxim values of signals of the phases a,
b, and c, respectively. Applying the trigonometric relationships (8), (9), (10), and
(11) on (5), (6), and (7) the following expressions are
In this paper the transient function is given as (4): obtained:
− (t − μ )2 3
2 (4) v0 = k1 cos(ωt − δ) + f (t ), (12)
f (t ) = FT e 2⋅σ cos (ω t), 3 T
T T
where:
where: 2 2
k1 = k 2 + k3 , (13)
ωT is the angular power frequency of transient;
μ is the time position of transient;
σ represents the temporal dispersion of the Gaussian function. 3 V V
k2 = (Vma − mb − mc ), (14)
The application of Park’s transformation according to (2) 3 2 2
on (3) results in:
1
2 1 k3 = (Vmb −Vmc ). (15)
v0 = {[Vma cos (ωt) + 2
3 2 and
Vmb cos (ωt − 2π / 3) + −1
δ = cos ( k 2 / k1 ). (16)
Vmc cos(ωt + 2π / 3)] + f (t )},
T
 Similarly, (6) can be rewrite in the form: III. ANALYSIS OF ANALYTICAL APPROACH
2 This section will analyze some important operation
vd = Vx + k 4 cos (2ωt + φ) + f (t ) cos(ωt ), (17)
3 T conditions of power system.
where:
A. Balanced System Operating without Eletromagnetic
6 Transient
Vx = (Vma +Vmb + Vmc ), (18)
6 For a balanced system operating without electromagnetic
transient, in (3) Vma = Vmb = Vmc = Vm and fT (t) = 0. Where Vm
2 2 is the maximum phases a, b, and c value.
k 4 = k5 + k 6 (19)
Analyzing (26), (27), and (28) for the condition described
above, it is possible to verify that:
6 V V − for the zero voltage component:
k5 = (Vma − mb − mc ) (20)
6 2 2 v0 = 0, (29)
because k1 is zero;
2 − for the direct axis voltage component:
k6 = (Vmb − Vmc ). (21)
4 6
and vd = Vx = V , (30)
2 m
−1
φ = cos ( k5 / k 4 ). (22) because k4 is zero;
− for the quadrature axis voltage component:
Similarly, (7) can be rewrite in the form:
vq = 0, (31)
2 because k4 is zero.
vq = − [ k 4 cos (2ωt + ϕ ) + f (t ) sin (ωt )], (23)
3 T
In other words, the application of Park’s transformation to
where k6 is given by (19) and a set of signals at the fundamental frequency of a balanced
−1 system results only in a DC level in the direct axis voltage
ϕ = cos ( k6 / k 4 ). (24)
component.
Thus, the application of Park’s transformation on (3) B. Balanced System Operating with Electromagnetic
results in: Transient
⎡ k1 cos (ωt − δ ) +
3
f (t )
⎤ For a balanced system operating with electromagnetic
⎢ 3 T ⎥ transient in phase a, in (3) Vma = Vmb = Vmc = Vm and fT (t) ≠ 0.
⎡ v0 ⎤ ⎢ ⎥
⎢v ⎥ = ⎢V + k cos (2ωt + φ) + 2 f (t ) cos (ωt ) ⎥ . (25) Analyzing (26), (27), and (28) for the condition described
⎢ d⎥ ⎢ x 4 above, it is possible to verify that:
3 T ⎥
⎣ vq ⎦ ⎢ ⎥ − for the zero voltage component:
⎢ −[ k 4 cos (2ωt + ϕ ) + 2 fT (t ) sin (ωt ) ] ⎥ − (t − μ )2
⎣ 3 ⎦ 3 2 (32)
v = 0 F e 2⋅σ
T cos(ω t),
Substituting (4) in (25) results: 3 T

because k1 is zero;
− ( t − μ )2
3 − for the direct axis voltage component:
2 (26)
v0 = k1 cos(ωt − δ) + F e 2⋅σ cos(ω t) − ( t − μ )2
3 T T
6 6 2
vd = Vm + FT e 2⋅σ [cos((ω − ω ) t) + cos((ω + ω ) t)],
2 6 T T
vd = Vx + k4 cos (2ωt + φ) + (33)
− (t − μ )2 because k4 is zero;
6 2
F e 2⋅σ [cos((ω − ω ) t) + cos((ω + ω ) t)] − for the quadrature axis voltage component:
6 T T T
− ( t − μ )2
(27) 6 2
vq = − FT e 2⋅σ [sin ((ω + ω ) t) + sin ((ω − ω ) t)],
vq = −{ k 4 cos (2ωt + ϕ ) + 6 T T
(34)
− ( t − μ )2
6 2
because k4 is zero.
F e 2⋅σ [sin ((ω + ω ) t) + sin ((ω − ω ) t)]}
6 T T T
That is, the application of Park’s transformation to a set of
(28) signals at the fundamental frequency of a balanced system
operating with electromagnetic transient in phase a results:
 • zero voltage component is the same as the modified on voltage signals, but the transformation could be applied to
amplitude transient signal; current or any other set of three-phase signal. For all the cases
the fundamental frequency was defined as 60 Hz.
• in a DC level and direct axis voltage component
modification in amplitude and natural frequency; The analyzed cases may be summarized as:
• only transient signal in quadrature axis voltage • Case I Park’s transformation to a balanced set of three-
component with natural frequency and amplitude phase voltages with fundamental frequency;
modification. • Case II examines a three-phase unbalance system
without an electromagnetic transient in phase a;
C. Unbalanced System Operating without Eletromagnetic
Transient • Case III analyzes a balanced system, however with an
electromagnetic transient in phase a;
For an unbalanced system operating without
electromagnetic transient, in (3) Vma ≠ Vmb ≠ Vmc and fT (t) = 0. • Case IV analyzes the voltages measured at any
substation during a momentary fault in load’s phase a
Analyzing (26), (27), and (28) for the condition described with simulation’s time duration of two cycles from
above, it is possible to verify that: fundamental frequency with the fault’s connection at
− for the zero voltage component: 10.5 ms and disconnection at 17 ms.
v0 = k3 cos(ωt − δ); (35) An illustration of the power system circuit simulated is
− for the direct axis voltage component: presented in Fig. 1.

vd = Vx + k6 cos(2ωt + φ); (36) 1 2

Z1 Transmission Line Z2
− for the quadrature axis voltage component:
vq = − k 6 cos (2ωt + ϕ ). (37) MP

Z 1: Source impedance 1;
The application of Park’s transformation to a set of signals Z 2: Source impedance 2; Zf
at the fundamental frequency of an unbalanced system Z f : Fault impedance;
operating without electromagnetic transient results: MP: Measurement point.

• the fundamental frequency in zero voltage component;

Figure 1. Transmission line simulated in ATP-EMTP.
• in a DC level and second harmonic in direct axis voltage
component;
V. TEST RESULTS ANALYSIS
• the second harmonic in quadrature axis voltage
component. A. Case I
D. Unbalanced System Operating with Eletromagnetic Park’s transformation was implemented as a system where
Transient the phase voltages were the inputs and the dq0 voltages the
outputs. Each phase voltage, which was represented by (3),
For an unbalanced system operating with electromagnetic was characterized by the phase amplitude. However dq0
transient, in (3) Vma ≠ Vmb ≠ Vmc and fT (t) ≠ 0. voltages also need to be characterized by frequency and
• the zero voltage component is a fundamental frequency continuous components. Table I shows the numerical results
and transient signals combination; that are according with (29), (30), and (31). In Fig. 2 the
results are presented in a friendly manner and easily identify
• the direct axis voltage component is the following
the transformation behavior in a balanced system.
combination:
− DC level;
TABLE I. VALUES OF CASE I
− second harmonic of fundamental frequency;
Input Output
− modified frequency and amplitude transient signal.
Amplitude Amplitude Frequency DC
• the quadrature axis voltage component is the same as Signal Signal
[PU] [PU] [Hz] [PU]
direct axis voltage component without the DC level. va 1 v0 0 0 0
IV. CASE STUDY
vb 1 vd 0 0 1.225
To achieve an understanding of the Park’s transformation
vc 1 vq 0 0 0
applied on power system signals and illustrate the analytics
results obtained before, four cases were analyzed. The first
three cases consists in simulations of Park’s transformation
carried out at Matlab [12] environment, using synthetics
signals to represent the three-phase voltages of a power
system. The fourth case is a simulation of a power system
performed in ATP-EMTP [13]. This paper’s approach based
 multiplications of the transient with a sine and cosine
according with (25), each one in the fundamental frequency.

Figure 2. Park’s transformation of Case I:

(a) Phase voltages; (b) v0; (c) vd; (d) vq.

Figure 4. Park’s transformation of Case III:

B. Case II (a) Phase voltages; (b) v0; (c) vd; (d) vq.
The results were analyzed as Case I, showing in Table II
that in an unbalanced set of voltages the dq0 voltages have In order to illustrate in a better way the frequency
oscillations and amplitudes according to (35), (36), and (37). distortion, the Gaussian envelope variance was extended to 5
Fig. 3 shows the graphical results of the Park’s transformation. ms and a Fourier analysis was performed. In Fig. 5 are shown
the Discrete Fourier Transform (DFT) [14] of the v0 and vq
signals. The peak seen on v0 DFT has the same value that the
TABLE II. VALUES OF CASE II characteristic frequency of the original transient (1000 Hz).
However the vq DFT is composed of two peaks located within
Input Output
60 Hz of the transient characteristic frequency. Similar
Signal
Amplitude
Signal
Amplitude Frequency DC behavior is noticed vd signal spectrum continuous component.
[PU] [PU] [Hz] [PU]

va 1 v0 0.2 60 0

vb 0.8 vd 0.141 120 1.225

vc 1.2 vq 0.141 120 0

Figure 5. Discrete Fourier Transform of signals:

(a) DFT of v0; (b) DFT of vq

C. Case IV
Fig. 6 shows the phase voltages and dq0 signals from the
Figure 3. Park’s transformation of Case II:
power system simulation. This is a more real case and shows
(a) Phase voltages; (b) v0; (c) vd; (d) vq. the combination of two related disturbs: a fault insertion
transient and an unbalance produced by the voltage drop in the
In this case was considered a set of balanced voltages with faulted phase.
a momentary oscillatory transient in phase A. Such transient
was mathematically represented by a Gaussian envelope
characterized by a variance of 1 ms, centered at 18.5 ms and
modulated by a cosine function of 1000 Hz:
2
1 ⎛ t − 0.0185 ⎞
− ⎜ ⎟
fT ( t ) = 0.3 ⋅ e 2 ⎝ 0.001 ⎠
⋅ cos ( 2π ⋅1000 ⋅ t ) . (38)
As shown in Fig. 4, the v0 signal is the proper transient
with only an amplitude modification exactly as (32) predicts.
The vd and vq signals were modified according to (33) and
(34), not only by an amplitude modification but also by a
frequency distortion. This distortion is given by Figure 6. Park’s transformation of Case IV:.
(a) Phase voltages; (b) v0; (c) vd; (d) vq.
 A DFT was simulated on the v0 and vd voltages with location in distribution networks: theory and experimental validation,”
results shown in Fig. 7. In v0 spectrum occur two means peaks IEEE Transactions on Power Delivery, Piscataway: vol. 25, pp. 3139-
3146, October 2010.
located at 60 Hz and 1440 Hz, the first due the unbalance and
[6] L. U. Iurinic; A. S. Bretas; E. S. Guimarães; D. P. Marzec, “ Analysis
the second due the oscillatory transient. The vq spectrum of single-ended traveling-wave fault-location based on continuous-
analysis indicates that the first peak occurs at 120 Hz, whereas wavelet transform,” In: International Conference on Developments in
the peak corresponding with the transient was degenerated in Power System Protection, Birmingham, 23-26 April 2012.
two near peaks located approximately at 1500 and 1380 Hz. [7] J. Han, W. K. Kim, J. W. Lee, C. H. Kim, “Fault type classification in
transmission saturation,” In: International Conference on Developments
in Power System Protection, Birmingham, 23-26 April 2012.
[8] F. V. Lopes, D. Fernandes Jr., W. L. A. Neves, “Fault location on
transmission lines Based on travelling waves,” In: International
Conference on Power Systems Transients, Delf, 14-17 June 2011.
[9] F. V. Lopes, W. C. Santos, D. Fernandes Jr., “An adaptive fault
location method for smart distribution and transmissions grids,” In:
IEEE PES Conference on Innovative Smart Grid Technologies (ISGT
Latin America), Medellin, 19-21 Oct. 2011.
[10] P. Kundur, “Power System Stability and Control,” New York:
McGraw-Hill, 1994.
[11] R. H. Park, “Two-reaction theory of synchronous machines,
generalized method of analysis – Part I,” Trans. of the American
Figure 7. Discrete Fourier Transform of signals. Institute of Electrical Engineers, New York: vol. 48, pp. 716-730, July,
(a) DFT of v0. (b) DFT of vq. 1929.
[12] The Mathworks Inc., “Mathworks Matlab.” [Online]. Available:
http://www.mathworks.com.
VI. CONCLUSIONS [13] Bonneville Power Administration, “Alternative Transient Programs:
A basic analytical study of Park’s transformation has been ATP/EMTP,” 2002. [Online]. Available: http://www.emtp.rog/.
presented within the context of transient signal analysis in [14] A. V. Oppenheim, R. W. Schafer and J. R. Buck, “Discrete-time Signal
electric power systems. By means of analytical results and Processing” New Jersey: Prentice Hall, 1999.
numerical validations, the following conclusions are possible:
• balanced systems, present all the dq0 signals constants
values, equal or not to zero;
• unbalanced systems present sinusoid of fundamental
frequency v0 component, whereas vd and vq components
are second harmonics of fundamental frequency. that can
include a DC level;
• balanced systems with a transient presented it in the v0
component with only an amplitude modification.
However in vd and vq the single-phase transient has a
natural frequency modification in addition to an
amplitude modification.
• unbalanced systems with a single-phase transient has as
result a superposition of the two cases above.
It is thus clear that Park’s transformation is a useful and
valid approach for transient detection in electrical power
systems.

REFERENCES
[1] Institute of Electrical and Electronics Engineers, “Recommended
practice for monitoring electric power quality: Std. 1159”. New York:
IEEE Power Engineering Society 1995.
[2] G. W. Swift, “The spectra of fault-induced transients,” IEEE Trans. on
Power Apparatus and Systems, Piscataway: vol. 98, pp. 940-947, May
1979.
[3] E. Shehab-Eldin and P. McLaren, “Travelling wave distance
protection-problem areas and solutions,” IEEE Trans. on Power
Delivary, Piscataway: vol. 3, pp. 894-902, July. 1988.
[4] A. T. Jonhs and S. Jamali, Accurate fault location technique for power
transmission lines,” IEEE Proceedings on Genereration, Transmition
and Distribution, vol. 137, pp. 395-402, November. 1990.
[5] A. Borghetti, M. Bosetti, C. A. Nucci, M. Paolone, A. Abur,
“Integrated use of time frequency wavelet decompositions for fault

View publication stats