2009 6th International Multi-Conference on Systems, Signals and Devices
A NEW COMPUTER BASED PHASOR MEASUREMENT
UNIT FRAMEWORK
Abderrahmane Ouadi1, Hamid Bentarzi1, and Jean Claude Maun2
1
Signals and Systems Laboratory, DGEE, FSI, Boumerdes University, Algeria.
e-mail: sisylab@yaho.com
2
Beams, ULB, Brusselles, Belgium
ABSTRACT
A Phasor Measurement Unit (PMU) is one of the most
important measuring devices that can provide
synchronized phasor measurements of voltages and
currents in real time from widely spread locations in an
electric grid. Indeed this device is crucial to the detection
of disturbances and characterization of transient swings.
In this work, a new PMU has been designed and
implemented including the hardware architecture and the
software program using a recent developed algorithm of
phasor measurement and frequency estimation. The
developed PMU is based on computer associated with an
acquisition card AD512 using Matlab as software tool for
developing its running program as well as its graphical
user interface. Finally, this PMU has been tested in
laboratory using network simulator that gives good
measurements with an acceptable errors even at off-
frequency.
Index Terms-- Wide area measurement system,
PMU, Phasor measurement techniques, DSP, frequency
estimation algorithm.
1. INTRODUCTION
When disturbances occur in a large scale and complex
power systems, protection and control systems are
required to stop the system degradation, restore it to a
normal state, and hence minimize the effect of these
disturbances. Using wide area measurement system based
on PMUs, different systems can achieve their tasks and
the black out can be avoided. In a wide area
measurements system, the Phasor Measurement Unit
(PMU) is considered to be one of the most important
measuring devices that can provide synchronized phasor
measurements of voltages and currents from widely
dispersed locations in an electric power network. These
PMUs should be associated with a reliable high-speed
communication system for transferring all measurements
and indicators to a central position, e.g. a control center,
for evaluation and decision. From this central position,
action orders are then sent to different parts of the power
system. Figure 1 shows applications where PMUs send
978-1-4244-4346-8/09/$25.00 ©2009 IEEE
data of the power system from different places to data
concentrator which will be used in control, monitoring
and protection system.
The purpose of phasor measurement is to produce a
simplified representation of power system signals that
accurately represents the actual system status.
The measurement rate, measurement accuracy,
conversion technique, and final data rate all have to be
integrated for optimal performance.
The phasor measurement process can be divided into two
sub-processes. The first is sampling the voltage or current
waveforms with an analog-to-digital converter (ADC).
This process requires pre-filtering the analog signals to
prevent high-frequency signals above the Nyquist rate
from being aliased into the converted data. The sampling
system also provides sufficient electrical isolation to
prevent interference and damage due to transient
phenomenon.
The second sub-process is the phasor calculation process.
It projects the power signal onto cosine and sine
functions, which are the real and imaginary components,
respectively, of the phasor representation. This process in
effect filters out all the 50-Hz harmonic components.
The accuracy with which these functions represent the
real-time power system depends greatly on the algorithm
used for phasor estimation.
There are four main considerations for choosing an
appropriate phasor computation algorithm: the phasor
sample rate, accuracy of the representation of included
information (the pass-band), the ability to reject
undesirable information (rejection outside the pass-band),
and measurement time delay.
There are two sample rates to be considered: the rate at
which the analog waveforms are digitized, and the rate at
which the phasors are calculated. The waveform-
sampling rate must be high enough to make a good
reproduction of the waveform. It requires analog anti-
aliasing filtering before sampling. This sample rate will
generally be much higher than the power system
frequency and will often be dictated by the phasor
derivation algorithms.
The phasor “sample rate” is the rate at which phasors are
produced by the phasor estimation algorithm. Nyquist
criteria limits the information bandwidth to half the
sample rate, so additional filtering may be required to PMU
PMU
prevent aliasing signals in the phasor estimations. This
filtering can be before, during, or after the phasor
estimation process. Since phasors are estimated for the
fundamental system frequency, filtering before phasor
estimation is restricted to frequencies much higher than
Protection
the fundamental [1]. Data Advanced control,
concentrators application system
2. HARDWARE ARCHITECTURE software archival
data base
The design task has been started by developing the
hardware specifications of the phasor measurement unit
that has the ability to track the phasor values of voltage
PMU PMU
and current synchronously on power system in real time.
The PMU that is a crucial to the detection of disturbances
and characterization of transient swings is based on the Figure 1. Application of PMU in power system.
system hardware shown in Fig.2.
The PMU hardware consists of: form where it will be processed.
- Signal transformation: potential transformer (PT) and
current transformer (CT) transform voltages and currents 3. SOFTWARE STRUCTURE
of the power system to low safe values.
- External clock: A clock signal that is derived from a A phasor may be calculated over any number of
phase locked oscillator to generate sampling clock and waveform samples, but most algorithms use a block of
provide a time tag data that is to be transmitted together samples that span an integer number of cycles. This span
with the phasor measurement information. of time over which the phasors are calculated is called the
- Synchronized sampling boards: the measured values of “window.” The 50-Hz waveform is constantly changing
the power system parameters fed from PT and CT in so the phasor calculation will represent an average value
analog forms are passed through an anti-aliasing filter over the calculation window. Information bandwidth and
amplifier (low pass filter). Sample and hold circuits and phasor sample rate are basic considerations that are
analog multiplexed are used to sample the six different characterized by data-handling capability. The graphical
signals supplied by instrument transformers at the same user interface is developed using Simulink/Matlab
time. The sampled signals are fed to the PC data software for visualizing the real-time signals as well as
acquisition board that converts these samples into digital the estimated phasors and frequencies.

Satellite

GPS receiver Phase locked

oscillator

Analog filters A / D converters DSP processor

(anti-aliasing)

Potentiel & current

transformer Communication
circuit

Power system

Figure 2. General Block Diagram of the Phasor.

A synchro-phasor is estimated from data samples using a
standard time signal as the reference for the sampling
processor for the whole system. In our case, the phasor
values from remote sites have been determined using
common phase relationship.
The synchrophasor representation X of a signal x(t)
current or voltage is the complex value given as follows:
X=Xr + jXi , (1)
Or,
X= (Xm / √2)(cos (ф) +jsin(ф)
Where (Xm / √2) is the RMS value of the signal x(t), and
ф is the instantaneous phase angle of x(t) relative to a
cosine function at nominal system frequency
synchronized to coordinated time universal (UTC). This
angle is defined to be zero when the maximum of x(t)
occurs at the UTC second rollover (one pulse per second
(PPS) time signal ), and –90° when positive zero crossing
occurs at the UTC second rollover [2].
Synchronized phasor measurements have the following
properties:
- Phasor technology provides time-synchronized data,
- The precise timing of the phasor data makes it useful
beyond the local bus,
- Phasor measurements directly provide the phase angles
at the high sub-second rate.
3.1. Phasor Estimation Algorithm
The traditional recursive DFT algorithm [3] is widely
used to compute phasors. Its advantages include fast
speed, small computation burden and good immunity to
normal harmonics. However, measurement errors are
incurred when system frequency deviates from nominal
frequency due to leakage effect of DFT. Therefore, a
method is required to correct the computed phasors in the
presence of frequency offset and thus to enhance the
accuracy of phasor measurement.
By considering a sinusoidal signal x(t) current or voltage:
x(t)= Xm cos(ωt+ ф), (2)
Where: ω =2л fo, fo is the nominal frequency and Xm is
the amplitude of the voltage /current signal.
x(t) can be represented in phasor form as:
X = (Xm / √2)exp(ф) (3)
The phasor X of Eq.(1) can be rewritten as:
X = (1/√2)(Xr – jXi) (4)
Where Xr and Xi are the real and imaginary components
of the DFT respectively.
If Xk is the sequence obtained when sampling x(t) at
sampling frequency fs = N*fo (N: is the sampling rate),
then the recursive DFT of Xk is [4]:
Xrk = Xrk -1 +(2/N)(Xk –Xk-N )cos(k) (5)
Xik = Xik -1 +(2/N)(Xk –Xk-N )sin (k ) (6)
With:
0 =< k =< N-1
Where:
X-1 ,X-2 ,…………………X-N and Xr-1,Xi-1
are set to zero.
The RMS value of the amplitude and the phase angle of
the phasors can be obtained as follows:
XRMS = (Xr +Xi )/2 (7)
ф = arctan(Xi / Xr) (8)
Phasor estimation algorithm steps:
a) The sinusoidal signal x(t) is sampled at a frequency fs
= number of samples *fo=12*50=500 Hz,
b) Calculation of the real and imaginary parts of the
phasor X using the recursive representation of the
Discrete Fourier Transform,
c) Compute the RMS and phase angle values of the
phasor X.
3.2. Frequency Estimation Algorithm
Xa , Xb and Xc are voltage phasors: The positive
sequence voltage phasor representation is given by:
V1=(1/3)*(Xa +α*Xb +α *Xc) (9)
Where: α = -0.5+j0.866
So, phasors Xa, Xb and Xc have a phase angle difference
of 120° respectively.
When the frequency of the power system is offset by Δf
and the sampling frequency is constant. The phase angle
of the positive sequence phasor at kth instant which is
shown in Fig.3, can be represented as [5,6]:
Δf=(1/2Π)*(θk- θk-N)*fo (10)
Thus, the rate of change of the angle of the positive
sequence phasor is directly related to Δf. The frequency
may be estimated once every cycle. The used frequency
estimation algorithm flow chart is illustrated in Fig.4.
 V1k Ψ/2
Frequency

A + jB = V − V
1K 1K− N
θκ V1k-N C + jD = V +V
1K 1K− N
θ k-N

ψ A
= x = C +
D
tan
2 + B
Figure 3. Successive positive sequence phasors.

ψ
4. TEST RESULTS AND DISCUSSION = tan − 1 x
2
The power signal is usually near 50 Hz and constantly
changing. The phasor derivation technique needs to
produce an accurate measurement of amplitude and phase
at 50 Hz, and a range of frequency around 50 Hz. The
typical amount of deviation from 50 Hz and the rate of f ⎛ψ ⎞
Δf = 0
change depend on the power system. The unwanted or
corrupted (due to governors, shaft dynamics, stabilizers,
π ⎜
⎝2 ⎠
and other system elements) information in the power
signal can range from low frequency (2 Hz or less) to
high, say 25 Hz, and even approaching the 50 Hz Yes
fundamental. The phasor algorithm pass band around 50
(AD-BC) >=0
Hz needs to be wide enough to represent the signals of
interest. No
The experiments have been performed for testing the
developed PMU under different conditions such as Δ f =−Δf
frequency and amplitude variations. In the first case, the
frequency is subjected to variations 45<f<55 Hz as shown
in Fig.7a. It can be noticed that measured values by the f = f +Δ f
PMU is in good agreement with measured one by scope. 0
In the second case as shown in Fig.7.b, the measurement
performed when the frequency is fixed at the rated value
fo=50Hz and the amplitude is varied with +/-10% from Return
the rated frequency fo. It can be observed that there is no
error between the actual amplitudes and the measured
Figure 5. Frequency estimation algorithm flow chart.
ones by PMU, besides the measured angle will not
undergo changes. Finally, when the variations applied to AD512 and it has been tested. Important notes deduced
both frequency and amplitude, the measured values are in from the test results:
a good agreement with the actual ones when the -Sampling at off-nominal frequencies leads to
frequency offset Δf is small but the error increases as Δf incomplete sample sets.
increases as illustrated in Fig.6. -Even though the amplitude of the input signal is
constant, the amplitude of the positive sequence phasor
5. CONCLUSION attenuates.
-The angle of the phasor is directly proportional to the
A PC based phasor measurement prototype has been frequency.
developed in this work using data acquisition card
 agreement with the acquired ones when the frequency
-When the amplitude is variable and the frequency is kept offset Δf is small but the error increases as Δf increases.
constant; the measured angle will not undergo changes. In order to reduce this error at the frequency offset a
- If the sampling rate is increased, more accurate results correction may be applied to the magnitude as well as the
can be achieved and errors may be reduced. phase angle estimation algorithm.
Finally, when the variations applied to both frequency
and amplitude, the measured values are in a good
55 Actual freq.
Estimated freq.
54
53
52
f
(Hz) 51

50
49
48
47
46
45
0 0.5 1 1.5 2 2.5 3
Time (Sec.), (a)
220

219.5 Actual Vrms

Estimated Vrms
219

Volt.218.5
(v)
218

217.5

217

216.5

216
0 0.5 1 1.5 2 2.5 3
Time (Sec.), (b)
Figure 6. At off-nominal frequency (a) Comparison between actual and estimated
frequency, (b) comparison between actual and estimated phasor amplitude.

.
 235
Actual Vrms
Estimated Vrms
230

225
Volt.
(V)
220

215

210

205
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (Sec.) (a)
230
Actual Vrms
Estimated Vrms

225

Vrms 220
(V)

215

210

205
0 0.5 1 1.5 2 2.5
Time (Sec), (b)
Figure 7. Comparison between actual and estimated voltage when variation of (a) voltage, (b) both freq. and Volt.
. Phasors, Local System Frequency, and Rate of Change of
Frequency,” IEEE Tran. on Power Systems and App.,
6. Vol. PAS-102, No. 5, pp. 1025-1038, May 1983.
REFERENCES [5] S.A.Soman, “Mathematical Basis and Algorithms for
Numerical Relaying”, 2005.
[1] IEEE power system relaying working group, [6] H. Karimi, “Estimation of frequency and its rate of
“Synchronized Sampling and Phasor Measurements for change for applications in power systems,” IEEE Trans.
relying and Control”, IEEE Transaction on Power Power Del., vol. 19, pp. 472–480, 2004.
delivery, Vol.9, pp442-452, 1994.
[2] IEEE Standard for synchrophasor for Power Systems,
IEEE Std 1344-1995 (R2001).
[3] A. G. Phadke, “Synchronized phasor measurements in
power systems,” IEEE Comput. Appl. Power, vol. 6, no.
2, pp. 10–15, Apr. 1993.
[4] A. G. Phadke, J. S. Thorp, and M. G. Adamiak, “A
New Measurement Technique for Tracking Voltage