2004 11th International Conference on Harmonics and Quality of Power

A Matlab-Simulink Flickermeter Model for

Power Quality Studies
A. Bertola, G. C. Lazaroiu, Student Member, IEEE, M . Roscia, and D. Zaninelli, Senior Member,
IEEE

at 50 Hz, the purpose of the flickermeter is to separate the

Abstract - The paper deals with the representation of thecarrier from the modulating wave, weight the effects of the
flickermeter operation in the Simulink software program. This latter based on human sensitivity to the disturbance,and return
can be useful in case of simulation studies related to power plants
the So) signal.
where flicker phenomenon can rise up due to the presence of In the history of electronics, many kinds of architecture
heavy disturbing loads (i.e. arc furnaces). The Standard block
diagram of the flickermeter is analyzed and reproduced in have been used to create such instruments. Some of them are
software blocks dedicated to the modeling of its different based on analog filtering of the voltage at the common point
of coupling in order to extract the instantaneous disturbance
components. A validation procedure on a real case is adopted for
testing the proposed model. level and apply it to a predefmed weighting curve (shown in
Fig. 1) for weighting its incidence and ascertaining its
Index Terms - Light Flicker, Flickermeter, Simulation tools severity.
Others are based on an approach with digital
I. INTRODUCTION implementation of the filter and use of the Fourier transform
n the study of disturbance propagation in the eleceicity PI, [61, VI, PI, P I , [lo].
I network in which there are highly pollutant loads, such as Others again, on an approach based on the Kalman filter
for example arc furnaces [I], [Z], [3] particular importance [ 1I], which makes it possible to estimate the amplitude and
attaches to the flicker phenomenon. phase of the voltage waveform present in the network by
To quantify the amplitude of such disturbance, reference is breaking it down into constant elements and fluctuating
made to the instantaneous flicker sensation So). elements in its envelope.
The value of So) is measured by means of an instrument, Fv)
the so-called flickermeter. 0,9
This paper deals with the setting-up of the flickermeter 03
model in Simulink, the simulation tool, in the MATLAB [4]
0,7
environment. The main advantage of this software product is
that it makes it possible to support an object programming 05
procedure, with a circuit representation obtained by means of 0,s
graphic blocks associated with the different power system
components. 03
Simulation in the MATLAB domain makes for simple,
02
powerful mathematical processing of output results, both on
0,I
line and in parallel to the simulation process.

11. THEUIE FLICKERMFTER Fig. 1. Weighting m e proposed by UIE (International Union for Electro-
Since, from the point of view of analyzing the signal, the thermal Applications)
flicker is a low-bequency modulation of the network voltage
Although the results produced by digital flickermeters are
comparable to, or even better than those based on analog
A. Bertola is With the Depaitnent of Elechicd Engineering, Polittecluco di
Milano, p i m a Leonardo da ViOci 32, Milano 20133, Italy (e-mail filtering, in this paper we have preferred to deal with the
[email protected]). conventional analog instrument. There are two reasons for this
0. C. Lazaroiu i s with the Department of Electical Engineering, choice: direct conformance to the models proposed in current
Politecnico di Milano, pi- Leonardo da Vinci 32, Milano 20133, Italy ( E .
maik cristian.lazaroiu~ppolimi.it).
standards and the setting up of an instrument model that is a
M. Roicia is Wilh the Electrical Engioeeriog Department of the Universia repeat of the classic layout traditionally used for flicker
di Napoli "Federieo II", Napoli, Italy (email: [email protected]). studies. The latter aspect is particularly important in the case
D. Zaninelli is with the D e p a r r " of Elechical Engineering, Politecnico of comparative analyses between innovative equipment and
di Milano, p i m a Leonardo da Vinci 32, Milano 20133, Italy (e-mil:
[email protected]). conventional architectures, whose disturbanceswere measured

0-7803-8746-5/04/$20.00 02004 IEEE. 734

using analog instruments Only in this way is it possible to dual frequency in respect of network Gequency. The second
compare the results obtained and to evaluate the filter, centered at 8.8 Hz,imposes the form of the response in
improvements achieved quantitatively without being assailed flickermeter Gequency on the modulating fluctuation. The
by the doubt that the values measured may also vary “weighting filter” block simulates the response in Gequency
depending on the type of instrument used. on the sinusoidal fluctuations in the voltage of the chain
With reference to [12], Fig. 2 shows the layout of the UIE consisting of a lamp filled with inert gas with a spiral filament
flickermeter, a standardized instrument for measuring the (60 W-230 V), followed by the human eye. The response
flicker obtained by simulation and by stochastic analysis of function is based on the perceptibility threshold ascertained,
the response of the lamp-eye-brain chain to voltage for each frequency, ont of 50% of the persons taking part in
fluctuations. the experiment.
The transfer function that describes the behavior of the
The input signal consists of the 50 Hz network on which
weighting filter described above is shown in equation 1:
flicker is superimposed in the form of amplitude modulation
and enters Block 1, the so-called “normalization block”, after
which a transformer reduces its voltage level to values
compatible with the electrical specifications of the elecixonic
components located downstream.
Block 1 contains a signal generator for checking the
flickermeter setting in the field and a circuit for normalizing,
at the intemal reference level, the rms value of the input in which s is the complex Laplace variable and the values of
voltage at network frequency. the parameters, taken from [12], are as follows:
In this way, flicker measurements are made independent of k = 1.74802
the actual input voltage, whose variations (AV) are expressed h = 2 n 4.05981
as a relative value of the rated voltage AV/V,. This function ol= 2 n 9.15494
is obtained by acting on the gain of an amplifymg stage by 02= 2 n 2.27919
means of a suitable control signal obtained for comparison os=2 x 1.22535
between the effective output voltage value and a reference on=2 x 21.9
signal. Fig. 3 shows the response in frequency of function (1).
The subsequent blocks (from 2 to 4) react to the signal by The “range selector” determines the sensitivity of the
simulating the behavior of the lamp-eye-brain chain. instrument, by varying the gain as a function of the amplitude
Block 2 is a quadratic demodulator for separating the of the voltage fluctuation to be measured.
modulating fluctuation (AV) Gom the carrier wave (network Block 4 performs two functions:
frequency). To avoid making the subject too confusing, the - Squaring of the flicker signal in order to simulate non-
detailed analysis of this block and of that regarding the fmt linear eye-brain perception;
filter in Block 3 will follow in the next paragraph. Block 3 - Leveling to a sliding mean of the signal in order to
consists of two filters in cascade and by a scale selector of the simulate the effect of memorization in the brain.
sensitivity that may precede or follow the selection filter
circuit. The fust filter eliminates from the output voltage of
the quadratic demodulator the dc component and that of the
E.mYh<lr)” ~rL.mp.cys.h,.>n I C .pon.r

135
 1.2 interface, it is possible to obtain direct from this block
I analog graphs of the accumulated probability function.

.
>- 08
>=0.6 111. THE QUADRATIC DEMODULATORAND THE FIRST
FILTER IN BLOCK 3
0.4
The quadratic demodulator is an apparatus whose job it is
0.2
to separate, in a modulated signal, the carrier from the
0 modulating wave by squaring the wave to be processed.
0 5 10 15 20 25 30 35
f m1 In the case in point, the voltage signal fed into the
Fig. 3. Response in frequency ofthe fitter simulating the reaction of the demodulator (VJis expressed as follows:
human eye lo luminosity fluctuations

To set up the above functions, the block in question

contains a squarer associated with a low-pass filter of the fmt
qct, = v .[L :
+ k=l (mk .C0S(Rk . t ) ) l . cos(oo .t) (2)

order with a time constant equal to 0.3 s. in which Vis the amplitude of the waveform at 50 Hz; mn is
The exit of Block 4 represents the instantaneous sensation the index of modulation of the component k; 0, is the
of tbe flicker So). pulsation of the component k of the voltage modulating signal;
Block 5 incorporates the microprocessor that performs in oois the angular frequency of the voltage at network
real time the analysis of the flicker level, thus making possible frequency.
direct calculations of the significant evaluation parameters. Since the demodulator receives on input a voltage signal
Altematively, it is possible to record the time trend of S(t) and whose amplitude is expressed as a per unit value @,u.), the
measuring tbe disturbance. quantity Vis equal to 1.
Before completing the process, it is advisable to make a Considering separately each k-th component of the
brief analysis of the exit signals shown in the layout io Fig. 2. modulant, by squaring and exploiting the trigonometric
The standard [12] says that, in the flickermeter, the exits relationships, we find that a general component of the voltage
marked with an asterisk are not essential, but may enable tbe signal exiting from the quadratic demodulator is equal to:
potentialities of the instrument for recording voltage
fluctuations to be exploited to the full. Obviously,
consideration may also be given to other optional exits.
The purpose of optional Exit 1 and of the effective
value measured associated with it is to make it possible
to visualize the waveform of the voltage fluctuation
(i
. 2 + "(2.
2 WO .4 (3)

starting from the variations in the m s value of the For each k-th component of the signal exiting from the
input voltage. This may be obtained by squaring the demodulator the part that is relevant for the flicker study is
signal, integration between the passages through zero represented by the term: ntycos(C&-f).
of each half-period, and extraction of the square root of To eliminate the remaining part of the starting signal it is
the signal. To be able to observe with a satisfactory necessary to introduce a high-pass filter (which eliminates the
degree of resolution any small voltage variations, it is dc component) in series with a low-pass filter (which
advisable to compensate the dc component; eliminates the carrier wave and the high-frequency
Exit 2 is intended mainly for controlling the response components). These functions are assembled in the first filter
of Block 3 and for regulating the latter; in Block 3, exiting from which we obtain the single
Exit 3 gives an instant linear indication of the relative component of the voltage modulant:
voltage variation AVW, expressed as a percentage
(4)
equivalent to a modulation with a sinusoidal wave with
a frequency equal to 8.8 Hz. This exit is useful for The modulant will therefore consist of the series:
selecting the appropriate measuring range;
m
Exit 4 supplies the integral on one minute of
instantaneousperception of the flickers; k=l k=l
Exit 5 represents the instant flicker perception and may The above assumption that each modulant component can
be recorded with an oscillograph (recording on paper) be considered separately after which the results obtained can
for a rapid evaluation on site or with a tape recorder for be considered together is not formally correct because, in this
measuring a longer period and for subsequent way, no account is taken of the mixed products generated by
processing; squaring the starting signal. However, since the mixed
.Exit 6 of Block 5 is connected to a digital serial products produce high-frequency signals, the result obtained
interface suitable for a printer and for recording on by applying the aforesaid assumption is analogous to that
tape. By using another digital-analog conversion obtained by considering all the mixed products. Indeed,

736
downstream of the quadratic demodulator there is a low-pass the steps necessary for setting up efficient electronic
filter that cuts off the signals due to the presence of the mixed implementation: this slightly modifies the logical structure of
products together with those due to the carrier wave and the the device shown in Fig. 2, but has the advantage of already
high-frequency components. Therefore the overall result having the block structure included in Simulink.
exiting from the filter is, in any case, that given in (5). For effective clarification of the flickermeter structure
It now remains to clarify the characteristicsof the filter: if supplied with the simulation program, we now describe,
the latter were not well designed, there might be a risk of briefly, the function of each block. In this way, it will also be
omitting pari of the flicker, or of considering disturbing possible to check the exact conformance of the model
components that are, in fact, nothing of the kind. simulated with that described by the regulation.
For this reason, the standard [12] lays down the design in Block A has the dual function of expressing the instant
detail; this should consist of a high-pass filter of the fust order input voltage as a p.u. value and that of simulating the
with a cutoff frequency of 0.05 Hz,followed by a Butterworth behavior of the quadratic demodulator. In practice, it
filter of the sixth order with a cutoff frequency of 35 Hz that performs the functions that, in Fig. 2, had been
performs the low-pass function. performed by Blocks 1 and 2 jointly;
The transfer function (in the Laplace transform) of a Block B has the job of cutting off the dc voltage
Butterworth filter can be expressed, in mathematical terms, by component, eliminating the signal of the carrier wave
means of the following series of polynomials: and the high-frequency fluctuations found at the exit of
the quadratic demodulator. It performs the function of
the first filter in Block 3, shown in Fig. 2;
Block C simulates the response in the frequency of the
human eye to the voltage fluctuations of an
incandescent lamp supplied by a variable sinusoidal
voltage. This block performs the function that, in Fig.
2, had been performed by the second filter in Block 3;
in which s is the complex Laplace variable, oBis the cutoff
angular frequency, and the remaining parameters, depending
, Block squares the signal exiting from Block by
simulating the non-linear perception of the flicker in
on the order of the filter chosen, are given below (in the case the eve-brain chain. This has the same function as had
..
of a sixth-order filter):
nreviouslv been nerformed bv the fmt D a r t of Block 4:
01 = wz= 0,= 1
Block E is a low-pass filter that performs the second
cl = 0.26
part of Block 4. This block is the modeling of the
c2= 0.71 flicker memorization in the brain.
G3= 0.97
By replacing the above parameters given in (6) and As will he noted from the foregoing, there is complete
multiplying the terms as indicated, the filter's transfer function conformance between the block diagram in Fig. 4 and that
is obtained. proposed by the standard in Fig. 2.
In Fig. 5 we find implementation of the functionalities of
n'.IMPLEMENTATION OF THE UIE FLICKERMETER AND OF THE Block A, in Fig. 6 those of Block B, in Fig. 7 those of the
INSTRUMENTS CALCULATING THE SEVERITY INDICES OF THE Butterworth filter, in Fig. 8 those of Block C, in Fig. 9 those
DISTURBANCE of Block D, and in Fig. 10 those of Block E. In the latter there
In the light of software implementation, the considerations is the only discrepancy between the block diagram of Fig. 4
expressed in the foregoing regarding the structure of the and the procedure implemented in Simulink: indeed, in the
flickermeter may be represented through the transfer functions model supplied to Simulink, a block has been added to obtain
contained in the block diagram in Fig. 4. the absolute value of the signal So). Justification for this
choice lies in the fact that, for the subsequent processing of
S(rJ and for determining the disturbance indices, where it is
preferable to use a signal whose instantaneous values are
expressed in real numbers, rather than complex ones.
Once we have obtained the trend in respect of the
instantaneous flicker sensation, we have, kom the latter, to
L calculate the indices that make it possible to evaluate the
Fig. 4. Block diagram ofthe flickermeter simulated with Simullnk severity of the disturbance injected into the network.
This may be done by memorizing on disc the sampling of
This condenses all the signal manipulations required for the signal measuring the instantaneous flicker trend during
obtaining in exit the trend of the instantaneous flicker simulation and the subsequent deferred analysis of the results
disturbance in accordance with the information supplied by by means of the mathematical functionalities of MATJAB. In
the current standard [12]. In addition, this diagram considers this way it is possible to enable the program to distribute the

131
instantaneous flicker perception levels, based on kequency, to with reference to the International Standards on this subject.
weight the disturbance generated based on their duration, and Comments and discussion on the model construction,
to r e m the accumulated flicker probability curve and the together with some validation procedures are described along
index P,,. the paper.

VII. REFERENCES
L. di Stasi, Electric furnaces (in Italian), Pahon Editore, Bologna, Italy,
1976.
Fig. 5. DiapmofBloekA E. Ti”, Elecnothermol Applications (in Italian), CUSL, Milan, Italy,
2000.
Connection cyiteria for user electrical plontr wirh ac arc fumocer,
Standard ENEL DK5550, Rome, Italy, March 1991
F. Par&, Electrical Devices. Theory, construcrion, opplicarion (in
.
---a-r-
cutoffkmur0cyo.OSHz
Italian), vol. 1, CLUP Milan, Italy, 1996.
L. Toivonen and J. Momky, “Digital multirate algorithms for
Fig. 6. Hi&-pass tilter measuement of voltage, current, power and flicker”, Proc. 1994 IEEE
Power Engineering SocieIy Tronrrnisrlon and Disrribution Conf, pp.
330 - 340.
S . Nuccio, “‘A digital instrument for measurement of voltage flicker”,
Inrtrumentnrion and Measurement Technology Conjerence 1997
Fig. 7. Bunenvorth Filter Diagram IMTC/97, vol. 1, 1997, pp. 281 - 284
S . Caldara, S . Nuccio and C. Spataro, “A h a 1 instrument for
100.5499.~ measurement of flicker”, IEEE Trmroctionr on Instrumentorion and
1 Measurement, vol. 47, no. 5, Oct. 1998, pp. 1155 1158.
ue
~

~2+51.017077-r+3308.8017
S . Caldara, S . Nuccio and C. Spataro, “ A v h ~ a linstrument for
Lamp - Eye. Brain - Chain I Lamp - Eye - Brain - Chain 2 measurement of flicker”, IEEE Instrumentorion and Measurement
Fig. 8. Diagram of Block C Technolog) Confirence 1998, vol. 1, pp. 342 345.
~

S . Caldara, S . Nuccio and C. Spataro, “Digital techniques for flicker

Q “2 .Q measurement: Algodthms and implementations analysis”, 16th IEEE
Insrrumentotion ond Measurement Technolog) Conference 1999, vol. 2,
Ud ~ d i n g ~ e - b m i nUe pp. 656 - 661.
perception .no1. 1. P. de Sousa Rocha and S . M. D e c k ” , “Di.qitd flickermeter
Fig. 9. Diagram of Block D. implementation”, Proceedings of the 38th Midwest Symposium on
Circuits and Syslems, vol. 2, 1996, pp. 157 - 760.
1 [I I] C. M. Fallen and B. A. McDermon, “Development and testing of a real-
I ~

IYI 1 lime digital voltage flickmeter”, Proc. 1994 IEEE Power Engineering
0.3.stl 4
-‘ Sociev Transmission nndDislriburion Cmf,vol. 2, pp. 3 1 - 36.
Memorization in “ah
hrain
~~~~
[ 121 Flickermeter Functional and design speclfications, IEC Standard 868,
1996
Fig. IO. Diagram of Block E
[I31 Flickermeter - Functional and design speclficarionr. IEC Standard
610004-15, Feb. 2003
V. CASE OF APPLICATION [I41 A. Benola, DC eledricfurnoces: modeling, conrrol and disturbance
onolysis (in Italian), M. Sc. Thesis, tutor D. Zaninelli and G. Carpinelli,
The prepared flickermeter model has been tested with Polilecnico di Milano, Milan, Italy, Academic Year 199912000
measurements performed on a real plant: a 60 M V A dc arc
furnace supplied at 220 kV and located close to Brescia in the VIII. BIOGRAPHIES
North of Italy. Data on this plant are reported in [ 141. Andrea Berlola received the M. Sc. Degree in Electrical Engineering
Depending on different control system techniques, different horn Politeenico di Milano, in 1999, and he is now working in a consulting
values of P,, factor were measured during the arc furnace firm on automation and information technology fields. His areas of interest are
s o h a r e application to industrial plants and power quality.
operation. The results of the simulation procedure give rise to
a very interesting agreement with the recorded measurements. George Cristian Lnraroiu received B. Sc and M. Sc. degrees from
The obtained deviation from the measured values are less than Department of Electrical Engkeerhg, University Polytechnic of Bucharest, in
2002 and respectively, 2W3. Now he is a PhD student at Deparlment of
8%, limit that is considered reasonable considering that the Electrical Enginerhg, Politeenico di Milano. His are- of research include
simulation in Simulink environment is performed with the distributed generation and power quality.
model of all the parts that constitute the dc arc furnace circuit,
both power and control systems. If the flickermeter model is Mariacriitina Roscia, received M.S. degree (1999) and she obtained !he
degree PbD in Electrical Engineering at the University of Naples “Federica
directly supplied by the waveform recorded on the real plant, n . Her areas of research include Energy and Environment, Sustainable
the error introduced by the proposed model fall down close to Development and renewable resources.
1%.
Dsrio Zaninelli, (SM’97) received the Ph.D dzgre-ee in Electrical
Engineering from Politemiso di W a n o , in 1989, and he is now Full Professor
VI. CONCLUSIONS in the Electrical Engineering Department of the Politecnico di Milano. His
areas of research include power system harmonics and power system analysis.
The spread of the use of the program S i m u l i in the Dr.Zaninelli is a senior member of IEEE, a member of AEI and a member of
Matlab environment encourage the Authors to develop a study the Italian National Research Council (C.N.R.) group of Elecmcal Power
for simulatingthe flickermeter block in this software tool. The system.
paper present this model and justify the introduced parameters

738