IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 44, NO.

3, “E 1995 683

Harmonic Power Flow Analysis for the

Measurement of the Electric Power Quality
Loredana Cristaldi, Student Member, ZEEE, and Alessandro Ferrero, Member, ZEEE

Absh.acf-The number and power of loads that pollute, from single sinusoidal generator with nonzero source impedance
an electric point of view, the electric network are constantly was widely discussed in [3] for single-phase and balanced
increasing. Their connection to the electric network causes a three-phase systems. This discussion leads to two important
significant distortion in the line current and, when their overall
power is no longer negligible with respect to the network power, conclusions.
also some distortion in the line voltage, thus deteriorating the 1) If only one polluting load is C O M I X ~ ~to~ the bus-bar,
electric supply quality. The paper proposes a method, based on it delivers a distorted current into the bus-bar, so that
the evaluation of the harmonic powers, to identify whether the the active power associated with each current harmonic
source producing harmonic distortion in the line currents and
voltages of a three-phase system is a polluting load C O M W ~ ~ component is always flowing out of the load branch into
after the metering section or is the distortion of the line voltages. the bus-bar.
Some parameters are also dehed in order to quantify the effects 2) If several polluting loads are connected to the bus-bar,
of the polluting load on the power quality. A V i Instrument each load delivers a distorted current into the bus-bar and
is also described, realized to implement the proposed method and also consumes a distorted current injected by the other
to verify it experimentally. The instrument accuracy is discussed.
polluting loads. The active power associated with each
current harmonic component and measured on each load
I. INTRODUCTION
branch is hence a function of the net current delivered by

T HE STUDY of the electric power systems under nonsinu-

soidal conditions received a great number of significant
contributions during the last decade and now state-of-the-
the branch into the bus-bar and may generally flow out or
into the branch. Its direction can nevertheless represent
whether the load is actually polluting the network or is
art is well summarized in [l]. In spite of the great number a “polluted” load.
of contributions devoted to the interpretation and compen- When unbalanced three-phase systems are concerned, the
sation of the useless current components, few contributions above approach must be modified in order to take into account
can be found dealing with the identification of the sources that harmonic active powers flowing out of the load branch are
producing harmonic pollution in the electric network and the generated also because of the presence of unbalance currents
quantification of their effects on the electric power quality. [2]. Therefore, a method should be provided in order to
The definition of a method and related instrument for distinguish between the reflected harmonic powers due to
such an identification and quantification is nevertheless of polluting loads and those due to unbalanced loads.
utmost importance for the correct insertion and operation of In [2] a method was proposed, based on the linear, orthog-
the compensating devices and for the application of proper onal Park transformation. The instantaneous values of the line
tariffing policies in order to discourage the connection of voltages and currents are transformed into the Park quantities
uncompensated polluting loads. by:
This paper discusses a method for the identification of
the source of harmonic distortion in three-phase networks
based on the analysis of the harmonic active power flows [2]
and describes an instrument specially realized for real-time
measurement of quantities that allow identifying the prevailing with [TI being the orthogonal matrix:
source of distortion (supply or load) and, in case the source is
the load, allow quantifying its effect on the supply quality.
The results of some experimental work are reported, and the
instrument accuracy is discussed.

11. METHODFUNDAMENTALS
The practical situation of different loads (linear and non- The voltage €‘ark Vector v(t> = wud -k jvq and the Current
linear) connected in parallel to the same bus-bar fed by a park Vector i ( t ) = i d -!- j i q describe the three-phase System
completely if the zero sequence components w, and io are nil
Manuscript received May 10, 1994; revised February 9, 1995. [4]. If this is not true, they describe the “pure” three-phase
The authors are with the Dipartimento di Elettrotecnica, Politecnico di
Milano, 20133 Milano, Italy. component of the system, that can still be used for the correct
IEEE Log Number 9411478. identification of polluting loads [2].
001&9456/95$04.00 0 1995 IEEE
684 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT. VOL. 44, NO. 3, JUNE 1595

Fig. 1. Block diagram of the realized instrument.

It was also proved [4] that, under periodic conditions

(period T = 2 ~ / w ) the
, generic Park vector w ( t ) can be
decomposed into its complex Fourier series components Wk
that are related, at each harmonic frequency hw, h = lkl, to Fig. 2. Front panel of the realized V i Instrument.
the positive (Wph)and negative (Wnh)sequence phasors by:
-
Wph= W k , fork > 0 Wnh= Wi, fork < 0. (3) m. THE MEASURING INSTRUMENT
Starting from the frequency-domain components of the The theoretical discussion reported in the previous section
voltage and current Park vectors, the following harmonic shows that an instrument for the evaluation of the effects
active powers can be measured: of a three-phase load on the supply quality should measure
the energy E; and E,i. In order to provide more complete
Pk =Re(Vk.Ii) for k < O and information, the active power P, the apparent power S, the
Pk =Re (Vi. Ik) fork < 0. (4) power factor X [4] and the total active energy flowing through
the metering section should also be measured.
The following sets can be defined for the harmonic indices The instrument schematically depicted in Fig. 1 was real-
k , according to the signs of the harmonic active powers Pk [2]: ized, based on digital signal-processing techniques.
K , = {kI(Pk < 0) A (P-k > 0)) (5) The line signals are acquired by two ADC boards, equipped
with 4 input channels each, f 5 V range, 12-bit resolution
Kni = {k((Pk< 0) A (P-k < 0)). (6)
and sampling rate up to 1 MHz/channel. Since synchronous
In agreement with the discussion in [3], it can be proved sampling techniques have proved to be very effective to
[2] that the active power P; = C k E ~ , , p<k0 Pk flows out prevent leakage errors [5], the sampling clock is synchronized
of the load branch and represents the effect of an unbalanced to the signal fundamental frequency by means of the frequency
load on the network. For the same reason [2], the active power multiplier CK. A sampling rate of 256 samples/period is
P i = C k E K , , Pk flows Out Of the load branch and represents generated.
the effect of a polluting load on the network. The sampled values of the input signals are acquired and
The active powers P; and P i represent the actual effect processed by a TMS32OC30 floating-point DSP, in order to
of an unbalance and polluting load on the deterioration of evaluate the apparent power, the active power, the harmonic
the supply quality. However, their values depend also on the active powers and the reflected powers P; and P i .
actual network configuration [3]. The DSP averages the above quantities over the 1 s period
For this reason, the energy associated with P; and P,i and sends the resulting values to the host computer. A Virtual
gives a more suitable representation of the global, round-the- Instnunent (VI), whose front panel is shown in Fig. 2, was im-
clock effect of these loads on the supply quality. Therefore, plemented on this computer to display the measurement results
the following quantities have to be measured: and integrate the active power and the reflected powers, for the
measurement of the total energy, and E; and E,i energy flow-
E l = LtPU(.)d. (7) ing through the metering section. The input signals are used as
clock signals for the determination of the integration interval.
E; = L t P n l ( T ) d T . (8) The computation time required by the DSP to evaluate the
above-listed quantities is short enough to allow real-time mea-
If an energy meter for the measurement of E; and E; is surements with signal fundamental frequency up to 400 Hz.
inserted in the branches with only linear loads connected, no Particular care was taken in designing the instrument input
E,i energy flow is counted. If the loads are balanced too, no stage, since it is supposed to work with heavily distorted
W; energy flow is counted as well. On the contrary, if the signals, and the traditional measurement transformers are not
same energy meter is inserted in a branch with a polluting load expected to provide enough accuracy. High-accuracy voltage
connected, some E; energy flow is counted. In case the same and current transducers have been implemented, based on
load consumes unbalance currents, some E; energy flow is high-precision, noninductive resistive dividers and shunts and
also counted. electronic isolation amplifiers.
CRISTALDI AND FERRERO HARMONIC POWER FLOW ANALYSIS FOR THE MEASUREMENT OF THE ELECTRIC POWER QUALITY 685

TABLE I
MEASUREMENT
UNCERTAINTIFS FOR THE DEFERENT MEASUREMENTS P!3FORMED BY THE REALIZED hSTRUhENT

*referred to the instrument voltmetric/ampermetric full-scale value.

Oreferred to the instrument three-phase wattmetric full-scale value Pfs
“ P : actual active power value flowing through the metering section.

IV. MEASUREMENT
ACCURACY
The behavior of the realized instrument was tested by insert- [l] All papers in ETEP, no. 1, Jan./Feb. 1993: dedicated issue to the First
ing it in a three-phase test circuit with two loads connected International Workshop on Power Definitions and Measurements under
Nonsinusoidal Conditions, Como, Italy, Sept. 10-12, 1991.
to the same bus-bar fed by a three-phase sinusoidal gener- [2] L. Cristaldi and A. Ferrero, “A digital method for the identification of
ator with 4 R balanced series resistors, 50 Hz fundamental the source of distortion in electric power systems,” IEEE Trans. Instnun.
Meas., vol. 44, no. 2, pp. 14-18, Feb. 1995.
frequency and selectable line-to-ground voltages up to 220 V. [3] P. H. Swart, M. J. Case, and J. D. van Wyk, “On techniques for
Linear, balanced and unbalanced, nonlinear and periodic time- localization of sources producing distortion in electric power networks,
variant loads have been connected to the bus-bar, in order ETEP, no. 6, pp.485490, Nov.lDec. 1994.
[4] A. Ferrero and G. Superti-Furga, “A new approach to the definition of
to reproduce different operating conditions. The instrument power components in three-phase systems under nonsinusoidal condi-
was inserted in a metering section placed in the load branch, tions.” IEEE Tram. Instrum. Meas., vol. 40,pp. 568-577, June 1991.
immediately before the load itself. [5] A. Ferrero and R. Ottoboni, “High-accuracy Fourier analysis based on
synchronous sampling techniques,” IEEE Trans. Instrum. Meas., vol.
The measurement results were always found in agreement 41, pp. 780-785, Dec. 1992.
with the results of the theoretical discussion reported in the
previous sections. In particular, the E; energy was counted
only when the load connected after the metering section was Loredma Cristaldi (S’93) was born in Catania,
a nonlinear or periodic time-variant load, and the E; energy Italy. She received the M.S. degree in electrical
was counted only when the load connected after the metering engineering from the University of Catania in 1992.
She is now working towards the Ph.D. degree in
section was an unbalanced load. electrical engineering at the Dipartimento di Elet-
The accuracy of the realized instrument was determined trotecnica of the “Politecnico di Milano” University.
experimentally by evaluating the measurement uncertainties Her current research interests are in the field of
the measurement of the power components under
introduced by the single steps of the measurement procedure. nonsinusoidal conditions.
The results in Table I have been attained.

V. CONCLUSION Alessandm Ferrem (M’88) was born in Milano,

Italy, on December 9, 1954. He received the M.S.
The method and related instrument presented in this paper degree in electrical engineering from the “Politec-
proved to be effective in identifying, in three-phase networks, nico di Milano” University, Milano, in 1978
the loads producing harmonic distortion. The active energy In 1983 he joined the Dipartimento di Elettrotec-
nica of the Politecnico di Milano as an Assis-
flowing backward through the metering section is proposed as tant Professor of Electncal Measurements. From
a good index of the power supply quality deterioration caused 1987-1991 he was Associate Professor of “Mea-
by the polluting load. surement on Electrical Machines and Plants” at the
“Dipartimento Elettrico, Elettronico e Sistenustico”
A Virtual Instrument was realized for the execution of of the University of Catania, Catania, Italy. From
real-time measurements of all quantities needed to monitor 1991-1994 he was Associate Professor of Electrical Measurements at the
the effects of a load on the supply quality, for fundamental Dipartimento di Elettrotecnica of the “Politecnico di Milano” University,
Milano, Italy. He is presently Full Professor of Electronic Instrumentauon at
frequencies up to 400 Hz. the Dipartimento di Elettrotecnica of the “Politecnico di Milano” University
The results of some experimental tests performed on a His current research interests are concerned with the application of digital
three-phase test circuit are in agreement with the theoretical methods to electrical measurements.
Mr. Ferrero is a member of the Italian C.N.R. Informal Group on Electrical
results and allow for direct implementation of the method in and Electronic Measurement and he is Member of the North Italy Chapter of
a specialized integrated circuit for large-volume applications. the IEEE JM Society.