EEE Transactions on Power Delivery. Vol. 7, No. 4, October 1992.

1793
EVALXJATIONOF REACTIVE POWER METERS
IN THE PRESENCE OF HIGH HARMONIC DISTOdTION

P.S. Filipski, Sr. Member IEEE P. W. Labaj, Member IEEE

National Research Council of Canada Ontario Hydro Research Div.
Ottawa, Ont., Canada K1A OR6 Toronto, Ont., Canada M8Z 5S4

- Four reactive power meters, operating on different these readin s can be interpreted as approximate measures of the

-
principles, were tested under nonsinusoidal conditions. Because different dekitions.
the definition and meanin of the nonsinusoidal reactive power are VE-oP
still being active1 debate$ the readings of the tested meters were
compared with &ur nonsinusoidal reactive powers and the first
harmonic reactive power. A digital instrument, programmed to
measure each of these powers, was used as a reference. When the voltage and current are sinusoidal, U ( t ) = 6
V s i n 03 t
and i ( t ) = $2 I s i n ( w t + 0 ) , the product of voltage and current,
The performance of revenue meters under distorted waveform known as instantaneous power A is the rate at which the electric
condiuons is no longer of theoretical interest only. There is a energy is transmitted into or out of the circuit.
proliferation of electronic loads and their number is expected to
maease at an even faster rate as energy-saving devices are ro p ( t ) = u ( t ) . i(t) = V I c o s @ - V I c o s @ c o s 2 w l +V I s i n 9 s i n 2wt
moted. Some of these devices are very nonlinear in nature, & a i = P ( 1 -cosZwt)+QsinZcut (1)
currents very rich in harmonics and promote voltage distortion.
The electrical industry is working to minimize the effects of this The average value of p , the active power P,
situation by proposing standards restraining harmonic injection
into the system. However, in the meantime it is important to P=Vlcos+ (2)
examine whether current instrumentation meets the requirements
of a changing situation [11.
Most of the practical tests to date have concentrated on energy is the average value of energy transfer.
and VA meters [2], [3]. There is very limited data on behavior of
reactive power meters [4]. There are two reasons for this. First, The reactive power Q is defined as
the definition and meanin of the reactive power under nonsinu-
soidal conditions are still keing actively discussed [l]. However, Q = V Isin+ (3)
there is an increasing recognition between engineers that the tra-
ditionally used definition of reactive ower ("Budeanu reactive The reactive power is added geometrically to the active power to
power") has no physical meaning a n z at least theoretically, its obtain the apparent power S
application can either lead to misleading conclusions or does not
supply the requiredinformation. There isno agreement on whether
S =V I = JP2+Q2 (4)
thls traditional definition should be discarded and what new defi-
nition should be used instead. Several proposals have been for-
warded, each havin its merits. Practically none have been used The efficiency of the system utilization is expressed as the power
in practice or impfemented in a commercial instrumentation. factor P F.
Second, reactive power meters are more difficult to test; there are
no reactive power standard meters on the market which would PF = P / S = COS@ (5)
operate under nonsinusoidal conditions (regardless of the defini-
tion). The sinusoidal reactive power has several interpretations,related
The work described in [4] is a good example of these difficulties. to its mathematical representation and the physical phenomena it
The authors experienced unexplained problems with calibrating expresses. Themost important properties and interpretationsare
an inductior reactive power meter. The reported results are for as follows.
sinusoidal voltage only, the test conditions were restricted to
available nonlinear load, and Q ("Fryze reactive power") was used (a)The reactive power is the quantity expressed in terms of V /sin+ .
as a reference, a quantity which is not universally recognized as the b) The reactive power is a signed quantity.
nonsinusoidal reactive power. c) The reactive power in the system can be balanced; the reactive
This paper discusses different definitions of the reactive power powers flowing into the node are equal to the reactive powers
and gives the reasons behind each proposal. Results of nonsi- ex orted from the node, such that their algebraic sum is zero.
nusoidal tests of four reactive wer meters, operatin on different (dr The reactive power is the magnitude of the bidirectionally
principles, are presented. gadings of the teste$ meters are pulsating component of p , and the active power is the magnitude
compared with different reference quantities to verify whether of the unidirectionally pulsating component.
(e) The reactive ower is proportional to the amount by which the
92 WM 192-5 PWRD A paper recommended and approved mean value of tLe stored electrostatic ener exceeds the mean
by the IEEE Power System Instrumentation & Measure- value of the stored electromagnetic energy 8 r i n g a cycle.
ments Committee of the IEEE Power Engineering Society ( f ) The reactive power can be compensated using a linear
for presentation at the IEEE/PES 1992 Winter Meeting, shunt/series reactive element, capacitor or inductor. It can be used
New York, New York, January 26 - 30, 1992. Manuscript to calculate the compensating shunt capacitor or inductor.
submitted August 26, 1991; made available for (g) The compensation of the reactive power increases the load
printing November 20, 1991. power factor to unity.
(h) The geometrical sum of the reactive and active powers is the
(7
a parent power.
I The line voltage drop in transmission lines, predominantly
inductive in Dractice, depends only on the reactive power. (The
active poweideterminesthe angle of transmission.)
0885-8977/9u$3.0001992 IEEE
1794

MAD LOAD
The definition of the nonsinusoidal reactive power is an extension
of the sinusoidal definition. However, in a nonsinusoidalsituation
there is no one quantity which has the same properties as the
sinusoidal reactive power. All nonsinusoidal extensions of the
reactive power definition preserve only certain properties, but do
not preserve the others. The main disagreement between the
different proposals can be explained as a disagreement of opinions
on what is the nature of "reactiveness"of the reactive power. Is it
related to the oscillation of energy? Or is it related to the method
of compensation (i.e improving the load ower factor)? Or is it
related to some other property? Some orthe definitions applied
in the nonsinusoidal situations are discussed below.
LOAD LOAD

The energy transfer from a Utility to its customers is conducted

at the mains frequency. Energy relations at this frequency are of
primary importance, the rest of the frequencies can be treated as
noise or pollution [l]. (This may not be the case for some
switching loads.) If such an approach is taken, then only the value
of the first harmonic reactive power Q I is important.

Q 1= I ' , I l s i n 9 , (6)

+
where 1. I , I I and I are first harmonic rms voltage, current and Fig. 1. Load-current splitting,
phase shift angle.
It should be noted that no reactive power meter available on the
market, apart from expensive harmonic power analyzers, specifi- When the current splitting approach to the reactive power defi-
cally measures the reactive power of the first harmonic. nition is taken then the reactive power is expressed in terms of the
ve Power rms values of the voltage and the current I Q associated with the
The classic definition of nonsinusoidal reactive power is due to given definition Q.
Budeanu [l], and is often called the Budeanu reactive power Q .
Q=V.I,
Q*= 1I ' h I h s i n + h
h
(7)
The drawback of this approach is that it does not take into account
the internal impedance of the source. In consequence, when a
where h is the harmonic number. Definition (7) has been struc- compensating shunt is connected the voltage and current wave-
tured to have a form analogous to the definition of nonsinusoidal forms and/or the configuration of the optimal circuit changes.
active power P. ive Pow=
The most general definition of the reactive power, based on the
concept of the load current splitting is due to Fryze (c.f. [6] and
[7]). The load current is decomposed into two components, Fig. 1a,
called the active and reactive currents
There is a perception among engineers that Q expresses oscil-
lations of energy between the source and the load, energy which is i = ia + ir (9)
not absorbed by the load but causes transmission losses and volta e
drops. This perception is false [SI. It presumes that the phase shifts The reactive current I is the "useless" current component which
between the voltage and current harmonics are caused only by
reactive, energy-storing devices, capacitors and inductors. In does not contribute to the net transfer of energy, but which has to
practice nonlinear elements, such as thyristor controlled loads, also be delivered to the load together with the active current. This
cause phase shifts between voltage and current harmonics, but do "useless"component causes transmissions losses and voltage drops.
not cause energy oscillations. Generally it is impossible to dis- The active, reactive and apparent powers are expressed in terms
tinguish whether the phase shift has been caused by a reactive of the rms values of voltage V and these two currents, I I .
element or a nonlinear load.
This definition of reactive power preserves properties (a), (b), Q F = V I r= , / ( V I ) ' - ( V I a ) ' = ~ ~ (10)
(c), and (h), i.e. it is expressed in terms of VIsinQ, it is a signed
quantity, its algebraic sum in the node is zero. In general it cannot Q &an be calculated directly from the apparent and active powers;
be used to calculate the compensating element. No commercial there is no need to have a separate reactive power meter. The
instrumentation is available for measuring Q . drawback of this ap roach is a loss of accuracy due to the calcu-
lation using the difkrence of squares. The active and reactive
powers add geometrically to produce the apparent power.
Q is always positive; leading and lagging power factors cannot
The followingdefinitions of reactive power are based on a concept be distinguished. It retains properties (8) and (h). Its compensa-
of dividing the load current into two or more components, pres-
umably responsible for different energy phenomena. The load is tion, by injection of the compensating current - i r , leads to the unity
presented in an equivalent form consisting of a linear equivalent power factor. Several compensators operating on this principle are
resistor R e = 1.' / P and a parallel combination of linear or non- discussed in the literature.
linear components or current sources, Fig. 1. The "basic" current Moore Reactive Power
component is called the "active current" i a. This is the current The compensation of the reactive power is carried out in practice
flowingthrough the equivalent resistor R e , Fig. 1 a, dissipating the by installing a bank of capacitors close to the load. The optimal
load active power. The remaining current components are asso- value of this capacitor bank, for given voltage and current wave-
ciated with the reactive power and/or other power components. forms, can be determined as follows.
 1795

The load current is decomposed into three components, active, of non-dissipated energy (i.e. (d) and (e)). At present the com-
capacitive reactive and residual currents. The capacitive reactive pensation aspect appears to be stressed (properties ( f ) and (g)).
current, i Q-KM in Fig. 1 b, has the same waveform and phase as In practice what a given uantity is called is not important. It is
that of the current in a capacitor with the same voltage across it. important to understand wlat a eiven quantity expresses, what it
The magnitude of this current is such that it minimizes the rms does not express and to use the right indicator for the given task.
value of the residual current. If an equivalent capacitor C e , drawing This is why the classic definition of the reactive power Q is mis-
this capacitive reactive current, has a ne ative value then a com- leading. In the nonsinusoidal situation it does not reflect any
pensating capacitor connected in parallef with the load will com- important property of the network.
pensate completely its loading effect. The compensating capacitor Aside from the theory one can distinguish three general areas
is the optiir.um capacitor for the given voltage and current wave- where the reactive power is used: power system management, load
forms. compensation and customer billing.
The power of this equivalent capacitor, called the capacitive In system management the first harmonic active/reactive energy
reactive power Q K M , and calculated as generation, flow, and absorption is of interest. The nonsinusoidal
reactive powers proposed to date would not help in determining,
for example, how to control the system voltage through reactive
power injection. The first harmonic reactive power appears to be
the most appropriate indicator.
In the nonlinear load compensation the primary objective is
bringing the load current to the sinusoidal-like zero-phase-shift
condition. How this objective is achieved and what indicator is
was proposed by Kusters and Moore [8] as the general definition selected to monitor the difference between the compensated and
of the reactive power. non-compensated state depend on the design. To varying degrees
If the capacitive reactive power is positive then no capacitive the nonsinusoidal reactive powers Q F , Q K M ,Q sharesuch indicators.
compensation is possible. In this situation an inductive reactive In customer billing two factors appear to be important. First,
current and an inductive reactive power can be defined 81. how much of the first harmonic reactive power is the customer
The Kusters-Moore definition preserves properties (b\ and (f), drawing? This causes transmission and distribution losses, upsets
with the reactive power bein a signed quantity compensable by a the reactive ower balance and the node voltage. Second, is the
single linear shunt capacitor$inductor. client's IoaJcurrent distorted, i.e. plluting the system and
ve Powa increasing the distribution losses? The irst factor canbe monitored
by measuring the first harmonic reactive power or displacement
The definition of the reactive current as a current drawn by an factor D F , the second by measurement of the load current total
optimum reactive shunt compensator of given configuration can harmonic distortion ( T H D ) . It is doubtful that a nonsinusoidal
be further extended for a two-element shunt, Fig. 1 c, Page [9], or reactive power can be used for this purpose.
to the maximum reactive current which can be compensated by a
theoretical optimal combination of capacitors and inductors, w l 3 2
Fig. 1d. No better compensation can be achieved by the use of A sample of four meters, using different operating principles,
linear reactive elements. The reactive power resulting from such representative of Ontario Hydro practice, were tested. The meters
a proach was first proposed by Shepherd in [lo], [1I], and gener- are denoted below as DIG, ANA, IWHM, IWHT 2e.
$ k d by Sharon [12]. Czarnecki interpreted this quantity using DIG - This is a sampling (i.e. purely digital) multi-function
the load-current-splitting concept [131. polyphase meter. The instrument takes instantaneous samples of
The following formula is used to calculate the Sharon reactive three voltages and three currents and calculates quantities such as
power Q S h . active power/energy, reactive power, voltage and current rms
values, power factor, etc., from these digital samples. Apparent
Q s h = V J V (12)
power (VA) is calculated usingvoltage and current rmsvalues. The
This definition preserves in a general sense the property (f). manufacturer does not provide particulars as to how the reactive
power is calculated. Assuming, that the 90" phase shift is obtained
Ibmd3mK by quarter-period time-delay of samples of one of the inputs, the
The loa" power factor P F is considered to be a measure of reactive power reading of the meter, expressed in terms of reac-
efficiency of the energy utilization by the load. In a sin8le phase tive/active harmonic powers, is proportional to
sinusoidal situation only a purely resistive load has umty power
factor, all other loads have a power factor lower than one. In a
sinusoidal three-phase system only a symmetrical resistive load can
QOlc =+I 0
u ( t ) i ( t - $ ) d t = Q , - P 2 - Q 3 + P,+ Q s . . .

have a unity power factor, [7].

The power factor is defined in terms of the active and apparent ANA - This is a solid-state (i.e. electronic) Wh/Lag/Lead Varh
powers; however, in practice it is often calculated from the readings demand meter. The active power is measured by Time-Division
of active and reactive power/energy meters. The result of this Multiplier (TDM) elements and the reactive power is measured
approach P F ' is always larger than P F (thus disadvantageous to by multiplying current by voltage shifted by 90". An active (inve-
Utilities). rting) integrator is used to perform the phase shift. The reading
of the reactive power element is thus proportional to
D D D

QANA = +J 0
w , [ - j u ( t ) d t ] t ( t ) d t = Q 1+ Q2 / 2 + Q 3 / 3 . . .

where Dis the "Budeanu distortion power".

The apparent power is calculated from \i P * + .a
,:
In a nonsinusoidal situation P F has no relation to the phase shift
IWHM - This meter consists of two separate three-element
(i.e. P F f cos Q). For this reason some commercial instruments inductionwatthourmeters and a microprocessor-controlled calcu-
also display displacement factor D F , which is the power factor lator and display. This is essentially an induction watthourmeter
calculated for the first harmonic ( D F = cos+ ,). with a digital display. The potential coils of the second watt-
ve P o w hourmeter are lagged by 60,. (rather than by 90 ") by cross-phasing
and the meter measures so called Qhour Qh. The reactive power
The discussion about the nonsinusoidal reactive power is a is calculated as
discussion about the "reactiveness",which quantity inherits suffi-
cient number of the sinusoidal reactive power properties to be 2Qh - P
called the nonsinusoidal reactive power. In the past the most Q I W H M = ___
important properties were considered as related to the oscillations 6
1796

The meter displays the active power, reactive power and apparent The test conditions, i.e. voltage waveform and current waveform
<
power demand calculated from P Q fVH,,,,. In a nonsinusoidal
+
numbers, amplitudes, and phase shifts, are entered from the key-
board. The program loads the required waveforms, stored pre-
situation the calculated reactive and apparent powers are complex viously on a hard disk, into the arbitrary waveform generators and
functions of active and reactive harmonic powers. sets the appropriate magnitude of their outputs. This outputs are
IWHT 2e - This is a two-element combination energy and kVA then amplified to the required levels by set of three voltage and
thermal demand meter. An inductionwatthourmeter is used to three current amplifiers. The tests were performed using the
measure energy. The apparent power is measured using a recti- "phantom load" technique.
thermal register; the voltage and current waveforms are rectified, Voltages and currents, converted by a set of shunts to voltages,
filtered and
- then- multiplied in a thermal quarter-square multiplier, are measured using two wideband thermal ac voltmeters. Active
s,,,,-I L ( t > I I t ( t > I. power is measured using a standard Time-Division-Multiplier
P (TDM) wattmeter. Reactive powers (a,,
Q B , Q F ,Q S h , Q X M ) are
measured using an instrument marked as a digital recorder, shown
mS3PE.m in detail in Fig. 3. The digital recorder consists of two digital
The schematic diagram of the NRC test system is shown in Fig. 2 multimeters and a PC-AT 386 type computer. The multimeters
[2).It consists of six computer-controlled digital arbitrary waveform sample the test signals. These samples are sent via the IEEE-488
generators, three voltage amplifiers and three current amplifiers. bus to the computer which calculates the required quantities. This

VOLTAGE
AMPLIFIERS

AMPLIFIERS - - _

Fig. 2. Three-phase nonsinusoidal calibration system.

INDUCTIVE
VOLTAGE DIVIDER
r----------i

Fig. 3, Digital recorder.

 1797

is done either by operation on the time samples (Q ) or on fre-

quency components obtained after a fast Fourier transform of the
data (QI ,Q B ,Q S h , Q KM). Sampling is synchronized with the fun-
damental harmonic frequency of the test waveforms, which in turn
is synchronized with the main frequency, Fig. 2. The harmonic
content of the test waveforms was limited to 128 harmonics. For
this reason the sampling frequency was selected to be equal to 512
times the fundamental frequency, i.e. four times the maximum
harmonic frequency.

The meters were tested under sinusoidal conditions and eight

nonsinusoidal conditions. The nonsinusoidal test waveforms are
shown in Fig.4. These waveforms were acquired in the field
harmonic problem locations and, after limiting their spectra to 128
harmonics, stored in the NRC waveform library. The meters were
thus tested under simulated field conditions. Table 1 shows vol-
tagelcurrent test waveform numbers, test active power, power VOLTAGE NRC0139 CURRENT NRC0140
Ir I1
factor, reactive powers as well as voltage and current harmonic
distortions. The first two tests (001/001) refer to the sinusoidal
conditions with the nominal phase angles *30". As the data in
Table 1 indicate the test conditions generated in practice differ
slightly from the nominal test conditions. The test voltage was
120 V rms, the test current 5 A ,,s, except for conditions 2501251
and 2521253 where it was 4 A,,, and 4.5 A,,,, respectively.
VOLTAGE NRC0179 CURRENT NRC0180
Table 1. Test Conditions.

m
CURRENT NRC0193

VOLTAGE NRC0250 CURRENT NRC025 1

Test UncemhtY
The uncertainty of the test results de ends on the accuracy of the
reference instruments, the stability orthe test conditions and the
repeatability and resolution of readings of the tested meters. It
varieswith the measured quantity, the test conditions and the tested
meter.

m
The uncertainty of the reactive power reference meter is difficult
to estimatein nonsinusoidal situations. However, assuming alinear VOLTAGE NRC0252
o eration, and taking advantage of the known harmonic content
ofthe computer generated waveforms, the influence of a particular
error source can be computed from the frequency characteristic of
the error. Linear operation of the instrument must be verified by
comparison with a different reference.
In order to simplify the presentation of results the maximum

mi
calibration system uncertainties and meters' uncertainties are cited
below.
The calibration uncertainties include combined uncertainties of VOLTAGE NRC0254
the standard instruments such as wattmeters, voltmeters, and
digital recorder as well as the statistical uncertainty due to results
averaging. At the three standard deviation level (3- a) the reactive
powers were measured with uncertainty of 0.09% of full scale i.e.
1800 VA.
The meters' uncertainties originate from a non-repeatability of
readings, which varies with the test conditions, and a resolution
uncertainty. For example, IWHT meter does not indicate the VOLTAGE NRC0256 CURRENT NRC0257
reactive power, it has to be calculated as Q I W ~ : T= J S * - P *. Error
propagation in this equation is very unfavorable. For example, the Fig.4. Test waveforms.
resolution uncertainty of a IWHT 2e meter equals *OS division i.e.
10 W or 0.55% at 1800 VA level. The resolution uncertainty of the
reactive power calculations can attain

1 -
1798

\ (1800+ 10)'- 1800'= 190, i.e. 11% of the full scale at the Table 3. Reactive Power Differences.
1800 VA level. Al.,=(Q,-QA)/ F u l l Scale VA
The combined uncertainties of the test results were estimated as
not exceeding: DIG,0.10%, ANA,O.lO%, IWHM, 0.2%, IWHT 2e, A = B , F , KM,S h
1.4%.
TEST RESULTS TEST Al-t! Al-F Al-KM Al-Sh
% % % %
The test results are presented in Table 2. All errors are referred
to the full scale of the test level, i.e. 1800 VA. For each meter the 137/138 t0.0 -5.8 t 7.5 -2.8
first column shows the relative difference between the readings of 139/140 t1.6 -41 -2.7 -26
the meter and the first harmonic reactive power Q I . The second 179/180 t 0.2 -8.7 t 2.7 -5.4
and the third columns, use, respectively Q and Q as the reference 1 192/193 1 -0.0 1 -3.4 1 t31 I -1.2 1
reactive powers.

Table 2. Test Results. Reactive Power Measurement Errors. 252/253 I +4.8 I -24 t38 -14
254/255 I t1.4 I -8.4 I t25 I -2.0
1= (Q,,,,, - Qreference)/ F u I ~Scale 1 L 2S6/257 I t1.6 I -8.6 I +68 I -6.0 I
meter = DIG, ANA, IWHM, IWHT
Q reference = Q i . Q B Q F
3 specific to the meter design or to the selected definition. It is clear
that in most cases the selected definition is the main source of the
error. For example, under conditions 2501251 there is a very large
difference between the readings of the meters DIG, ANA and
IWHM, close to -40%, and Q F . Table 3 shows that this is due to
the difference between Q and Q F , equal to -41%, rather than to
any particular meter design.
Table 3 additionally shows how the readings of the meters
compare with the two others definitions of the reactive power, Q K M
and Q S h . It is obvious from the magnitude of these differences,
often larger than 10% and even approaching + 68%, that none of
the tested meters could be used to measure these reactive powers.
Energetic relations in the load, expressed by these quantities,
cannot be estimated from the readings of conventional reactive
power meters.
CONCLUSIONS
256/257 -0.5 I t 1.1 I -9.1 -0.2 I t1.4 1 -8.9
Indication of reactive power meters in nonsinusoidal situations
IWHM IWHT depend on their operating principles. However, in practical situ-
001/001 t0.2 I t0.2 I t0.2 13.4 I 13.4 I +3.s ations, such as discussed in this paper, there is a relatively small
I 001/001 I -0.7 I -0.7 I 1-0.7 1 t3.4 I t3.4 I t3.4 I difference between the indicated reactive power and Q I . With
137/138 I t0.2 I t0.2 1 -5.6 1 -8.2 I -8.2 I ~~

-14 errors not exceeding a few percent (-3% in this paper) readings of
the DIG, ANA and IWHM meters can be considered to reflect Q .
139/140 I -1.3 I t0.3 I -42 I +34 I t35 I -6.9
It should be recognized that for certain applications this is the
I 179/180 1 +0.2 I t0.4 1 -8.6 1 +4.7 1 t4.9 t -4.1 1 reactive power of interest. An effort should be made to measure
192/193 tO.l tO.l -3.4 t2.4 t2.3 -1.0 this quantity accurately. Only expensive harmonic analyzers
250/251 +OS t1.6 -40 +32 +31 -9.2 measure Q I quantity explicitly.
252/253 t2.2 t7.0 -22 -1.6 13.1 -26 The classic definition (7) of reactive power Q has no physical
254/255 I -0.4 I t1.0 I -8.8 I +14 I t15 I -5.0
interpretation; there is no need to increase accuracy of these
measurements. None of the meters tested actually measured this
256/257 I t1.4 I +3.1 1 -7.2 1 -4.0 1 t5.6 I -4.2 quantity.
IWHT type meters should not be used in nonsinusoidal situations.
of -T Their VA demand registers can exhibit large errors under these
conditions [2]. The root-difference-square method of calculating
All meters were designed to operate properly under sinusoidal the reactive power from their readings is prone to large errors.
conditions, thus to measure correctly the first harmonic reactive There can be a large difference, exceeding several percent,
power. The difference between the reactive power registered by between the reactive power as indicated by the tested meters and
the meter and the first harmonic reactive ower is therefore the
most meaningful indicator of the accuracy otthe meter. DIG, ANA quantities Q F , Q S h , Q K M . It means that the readings of the tested
and IWHM type meters register the first harmonic reactive power meters cannot be used to determine an optimal compensating
with errors exceeding the 1%level in only a few cases. The largest capacitor (QK M ,errors up to t 68%), or the rnaximumcompensable
errors, reaching -3% for the DIG meter, are for conditions where reactive current/power (Q F , errors up to -41%) or the maximum
Q , is small, i.e. 250/25 1,c.f. Table 1. Similar comment can be made reactive current/power compensable by linear shunt elements
about the results obtained when Q was used as a reference. (Q S h , errors up to -26%).
Very large errors, reaching +34%,were recorded for the IWHT When monitoring nonlinear loads, one should take into account
e meter. Here the main sources of error are a poor resolution not only the consumption of the first harmonic power but also
F t h e VA register and the method of calculating the indicated harmonic "pollution" of the system by the load and harmonic-
reactive power from the root-difference-square equation. These related increased distribution losses. There is no agreed upon
errors are largest when the difference between the active and indicator which accentuates the second factor. In this application
apparent owers is smallest. the DIG type meter appears to be the best of the tested meters. It
To clarig the source of the errors, Table 3 shows the differences measures Q which approximates Q I and can be thus used to
between the reference first harmonic reactive power Q and the estimate the displacement factor DF.It also measures rms values
four other reactive powers, Q , Q F , Q S h , Q h h f . By comparing the (harmonics included) of voltages and currents and P F based on
figures in Tables 2 and 3 one can find whether the meter error is these values.
 1799

ACKNOWLEDGEMENTS [8] N.L. Kusters,and W.J.M. Moore, "On Definition of Reac-

The authors wish to thank Ontario H dro Energy Management tive Power Under Nonsinusoidal Conditions," IEEE
Branch, Technical Services and Devefopment for their financial Trans. Power App. Systems, Vol. PAS-99, pp. 1845-1850,
support of this research project. Sept 1980.
[9] C. H. Page, "Reactive Power in Nonsinusoidal Situations,"
IEEE Trans. Instrum. Meas. Vol. IM-29, pp. 420-423,
REFERENCES Dec. 1980.
[l] GE. Emanuel, "Powers in Nonsinusoidal Situations. A [lo] W. Shepherd, and P. Zakikhani, "Suggested Definition of
Review of Definitions and Physical Meaning," and Discus- Reactive Power for Nonsinusoidal Systems," Proc. IEE,
sion by J.M. Willems, T. Tang, J.H.R. Enslin, P.S. Filipski, Vol. 119, pp. 1361-1362, Sept 1972.
Y.Baghozouz, LS.Czarnecki, IEEE Trans. on Power
Delivery, Vol. PWRD-5, pp. 1377-1389, July 1990. [ll] W. Shepherd, and P. Zand, Energy Flow and Power Factor
in Nonsinusoidal Circuits, Cambridge University Press,
[2] R. Arseneau, P.S. Filipski, "Application of a Three Phase Great Britain, 1979.
Nonsinusoidal Calibration System for Testing Energy and
Demand Meters under Simulated Field Conditions," [ 121 D. Sharon, "Reactive Power Definition and Power Factor
IEEE Trans. Power Deliv., Vol. PWRD-3, pp. 874-879, Improvement in Nonlinear Systems," Proc. IEE, Vol. 120,
July 1988. pp. 704-706, June 1973.
P.S. Filipski, R. Arseneau, "Behavior of Wattmeters and [ 131 L.S. Czarnecki, "Considerations on the Reactive Power in
(31 Nonsinusoidal Situations,"IEEE Trans. Instrum. Meas. Vol.
Watthourmeters under Distorted Waveform Conditions," IM-34, pp. 399-404, Sept. 1985.
IEEE Tutorial Course, Nonsinusoidal situations: Effects on
the erformance of meters and definitions of power, IEEE
Pdkations, New York, NY, 1991, Publication #
90EH0327-7-PWR, pp. 13-22. BIOGRAPHY
(M84 - SM90) received M.Sc. and Ph.D. degrees
[4] M.D. Cox, T.B. Williams, "Induction Varhour and Solid
State Varhour Meters Performances on Nonlinear Loads," Y
in electrica engineering from the Silesian Technical University,
Poland, in 1972 and 1977, respectively. From 1972 to 1981 he
and Discussion by AE. Emanuel and P.S. Filipski, IEEE
Trans. on Power Delivery, Vol. PWRD-5, pp. 1678-1686, worked at the Silesian Technical University. In 1981 he joined the
October 1990. National Research Council of Canada, and at present is a Senior
Research Officer in the Institute for National Measurement
[5] L.S. Czamecki, "What is Wrong with the Budeanu Conce t Standards.
of Reactive and Distortion Power and Why It Should i e Dr. Filipski is a past chairman of the IEEE Ottawa Section and
Abandoned," IEEE Trans. Instrum. Meas. Vol. IM-36, pp. a member of the Association of Professional Engineers of Ontario.
834-837, Sept. 1987.
(S'76 - M78) received his B.ASc. degree in
[6] P.S. Filipski, "A New Approach to Reactive Current and ring from the University of Windsor, Ontario. He
Reactive Power Measurement in Nonsinusoidal Systems," has since worked as a research engineer at Ontario Hydro,
IEEE Trans. Instrum. Meas. Vol. IM-29, pp. 423-426, Dec. Research Division, in the Electrical Metrology group.
1980. Mr. Labaj is a member of the Association of Professional Engi-
P.S. Filipski, "PolyphaseApparent Power and Power Factor neers of Ontario.
(71
under Distorted Waveform Conditions," IEEE Trans.
Power Delivery, Vol. PWRD-6, pp. 1161-1165, July 1991.