Passive filter design for harmonic reactive power

compensation in single-phase circuits supplying

nonlinear loads
E.F.El-Saadany, M.M.A.Salama and A.Y.Chikhani

Abstract: The proliferation of dil'fcrcnt typcs 01' nonlinear loads into a distribution system has the
potential to increase tlie harmonic levels in the system. The cTfcct of attenuation on reducing the net
current harmonics is reported for different nonlincar loacis. Calculation of nonlincar load
stisceptanccs at different harmonic frcquencies is prcscnted. Thc design proccdurc of a reactance onc-
port compensator capable of reducing tlic voltage and current harmonics produced by nonlincar
loads supplied from a sourcc having interiial inductive impcdancc is offcrcd. The reactancc one-port
compensator application to a secondary distribution system loadcd with different load types is
presented. The analysis uses the clcctromagnetic tl'ansient progfinii (EMTP) to model diiTcrent loads
as well a s tlic distribution system. Finally, physical experiments liavc been conducted to verify tlic
results obtained by simulation.

1 Introduction filters [3], the electronic vollagc regulator and the adaptive
VAR compensator (AVC) can cause further distortion in
Different types of nonlinear loads arc expected to prolifer- thc distribution system. This is attributcd to tlie inherent
ate into tlie distribution systems, atusing the harmonic lev- nonlinearity associated with the semiconductor elements
els on thcsc systems to increase. Among the diffcrcnt single- and by switching operations involved in their usc.
phase loads connected to the distribution system, thrcc load Tlie passive circuits used to reduce tlie distribution sys-
groups arc of significant importance. kin current and voltage harmonics alii be divided into
Tlie first group contains those loads that employ the three categories. The first is to use shunt capacitors [4, 51;
capacitor filtcrcd diode bridge rectifier (DBR) a s their however, this inay generate unexpected harmonic voltages
power supply. The cnrrcnt topal harmonic distortion
due to tlic scrics resonance.
(THDI) is normally in the range 01' 100'%~Examples of
such loads include desktop computers, television sets, bat-
The sccond category is lo use a series LC compensator
[6].This nictliod provides a higher power factor compared
tery chargcrs and adjiistablc speed dl-ivcs (ASD) Tor heat
pumps and air conditioning. to the pure capacitive compensation. However, in [7] it was
The second load group conlains the compact fluorescent shown that despite tlic higher power kictor value gained by
lamps (CFL) that employ either magnetic ballast or clcc- the scrics LC compensator, the source current harmonics in
Ironic ballast. The CFL ciin be Ttirtlncr divided into thrcc tlie circuits having /,C compensators are higher than in the
categories depending on the dcgrcc of current distortion uncoinipcnsated circuits. Only the fiindanicntal harmonic is
[I.21, which rangcs from I5 to 140'%1. red uced .
The third group of single-phase loads connccted to the The third catcgory of passive circuits that arc utilised to
distribution system contains those loads utilising the phase rcducc the system liartnonics is thc rcactive one-port nct-
control of thyristors to control both the input A C voltage work. In [7, X I a rcaclance one-port compcnsator has been
and the input power. The major loads appearing in this employed succcssfully to rcdttcc the current harmonics pro-
group are light dinuners, heating loads and controls of sin- duccd in circuits having linear loads and fed rrorn 21 sotircc
gle-phase induction motors. of periodic nonsinusoidal voltage and zero internal iinpcd-
Different techniques have becn iniplemcntcd to reduce inice. However, two important concerns have not been
both current tunc1 voltagc Iiiirmonics. Thcsc lechniqucs addrcsscd. The distribution system is hcavily loadcd with
depend on using auxiliary circuits, active or passive. nonlinear loads and thc source inductivc impedance cannot
connccted in parallel at the points of common coupling be neglected in any practical system.
(PCC). Tlie use of active electronic devices sticli as active Prcvious distribution system harmonic studies 19, IO]
cniploy harmonic current sourccs with constant magnitudes
to rcprcscnt difCercnt typcs of nonlinear loads. I-lowcver,
these studies did not consider the c1'fcct oT the interactions
hctween the load voltage and currcnt harmonics duc to tlic
presence of the supply impedance. Neglecting these interac-
tions could lead to wrong cstimation for both voltage and
current harnionics. Tlie clcctromagnetic transient program
(EMTP) will be tiscd to niodcl the nonlincar load,
a s thc ovcrall systcm [I I].

371
2 Load modelling R I ,= 1 0 9 as shown in Fig. 4. The snubber circuit consists
or a series RC circuit where R = 19 and C = 2Ok.P. The
For the purpose of load voltage and current harinonic firing angle of the thyristor is adjusted at 45 dcgrees.
interaction study, the time domain method is used to repre-
sent the load nonlinearity and voltage dependency. IQ 200PF

2. I Diode bridge rectifier (DBR)model

The circuit model of the DBR load is given in Fig. 1. The
load consumes 0.67kW and is energised from a 1 IOV
supply that has a fundamcnlal frcqticncy equal to 60Hz.
The DBR unit has a smoothing capacitor or 100OpF. The Rload
snubber circuit consists of a series RC branch connected in
parallcl with cach diode, whcrc R = I 9 and C = 33kF. The
load resistance = 209.

2.4 System model

Thc system shown in Fig. 5 is considered. Thc supply volt-
age is sinusoidal with 1 10V (RMS) and the supply inipcd-
ance Z,, = R,, + ,jX', = 0.8 + .j0.9439. The X/R ratio of the
supply impedance was chosen to be 1.2 to represent a low
voltage transforiner.

'I
4 lood I h

Fig. 1
+=4 ----t

Lood 2

lood 3
2.2 Compact fluorescent lamp (CFL)model
The second load represents a single-phase fluorescent lamp
(CFL) that employs magnetic ballast. The CFL consunics
40W with IIOV (RMS) rating. In this paper the CFL load Fig. 5 DRR, CFL mid P A VC lads sIi(iriii,q. \ m e ,siipp/i~iiwpe(1iiiim
voltage and current characteristics (results of experiment)
that are given in Fig. 2 will be utiliscd to obtain a rcprc- 3 Solution algorithm
sentative model. The CFL model is formed from two
switched nonlinear resistors connected in parallel as shown In practical distribution systcms, thc source impedance can-
in Fig. 3. Each resistor will have the V/I characteristics of not bc neglected. Thc harinonic voltage drop crcatcd on
one-half of the CFL load characteristics. this impedance due to the current harmonics injected by
diffcrent types of nonlinear loads will distort thc load volt-
age. To account for the interactions betwccn the changes of
current harmonics injected by the nonlinear loads and the
rcsulting variations in the load voltagc, an iterative tech-
=? nique is employed. This technique can be described accord-
ing to thc following flow chart given in Fig. 6. This
2 4 6 8 10 nicthod converges typically within 10-20 iterations dcpcnd-
ing on the distortion degree. Kill, VLlS,and VI-,,,,are the
voltage drops on the supply impedance, old load voltage
and new load voltage, respectively.

-150 - 4 Voltage and current harmonic interactions

To investigate the cfl'ect of the load voltage and current

harmonics interactions on reducing the total current har-
monic distortion, cach load type is energised once consider-
ing an ideal power supply (having no internal impedance)
and another time assuming a practical power supply having
a source impedance equal to 0.8 + j0.943Q.

I Computer simulation 4.

Fig.3
,
,Siiig/e-pli~i,w,
tiiqy& CFL loin1 iiiodd
liiill~i~st,
It was observed that the system total current harmonic dis-
tortion (THDI) could bc rcduced due to the attenuation
effect. These interactions will affect the vollagc and current
2.3 Phase angle AC voltage controller (PAVC) waveforms as well. The waverorins of both load voltage
model and ctirrcnt for DBR, CFL and PAVC loads with and
The third nonlinear load represents a 0.9kW singlc-phasc without the iuleraction effects are given in Figs. 7-9,
A C voltage controller (PAVC) supplying a resistive load respectively.
374 /El< I'rm - f h w r . 7'rmiw?. Di.i.iril>, Vol. 147, No. 6. N,wcwiho. ZlJllO
 set load voltage at bus 2

calculate tatal supply current Is

i
draw load voltage and
supply current waveforms
4
perform FFT on
supply current

150

;-
al
L
100

5
U
50
U
s
>
o
a- -50
m
0
5
-100
current woveforms
-150

Percent of fundamental current

Load type l3 l5 ,1 19 Ill Il3 Il5 THDl

DER wloint. 72.5 34.1 7.8 10 6.16 2.9 4.16 81.5
DER wiint. 57.1 12.1 6.65 2.8 2.8 1.3 1.47 58.9
CFL wioint. 10.5 7.8 4.7 2.1 0.7 0.4 0.6 14.25
CFL wiint. 9.1 5.773.1 1.7 1.2 0.7 0.35 11.5

5 Reactive power compensation

In previous studies [7], thc reactance one-port was used to

minimise the cLirrent distortion in circuits supplying linear
loads from a nonlinear voltage sotircc. In this paper a novel
approach is developed to use the reactance one-port to
4.2 Physical experiments minimise the voltage and current harmonics in circuits with
Laboratory tests were conducted to support thc simulation nonlincar loads. To minimise the currcnt harmonics, the
results presented in Section 4.1, These tests were perrormed compensator stisceptancc should he equal to the negative of
on two typcs of loads, namcly, the DBR and CFL loads. It the load susccptance at each harmonic ' H ' . That is,
was found that the laboratory results are in lull agrccnient
with tlie simulation output. The net harmonic currcnts Bo, = - B h (1)
flowing through the common supply impedance for differ- where &,, is the reactance one-port compensator sus-
ent types of loads once considering tlie interactions and ccptaiice at harmonic n and B,,,, is the load stisceptmce at
another time without intcractions arc given in Tablc 1. harmonic 11.
 The first step towards achieving the desircd goal is tlie j0.94352 For this load (capacitivc load), and using the pro-
calculation of the nonlinear load susceptance. cedure described in Section 5.1, the load susceplances wcre
found to be: BI = 0.044S, B, = 27.88 and B, = 8.5s.The
5. I Load harmonic susceptance calculation plot of the required susceptance function YC(u)is givcn in
In [12] it is assumed that if under no load conditions the Fig. 10.
busbar voltage is purely sinusoidal, then the AC system The compensator will have such susccptance if its inipcd-
harmonic impedance can be evaluated by the ratio between aiice is in the form
the harmonic voltage and the harmonic current. I-lowcver,
this inethod is not accurate since it assumes that the noiilin-
ear load is represented by constant magnitude harmonic
current injection sources. This assumption implies that thc
wherc A is the gain, A > 0, Z , and Z, are the compensa-
voltage and current harmonics interactions arc neglected.
tor's, PI, P2 aiid P, are the compciisator's poles, and
T o account for the voltageicurrcnl harmonic interactions
when calculating the load susceptance, the following proce- 15 P, 5 2 1 5 3
dure will be followed:
(i) From the voltage and currcnt waveforms obtained in
3 5 12, 5 55 z,
(3)
Section 4. I , observe whether the current lcads or lags the 5 5 P:(
voltage in order to know whether the load is inductive or Assuming, Tor example, PI = 1.2, P, = 4.0 and P, = 8.0,
capacitivc. the numerator of eqn. 2 can be expressed a s follows:
(ii) Depending on the load impedance type insert either an
inductor or capacitor in parallel at the PCC.
h ( S 2 + +
Z f ) ( s Z 2;)= (1.5S5 (1:3,':+ u1.s (4) + +
The coefficients U>, + and can be calculated from eqn. 1
(iii) As an example, for a capacitive circuit, clxmgc the coil and substituting for s = jn, where n E { 1, 3, 5 ) . The equa-
value in steps and for each step calculate the supply currcnt tion
aiid perfomni FFT on it.
(iv) Kccp changing the coil value till a minimum current at
a specified harmonic frequency IZ is reached. In this case tlie
value of the coil susceptaiice at that frequency n will be results in
equal to tlic negative of the load susccptancc at the sainc
frequency.
(v) Tune the coil again till a minimum current at a different
harmonic frcquency is reached. Repeat for all the interested
harmonics. The solution for these equations' results in the compensator
impedance is givcn a s
5.2 Numerical example 9.5GGs 4.373 3G.96.3
After obtaining the load susccptancc, the required one-port
compensator complexity should be identified. In [I31 it was
Z,(S) =
(.s> + 1.22) + (9+ 4.09 + (,+ + 8.02) (7)
stated that the compensating reactance one-port should Thus the resulting compensator will have the structure
have a complexity equal to M = N(2N - 1) for N harmon- shown in Fig. II.
ics. However, in [7] it was shown that Tor A' haimonics tlic
required complexity is much lower and not higher than N c 17.6mH 0.72L mH 1.5 mH
M B 2N.
The compensator complex admittance Y,.(,F) or complex
impedance Z,.(s) is determined by the number of paramc-
Z,(W) 277pF 607uF 71.8pF
ters equal to its complexity, i.e. 2N, whcre s is the complex ___c
variable. According to eqn. 1 we have only N equations. ~1:377rod/s
The remaining N parameters can bc choscii arbitrarily by
assuming the values of the compensator poles. Lct us illus-
trate the synthesis procedurc [ 141 for a one-port compcnsa- 5.3 Computer simulation results
tor capable of reducing the current harmonics injectcd by ii Inserting the onc-port compcnsator shown in Fig. 11 in
DBR load. parallel with the DBR load will reduce the current total
compensator harmonic distortion (THDL) from 58.9 to 7.86% Also the
suscept once load voltage total harmonic distortion (THDV) will be
decrcased from 14.35 to 5.15'%. Thc waveforms of the sup-
ply current and thc load voltage before and after inserting
the compensator are given in Figs. I 2 and 13, respectively.
The Same design procedure was followed to design a
reactive one-port compensator capable of reducing the
current aiid voltage harmonics for a circuit having a PAVC
load. The thyristors were supplied from a l l 0 V supply
having internal impcdance equal to 0.8 + j0.943Q. The
Fig. 10 l'lot .JC ~ I ~ I ~ ~ I , S ( I ~ I.sii,srepiiiiiw
I~ lnnisfi,r Isl, 3r.d oial51lr li(i~'i~ionk. load is resistive with R = 5'2. Thc compensator succeeded
KWl'liVl~/J(Jll'('? C ~ l i 7 l ~ l l ~ i ~ , \ i i / ~ ~ l l 7 in reducing the THDI Tram 20.65 to 7.1'%1and the load
voltage THDV was reduced from 17.2 to 10.3'%~Tlic
A 0.34kW DBR load having a sniootbing capacitor of compensator construction along with the waveroims or
lO00pF and a load resistance of 20Q is supplied Trom a both the supply current and the load voltage before and
l l 0 V supply having an internal impedance of 0.8 + after compensation are givcn in Figs. 1 4 1 6 , respectively.
376 1lCE Proc.-ti~,icr..I k n . v > ~nirii
. ill.. I'd. 147, No. 6, Noaciid>er 2000
 150r

30r

20

0.03 0.04 0.05 0.06 0.07 0.08 0.09

I
0.10 :- 10
e
2 0
-ax
?
VI
-10

55.8mH 2.32mH L.71mH -20

2, I W I 87.5pF 190.0 PF 23.L W F

___c
W1.377 rad/s

30r 1501-

-301
0.0167 0.0267 0.03670.04670.05670.06670.0767 0.08670.0967 0.0167 0.0267 0.0367 0.0467 0.05670.0667 0.07670.08670.0967
time,ms
 Furthermore, another compensator was designed to R3 = 125.015' and B, =75.76S. The reactance one-port
reduce the current and voltage harmonics for a circuit hav- compensator was found to have impedance
ing a combination of DBR and CFL loads. The DBR load 1.4.318s 6.6085 52.423s
has a 20Q resistor on the DC side and draws 8.176A = (sZ + 1.22)
zcl(s) + (9+ 4.02)+ (,$a + 8.02)
(RMS) and the CFL load draws 8.57A. This compensator
reduces the THDI from 24.86 to 2.7% and the voltage (8)
THDV was reduced from 12.9 to 2.4'%~The compensator The compensator has the structure and parameters given
construction along with tlie waveforms of both the supply i n Fig. 21. Inserting this filtcr at PCC succeeded in rcduc-
current and the load voltage before and after inserting the iiig the T H D of both supply current and load voltage to
filter are given in Figs. 17-19, rcspectively. 18.51 and 13.19'%1,respectivcly. However, these distortion
le\& violate the recommended values [I 51, and thus
5.4 Test distribution system another compensator stage had to be designed in conjunc-
The reactance one-port compensator was additionally tion with the first stage to further decrease tlie harmonic
examined using a more complex distribution system that is distortion levels to the acceptable values.
given in Fig. 20. The distribution system feeder section data
are listed in Table 2. The proposed 3-phase secondary 26.37 mH 1.095mH 267mH
distribution system supplied a mixture of nonlinear loads
and the system was loaded until it reached its rated capac-
ity. The load sharing percentages will be equal to DBR =
40%, PAVC = 20'%1,CFL = 20% and the remaining 20'%1is z,, Is) 185.2pF 401.37kF 50.6uF
___c

3-phase star-connected linear loads that have 90% power WIZ 371 rad/s
factor lagging. The resulting T H D for both supply currcnt
and load voltage were found to be equal to 30.18 and
19.9%1,rcspectively.

1500 KVA
13.8/L.I6KV
2lY/Y)
z5,5~0
e tion on
transformer
The susceptance of the equivalent load (including the
first compensator stage) was recalculated and found to be
B, = 0.2798, B, = XX.94S and E, = 0.885s. The impedance
function of the second compensator stage was obtained
following the Same design procedure and was found to be
1.38,s 0.1,s
+ 11.215s
1 1 1

Q secondary
other secondory lines L ' 1 6 < ~ ~ ~ ~ K distribution
combinations
100mt f
V
w i t h different b o d Z=L,3%,R=1.Lo transformer
= (9 1.22)
zc2(s) +
The second compensator slage has the structure and
parameters shown in Fig. 22. Applying the second stage in
(9)
+ (,SZ + 4.02) (s' + 8.02)

parallel with the first slage resulted in further reduction in

tlie system harmonic distortion levcls. The THDI was
reduced to 8.66'%1 and the THDV was reduced to 8.01'%,.
These results indicate that the nonlinear susceptance calcu-
lation was not very accurate, leading to km unoptimum
compensator design. This can be attributed to the system
complexity, since the voltages cross different loads con-
L nected at different buses arc diuferent because of the prcs-
cnce of the feeder section impedance.
Rg DBR3 PAVC3
2.551mH 0.0165mH 0.L65mH
Fig.20

Table 2 Secondary distributionsystem feeder data

Zczls) 1915.0pF 26.525 p F 236.51F
P
Section Type WIZ 377 rad/s
Fig.22 .%wn,nl/ .si(i,ye wimiice o~re-i~ori I O ~ im s y i e m riwc-
C O I ~ I ~ I ~ I W .fi,r
lIW / l O l l ' U ~ f I l i l / J ~ ~ l , S ~ l l ~ ~ J l l
1 3-phase 100.0 16.0 1.42 0.106
2 3-phase 100.0 95.0 0.239 0.081
To minimise tlic filter complcxity and cost, the two com-
3 Single-core, PVC 50.0 50.0 0.464 0.112
pensator stages were combined together, leading to the
4 2-core, PVC 30.0 16.0 1.38 0.08 combined compensator transfer function given as
5 2-core, PVC 20.0 16.0 1.38 0.08
1.283.5 0.582,s 4.805,s
zr,<>,c,l= (s2 + 1.22) (.s2 + 4.P) ( s 2 + 8.P)
The supply impedance, which is equal to tlic secondary + +

distribution transformer impedance plus the impedance of (10)

the line connecting the transformer to the distribution The parameters and structurc of the combined compensa-
panel, was equal to 0.032 + jO.I169Q, with the XIR ratio tor are given in Fig. 23. Examining the performance of the
3.65. For harmonic frequencies 11, where n E { I , 3, 5 ) and combined compensator leads to a great reduction in the
using the nonlinear susceptance calculation proccdure system distortion levels. The TH DT was reduced to 2.35'%1
described in Section 5.1, it was found that the equivalcnt and the THDV was reduced to 3.87'%1,which complies with
load susceptance seen at tlic PCC wcre cqLd to BI = 0.028, thc recommended standards. The supply currcnt and the
378 IEE I'wr -Gnier. l'hisiii. Divirih., I ' d 147, Mo 6. No~wmher20110
PCC voltage after inserting the compensator arc given in the available ratings in the market did not match the values
Figs. 24 and 25, respectively. used in the simulator.
For the DBR load the THDI was improved Cram 58.9 to
2.363mH 0.0965mH 0.199mH
23.5%1and the THDV was improved from 14.35 to 5.35%.
Figs. 26 and 27 show the current and voltage waveforms
after inserting the filter as obtained experinientally. For the
Z,M 2067.0pF 4557.0pF 552.0pF DBR and CFL load combination, the stipply current T H D
___c

U,:377rad/s was improved from 24.9 to 15.3%,and the voltage THD

was improved from 13.0 to IO. I%L Figs. 28 and 29 show
the currcnt and voltage bclore and after inserting thc com-
pensator.
4OOr.
300 -
20r .
200
?
; 100-
E
L 0-
Y

-a
h -100-
4 -200-
-300 -
-400-
0 0.02 0.04 0.06 0.08

2OOr
150

100
.& 50
0
z o
D
-50
-o
0

-100

-150
-2001 I
0 0.02 0.04 0.06
time,s

It is important to notice that the triinsformer internal

impcdamcc and all other impedances appearing bcforc thc
35

25
m
PCC from the supply side will not affect the rcsultant sys-
tem stisccptancc seen at thc PCC. <. 15-
I
C
5.5 Experimental verification E 5-
3
To support the computer simulation results, laboratory Y

w -5-
experiments havc bccn conducted. These tests are per-
0
L
formed on two loads. The first is the DBR load and the a -15-
second is a combination of DBR and C F L loads. It was
observed that the experimental results are in agreement -25 -
with the simulation results. Only the improvement in the -5
3-
load voltage and current waveforms and thc rcduction in 0 0.02 0.04 0.06 0.08 0.10
the current THDT and the voltagc THDV arc not the time,s
same. This is attributed to the fact that the components
that arc tiscd in the laboratory do not havc exactly the
same values as those uscd in thc simulator. This is because
/ / i l l lro,-Cwicw ?rir,inn. />i.sird., Vol. 147, N o . 6 , N O I W I ~ W2001~
I
 load voltagc and current. Thc design of these compensators
is highly dependent on the accurate calculation of the non-
150 linear load susccptances at different harmonics.
100
1 7 References
x
0
50
2z 1 EMANUEL, A.E., GENTILE, T.J., IILEGGI. D.J., GULACIHEN-
? O SKI, EM., riiid ROOT, C.E.: Thc effect of modcrn coiiipact fliiorcs-
cent lights 011 vollagc distortioii, IEEE T w a . Puiver Dclis., 1993,
-50 PWRD-8, (3), pp. 1451Ll459
c
a 2 DWYER, R., KHAN, A.K., MCCLUSKEY, ILK.. atid SUNG, R.:
-100 Evaluation of harmonic iinpxts from conpicl Iluorcscent liglils on
POMW YJrSt., 1995, 10, (4). pp. 1772-
-150
, H., LONG, W., DREIFUERST, G.,
ive fillers Ibr AC haiinonic supprcs-
iter Powcr Mceting, New York, 1980,
pp. 1-6
4 SI-IEIHERD, W., aiid ZAI<lI<HANI, P.: Suggested dclinitiotis of
rcactive power for non-sinusoidal syslems, PRJ?. IEE, 1972, 119, (9),
pp. 1361~I362
5 KUSTERS, N.L., kind MOOICE, W.J.M.: On tlic delinition of reiic-
live power under nonsiiiusoidal conditions, I i L W T~NII,?. Power Appl.
S v s t . , 19x0, PAS-99, pp. 1845 1x54
6 Conclusions 6 IIARTANA, P K , and RICHARDS, G.G.: Coiiipal.ing capacitive
atid LC compensators for power factor correction and voltage har-
This paper discusses two issues: the first is the altcnuation iiiotiic reduction, Elec!r. Poiiw Sysl. Re,?.,1989, 17, pp. 51-64
effect that refers to the interactions between the load volt- iiiimizatioii of I-cactive power under ~ionsiiiu-
~ k l i i Sh S ~ I . k i I ? I . MHI,S., 1987, IM-36, (I),pp,
age and currcnt harmonics due to the presence of thc sup-
ply interim1 impcdance; the second is a reactive one-port 8 CZARNECKI, L.S.: Scattered and reactive currcnt, voltage and
compensator design to decreec the voltagc and current powcr in circuits with iioiisiiisoidal wavcfoiins and their coiiipciisri-
lion, 1EEE Trim. ln,sIriitii, Meii.s., 1991, 40, (3), pp. 563-867
harmonics injected into thc distribulion system by difrereat 9 SHARON, D.: Powcr lact efinitions and powcr lr>insferqiiiility i n
types of nonlinear loads. An iterative approach is dcvel- nonsinusoidal siluation, li niii1s. I~l.struii7. M e m 1996, 45, (3),
opcd to account for thc interaction effcct. A new techniquc pp. 728~733
10 GIROIS, A.A., QUAINTANCE. W.II., QIU,.I:, and MAKRAM,
to calculate thc nonlinear load susccptance at diffcrcnt har- EB.: A tiinc d o i n m lhrcc phase power system iiupcdaiicc modeling
monic frequcncies is discusscd. The efficiency of a rcactive approach for hamionic filicr analysis, I i X E Trms fmw Deliv.,
one-port compciisation reducing thc voltage and currcnt 1993, PWRpS, (2), ipp. 504 510
1 I DOMMCL, 13.W.: Elcclromagnctic trailsictit program EMTI refer-
harmonics is investigated. From the computer simulations ence maiiual and lhcol-y book. Bonncvillc power adniinistratv~n,
and the physical cxperiments, the following conclusions Dcpii-lniciil of Electrical Engineering, University of Brilisli Colombia,
have bccn reached: Augusl I986
12 DE OLIVEIRA, A., DE OLIVEIRA, I . C . , RESENDE, J.W., and
1. The interactions between the load voltage and currcnt MISKULIN, M.S.: Practia rmichcs for AC sysleiii hariiionic
h a ~ n i c s cannot bc ncglected in any practical systeni. impedwncc tncasurcinents, Trmi.?. Ioiier IIdii,., 1991, PWRII-
6,(4), p p 1721-1726
Neglecting these interactions could result in wrong cstima- 13 EMANUEL, A.E.: Suggested dcfinilion or rcaclive power in iionsiiiii-
tion for both load voltage and current harmonics. soidal systems, Prnc. IEE, 1974, 121, (7),pp. 705-706
14 TUTTLP, D.F.: Nclwork synthesis (McGraw-Hill, Ncw York, 1969)
2. The reactance one-port compensator can be imple- 15 Rccomiiicnded practices and scquireiiicnts for Irarmonic conirol iii
mented successfully to decrease thc nct harmonics of both elcctdcal power systems. ICEE Std. 519-1993. New York, 1984

380