72 Recent Patents on Electrical Engineering 2012, 5, 72-80

Single-Tuned Passive Harmonic Filters Design with Uncertain Source and

Load Characteristics

Hassan A. M. Ali2,*, Ahmed F. Zobaa1 and Essam E. A. El-Zahab2

1
School of Engineering and Design, Brunel University, Uxbridge, UB8 3PH, United Kingdom; 2Faculty of Engineering,
Cairo University, Giza, 12613, Egypt

Received: October 17, 2011; Accepted: November 1, 2011; Revised: November 10, 2011

Abstract: In this patent, the optimal single-tuned passive harmonic filter corresponding to the maximum expected value
of the annual gain due to the power loss reduction is determined. Voltage and load current harmonics, as well as source
and load impedances are represented as randomly time-varying quantities. The distribution characteristics of these random
variables at a supply bus can be determined by sampling measurements performed on the examined electrical plant, as
well as statistical analysis. The major advantage of this method is that it, unlike conventional approaches, guarantees con-
vergence to the optimal solution. The contribution of the developed method is demonstrated in examples taken from exist-
ing publications.
Keywords: Optimization, power systems harmonics, power factor, probability.

1. INTRODUCTION higher than that of a single tuned band-pass filter. A high-

pass filter will allow a percentage of all harmonics above its
Harmonic pollution can be found in all worldwide indus-
notch frequency to pass through. This results in large rating
trial facilities and with the incremental dominance of nonlin-
at fundamental frequency and high losses in the resistor. The
ear loads, degradation of the loads power factor, increase of
filter is commonly applied for higher frequencies. The com-
the transmission line losses, and reduction of the transmis-
posite filter consists of two branches of bandpass filters and
sion network efficiency are all expected, also amplification
a parallel branch of high-pass one for higher frequencies.
of harmonic currents and voltages can have detrimental ef- This configuration is commonly applied to arc furnace loads,
fects on other elements of the system [1].
generally, if significant harmonic currents are generated
Most distribution systems require reactive power com- locally by a nonlinear load, tuned filters may be installed to
pensation to improve the power factor, save demand charges, prevent the currents from being injected into the system.
or to release additional active power from existing equip-
The filter types shown in Fig. (1) are generally adequate
ment or for voltage support, i.e., the reactive power support for industry applications. The other filter types are third-
required to arrest the voltage drop on loss of a plant genera-
order damped filter and the recently introduced C-type pas-
tor. When power capacitors are used for reactive power
sive filter. The characteristics of passive filters are described
compensation, it becomes necessary to turn them into filters
in [2-6].
to escape from the expected harmonic resonance problems
with one of the nonlinear loads. Over the past several years, electric utilities have expe-
rienced and research groups have confirmed the nondeter-
Passive filters have been extensively used to simultane-
ministic nature of harmonics due to continual changes in
ously meet one or more objectives and also meet the re-
system configurations, in linear load demands, and in
quirements of IEEE Std. 519 with respect to total demand
operating modes of nonlinear loads. Taking into account the
distortion at the point of common coupling [1]. Passive fil-
time-varying nature of harmonics, it is useful to introduce
ters have also been extensively used in HVDC systems and
random variables and to apply probabilistic techniques of
arc furnace installations. analysis.
Figure 1 shows the common types of passive filters, their
Time variations of the harmonics and impedances must
configuration, and plots, in which the single tuned filter is
be considered in designing optimal filters [7-10].
most commonly applied; however, one filter may not be
adequate to filter effectively all the troublesome harmonics. Time-varying quantities to be merged with th e previous
Two single tuned filters will have characteristics identical to statement can be either found by parameters estimation [11-
a double bandpass filter. 13], or by direct measurements [14, 15].
The plot of a second order high-pass filter shows that the A microcomputer-based parameter estimator [11] for
minimum impedance of this type of filter in its passband is determining the distribution characteristics of a power
system is discussed in [11]. The use of this parameter
*Address correspondence to this author at the Faculty of Engineering, Cairo estimator is illustrated on an analog of a single-phase feeder.
University, Giza, 12613, Egypt; Tel: +201222708269; It is found that even in the presence of source harmonics and
E-mail: [email protected]

1874-4761/12 $100.00+.00 © 2012 Bentham Science Publishers

Single-Tuned Passive Harmonic Filters Design Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 73

Fig. (1). Common types of passive filters [2].

with a nonlinear load the system parameters can be quently, harmonic analysis of this system requires definitions
determined to a high degree of accuracy. Such measurements and mathematical relationships beyond those derived for
can be used in the design of a fixed components optimal fundamental frequency analysis. The reader should refer to
compensator or permit the dynamic control of an online Ref. [16] for detail discussion.
optimal compensator.
Ref. [17] demonstrates the representation of the power
In this patent, an optimization criterion of single-tuned system loads and extended networks. General loads in a
passive filter to maximize the expected value a customer can transmission or distribution systems are generally expressed
gain annually with the reduction the power losses in the by their active and reactive power respectively, which are
supply network while constraining the power factor (PF), the used to calculate the equivalent impedance for load flow
total voltage harmonic distortion (VTHD), and the values of studies at the fundamental frequency, assuming the system
the compensator which cause resonance. The presented voltage.
method is tested using two case studies taken from previous
In addition a measurable percentage of any general load
publi-cations.
nowadays is electronically controlled and needs to be pro-
The criterion for the design of the single tuned passive perly represented. Electronic loads are harmonic producing
filter is discussed by the penalty function search algorithm.it and consequently do not exhibit a constant resistance and
was chosen because it requires fewer steps and function inductance. Therefore, they cannot be included as part of the
evaluations. Penalty Function methods transform the basic passive component of the equivalent impedance. They
optimization problem into alternative formulation such that should be represented by a harmonic current source at all
numerical solutions are obtained by solving a sequence of frequencies of importance.
unconstrained minimization problems[10].
“The statistical characterization of the input data consists
The guaranteed convergence to the optimal solution is a of checking:
direct result from the improvement in the algorithm in which
(i) which of the components have to be considered as
provisions are made to identify and to avoid compensator
random in nature and which of them can be kept fixed;
values that would create resonance conditions, the problem
formulation in which the effect of the Thevenin impedance (ii) what are the statistical features to characterize the
on the load voltage is included. random nature of the variables identified in (i).
With reference to problem (ii), the random nature of the
2. SYSTEM MODELING AND SIMULATION non-deterministic components is analyzed by distinguishing
In a non-sinusoidal system, the distorted voltages and the cases of linear loads, generators, and converters; only
currents are considered to be periodic waveform, which may then is the probabilistic nature of the system admittance
be represented by Fourier series of harmonic terms. Conse- terms discussed.
74 Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 Ali et al.

Linear Loads
In the case of balanced power systems, the statistical
characterization of linear loads has been fully analyzed [18],
considering both dependent and total impedance between
load demands.

Converters
If the hypothesis of constant direct current is assumed,
then the DC load can be simulated by a direct current
generator, whose statistical characterization can be carried
out by the knowledge acquired of the expected operational
process [18].

Generators
No difference between balanced and unbalanced systems,
because in both cases the input data are the three-phase real
power. The statistical characterization of the three-phase real
powers has been fully analyzed, considering both inde-
pendent and dependent generation systems [19].

System Admittance
The uncertainties of the terms of the system admittance
are linked to the system structures and for an assigned Fig. (2). Configuration of the studied system.
structure, to the values of the electrical parameters of the
equivalent circuits of the power system components (line, To simplify the analysis, only the load model using the
transformers, etc.). the electrical parameter values of respective active and reactive powers at the fundamental
network components at the fundamental frequency could be frequency is considered while sizing the compensators. The
considered deterministic; at the harmonic frequencies the model shown is adequate where total voltage harmonic dis-
only uncertainties, which have to be taken into account, are tortion (VTHD) is less than 10 % [20].
those due to the statistical nature of the linear loads [18].” As shown in Fig. (3),

( )
[17].
VSK = ISK R TK + jX TK + VLK
3. THE SYSTEM UNDER STUDY VLK = (ISK
I LK ) * ZCLK

The configuration of the studied system is given in Fig.
(2).
The Thevenin voltage source representing the utility sup-
ply and the harmonic current source representing the nonlin-
ear load are
v S (t) =
v SK (t) (1)
K

and,
iL (t) =
iLK (t) (2)
K
Fig. (3). Single phase Equivalent circuit for K th harmonics.
where, K is the order of harmonic present. The K th har-
monic Thevenin impedance is By simplification of the above equation and using (3) to
(5), the supply current is obtained.
ZTK = R TK + jX TK (3)
VSK (AR + jBR) + I LK (CR)
and the K th harmonic load impedance and admittance are ISK = (6)
A IK + jA JK
ZLK = R LK + jX LK (4) The load voltage is given by
YLK = G LK
jBLK VSK (CR)
I LK (DR * ER)
VLK = (7)
The K th harmonic compensator impedance is A IK + jA JK
XC where,
ZCK = R + j(KX L
) (5)
K
Single-Tuned Passive Harmonic Filters Design Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 75

AR = R + R LK , In most power systems, one can generally assume that

the resistance does not change significantly when studying
XC the effects of harmonics less than the ninth [22].
BR = X LK + KX L
,
K
CR = R CLK + jX CLK 4. OBJECTIVE FUNCTION AND CONSTRAINS
 X  For randomly time-varying source and load harmonics as
DR = R + j  KX L
C  , well as impedances, the objective function to be maximized
 K
is the expected value of a customer can gain annually to re-
ER = R TLK + jX TLK , duce the power losses in the supply network, E(AG).
E(AG) must be expressed as function of XC and X L and of
XC
A IK = R TLK
(KX L
)(X LK + X TK ) + R(R TK + R LK ), the statistics of random variables. Because distribution sys-
K tem harmonic generators are generally current sources, a
X positive correlation exists between Thevenin harmonic im-
A JK = X TLK + (KX L
C )(R LK + R TK ) + R(X TK + X LK ). pedance components, XTK and RTK, and source harmonic
K
voltage VSK, which is a product of source impedances and
Norton equivalent current sources. Source voltage, Thevenin
Z LK ZCK R + jX CLK resistance and reactance are linearly correlated to each other
ZCLK = = CLK (8)
ZLK + ZCK ZLK + ZCK at each harmonic order K>1 in the most extreme case. Thus,
Z LK ZTK R + jX TLK VSK = G K X TK (14) a
ZTLK = = TLK (9)
ZLK + ZTK ZLK + ZTK nd,
 X  R TK = TK X TK
R CLK = RR LK
X LK  KX L
C  ,
 K where, G K and TK are constants for K>1, and X TK is a ran-
dom variable which linearly with frequency. Then, the ex-
 X  pected value of TL can be written as
X CLK = RX LK + R LK  KX L
C  ,
 K

d X 2 + e K X TK + f K 
R TLK = R TK R LK
X LK X TK ,
2
E(ISK ) = E  K 2TK  (15) b
 c K X TK + b K X TK + a K 
X TLK = R LK X TK + R TK X LK .
y definition,
The uncompensated supply current is given by +
)=  f(X
2 2
E(ISK )ISK d X TK (16)
VSK + I LK (R LK + jX LK ) TK
o
=


(R ) ( )
I (10)
+ R TK + j X LK + X TK
SK
LK where, f (XTK) is the probability density function of X TK. If
The expected value of the annual gain due to the power the Thevenin reactance X TK is assumed to have a uniform
loss reduction is given by distribution function with minimum value
K and maximum
value K, E(ISK
2
) from (16) becomes
AG = C F  E ISK ( )

E ISK  ( )
2 2
o
(11)
  1  K  d X2 + e X + f 
 
K TK K TK K
where, CF is the utility discount factor. In this study, CF = RT, dX (17)
 K
 K K  c K X 2TK + b K X TK + a K
TK
where RT is the fundamental value of the Thevenin resis-
tance. d 1 K  bb K X TK + aa K 
= K
  dX , (18)
The PF at the load is given as c K  K
 K K  c K X TK + b K X TK + a K
TK
2

P
G LK
2
VLK where,
PF = L = K
(12)
VL IS
I
V 2
SK
2
LK
a K = b2K1 + b2K 2 ,
K K
b K = 2a K1b K1 + 2a K 2 b K 2 ,
c K = a 2K1 + a 2K 2 ,
The VTHD at the compensated load terminals is defined  X 
as a K1 = TK R LK
X LK
 KX L
C  ,
 K

V 2
LK  X 
VTHD = K>1
(13) a K 2 = TK X LK + R LK + TK  KX L
C  ,
VL1  K

In the optimization process, the resistance of the compen-  X 

b K1 =
X LK  KX L
C  ,
sator reactor has been neglected due to its small value with  K
respect to its fundamental reactance [21].
76 Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 Ali et al.

 X   X 
2
 X 
b K 2 = R LK  KX L
C  , A1  KX L
C  + A 2  KX L
C  + A 3 = 0 (21)
 K  K  K
2
 X 
( )
where:
d K = G 2K R 2LK + X 2LK + G 2K  KX L
C  A1 = XTK + X LK
 K
A2 = X LK
2
+ RLK
2
+ 2 XTK X LK A3 = XTK (R + RLK )2 + R 2 X LK + XTK X LK
2

 X 
+ 2G X LK  KX L
C  ,
2 Solving (21) for finding KXL, XC/K:
 K
K

 X C 
A 2 ± A 2
4A1A 3
2
2
 X   L K
KX
= (22)


e K = 2G K I LK R LK  KX L
C  ,   2A1
 K
Hence using only the set values for shunt capacitors [24],
2
 X  one can obtain values for the inductive reactance since (21)
( 
)
f K = I 2LK R 2LK + X 2LK  KX L
C  ,
K
will then become a one variable equation in XL only.

d K bK 5. FORMULATION OF THE PROBLEM

bb K = e K
,
cK The AG, the PF, and the VTHD can be expressed as
dKa K functions of XC and XL, using (11), (12) and (13).
aa K = f K
. Each value of the reactive power ratings QCi of the par-
cK
ticular voltage [24] is used to calculate the corresponding
2
Integration of E ( I SK ) for K>1 can be calculated [23]. value of XCi. This value is then substituted into the objective
o2
One can calculate the integration of E ( I SK ) for K>1 as E function to become one variable equation in XL, which can
2
( I SK ) knowing that: be solved by using the Penalty Function method.

(
a K = R 2LK + X 2LK , ) After formulating the objective function and the con-
straints, the problem addressed in this study becomes:
b K = 2TK R LK + 2X LK , Maximize E (AG) (XCi, X L) (23)
c K = TK2 + 1, Subject to: 90%
PF (XCi, X L)
100%
dK = G , 2
K VTHD (XCi, X L)
5%
e K = 2G K I LK R LK , XCi, X L is not part of solution of (21).
fK = I 2
LK (R 2
LK
+X 2
LK ). This problem can be rewritten in the form
then, the expected value of AG can be expressed from (11) Maximize E (AG) (XCi, X L) (24)
as Subject to: g1 (XCi, X L) = 0.9 – PF (XCi, XL)
E(AG ) =
 E ISK ( ) - E (I ) R
2 2
o
(19) g2 (XCi, XL) = PF (XCi, XL) – 1.00

K 
SK
 T
g3 (XCi, XL) = 0.0 – VTHD (XCi, XL))
It was pointed out that the passive filter on a network g4 (XCi, XL) = VTHD (XCi, X L)) – 0.05
could create undesirable resonant conditions. System reso-
nant conditions are the most important factors affecting sys- (XCi, XL) is not part of solution of (21)
tem harmonic levels. Parallel resonance is high impedance to Note that, in order to have the algorithm to guarantee
the flow of harmonic current, while series resonance is low convergence, the objective function has to be a unimodal
impedance to the flow of harmonic current.
At resonance, large oscillating harmonic currents can function [25, 26]. Due to the resonant conditions, there might
exist and overload circuit elements causing failure on opera- be local maximums to which the solution will converge. To
tion of protective equipment. To impose this constraint on avoid this problem, the pre-calculated compensator values
the formulation of the optimization problem, the compensa- for series resonance will be used to subdivide the entire
tor values resulting in resonant phenomena have to be identi- search region into numeral small regions. Within these re-
fied and omitted from the solution process. gions, the local maximums are identified; hence, the global
maximum is voted.
As shown in Fig. (3), the driving point impedance can be
determined by the following equation.
6. THE SEARCH ALGORITHM
Z = (R TK + jX TK ) + (R CLK + jX CLK ) (20)
The suggested search algorithm is discussed below:
If a resonant circuit occurs at a given harmonic order (K),
this impedance will behave like a pure resistor. Setting the Step 1: Choose the first value of the standard manufactured
imaginary part of this impedance equal to zero, series reso- reactive power rating of capacitors in Kvar [24].
nance, and results in the following quadratic equation: QCi = {QC1, QC2… QCn} (25)
Single-Tuned Passive Harmonic Filters Design Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 77

where, n is the number of discrete values available for the Table I. System Parameters and Source Harmonics
particular voltage rating used and (i) has a starting value of
1.
Parameters & Harmonics Case 1 Case 2
Step 2: Using only the selected value of Qci, calculate Xci
from the following equation RT1 [
] 0.01154 0.01154
V2
X Ci = S1 (26) XT1 [
] 0.1154 0.1154
QCi
E(XT5) [
] 0.5570 0.5570
Step 3: Substitute the value of XCi into the objective function
E(XT7) [
] 0.8078 0.8078
to become one variable problem in X L.
E(XT11) [
] 1.2694 1.2694
Step 4: Using Penalty Function algorithm [26] to solve (24)
for optimal XL. E(XT13) [
] 1.5002 1.5002
Step 5: If i=n stop otherwise replace i by (i+1) and go to step RL1 [
] 1.7421 1.7421
1.
XL1 [
] 1.6960 1.6960
Step 6: After stopping, scan through to get the global maxi-
mum. VS1[V] 2400 2400

The algorithm is discussed below: E(VS5 ) [%V S1] 5.00 1.00

Step 1: Let penalty parameter (μ > 1), scalar ( > 0), and E(VS7 ) [%V S1] 3.00 7.00
(J=1).
E(VS11) [%V S1] 2.00 2.00
Step 2: Starting with X L (J), which is a solution of (21), to
solve the following problem to get an optimal solution XL E(VS13) [%V S1] 1.00 1.00
(J+1)
. IL5 (A) 49.39 49.39

( )( )
Max. E TL X Ci , X L +
μ m (max[0, g m (X Ci , X L )]) IL7 (A) 29.64 29.64
m
IL11 (A) 19.67 19.67
Step3: If μ m (J ) (max[0, g m (X Ci , X L (J+1) )])

stop: otherwise
let
m (J+1) =
m (J) IL13 (A) 9.88 9.88

Step 4: Replace (J) by (J+1) and go to step 2.

Table III shows a summary of the results for the cases
Ref. [27] shows that the starting penalty parameter value under study. The presented method results in lower supply
has no measurable effect on the efficiency of the algorithm. current lower transmission loss, higher transmission effi-
It is always started with (
(0) = 10). However, the way ciency, and higher power factor than the uncompensated
should be changed at each iteration depends on the value of system, Table II. So, the main advantage of the presented
the parameter (). Values of () are in the range (0.1 - 0.5) in method consists in less harmonic power in secondary cables,
order to work well for most problems. lines and switchgear owned by the end-user. Also, Compari-
Many stopping criteria can be used. Here, the algorithm son of the results, it is shown that an additional harmonic
will stop when a feasible point will be reached or when the contents result in lower power factor. This is caused by the
relative change in the objective function is small (less than  increase in compensated line current due to the additional
= 10-6). harmonics.
Table IV presents the harmonic distortion levels for the
7. EXAMPLES AND SIMULATED RESULTS cases under study.
Two cases of an industrial plant were simulated using the To confirm the reduction of the harmonic distortions
optimization method. The numerical data were primarily within the standard limits [1], two factors can be modified:
taken from an example in Ref. [28] where the inductive the level of harmonic currents, and transformer impedance.
three-phase load is 5100 kW with a displacement factor of Using phase-shifting techniques may reduce the level of
0.717. The 60-cycle supply bus voltage is 4.16 kV (line-to- harmonic currents, and low impedance plays a crucial role in
line). The resistance and fundamental reactance values are reducing voltage distortion. Now, low-impedance phase-
given in Table I. The short circuit capacity is 150 MVA. shifting transformers [29] have been designed. They allow
the treatment of harmonic currents while providing a path of
Table II shows the simulated results of the compensated low impedance. The quality and reliability of the electrical
system for the different cases under study including: (PF), system can thus be considerably improved through the use of
(IS), (), (TL), (VTHD) and (ITHD). The results of the cases a single piece of equipment.
under the study using the presented method are shown in
Table II.
78 Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 Ali et al.

Table II. Simulated Results of the Uncompensated System

Case PF (%) IS (A)  (%) TL (kW) VTHD (%) ITHD (%)

1 71.51 953.64 99.34 10.49 2.19 6.00

2 71.51 953.70 99.34 10.50 1.83 6.10

Table III. Simulated Results of the Presented Method

Case XC(
) XL(
) PF (%) IS(A) (%) TL(kW)

1 3.84 0.2728 99.78 713.88 99.65 5.88

2 3.97 0.2744 99.73 710.11 99.65 5.82

If the resultant values are greater than standard limits, it

8. CAPACITOR BANK SPECIFICATIONS is a good idea [24] to use capacitors with a higher voltage
IEEE Std. 18-2002 [24] specifies the following continu- rating.
ous capacitor ratings: Table VI shows the calculated capacitor limits compared
• 135 percent of rated rms current (IC) including funda- with the standard limits for case 2.
mental and harmonic current, Comparison of the calculated and standard limits, Table
VI, shows that all values lie within the standard limits. It is
IC =
I 2
CK
(27) usually a good idea to use capacitors with a higher voltage
K
rating in some cases because of the voltage rise across the
where the capacitor current at harmonic number K, ICK, is reactor at the fundamental frequency and due to the har-
given monic loading. Also, we can take into consideration VC, IC,
V (R + jX LK )
I LK (R TLK + jX TLK ) and QC as constraints in the optimization method to confirm
I CK = SK LK that the results will be lie within the standard limits.
A IK + jA JK
• 110 percent of rated rms voltage (VC) including harmon- Finally, Source and load harmonics as well as imped-
ics but excluding transients, ances are generally time-varying at random. An optimal pas-
sive filter designed for the mean values of those random
VC =
V 2
CK
(28) w variables neglecting the uncertainty of the source harmonics
K and load characteristics does not render optimum average
here the capacitor voltage at harmonic number K, VCK, is power factor and may even be overloaded when imple-
given mented to the time-varying case. Therefore, time variations
X of the harmonics and impedances must be considered in de-
VCK = I CK CK signing an optimal fixed passive filter.
K
• 135 percent of nameplate KVAR (QC), 9. CONCLUSIONS
QC = VC
I C (29) In this patent, a method is presented for finding the val-
Table V presents the harmonic levels of the capacitor ues of the parameters of the passive filter to maximize the
current and voltage for case 2. expected value of the annual discount gain due to power
losses reduction where it is desired to maintain a given
The following is the illustration of the calculation of the power factor with uncertain source harmonics and load char-
capacitor duty of the compensator for case 2: acteristics.
Capacitor bank rating = 4350.00 KVAR Two cases are tested, and the general performance of the
Capacitor fundamental current rating = 603.72 A proposed method is satisfactory, providing reduction in the
transmission loss and improvement of in the power factor
Capacitor fundamental voltage rating = 4160.00 V
compared with other published techniques.
Capacitor current distortion = 12.78 %
Root –mean-square capacitor current = 649.80 A CURRENT & FUTURE DEVELOPMENTS

Capacitor voltage distortion = 1.49 % The recent trend of harmonic system standards, in fact, is
in deeply considering the random nature of voltage and cur-
Root –mean-square capacitor voltage = 4440.02 V rent harmonics. The IEC publication 1000-3-6 uses probabil-
istic approaches both in comparing the actual current and
Single-Tuned Passive Harmonic Filters Design Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 79

Table IV. Harmonic Distortion Levels for the Different Cases Under Study

Case 1 2

VTHD (%) 3.65 4.70

ITHD (%) 16.97 13.28

Table V. Harmonics Levels For Capacitor Used: Case 2

K VC (V) VC (%) IC (A) IC (%)

1 4439.30 106.71 645.20 106.87

5 41.70 1.00 30.30 5.02

7 69.00 1.66 70.20 11.63

11 6.00 0.15 9.70 1.60

13 2.00 0.05 3.70 0.62

Table VI. Capacitor Limits (IEEE Std 18-2002) for Case 2

Item Calculated (%) Limit (%) Exceeds limit

rms voltage 106.73 110.00 No

peak voltage 109.57 120.00 No

rms current 107.63 135.00 No

KVAR 114.88 135.00 No

voltage harmonic levels with the planning levels and in as- Egypt for his technical support during the preparation of this
sessing the emission limits for distorting loads; the IEEE Std. work.
519 only slightly addresses the probabilistic application of
harmonic distortion limits, but various efforts are in progress NOMENCLATURE
to more extensively include the probabilistic aspects.
RLK, XLK = Load resistance and reactance in ohm at
For nonlinear loads it is necessary to use passive filter. harmonic number K
Such compensators have dual purposes. The first is that it
acts as a compensator to improve the power factor of the GLK, BLK = Load conductance and susceptance in
nonlinear loads. Secondly, it acts as a filter of the harmonic mho at harmonic number K
load currents thus preventing the proliferation of the network RTK, X TK = Transmission system resistance and re-
with these currents. actance in ohm at harmonic number K
The selection of the optimal passive filter is a multi- XL, XC = Magnitude of the fundamental inductive
objective optimization problem. One of the known ap- and capacitive reactance in ohm of the
proaches for solving a multi-objective optimization problem compensator
is to transform it into a single objective function. For this
reason a solution algorithm has to be developed taking into R = Resistance of the compensator reactor in
ohm
consideration the discrete nature of the standard values.
ISK = Average value of supply current in am-
CONFLICT OF INTEREST pere at harmonic number K
The authors declare no conflict of interest. IS = RMS value of supply current in ampere
ILK = Average value of load harmonic current
ACKNOWLEDGEMENTS in ampere
The authors gratefully acknowledge and thank Mr. S. H. PL = Load active power per phase in watt
E. Abdel Aleem, 15th May Higher Institute of Engineering,
PS = Supply active power per phase in watt
80 Recent Patents on Electrical Engineering, 2012, Vol. 5, No. 1 Ali et al.

VLK = Average value of load voltage in volt at [11] P. Klinkhachorn, R. K. Hartana, B. Huner, B., “Microcomputer
harmonic number K based estimator for power system parameters in the presence of
harmonics,” IEEE Southeastcon '88, 11-13 April 1988, pp.529 –
VSK = Average value of supply voltage in volt 532.
at harmonic number K [12] M. El-Amin, I. Arafah, “Artificial neural networks for power sys-
tems harmonic estimation,” 8th International Conference on Har-
VL = RMS value of load voltage (line to neu- monics And Quality of Power, 14-16 Oct. 1998, vol.2, pp. 999 –
tral) in volt 1009.
[13] T. Lobos, T. Kozina, H. J. Koglin, “Power system harmonics esti-
VS = RMS value of supply voltage (line to mation using linear least squares method and SVD,” IEE Proceed-
neutral) in volt ings -Generation, Transmission and Distribution, vol.148 , issue
6, Nov. 2001, pp.567 – 572.
VTHD = Voltage total harmonic distortion in per- [14] L. Siqueira, A. M. Carneiro, M. De Oliveira, “Assessment of har-
cent monic distortions on power and energy measurements,” Ninth In-
ternational Conference on Harmonics and Quality of Power, 1-4
ITHD = Current total harmonic distortion in per- Oct. 2000, vol.1, pp.105 – 109.
cent [15] R. Yacamini, “Power system harmonics. II. Measurements and
calculations,” Power Engineering Journal, vol.9, issue 1, Feb. 1995,
PF = Load power factor in percent pp.51 – 56.
[16] IEEE Working Group on Nonsinusoidal Situations, “Practical
definitions for power in systems with nonsinusoidal waveforms and
DISCLOSURES unbalanced loads: A discussion,” IEEE Trans. on Power Delivery,
vol.11, no.1, 1996, pp.79-87.
The part of information included in this article has been [17] P. Caramia, G. Carpinelli, F. Rossi, and P. Verde, “Probabilistic
previously published in Journal of Electric Power Compo- iterative harmonic analysis of power systems,” IEE Proceedings on
nents and Systems Volume 37, Issue 3, 2009. Generation, Transmission, and Distribution, vol.141, no.4, July
1994, pp.329-338.
[18] M. Schilling, A. M. Leite da Silva, R. Billington, and M. A. El-
REFERENCES Kady, “ Bibliography on power system probabilistic analysis,”
[1] IEEE Recommended Practices and Requirements for Harmonic IEEE Trans. On Power Systems, vol.5, no.1, 1990, pp.1-11.
Control in Electrical Power Systems, IEEE Std. 519-1992, 1992. [19] J. Deaver, W. O. Radtke, and W. H. Berkman, “Method and system
[2] J. C. Das, “Passive Filters-Potentialities and Limitations,” IEEE for providing power factor correction in a power distribution sys-
Transactions on Industry Applications, vol.40, no.1, Jan/Feb 2004, tem,” US7804280B2, September 28, 2010.
pp.232-241. [20] M. F. McGranaghan, R. C. Dugan, J. A. King, and W. T. Jewell, “
[3] F. D. Graza, “Mitigation of Harmonic Currents and Conservation Distribution feeder harmonic study methodology,” IEEE Trans. on
of Power in Non-Linear Load Systems,” US20110121775, May 26, Power Apparatus and Systems, vol.103, no.12, Dec 1984, pp.3663-
2011. 3671.
[4] D. A. Gonzalez and J. C. McCall, “Design of filters to reduce har- [21] G. G. Richards, O. T. Tan, P. Klinkhachorn and N. I. Santoso,
monic distortion in industrial power systems,” IEEE Trans. on In- “Cost-Constrained power factor optimization with source harmon-
dustry Applications., vol. IA-23, pp. 504–512, May/June 1987. ics using LC compensators,” IEEE Transasctions on Industrial
[5] J. K. Phipps, “A transfer function approach to harmonic filter de- Electronics, vol.34, no. 2, May 1987, pp.267-270.
sign,” IEEE Industry Applications Magazine, vol. 3, pp. 68–82, [22] R. C. Dugan, Surya Santoso, M. McGranaghan, M. F. McGrana-
Mar./Apr. 1997. ghan, H. Beaty and H. W. Beaty, Electrical Power Systems Quality,
[6] Ludbrook, “Harmonic filters for notch reduction,” IEEE Trans. on Second Edition, Mc-Graw Hill, USA, 2002.
Industry Applications, vol. 24, pp. 947–954, Sept./Oct. 1988. [23] S. Gradshteyn and L. M. Ryzhik, Tables of Integrals, Series and
[7] A.F. Zobaa, and W. J. Lee, “The optimal LC compensator corre- Products, Academic Press, Cambridge University Press, California,
sponding to maximum annual reduction in the source losses” IEEE USA, 1980.
/ IAS 41st Annual Meeting, IAS 2006, Tampa, Florida, October 8 - [24] IEEE standards for shunt power capacitors, IEEE Std. 18-2002,
12, 2006, vol.3, pp.1432-1437. 2002.
[8] A.F. Zobaa, “Voltage harmonic reduction for randomly time- [25] Ross, Applied Optimization: Formulations and algorithms for engi-
varying source characteristics and voltage harmonics,” IEEE neering systems, First Edition, Cambridge University Press, New
Transactions on Power Delivery, vol.21, no.2, April 2006, pp.816- York, USA, 2006.
822. [26] G. H. Hadley, Nonlinear Proframming and Dynamic Programming,
[9] O. M. Aloquili, A. F. Zobaa, and H. H. Zeineldin “Power factor Addison-Wesley Publishing Company Inc., Reading, Mass., 1964.
correction based on transmission loss minimization with uncertain [27] M. Ben-Daya and K. S. Al-Sultan, “ A new penalty function algo-
source harmonics and load characteristics”, Electric Power Com- rithm for convex quadratic programming,” European Journal of
ponents & Systems Journal, vol.37, no.3, March 2009, pp. 331-346. Operational Research, vol.101, 1997, pp.155-163
[10] Cavallini, G. Mazzanti, G. C. Montanari, C. Romagnoli, “Design of [28] R. K. Hartana and G. G. Richards, “Comparing capacitive and LC
shunt capacitor circuits for power factor compensation in electrical compensators for power factor correction and voltage harmonic re-
systems supplying nonlinear loads: a probabilistic approach,” IEEE duction,” Electric Power Systems Research, vol.17, 1989, pp.57-
Transactions on Industry Applications, vol.34, issue 4, July-Aug. 64.
1998, pp.675 – 681. [29] IEEE Guide for the Application, Specification, and Testing of
Phase-Shifting transformers, IEEE Std C57.135-2001, 2002.