Sie sind auf Seite 1von 8

IEEE TRANSACTIONS ON INDLUSTRY APPLICATIONS, VOL. IA-23. NO. 3.

MAY/JUNE 1987

504

Design of

Filters to Reduce Harmonic Distortion


Industrial Power Systems

DAMIAN A. GONZALEZ.

MEMBER, IEEE, AND

Abstract-Shunt filters are effective in minimizing voltage distortion


caused by nonlinear loads in industrial power systems. Different alternatives of filter design should be considered before making the final decision
on filter configuration. Among the criteria used for performance
evaluation are losses, current, and voltage ratings of each of the filter
components, and the effect of filter and system contingency conditions.
The design and performance of single-tuned and high-pass filters and the
methodology used for the analysis will be discussed.

in

JOHN C. MCCALL, MENMBER, IEEE

x
(OHMS l0 2)
2.00 1

INTRODUCTION
ARMONICS have been shown to have deleterious effects
on equipment including transformers, rotating machines,
switchgear, capacitor banks, fuses, and protective relays.
Transformers, motors, and switchgear may experience increased losses and excessive heating. Induction motors may
refuse to start (cogging) or may run at subsynchronous speeds.
Circuit breakers may fail to interrupt currents due to improper
-2.00 J
operation of blowout coils. Capacitors may prematurely fail
from increased dielectric stress and heating. The time-current Fig. 1. Typical R-X locus plot of driving-point impedance of arc furnace
installation.
characteristics of fuses can be altered, and protective relays
may experience erratic behavior [11.
Electronic power converters have become one of the major less damping in the circuit, and often have multiple voltage
sources of harmonics in the industrial environment. These levels with capacitors at each level. Fig. I shows the
converters operate at a low power factor, usually making it impedance characteristics of a typical steel mill [3].
The accurate representation of the supply system impedance
necessary to apply shunt capacitors for reactive compensation.
a strong influence on the results of a harmonic analysis. A
has
The general procedure in analyzing any harmonic problem
polar
plot of the supply system impedance, showing its value
is as follows.
for each harmonic frequency, is ideal. It will aid in the
* Identify the worst harmonic condition.
selection of a filter that will not adversely interact with the
* Design a suppression scheme.
system. When the supply system consists mainly of transform* Recheck for other conditions.
ers and/or short overhead lines, the system short-circuit
Some of the methods used to solve harmonic problems are impedance can be used.
The X/R ratio of the supply system impedance is usually
phase multiplication, harmonic injection, shunt filters, and
It should be noted that the resistance portion of this
3:6.
control system operation conditions [2]. Shunt filters have the
is also frequency-dependent. If the impedance is
impedance
advantage of providing the needed reactive compensation
transformers, a good approximation is to assume
dominated
by
required by the harmonic producing devices. This paper
X/R ratio remains constant as frethe
transformer's
that
describes the design and performance of shunt filters to reduce
increases
quency
[4].
harmonic distortion in industrial power systems.
Due to the complexity and size of the problem, computers
Industrial power systems are frequently radial, similar to
needed for most harmonic analyses. A good computer
are
distribution feeders; however, they are more compact, have
program for harmonic analysis should be able to model several
Paper ICPSD 85-63, approved by the Power Systems Engineering types of nonlinear loads and be able to perform multipleCommittee of the IEEE Industry Applications Society for presentation at the source solutions and frequency scans [5].
1985 Industry Applications Society Annual Meeting, Toronto. ON, Canada.
October 6-1 I. Manuscript released for publication December 3. 1986.
D. A. Gonzalez is with Systems Engineering. McGraw-Edison Power
Systems Division of COOPER Industries, P. 0. Box 440. Canonsburg, PA
15317.
J. C. McCall is with Capacitor Products. McGraw-Edison Power Systems
Division of COOPER Industries, 140 Kingston Road, Greenwood, SC 29646.
IEEE Log Number 8613397.

VARIABLE-SPEED DRIVE HARMONIC CURRENTS


Six-pulse rectifier circuits are nearly approximated by a
rectangular wave with a duration of 1200 on both the positive
and negative half-cycles. The harmonic currents in per unit of
the fundamental are the inverse of the harmonic order for this

0093-9994/87/0500-0504$01.00 1987 IEEE

505

GONZALEZ AND MCCALL: FILTERS TO REDUCE HARMONIC DISTORTION

25.

olphoa 275

EdI 102 %

I
20w

.4

>

15

Fig. 2. Six-pulse rectifier circuit current.


w

10

461

3rd

4th

w
zZ

20

cr

6- PULSE

10 12-PULSE

15

2nd

U-

10 _

5th

6th

Bth

7th

9th

HARMONIC

Fig. 4. Harmonic content of arc furnace voltages in percent of fundamental.


5

I[
2nd

3rd

4th

5th

7th

11th

13th 17th 19th 23rd 25th

HARMONIC

Fig. 3. Suggested harmonic content of six- and twelve-pulse drive currents


for harmonic analysis.

waveform. However, the actual rectifier harmonic current is


be seen in Fig. 2. This
waveform was obtained from an actual six-pulse thyristorcontrolled drive.
During the actual operation of six- and twelve-pulse drives,
thyristor firing times may vary slightly from the desired 60
delay. This and system unbalance result in the existence of
some noncharacteristic harmonics, as well as a variation in the
magnitude of the characteristic harmonics. The bar chart in
Fig. 3 illustrates typical harmonic currents, which may be
used in harmonic analyses to account for these effects in
thyristor-controlled drives.
not a rectangular wave, as can

Fig.

5.

Typical voltage and current waveshapes of arc furnace.

D)
A)
B)
C)
A) Single-tuned filter
i) First order high-pass filter
C) Second order high - pass f i I ter
D) Third order high-pass filter

Fig.

6.

Shunt filters.

ARC FURNACE HARMONIC VOLTAGES


Measurements made of arc furnace voltage have indicated a full line voltage. In contrast, shunt filters carry only a fraction
highly variable harmonic output, e.g., 8 percent fifth-har- of the current that a series filter must carry. Given the higher
monic voltage at the beginning of meltdown, 5 percent at the cost of a series filter, and the fact that shunt filters may supply
end of meltdQwn, and 2.5 percent of the fundamental during reactive power at the fundamental frequency, the most
refining. Under unbalanced conditions of electrode arcing, the practical approach usually is to use shunt filters.
The most common shunt filters are the single tuned filter
third harmonic may occur in appreciable amounts. In addition,
and
the high-pass filter. These two filter types are the simplest
the fifth and seventh harmonics may increase significantly
to
and the least expensive to implement. The general
design
during unbalanced arcing conditions.
of
shunt
filters is shown in Fig. 6.
layout
The bar chart in Fig. 4 shows an example of harmonic
voltage levels used in an arc furnace harmonic analysis. These Single-Tuned Filter Design
values are based in part on field-measured waveforms and on
The single-tuned filter, or notch filter, is probably the most
estimates of worst-case harmonic distortion. Fig. 5 shows
common
shunt filter in use. The general layout of the filter is
example voltage and current waveforms at an arc furnace.
as shown in Fig. 6. A typical frequency response plot is as
shown in Fig. 7.
FILTER DESIGN
Examination of filter response reveals the following characUndesired harmonic currents may be prevented from
teristics.
flowing into the power system by one of two methods.
* It acts as a very low impedance at the frequency for which
* Use of a high series impedance to block them.
it is tuned and, as such, effectively shunts most harmonic
* Diverting them by means of a low impedance shunt path.
line quantities at that frequency.
Series filters must carry full load current and be insulated for
* When the source impedance is inductive, there is a

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-23, NO. 3, MAY/JUNE 1987

506
IMPEDANCE MAGNITUDE
(POWER OF 10)

0
LL

o
Z :0
(D0

-wW
E_
FREQUENCY

(a)

FREQUENCY

IMPEDANCE MAGNITUDE
( POWER OF 10 )

Fig. 8. Scan plots for 4.7th harmonic notch filter with various values of Q.

as
WO

QWWI -W2
FREQUENCY

(b)
Fig. 7. Typical frequency response of notch filter. (a) Filter alone. (b)
Filter and system.

where wo is the tuned frequency, and wI and w2 are the


dB points. This simplifies to

dLIC XLr XCr

(7)
-3

(8)

R R
R
resonance peak which always occurs at a frequency lower
than the frequency for which the filter is tuned.
where the reactances at the resonance frequency are given by
* There is a sharp increase in impedance below the tuned XLr and
Xcr.
frequency due to the proximity of the resonant frequency.
The following points concerning the Q of single-tuned
* The impedance rises with frequency for frequencies filters are of interest.
above that at which the filter is tuned.
* The value of Q is seldom considered in regards to
Design Equations: The impedance of the filter branch is
filtering action. This is due to the fact that the values of
given by
R, which must be used to significantly alter the filter
response, usually result in a significant increase in losses
(1)
Z= R +j[wL -l /(wC)].
within the filter.
* The higher the value of Q, the more pronounced is the
Resonance occurs when the imaginary part is equal to zero, at
valley at the tuned frequency. Fig. 8 consists of a number
which time the impedance is limited by the value of R. The
of overlaid plots for a 4.7th harmonic single-tuned filter
frequency for which the filter is tuned is given by the value of
with various values of Q.
w that results in series resonance. This frequency is given as
* Typically the value of R consists only of the resistance in
the inductor. In this case the Q of the filter is equal to R
I
times the X/R ratio of the tuning reactor. This usually
(2)
1/2
results in a very large value of Q and a very sharp
filtering action.
Defining the harmonic number n as the frequency of the
* The response for values of Q above 25 are essentially
harmonic divided by the fundamental system frequency allows
indistinguishable from the Q = 100 plot, except for the
the impedance of the inductive and capacitive reactances to be
magnitude of the peak.
stated as
Design Considerations: As previously mentioned, filter
(3)
nwL
XLn=
interaction with the source impedance results in a parallel
peak. For inductive source impedances (L,), this
(4) resonant
Xcn = Il/(nwC).
point occurs at a frequency below that at which the filter is
tuned. It is given by
Since the imaginary part is zero at the resonance harmonic n
r, then:
1
Hz.
(9)
f sys=
(5)
XLr = XCr.
2w

f=2(r)(LC)

(Ls+L)C

Solving for r results in the design formula


r2= (XC/XL).

(6)

Quality Factor: The quality of the filter is a measure of the


sharpness of tuning. Mathematically, quality, or Q, is defined

For multiple parallel single-tuned filter installations there will


exist one resonance peak for each filter. The actual calculation
of these exact frequencies is arrived at most easily via the
solution of a set of simultaneous equations. A method for the
formulation of these equations is given in [61.

507

GONZALEZ AND MCCALL: FILTERS TO REDUCE HARMONIC DISTORTION

S
Uwo

w00

1--W
0
0

0.

a,.

FREQUENCY

Fig. 9. Response of 4.7th harmonic filter for different values of MVA short
circuit.

FREQUENCY
Fig. 10. Typical frequency response of high-pass filter.

The resonance peak also has its own value of Q. This is used for all filtering. Two factors may discourage such an
given by
application.
* The minimum impedance of the high-pass filter in its
passband never achieves a value comparable to that of the
Q sys= (R
L
(10)
single-tuned filter at its notch frequency.
* The shunting of a percentage of all the system harmonics
From inspection of (2) and (9), it is evident that the proximity
through one filter may require that filter to be vastly
of the parallel resonance peak and the filter notch is dependent
overrated from a fundamental frequency point of view.
entirely upon the source inductance. This in turn is a function
Design Equations: High-pass filters typically take on one
of the available three-phase fault MVA. Fig. 9 shows how the
of
three forms, as shown in Fig. 6. These are the first-,
response for the same 4.7th harmonic filter as in Fig. 8 varies
and third-order implementations. The first order,
second-,
for different values of available short-circuit MVA.
which
is
characterized
by large power losses at fundamental
The problem associated with these adjacent resonance
is
used.
The second-order high-pass filter is
frequency,
rarely
points is one of filter detuning. If a filter is tuned exactly at the
the
to
while
simplest
apply
providing good filtering action and
frequency of concern, then an upward shift in the tuned
reduced
fundamental
frequency
losses. It is this filter that will
frequency will result in a sharp increase in impedance as seen
be
discussed.
While
loss
of the third-order filter
performance
by the harmonic. Should the resonance peak shift upward
is
to
that
of
the
second
it is less effective in its
superior
order,
enough to coincide with the harmonic of concern, the resulting
action.
filtering
voltage amplification may be disastrous. The four most
The impedance of the second-order high-pass filter is given
common mechanisms that may result in filter detuning are

(Rs)

* capacitor fuse-blowing, which lowers the total capacitance and thereby raises the frequency at which the filter
is tuned;
* manufacturing tolerances in both the tuning reactor and
the capacitor units;
* temperature variations;
* system variations.
With this in mind it becomes advantageous to tune a filter to a
frequency somewhat below that at which it is desired. This
will provide for sufficient harmonic filtering action, yet will
also allow for operation of the bank in the event of the removal
of a few capacitor units. Typically, filter banks are tuned to
approximately 3-10 percent below the desired frequency. In
consideration of the above, the importance of a properly
designed unbalanced circuit detection scheme to protect the
bank and the system becomes clear.

High Pass Filter Design


The high-pass filter is so named due to its characteristic low
impedance above a corner frequency. The typical frequency
response of a high pass filter is shown in Fig. 10. This filter
will shunt a large percentage of all harmonics at or above the
corner frequency.
Frequently, one high-pass filter whose corner frequency is
located at the lowest harmonic, which is to be eliminated, is

by

1 X-I
Z=- 1 + /1_ +-

jwC

R jwL /

(11)

The frequency at which the filter is tuned, its corner


frequency, is given by (2), and the Q of the filter is given as
R

Q((LIC)112 XXL XXC

(12)

Equation (6) is also valid for the high-pass filter. Fig. 11


shows how the response of a high-pass filter with a corner
frequency at the 10.7th harmonic varies with different values
of the resistor. Fig. 12 is similar but here the filter is in a
circuit with a source impedance. It can again be seen that the
system impedance interacts with the filter and produces a
resonance point.
For the high-pass filter, typical values of Q vary from 0.5 to
2.0. With a high Q, i.e., 2, the filtering action is more
pronounced at the corner frequency, while at higher frequencies the filter impedance rises steadily. For lower values of Q,
i.e., 0.5, the response at the corner frequency is not
noticeable, and as the frequency increases, the impedance is
roughly constant.
Other factors to be considered in the choice of Q include
* the frequency to which the filter is tuned;

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS. VOL. IA-23. NO. 3. NMAY/JUNE 1987

508

IrO= |1.4

-1

-Q= 2.0

FREQUENCY

Fig. 1 1. Response of high-pass filter for different resistor values.

KVASC/KVA (LOAD)

U-

Fig. 14. Capacitor size requirements for 4.7th harmonic filter as function of
short-circuit ratio.

La o
_D (r
z

3-

(D O
4 .

Fig. 12. Response of

40
35
30
25

2
15

10 [

"I'

F
0

circuit MVA versus total-load MVA exceeds 45 for a six-pulse


converter, the maximum possible THD is below 5 percent and
no filter is necessary. The same is true for a ratio of 32 for a
FREQUENCY
drive.
twelve-pulse
system with high-pass filter for different resistor
to calculate the actual capacitive var's
possible
is
It
values.
necessary for a minimum filter given proper constraints. Fig.
14 shows the capacitive kvar per kVA of harmonic load
necessary for limiting the THD to 5 percent. The constraints
used in obtaining these plots were as follows.
* The filter used was a single-tuned 4.7th harmonic filter.
* No harmonic current contribution below the fifth harmonic was considered.
* No other capacitors were on the system.
In Fig. 14 there are two curves representing two combinations of harmonic sources. They are as follows.
* The total harmonic load is composed of six-pulse drives.
_
* The total harmonic load is composed of equal amounts of
six- and twelve-pulse drives.
60
40
50
30
20

i0

MVASC / CONVERTER MVA

Fig. 13. Variation of THD

*
*

the

concern
losses.

as

function of short-circuit ratio.

for telephone interference (if any);

MINIMUM FILTERS
A minimum filter is one that "'adequately suppresses
harmonics at the least cost and supplies some reactive power
but perhaps not all that is requiired" [8]. The major consideration here is that of capacitor bank size. The capacitor must be
able to handle the harmonics it is shunting in addition to the
normal 60-Hz quantities. Arriving at the actual size is an
iterative process as the magnitude of the fundamental frequency current is dependent upon the bank size.
The graph in Fig. 13 illustrates the variation of the total
harmonic distortion (THD) with respect to the MVAsc per
converter MVA ratio using the harmonic current magnitudes
given in Fig. 3. The THD is limited to 5 percent in most
industrial power systems. Note that when the ratio of short-

Increasing the kvar beyond these values will reduce the THD
to below the 5-percent point. Decreasing the kvar will have the
opposite effect.
Fig. 15 shows the voltage multiplying factors as a function
of the kVA short circuit to kVA load ratio for minimum 4.7th
harmonic filters. The factors on the vertical axis should be
multiplied by the system voltage to obtain a capacitor voltage
rating that will be equal to the arithmetic sum of the
fundamental and harmonic voltages. Of all the ratings that will
be discussed in the next section, this has proven to be the

limiting one in most cases.

FILTER COMPONENT RATINGS


Capacitors
The allowable overload limits of capacitors based
standards are
kvar
rms
sum
rms

voltage
of peak voltages
current

135%
110%
120%
180%.

on

509

GONZALEZ AND MCCALL: FILTERS TO REDUCE HARMONIC DISTORTION

Tuning Reactor
a

EOUAL

The reactors used for filter applications are usually built


with an air core, which provides linear characteristics with
respect to frequency and current. A 5 percent tolerance in
a 1.18
the reactance is usually acceptable for industrial power system
0X 1.16
applications. The 60-Hz X/R ratio is usually between 50 and
150. A series resistor may be used to lower this ratio if
2 1.14
desired.
1.12
The reactor should be rated to withstand a short circuit at the
between the reactor and the capacitor. The insulation
point
1.10
I
(BIL) of the reactor should be similar to that of power
1.08
transformers connected at that same voltage level.
Parameters to include when specifying a reactor are
KVASC /KVA LOAD
* 60-Hz current;
Fig. 15. Voltage overrating factors for minimum harmonic filters.
* harmonic current spectrum;
* short-circuit current;
* X/R ratio;
All of these parameters should be checked when applying
*
system voltage;
capacitors in a harmonic environment, especially if the
BIL.
*
capacitors are parts of a filter.
The use of an inductor in series with a capacitor results in a
FILTER SELECTION AND PERFORMANCE EVALUATION
voltage rise at the capacitor terminals given by
Before any filter scheme is devised, a power factor study
should
be performed to determine if any reactive compensan2-I
tion requirements exist for the system. If so, the filter will be
designed to provide the correct kvar. If power factor correcWhen verifying the maximum voltage rise the worst condi- tion is not necessary, then a minimum filter should be
tions should be taken into account. Maximum system voltage designed; that is, one that can handle the 60 Hz and harmonic
together with maximum capacitance tolerance (typically 8 currents and voltages without considering the reactive output.
percent) and maximum inductor tolerance (typically 5 percent)
Sometimes more than one tuned filter is needed to improve
should be used. Taking these tolerances into account will yield the system harmonic performance. In this case the filter
the maximum voltage rise across the reactor.
system must be designed for the possibility of having specific
When using a capacitor bank in a system with a voltage filter branches out of service.
lower than the capacitor rating the following formula should
While determining the effectiveness of a filter installation
be used to determine the effective kvar:
ultimately depends upon the degree of harmnonic suppression,
it also involves consideration of alternate system configurations.
As the supplying utility reconfigures its system, the
kvarcap = ( cap L - L )
(14)
Zc
impedance, looking back to the source from the plant's
standpoint, will change. Similar effects will be seen with the
The presence of the filter reactor changes the effective kvar plant running under light versus heavy loading conditions,
output of the bank. The new output is calculated by
with split-bus operation, etc. Therefore the filtering scheme
performance must be evaluated under any and all reasonable
operating configurations.
kvarlt _=(VL -L)(
To consider the effects of varying source interaction, the
ZC-ZL
utility should be contacted to determine the minimum and
Therefore, if reactive compensation is also required from the maximum available three-phase fault MVA at the point of
filter, the designer may have to do several iterations before connection to the plant. This will allow the calculation of the
finally deciding on the capacitor bank ratings.
minimum and maximum values of source impedance. SecThe current limit, although 180 percent by standards, may ondly, a list should be drawn up consisting of all the
be lower because individual capacitor units are usually fused at reasonable operating contingencies that the plant may be
125-165 percent of their current rating.
expected to operate in. Frequency scans of the system in these
When designing a filter, the limits on the rms voltage and various situations should now be made.
currents, and the arithmetic sum of the peak voltages on the
In general, a frequency scan should be made at each
capacitor bank, should be close to 100 percent of rating for problem node in the system, with harmonic excitation at each
normal conditions. This is done so that the overrating point where harmonic sources are connected. This allows the
capabilities are available to cover system overvoltages and easy evaluation of the effects of system changes on the
bank unbalance conditions. The harmonic components may resonant conditions. The cause for any problem that may arise
increase significantly for bank unbalance conditions [11].
may then be discovered quickly by examination of these plots.
1.22

1.20

----6

12-PULSE

510

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL.

One item of particular importance that must be observed is


the variability of the parallel resonance points (the peaks) with
regard to alteration of system parameters. For example, if for
the maximum system load a resonance peak exists at the sixth
harmonic, but at 50-percent load it exists at the 4.8th
harmonic, then at some loading between these two points the
resonance will occur exactly at the fifth, a location at which
three- and six-pulse variable-speed drives produce significant
harmonic current. This observation will require either a
redesign of the filter or the implementation of special
operating rules for the system that will minimize the effect of
this resonance point. Another consideration is that if the
converter is connected to a dc motor with a high-performance
regulator, instability may occur if the resonant peak is at the
fourth harmonic [7].
Traditional performance yardsticks are the total harmonic
distortion (THD) factor and telephone influence factor (TIF).
THD is an rms value of the distortion component of the
fundamental frequency voltage wave due to harmonics. TIF
measures the likelihood of high-frequency harmonics on the
power lines inducing noise on adjacent telephone lines. This
factor is usually of more concern at the utility level, where
power and telephone lines may parallel each other for
extended distances and where the difference in power levels is
larger. In consideration of the above, TIF will not be discussed
here. The reader is referred to [8] and [9].
Total harmonic distortion is defined as

THD

Sum of the squares of all


harmonic voltages

Fundamental voltage squared

100%

IA-23, NO. 3, MAY/JUNE 1987

TABLE I

RECOMMENDED DISTORTION LEVELS


% Distortion
Dedicated System General System

System Voltage (kV)

10%
8%
1.5%

0.45
2.4 to 69
1 15 and above

NEE
RACTIVE
COMPENS ATION

No

5%
5%

1.5%

IGJNM
MINIMUMI
FILTERI
DES

YES

SELECT CAPACITOR
KV a KVAR RATINGS

(16)

According to IEEE Std. 519-1981 on industrial power


systems, the THD levels in a plant should not exceed those
given in Table I.
The THD should be evaluated at any and all buses at which
harmonic generating equipment is located and at which
harmonic sensitive equipment is loacted, including the main
plant bus. If at all locations in question the THD is within
proper limits, the filtering scheme is so far a successful one.
At this stage the duties that all filter components see must be
considered. If a device's ratings are exceeded, then that device
must be reselected and/or the filter it is a part of must be
redesigned.
After an analysis of the filter systems performance is
completed, the design process is iterated upon to eliminate any
potential problems. If no problems are found to exist, then the
filter scheme is a successful one.
Loss evaluation of a filter installation should include the
fundamental frequency losses as well as the harmonic components. For a lossy filter design the cost of losses could be very
significant. Decision flow charts for filter design are shown in
Figs. 16 and 17.
UNBALANCED CIRCUIT DETECTION
The purpose of an unbalanced circuit detection scheme is
the removal of a capacitor bank from the system in the event of
fuse operation or operations. This will prevent damaging

Fig. 16. Decision flowchart for single-tuned filter design when reactive
compensation is desired.

overvoltages from being impressed across the remaining


capacitor units in a group where the capacitor outage occurs,
thereby protecting against a situation that can be immediately
harmful to the capacitor units or associated equipment.
Generally, all unbalanced circuit detection schemes are set
to trigger an alarm upon an initial outage or outages within a
bank. If subsequent outages occur where critical overvoltages
are produced, the bank would be removed. If a capacitor bank
is a filter component, the outage of capacitor units could
detune the filter and cause harmonic amplification. Reference
[11] has a complete description of the different unbalanced
circuit detection schemes available.
CONCLUSION
1. Factors to consider when designing filters to correct
harmonic problems include

GONZALEZ AND MCCALL: FILTERS TO REDUCE HARMONIC DISTORTION

Fig. 17.

Decision flowchart for single-tuned minimum filter design.

a) capacitive kvar requirements for

tion,

2.
3.
4.

5.
6.

11]

power

factor

51 1
[2] J. C. McCall et al., "Harmonic simulation of cycloconverter operation
in a mine power system," presented at the IAS Annual Meeting,
Toronto, ON, October 6-11, 1985.
[3] R. C. Dugan, "Computer simulation harmonics study," draft of final
report prepared for U.S. Department of Energy, Subcontract No. 86X9501 IC, September 20, 1984.
[4] M. F. McGranahan et al., "Distribution feeder harmonic study
methodology," IEEE Trans. Power App. Syst., vol. PAS-103, no.
12, pp. 3663-3671, Dec. 1984.
[5] M. F. McGranahan et al., "Digital simulation of distribution system
frequency-response characteristics," IEEE Trans. Power App. Syst.,
vol. PAS-100, no. 3, pp. 1362-1369, Mar. 1981.
[6] N. G. Hingorani and M. F. Burbery, "Simulation of ac system
impedance in HVDC system studies," IEEE Trans. Power App.
Syst., vol. PAS-80, pp. 820-828, May/June 1970.
[7] D. E. Steeper and R. P. Stratford, "Reactive compensation and
harmonic suppression for industrial power systems using thyristor
converters," IEEE Trans. Ind. Appl., vol. IA-12, no. 3, pp. 232254, May/June 1976.
[8] E. W. Kimbark, Direct Current Transmission. New York: WileyInterscience, 1981, ch. 8, pp. 295-391.
[9J IEEE Guide for Harmonic Control and Reactive Compensation of
Static Power Converters, IEEE Std. 519-1981.
[10) P. R. Clement and W. C. Johnson, Electrical Engineering Science.
New York: McGraw-Hill, 1960, ch. 15, pp. 439-477.
[11] W. S. Vilcheck and D. A. Gonzalez, "Guidelines for applying shunt
capacitors on industrial power systems," presented at the Industrial &
Commercial Power Systems Conference, Denver, CO, May 13-16,
1985.

correc-

b) parallel resonant peaks resulting from interaction of


Damian A. Gonzalez (M'77) was born in Ponce,
Puerto Rico. He received the B.S. degree from the
filter and source impedance,
University of Puerto Rico, Mayaguez, PR, in 1973
| *X
c) filter component ratings and tolerances,
and the M.E. degree in electric power from Renssed) filter losses.
laer Polytechnic Institute, Troy, NY, in 1983.
From 1973 to 1982 he was employed by the
Minimum filters should be designed when reactive comRico Electric Power Authority. In 1983 he
Puerto
pensation is not a requirement.
joined the Systems Engineering Group of McGrawVariations in system load and source impedance should be
Edison in Canonsburg, PA, as a Senior Power
considered to determine the worst possible harmonic
Systems Engineer. His general field of responsibilities includes transmission, distribution, and induscondition.
trial power system studies. His interests include harmonic and transient
Elimination of a high percentage of system harmonics analysis.
Mr. Gonzalez is a Registered Professional Engineer in the Commonwealth
through a high-pass filter may result in highly overrated
of Pennsylvania.
capacitor units.
The frequency response of single-tuned filters does not
show significant changes for values of Q above 25.
Using the outlined converter model yields a THD of 5
percent or larger for MVA short circuit per converter
Jack C. McCall (S'80-S'83-M'83-M'84) received
the B.E.E. degree from Gannon University, Erie,
MVA ratios below 45 for six-pulse drives and below 32 for
PA, and the M.E. degree from Rensselaer Polytechtwelve-pulse drives. Minimum filter parameters are given
nic Institute, Troy, NY.
for limiting the THD to 5 percent for the lower ratios.
He joined the Systems Engineering Department
of McGraw-Edison Power Systems in Canonsburg,
PA, as a Power Systems Engineer in 1984. There he
REFERENCES
performed specialized system studies in the areas of
W. E. Reid and K. J. Petrus, "Harmonics and capacitors in the power
system," presented at the Pacific Coast Electrical Association Engi-

neering & Operating Conference, Los Angeles, CA, March 19-20,


1985.

harmonics, transients, overcurrent device coordination, and short-circuit analysis for both utility and
industrial systems. His current position is that of
Assistant Marketing Manager, Capacitor Products, in Greenwood, SC.

Das könnte Ihnen auch gefallen