Investigating power issues on a 2,500-hp pipeline

booster station adjustable frequency drive.

HE PIPELINE PROJECT DISCUSSED

T in this article involved building a new

70-mi, 24-in diameter section of a
petroleum products pipeline stretch-
ing from Beaumont, Texas, to Longville, Louisiana.
IEEE INDUSTRY APPLICATIONS MAGAZINE • NOV|DEC 2005 • WWW.IEEE.ORG/IAS

From Longville, the pipeline was linked with an

existing 720-mi, 26-in pipeline (originally used for
natural gas) to Bourbon, Illinois. The pipeline was
then linked to a new 2-million barrel refined petroleum
products storage terminal in Lick Creek, Illinois. Booster
stations with soft starters were located in Beaumont, Texas,
and Sugartown, Louisiana. Stations with adjustable frequency © ARTVILLE

drives (AFDs) were located at Kilbourne, Louisiana; Tutwiler, Mississip-

pi; Fisherville, Tennessee; and Obion, Tennessee. The booster station located in Fisherville, Tennessee, is the subject
of this article.
The application employed a 2,500 -hp, 4,160 -V, 18-pulse capacitor-assisted, pulse-width modulated (PWM), current
source inverter-type AFD. This topology uses three series silicon-controlled rectifier (SCR) bridges in the rectifier, fed from
a four-winding transformer. The secondaries of the transformer are delta or extended delta, and they are phase shifted by 20º
from each other to achieve harmonic cancellation. The predominant rectifier current harmonics in this configuration are the
17th, 19th, 35th, 37th, 53rd, and 55th harmonics.
Some power quality issues arose in the utility power system. Nearby utility customers reported such effects as clocks
running fast, a humming noise from 120-Vac residential sockets, and a humming noise on telephone lines. There was

BY ROBERT ELLIS & BOB GUIDRY

20

1077-2618/05/$20.00©2005 IEEE
also a problem with the voltage regulator on the distribu- capacitors also must be modeled since they affect the
tion line, and phase imbalance was noted as well. Harmonic impedance versus frequency characteristics of the system.
measurements were taken at the pump station to quantify In addition, any significant loads on the system need to
the voltage distortion on the power system. A computer- be taken into account. Where possible, it is helpful to
based harmonic model of the power system was created to obtain an impedance versus plot (60 Hz to, e.g., 3 kHz)
diagnose the problems and to act as a design tool in evalua- for the utility system. Unfortunately, the utility was
tion of alternative means to mitigate the problems. unable to provide such impedance information in this
case. Table 2 shows the harmonic voltage distortion pre-
Notes on the Power System dicted by the model at the 4,160 -V bus.
The pipeline booster station is fed from the utility at 12.47 There is a good correlation between the predicted and
kV. The 12.4-kV feed to the station is implemented with 6 actual voltage harmonics from the fifth harmonic up to
mi of 336-kcmil aluminum conductor steel-reinforced the 31st harmonic. Above the 31st harmonic, the predict-
(ACSR) cable from the secondary of an 18-MVA, 161/12.47 - ed harmonics are significantly lower than the harmonics
kV transformer. A 7.5 -MVA transformer steps the voltage
down from 12.47 to 4.16 kV. A 2.5 -MVA, four-winding
isolation transformer with three 1,400 -V secondary windings 161 kV, Z1 = 1.6 + j3.9 p.u.
feeds power to the rectifier of the AFD. Figure 1 provides a on a 100-MVA base
one-line diagram of the overall power system.

Harmonic Measurements
Made at Problem Onset
Voltage total harmonic distortion (THD) at the 4,160 -V
bus was measured without the AFD operating. It was
18 MVA
found to be less than 2%, which indicates that some 161/12.47 kV
background harmonics were present but were not the Z = 8.8%
source of the problems. The harmonic voltage distortion
was measured at the 12.47-kV feed to the pipeline boost-
er station as well as at the 4,160 -V bus. Table 1 gives the
magnitude of the individual voltage harmonics, which 6 mi of 336 kcmil ACSR 300-kVAR
were gauged up to the 55th harmonic, as measured at Z1 = 0.306 + j0.64 Utility Capacitor
the 4,160 -V bus with the AFD operating. Ω/mi

IEEE INDUSTRY APPLICATIONS MAGAZINE • NOV|DEC 2005 • WWW.IEEE.ORG/IAS

The 13.7% voltage distortion measured is well in excess
of the 5% maximum recommended by IEEE 519-1992 [3]
at the point of common connection between the harmonic 7.5 MVA
producing load and other loads. The high voltage distor- 12.47/4.16 kV
tion observed is usually indicative of a weak (high-imped- Z = 7%
ance) power system and/or a possible resonance condition
on the system. The main harmonics are the 17th, 19th,
35th, 37th, 53rd, and 55th, as would be expected for an
18-pulse rectifier. It is unusual, however, that the individ-
ual voltage harmonics do not decrease to any appreciable 355 Effective kVAR
degree as frequency increases. For example, the 35th har- Capacitor
monic is higher than the 17th or 19th harmonics. The
53rd harmonic is also higher than the 17th, 19th, 35th,
and 37th harmonics. This potentially indicates that there
may be one or more natural frequencies on the power sys- 2.5 MVA
tem, resulting in some degree of amplification of these har- 4.16/1.4/1.4/1.4kV
monic voltages. These natural frequencies typically occur Z = 7%
due to power factor correction capacitors or stray cable
capacitances in combination with the inductance of the
power system [tuned inductance-capacitance (LC) circuit].
AFD
Modeling the Power System
A harmonic model of the power system was created using
a commercially available software package and available
one-line information. This model treats the AFD as a
multiple-frequency harmonic current source. It is impor-
tant to model impedances on the power system, such as
transformers, and the short-circuit impedance of the utili- 2,500-hp Motor
ty power supply. The impedance of long cable runs 1
21
should be accounted for as well. Power factor correction Power system one-line diagram.
 that were measured. This is an indica- significant amounts of higher frequen-
tion that there are inaccuracies in the A COMPUTER- cy harmonics would still be present.
power system model. It is likely that While this type of filter has some mer-
there are capacitances in the system in BASED its, we felt this approach would not
the form of power factor correction guarantee elimination of all ill effects
capacitors or stray cable capacitance HARMONIC that had been observed. Table 3 shows
that we were not aware of and, there- the predicted voltage distortion at the
fore, were not taken into account in the MODEL OF THE 4,160 -V bus after the addition of a
model. Capacitance, in combination fifth/17th harmonic filter.
with the system inductance, has the POWER SYSTEM
effect of creating one or more natural Evaluation of Utility
frequencies. The natural frequencies
WAS CREATED TO Capacitor as a Potential Solution
amplify nearby harmonic voltages ATTEMPT TO We identified a 300-kVAR utility capac-
and/or currents. itor on the 12.47 -kV system. This
DIAGNOSE THE capacitor was normally energized during
Evaluation of Alternative Means the summer months and de-energized
for Further Harmonic Mitigation PROBLEMS. for the remainder of the year. It was
located at approximately the midpoint of
Addition of the Fifth the 6-mi, 336 ACSR feed to the station.
and 17th Harmonic Filters Since the AFD system was commis-
We considered implementing a multiple-leg harmonic fil- sioned in the winter months, this capacitor had not been
ter with fifth and 17th tuned series LC circuits. This type energized when the problems were encountered. A capacitor
of filter has the advantage of pinning resonant frequencies bank in shunt with the power system will act as a sink for
away from characteristic harmonics of the 18-pulse AFD. higher-frequency current harmonics since its impedance
The disadvantage is that it does not specifically target the decreases as frequency increases. By introducing this capaci-
higher-frequency harmonics (i.e., the 35th, 37th, 53rd, tor into the computer model, a significant reduction in the
and 55th harmonics), which appeared to be significant expected harmonic levels was predicted. It was proposed that
contributors to the problem. While there would be some this capacitor be energized to test its usefulness as a quick
reduction in specific harmonics, the modeling showed that solution to the observed effects. The utility had no issues
IEEE INDUSTRY APPLICATIONS MAGAZINE • NOV|DEC 2005 • WWW.IEEE.ORG/IAS

TABLE 1. VOLTAGE DISTORTION AT 4,160 -V BUS TABLE 2. VOLTAGE DISTORTION AT 4,160 -V BUS
BEFORE MITIGATION. BEFORE MITIGATION AS PREDICTED BY MODELING.
Harmonic Number % of Fundamental Voltage Harmonic Number % of Fundamental Voltage
5 1.37 5 1.03
7 0.88 7 0.72
11 0.59 11 0.72
13 0.54 13 0.61
17 5.58 17 6.33
19 3.37 19 5.30
23 0.55 23 0.43
25 0.59 25 0.37
29 0.76 29 0.38
31 0.53 31 0.35
35 5.19 35 2.93
37 4.98 37 2.06
41 0.83 41 0.76
43 1.06 43 0.80
47 0.65 47 0.44
49 0.63 49 0.46
53 7.21 53 1.48
55 5.01 55 1.02
22 THD 13.70 THD 9.61
with the capacitor being energized on a short-term or even a monic distortion (3.85 versus 13.7% measured original-
permanent basis. The capacitor did, in fact, significantly ly). This voltage distortion level is now well within the
reduce the voltage distortion on the power system, as illus- 5% maximum recommended by IEEE 519-1992 [3].
trated in Table 4. Total harmonic distortion was reduced to The capacitor also had a dramatic effect in reducing the
half the level observed without the capacitor. Significant magnitude of the higher-frequency harmonics. This
reductions were achieved in the levels of the 35th, 37th, effect can be seen by comparing the value of the 35th,
53rd, and 55th harmonics. While the symptoms were some- 37th, 53rd, and 55th harmonic voltages in Table 5 to
what improved, the capacitor did not eliminate the ill effects those listed in Table 1. The addition of this capacitor filter
experienced by other customers fed from the utility system. completely eliminated the undesirable effects that had
been noted by the utility’s customers.
Evaluation of the Effects of a 4,160 -V Capacitor
Consideration was then given to the addition of a 355 Contributing Factors to the Power Quality Issues
effective kilovolt-amperes reactive untuned capacitor The power system was relatively weak, meaning that it
bank on the 4,160 -V primary of the AFD isolation had a high impedance. This is primarily due to the 6-mi
transformer rather than the addition of a multiple leg length of 336-kcmil ACSR cable feeding the booster sta-
tuned filter. This approach is similar to the addition of tion. The result of a high-impedance power system is that
the utility capacitor, except that the harmonic currents even the small amount of current harmonics being gener-
would be addressed closer to their origin. A potential ated by the 18-pulse AFD resulted in a large amount of
pitfall of using an untuned capacitor is that a parallel res- voltage distortion on the utility system. Voltage distor-
onance will occur at a frequency determined by the tion is caused by a distorted current passing through the
amount of capacitance and the inductance of the power impedance of the power system.
system. Care must be taken to ensure that this parallel As is typical for a pipeline pumping application, the
resonance is placed far enough away from the characteris- AFD represents the vast majority of the plant load. Har-
tic harmonics of the AFD that amplification of harmonic monic modeling at the design stage is often warranted in
voltages or currents does not occur. In this case, the par- such applications to help identify potential power quality
allel resonant frequency fell at approximately harmonic issues early on.
number 8.5, and modeling showed that no significant Another contributing factor was the presence of stray
amplification should occur. The harmonic voltages capacitances in the system, which caused natural frequen-
observed at the 4,160 -V bus are shown in Table 5. It can cies and unexpected amplification of higher-frequency
be seen that this has considerably reduced the total har- harmonics. One would normally expect that voltage

IEEE INDUSTRY APPLICATIONS MAGAZINE • NOV|DEC 2005 • WWW.IEEE.ORG/IAS

TABLE 3. PREDICTED VOLTAGE DISTORTION TABLE 4. VOLTAGE DISTORTION AT 4,160 -V BUS
AT 4,160-V BUS WITH FIFTH/17TH HARMONIC FILTERS. WITH UTILITY 30-kVAR CAPACITOR CONNECTED.
Harmonic Number % of Fundamental Voltage Harmonic Number % of Fundamental Voltage
5 0.21 5 1.80
7 0.42 7 1.00
11 1.40 11 1.80
13 0.51 13 3.15
17 0.27 17 1.49
19 0.67 19 0.28
23 0.21 23 0.16
25 0.18 25 0.21
29 0.26 29 0.13
31 0.24 31 0.32
35 2.58 35 2.44
37 1.79 37 1.28
41 0.28 41 0.34
43 0.28 43 0.31
47 0.02 47 0.24
49 0.21 49 0.20
53 1.67 53 2.64
55 2.59 55 2.06
23
THD 4.77 THD 6.40
 TABLE 5. VOLTAGE DISTORTION AT 4,160-V BUS WITH Conclusions and Recommendations
4,160 -V, 355 -KVAR CAPACITOR CONNECTED. A computer-based modeling study should probably have
been conducted at the design stage of this project. The
Harmonic Number % of Fundamental Voltage AFDs are fairly large, and it is common for the power
5 1.72 feed to a pipeline booster station to be relatively weak. A
7 1.44
study was not conducted because the AFDs were origi-
nally intended only to be for soft starting, not for contin-
11 1.89 uous use. When it was decided at a later date that the
13 1.16 AFDs would be used on a continuous basis, the need for
a harmonic modeling study was then overlooked. Fortu-
17 1.98
nately, production was not drastically impacted due to
19 0.77 the speed with which the problem was analyzed and
23 0.06 additional harmonic mitigation measures were identified
and implemented.
25 0.12 It should be noted that the use of an AFD with a high-
29 0.07 er pulse number rectifier would not necessarily have pre-
31 0.05 vented the occurrence of these power quality issues. The
characteristic harmonics for a higher pulse number AFD
35 0.26 would be at different frequencies, but amplification of the
37 0.32 higher-frequency harmonics would still have occurred.
41 0.02
An AFD using an active front-end topology may have
alleviated the concerns that arose. The active front-end
43 0.00 topology uses PWM switching of the rectifier, as well as a
47 0.03 reactor/capacitor on the input, to limit harmonics being
generated back into the power system. The use of a capac-
49 0.03
itor at the input to the rectifier makes this design partic-
53 0.03 ularly effective for reducing the magnitude of
55 0.02 higher-frequency harmonics. An active front-end AFD
was not used in this case because the application used
THD 3.85 existing motors. Without the addition of an isolation
transformer, the motor is subjected to higher than usual
IEEE INDUSTRY APPLICATIONS MAGAZINE • NOV|DEC 2005 • WWW.IEEE.ORG/IAS

voltages from phase to ground. The isolation transformer

harmonics, such as the 35th, 37th, 53rd, and 55th, would would have added significantly to the cost of an active
be significantly lower than the 17th and 19th voltage front-end AFD system.
harmonics. This was not the case here; the 53rd harmonic IEEE 519-compliant harmonic levels were achieved
voltage was the highest of any individual harmonic. This with the use of a capacitor at the isolation transformer
can only be attributed to the existence of one or more nat- primary. In this manner, the utility customer power and
ural frequencies on the power system, resulting in ampli- telephone system concerns were completely alleviated.
fication of nearby harmonics. Even lower harmonic levels could likely be achieved if the
utility’s 13.2 -kV, 300 -kVAR capacitor were energized in
Other Criteria That Should Flag the addition to the 4,160-V, 355 -kVAR capacitor at the iso-
Need for Harmonic Modeling Analysis lation transformer primary. No investigation has been
Remote location of an installation may be an indication carried out on the combined effect of the 4,160 -V and
that the power system is weak. This may be due, at least 13.2 -kV capacitors.
in part, to long and relatively small conductors feeding
the installation. References
The power system being fed by a generator under [1] F.A. DeWinter and L.M. Benke, “Systems engineering for large induc-
normal or backup conditions may also flag the need for tion motor adjustable frequency drives,” in IEEE/IAS Petroleum and
Chemical Industry Conf. Rec., 1991, pp. 29–37.
analysis. It is typical for a generator to have approxi- [2] N. Zargari, S. Rizzo, Y. Xiao, H. Iwamoto, K. Sato, and J. Donlon, “A
mately 20% impedance. Depending on the size of the new current source converter using a symmetric gate commutated
generator relative to the AFD, substantial voltage dis- thyristor (SGCT),” IEEE Trans. Ind. Applicat., vol. 37, pp. 896–904,
tortion may result. May–June 2001.
[3] IEEE Recommended Practices and Requirements for Harmonic Control in Elec-
A constant torque load, such as a conveyor, mixer, or trical Power Systems, IEEE 519-1992.
positive displacement pump, may also necessitate a har-
monic study since harmonic currents typically increase at
reduced speeds in these applications. Robert G. Ellis (bgellis@ra.rockwell.com) is with Rockwell
Presence of power factor correction capacitors should Automation in Cambridge, Ontario, Canada. Bob Guidry is an
also highlight the need for analysis, since capacitors, in electrical power and instrumentation consultant in Youngsville,
combination with system inductance, introduce natural Louisiana. Ellis is a Senior Member of the IEEE. This article
frequencies in the power system that can result in amplifi- first appeared in its original form at the 2003 IEEE Petroluem
24
cation of harmonic voltages and currents. and Chemical Industry Technical Conference.