214 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO.

1, JANUARY 2003

An Investigation on the Validity of Power-Direction

Method for Harmonic Source Determination
Wilsun Xu, Senior Member, IEEE, Xian Liu, Member, IEEE, and Yilu Liu, Member, IEEE

Abstract—The power-direction method has been used widely

to identify the locations of harmonic sources in a power system.
A number of utility-customer disputes over who is responsible for
harmonic distortions have been settled with the help of the method.
A closer examination of the method, however, reveals that it is un-
able to fulfill the task of harmonic source detection. Case studies
can easily show that the method yields incorrect results. In this
paper, problems associated with the method are investigated using
case studies and mathematical analysis. The results show that the
power direction method is theoretically incorrect and should not
be used to determine harmonic source locations. The main cause of
the problem is that the direction of active power flow is a function of
the phase-angle difference between the two sources. The direction Fig. 1. Problem of harmonic source detection.
of reactive power flow, on the other hand, has a better correlation
with the source magnitudes.
matical analysis. Causes of the problem are identified. In addi-
Index Terms—Harmonics, harmonic source detection, harmonic
sources. tion, this paper proposes superposition-based indices to quantify
the contributions of harmonic sources.

I. INTRODUCTION II. POWER DIRECTION METHOD

W HENEVER significant harmonic voltage or current dis-

tortions are observed in a power system, it is always
useful to find the sources of the distortions. Correct identifica-
tion of harmonic source locations is essential for designing ef-
fective harmonic mitigation means and for determining the re-
sponsibility of the parties involved. The most common situation
that needs harmonic source detection is to resolve the disputes
over who is responsible for harmonic distortions at the point of
common coupling between a utility and a customer or between
two customers.
The most common method for harmonic source detection is
the power direction method. This method checks the direction of
harmonic power flow. The side that generates harmonic power is
considered to contain the dominant harmonic source or to have
a larger contribution to the harmonic distortions observed at the
measurement point. This method is apparently sound and has
been used in industry as a tool for many years [1], [2]. A number
of power quality monitors have included it as a key product fea-
ture. However, we have found concrete proofs that the method is
not suitable for harmonic source detection. The method has been
found theoretically incorrect. In this paper, problems associated
with the method are investigated using case studies and mathe-
Manuscript received April 25, 2001; revised March 2, 2002. This work was
supported by a Grant from the Natural Sciences and Engineering Research
Council of Canada (NSERC).
W. Xu and X. Liu are with the Department of Electrical and Computer En-
gineering, University of Alberta, Edmonton, AB T6G 2V4, Canada (e-mail:
wxu@ee.ualberta.ca).
Y. Liu is with the Department of Electrical Engineering, Virginia Polytechnic
Institute and State University, Blacksburg, VA 24061-0111 USA.
Digital Object Identifier 10.1109/TPWRD.2002.803842
0885-8977/03$17.00 © 2003 IEEE
The harmonic source detection problem can be explained
with the help of Fig. 1. In this figure, the disturbance sources
are the customer harmonic source and the utility harmonic
source . and are the harmonic impedances of the re-
spective parties. The circuit is applicable to different harmonic
frequencies (the values will be different). The task of harmonic
source detection is to determine which side contributes more
to the harmonic distortion at the PCC, subject to the constraint
that measurements can only be taken at the PCC.
To determine which side causes more harmonic distortion at
the harmonic order , the power direction method first measures
voltage and current at the PCC and then calculates the following
harmonic power index:
(1)
where and are the harmonic voltage and current at the
PCC for a particular harmonic number. Since this paper deals
with one harmonic at a time, the subscript that represents har-
monic number will be omitted throughout the paper to avoid
excessive subscripts. The direction of is defined as from
side to the side. Conclusion of the power direction method is
the following.
• If , the side causes more th harmonic distortion.
• If , the side causes more th harmonic distortion.
III. QUANTITATIVE HARMONIC CONTRIBUTION INDICES
While the power direction method could be intuitively sound,
the term “causing more harmonic distortions” is vaguely de-
XU et al.: VALIDITY OF POWER DIRECTION METHOD FOR HARMONIC SOURCE DETERMINATION 215

A similar index can be defined from the voltage distortion

perspective. The contribution to voltage distortions at PCC by
each harmonic source can be determined as follows

(4a)
Fig. 2. Determination of harmonic source contributions for hth harmonic.
(4b)

(4c)

Again, the phasors and need to be projected

onto the axis to provide scalar harmonic contribution in-
dices. The process is similar to what has been applied to the
current indices.

IV. SAMPLE TEST RESULTS

Fig. 3. Decomposition of I of the hth harmonic into two components. The validity of the power direction method can be tested with
the following simple experiment:
1) Select a set of practical , , and data that cor-
fined. There is a need to be more precise on this term so that the
respond to any particular harmonic number. In this paper, we
effectiveness of any harmonic source detection methods can be
chose the following case:
assessed. To this end, the principle of superposition is applied to
the th system shown in Fig. 1 [3]. According to the principle, p.u. p.u.
the contribution of each source to the th harmonic current
can be determined according to Fig. 2 and the following equa- angle p.u. p.u.
tions:
Let the phase angle of vary from 0 to 360 . Phase angle of
is set to 0 as a reference.
(2a) 2) Check the correlation between the power direction index
and the superposition indices. If the power direction method was
(2b) correct, the following condition should always hold:

If then (5)
(2c)
Fig. 4 shows the results. The figure depicts the variation of
where and are the contributions of respective normalized harmonic power , current , and its components
sources to the PCC current. Again, all quantities in Fig. 2 and (2) and with respect to the phase angle of . The normaliza-
refer to a particular harmonic, the th harmonic. The equations tion divides the currents and power by their respective maximum
shown before are phasor equations. There are still some ambi- values so that they reside between the range of 1 to 1 and can
guities to define harmonic contributions since (2c) is a phasor be easily plotted in one chart for comparison. It can be clearly
summation. A more precise method is to decompose into seen from the figure that the condition shown before (5) does
two scalar components as shown in Fig. 3 and (3) [4] not always hold true. For example, is always greater than
and the side should be considered as the main harmonic
(3) current contributor. The harmonic power, on the other hand, can
flow in either direction. When the angle is greater than 190 ,
where is the projection of onto , and is the the power direction contradicts to the principle of superposition.
projection of onto . Thus, is the algebraic sum- There is an approximate 50% chance that the contradiction can
mation of two scalar components, one due to customer harmonic occur. Fig. 5 is a practical case that can create a situation where
source and the other due to utility source. For example, if the phase angle between and is far apart. Since there is
is equal to 3 A and is equal to 7 A, the current is 10 A. no guarantee that the phase angle between and resides in a
One can say that the customer’s contribution is 70%. certain range, one can conclude that the power direction method
Both and are scalars and can have opposite signs. If is unsuitable for harmonic source detection.
they have the same sign, the customer and utility harmonics add Test results are also obtained for the voltage distortion index
up to form . If they have opposite signs, the negative one has defined in (4). The results are shown in Fig. 6. It is noted that,
the effect of reducing the harmonic flow at PCC. We believe that according to the voltage superposition index, the side con-
the and indices accurately characterize the contributions tributes more to the PCC voltage distortion. This conclusion ap-
of respective harmonic sources to the PCC current. They are plies to all customer current angles. The power direction index,
used in this paper to refute the validity of the power direction however, changes sign when the current angle is between 190
method. and 340 . The contradiction between the two indices is obvious.
216
Fig. 4. Correlation between the power direction and current superposition
indices.
Fig. 5. A practical case that can cause large phase-angle difference between
I and I .
Fig. 6. Correlation between the power direction and voltage superposition
indices.
V. MATHEMATICAL ANALYSIS
Figs. 4 and 6 can be produced easily for many test cases. The
evidence against the power direction method is overwhelming.
This finding is also supported by the results shown in [5] and
[6]. In this section, we try to demonstrate mathematically that
the contradiction does exist. The sources of the contradiction are
also identified and explained.
A. DC Circuit Case
The first case is a general dc circuit shown in Figs. 7 and 8. Al-
though the actual harmonic source detection problem involves
ac circuits, the dc circuit case can reveal key characteristics of
the power direction method. A dc circuit is much simpler to an-
alyze since there is no phasor involved. The dc case can also be
considered a special ac case where the system has resistances
only and the harmonic sources have the same phase angle.
IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 1, JANUARY 2003
Fig. 7. DC circuit to demonstrate the power direction method (I and I
oppose each other).
Fig. 8. DC circuit to demonstrate the power direction method (I and I add
up).
The PCC voltage and current of the circuit can be determined
as follows:
(6a)
(6b)
Since the voltage is always positive, the condition for the
power ( ) to flow from side to side is or
(7)
On the other hand, the superposition current index shows that
if
(8)
or
(9)
holds, the side contributes more current than the side. Equa-
tions (7) and (9) are therefore consistent, which implies that the
power direction method works well in this case.
If the customer current changes polarity as shown in Fig. 8,
however, the current becomes positive all of the time. The
condition for the power to flow from side to side is ,
or
(10)
[based on Equation (6b)]. The condition for is still
the same as that shown in (9). This equation does not always
agree with (10), depending on the relative size of and .
Therefore, a mathematical proof has been found to show that
there is a contradiction between the power direction method and
the superposition method. Another conclusion drawn from the
analysis is that the polarity (or phase) difference between the
two sources has more influence on the direction of power flow
than that caused by their magnitude difference.
B. AC Circuit With Reactance Only
The second case is an ac circuit shown in Fig. 1. To sim-
plify the analysis, the impedances are assumed to contain imagi-
nary parts only. The circuit corresponds to a particular harmonic
XU et al.: VALIDITY OF POWER DIRECTION METHOD FOR HARMONIC SOURCE DETERMINATION 217

frequency. Following the classic power-angle equation for two
source ac circuits, the power flowing from side to side can
be determined as:
(11)
reactive power. One would, therefore, wonder if the direction of
reactive power could indicate the relative size (i.e., magnitudes)
of two harmonic sources. This question can be analyzed by ex-
amining the reactive power flowing out of source

where and are the open circuit th

harmonic voltages of the and sides, respectively. is the (17)
phase-angle difference between utility and customer side th where is the phase-angle difference between the two sources.
harmonic current sources. The significance of this equation is Since is always less than 1 and if , condition
the following: the direction of power is controlled by , the automatically implies that the direction of reactive
phase-angle difference between two harmonic sources. power is from side to the side. In other words, the direction
For the superposition-based current index, we want to show of reactive power can be used as an (necessary but not sufficient)
at first that condition implies . indicator to determine which side has a larger voltage source.
According to Fig. 3 and the well known triangular formula, the The reason that the indicator is not a sufficient one is because
superposition currents and can be determined as the reactive power can still flow from side to the side if
follows is greater than but is less than . The phase-angle
, therefore, plays an important role in this case as well. Despite
(12a)
this restriction, the direction of reactive power is still a more
(12b) reliable indicator than the direction of active power in this case.
Another important prerequisite for using the direction of reac-
Therefore tive power as a harmonic source locator is . While
this condition is generally true at the fundamental frequency, it
(13a) may not be true at the harmonic frequencies. This is the main
problem associated with the reactive power direction method.

(13b) C. AC Circuit With Resistances Only

Subtracting the above two equations yields
(14)
This equation shows that if is satisfied,
will be true. Note that this is a general conclusion. It
does not rely on the assumption of the impedances having imag-
inary parts only. In the following step, the relationship between
the superposition currents and the source currents is determined.
The principle of superposition shows
It is interesting and important to examine the hypothetical
case where the system and customer impedances are entirely re-
sistive, namely, and . In this case, the branch
resistance consumes power. The amount and direction of the ac-
tive power flowing on the branch are dependent on the location
of measurement. To simplify the problem, we first consider the
power generated or absorbed by harmonic source , as shown
in Fig. 9. For this case, the power flowing out of source is
(18)

(15a) where is the phase-angle difference between the two sources.

It can be seen that (18) is very similar to (17). It means that the
(15b) direction of active power can be used as an (necessary but not
sufficient) indicator to determine which side has a larger voltage
The condition for or to hold source for the resistive circuit. The reactive power flow for this
becomes case takes the following form:

(16) (19)

The condition just mentioned is not related to phase-angle ,
while (11) is. The power direction index is therefore inconsistent
with the current superposition index. Hence, the invalidity of the
power direction index has been demonstrated analytically. The
same process can be used to show the inconsistency between the
power index and voltage-based superposition index.
It is common knowledge for power engineers that the phase
angles of bus voltages mainly affect the flow of active power
while the magnitudes of bus voltages mainly affect the flow of
This equation is very similar to (11). Comparing (11) to (19),
and (17) to (18), one can conclude that the characteristics of the
circuit impedance determine which power, or , has
more bearing on the source magnitudes instead of source phase
angles. If the circuit impedance is dominated by reactance, the
direction of reactive power is a better indicator on the relative
magnitude of the two sources. If the impedance is dominated
by resistance, however, the direction of active power is a better
indicator.
218
Fig. 9. AC circuit with resistive elements.
The second analysis deals with a general case where the me-
tering point is the PCC. As shown in Fig. 9, the PCC voltage
and current can be determined as follows
(20a)
(20b)
The active power flowing from side to side can be de-
termined as
(21)
The equation just shown reveals that the resistances and
could have a large impact on the sign of , or the direction
of active power. We can quantify the impact by considering the
special case of . For this case, Equation (21) can be
simplified as
(22)
The equation just shown demonstrates clearly that the direc-
tion of active power is affected by the relative size of and
or the location of the meter. If , the power flows
from side to side, even if the two sources have the same
magnitude. This analysis has revealed another impact factor, the
point of measurement, which can influence the reliability of the
power direction method.
VI. CONCLUSION
The concern for power system harmonics is mainly on the
distortion of sinusoidal voltage and current waveforms. Even
with little harmonic power, a distorted waveform can trigger the
malfunction of electronic circuits. It is, therefore, important to
define the contribution of each harmonic source based on the
current and/or voltage parameters. This paper proposes superpo-
sition-based current and voltage indices to quantify the contri-
butions of harmonic sources. Using these indices, the validity of
the power direction method for harmonic source determination
is investigated. Both case studies and mathematicaly analyses
have shown that the power direction method is not suitable for
harmonic source detection. Main findings of this work are sum-
marized as follows.
IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 1, JANUARY 2003
1) The direction of active power is mainly affected by the
relative phase angle between the two harmonic sources. It has
little bearing on the relative magnitude of the sources. Note that
it is the source magnitudes instead of phase angles that are of
main interest for the harmonic source detection problem.
2) The direction of reactive power, on the other hand, has
a closer relationship to the source magnitudes. If the circuit
impedance is purely reactive, the direction of reactive power is
actually a necessary (but not sufficient) condition indicating one
source has a larger magnitude than the other.
3) The conclusions shown above are applicable to circuits
dominated by reactances. If a circuit consists of mainly resistive
components, the conclusions mentioned before are reversed.
Namely, the direction of active power is mainly affected by
the source magnitudes and that of the reactive power by the
phase angles. The implication of this conclusion is that the
characteristics of the circuit impedance ( – ratio) will affect
the reliability of the active or reactive power-direction-based
harmonic source detection methods.
4) The metering point or the relative size of the source and
the customer impedances will also affect the direction of either
active or reactive powers. This is another important factor that
makes the powe-direction-based methods unreliable.
The (active) power direction method has been used frequently
as a practical method for locating harmonic sources. This work
has shown that the method does not work and there is an urgent
need to develop new harmonic source detection methods.
ACKNOWLEDGMENT
The authors wish to express sincere thanks to M. B. Hughes
of BC Hydro for valuable comments during the course of this
work.
REFERENCES
[1] P. H. Swart, M. J. Case, and J. D. Van Wyk, “On techniques for local-
ization of sources producing distortion in three-phase networks,” Eur.
Trans. Elect. Power Eng., vol. 6, no. 6, Nov./Dec. 1996.
[2] L. Cristaldi and A. Ferrero, “Harmonic power flow analysis for the mea-
surement of the electric power quality,” IEEE Trans. Instrum. Meas., vol.
44, pp. 683–685, June 1995.
[3] H. Yang, P. Porotte, and A. Robert, “Assessing the harmonic emission
level from one particular customer,” in Proc., 1994.
[4] W. Xu and Y. Liu, “A method for determining customer and utility har-
monic contributions at the point of common coupling,” IEEE Trans.
Power Delivery, vol. 15, pp. 804–811, Apr. 2000.
[5] A. E. Emanuel, “On the assessment of harmonic pollution,” IEEE Trans.
Power Delivery, vol. 10, pp. 1693–1698, July 1995.
[6] M. B. Marz, J. F. Witte, D. L. Williams, and P. M. Thompson, “Finding
and determining the influence of multiple harmonic sources on a utility
system using harmonic measurements,” in Power Quality 2000 Conf.,
Boston, MA, Oct. 2000.
Wilsun Xu (M’90–SM’95) received the Ph.D. degree from the University of
British Columbia, Vancouver, BC, Canada, in 1989.
He is presently a Professor at the University of Alberta, Edmonton, AB,
Canada. He was with BC Hydro from 1990 to 1996 as an Engineer. His main
research interests are power quality and voltage stability.
Dr. Xu chairs the Harmonics Modeling and Simulation Task Force of the
IEEE PES.
XU et al.: VALIDITY OF POWER DIRECTION METHOD FOR HARMONIC SOURCE DETERMINATION
Xian Liu (M’95) received the Ph.D. degree from the University of British Co-
lumbia, Vancouver, BC, Canada, in 1994.
He is presently an Assistant Professor at the University of Alberta, Edmonton,
AB, Canada. He was with Nortel Networks from 1995 to 1999 as an Electrical
Engineer. His main research interests are electric machines, optimization, and
real-time software.
219
Yilu Liu (M’91) is an Associate Professor of electrical engineering at Virginia
Polytechnic Institute and State University, Blacksburg.
Her current research interests are power system transients, power quality, and
power system equipment modeling and diagnoses.
Dr. Liu is the recipient of the 1993 National Science Foundation Young In-
vestigator Award and the 1994 Presidential Faculty Fellow Award.