1562

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012

An Intelligent Control Strategy for Power

Factor Compensation on Distorted Low Voltage
Power Systems
Shunfu Lin, Member, IEEE, Diogo Salles, Student Member, IEEE, Walmir Freitas, Member, IEEE, and
Wilsun Xu, Fellow, IEEE

AbstractDue to the proliferation of harmonic producing-loads,

harmonic resonance has become a major hurdle for performing
power factor compensation in commercial power systems, such
as office towers and shopping complexes. This paper presents an
intelligent power factor compensation controller that can perform
power factor correction without exciting harmonic resonance
under varying demand conditions. Practical and robust control
algorithms are proposed for the purpose of easy implementation
in a micro-controller. In addition, the controller relies on common
low cost sensing devices and does not require additional measurements. As a result, the proposed controller can be constructed as
a retrofitting device to replace existing power factor correction
controllers with little effort. Analysis of representative case studies
is conducted to illustrate how the proposed controller performs.
Index TermsCapacitor switching, commercial buildings, harmonic resonance, power factor control.

I. INTRODUCTION

LECTRICAL energy efficiency is of prime importance to

commercial facilities, such as office towers and shopping
complexes. The application of power factor (PF) compensation
has long been accepted as a necessary step to improve the efficiency of these low voltage electrical installations [1][3]. This
is usually achieved by installing capacitors downstream to the
supply transformer at the entrance point of the facility. Such capacitor units are switched in and out of circuits as the demand
for VAR compensation of the building load fluctuates. If applied properly and controlled, capacitors can improve the performance of distribution circuits since, by providing the reactive
current locally, less power needs to be provided by the distribution network resulting in lower losses, improved line voltage,
Manuscript received January 28, 2012; revised April 24, 2012; accepted May
13, 2012. Date of publication July 13, 2012; date of current version August 20,
2012. This work was supported by the Natural Resources Canada through the
Technology, Innovation Program as part of the Climate Action Plan for Canada
and by FAPESP, Brazil. Paper no. TSG-00041-2012.
S. Lin is with the Department of Electric Power and Automation Engineering, Shanghai University of Electric Power, Shanghai, 200090 China,
(e-mail: shunfulin@shiep.edu.cn).
D. Salles and W. Freitas are with the Department of Electrical Energy
Systems, University of Campinas, 13083-852, Campinas, Brazil (e-mail:
dsalles@ieee.org, walmir@ieee.org).
W. Xu is with the Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 2V4, Canada (e-mail: wxu@ualberta.
ca).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSG.2012.2201756

and, ultimately, reduced facility billing charges (utility penalties) [1][5].

However, facility operators have reported frequent failures
or trips of these PF correction capacitor banks. One important reason is the proliferation of harmonic-producing loads;
solid-state power conversion devices are prime examples. The
problem is that capacitors might aggravate the existing harmonic distortion since, although these devices do not generate
harmonics, they provide a network path for possible local
or general parallel resonance conditions, which contributes
to a significant amplification of harmonic currents produced
by the facility loads. In cases of resonance, this current may
be very large and may damage the capacitors. Therefore, the
consideration of PF compensation capacitor installation should
also include harmonic resonance analysis at the design stage
[1][3], [5], [6].
According to [5], possible solutions to avoid harmonic
problems include the following: 1) ungrounding grounded-wye
capacitors; 2) changing capacitor bank sizes and locations; 3)
adding a reactor to an existing bank; 4) adding a filter capacitor;
and 5) controlling the capacitor switching scheme to avoid
resonance. The objective of this paper covers the fifth approach
and consists of developing a practical and robust control algorithm for the capacitor bank switching scheme that is capable
of achieving both power factor correction and resonance avoidance requirements. Previous works proposed optimization
algorithms for this purpose; however, they are time-consuming
and it is not guaranteed they converge to the optimal solution
under time-varying load and system impedance conditions
[7], [8]. More recent works have focused on the installation
of passive and active harmonic filters [9][11]. This approach
can be complex (e.g., stresses in the filters need to be considered, the harmonic spectrum of the nonlinear loads need
to be determined, etc.) and be costly (e.g., installation and
maintenance costs) to the facilitys owner. Another common
solution is to add reactors in series with existing capacitor
banks [12]. However, the system parameters vary dynamically
with the power system configurations and loads. Therefore,
the harmonic resonance might occur even if a combination of
capacitors connected in series with reactors has been installed.
Reference [13] proposed replacing the reactors with power
electronics inverters.
The concept of the controller proposed in this paper is
simpler; whenever resonance conditions exist, the capacitor
bank should be changed in size to shift the resonant point to

1949-3053/$31.00 2012 IEEE

LIN et al.: AN INTELLIGENT CONTROL STRATEGY FOR POWER FACTOR COMPENSATION

1563

Fig. 1. Typical power factor compensation arrangement for commercial power

systems.

another frequency [6]. However, one important issue must be

solved; how to determine if the resonance condition exists
under varying demand conditions? It is not safe and intelligent to decide the suitable number of switched capacitors
units assuming constant system impedance since it varies for
different operating conditions. In order to solve this issue, the
proposed controller uses pre- and postdisturbance (due to the
capacitor switching) steady-state waveforms of the voltages
and currents at the interface point to estimate the harmonic
system impedance [14][16]. It must be highlighted that the
proposed controller relies on common low cost sensing devices
and does not require additional measurements. As a result, the
controller can be constructed as a retrofitting device to replace
existing power factor correction controllers with little effort.
This paper is organized as follows. Section II discusses the
conventional control approach for power factor compensation.
Section III describes the characteristics associated to the harmonic resonance problem. Section IV presents the development
of the proposed capacitor controller. Section V presents case
studies to evaluate the proposed approach. Section VI summarizes the main findings of this paper
II. COMMON REACTIVE POWER COMPENSATION SCHEME
As mentioned before, the traditional approach for power
factor compensation in commercial facilities consists of placing
capacitor banks in parallel with the load at the entrance point
of the facility, offsetting the inductive loading (lagging power
factor) of the equipments. Fig. 1 shows a common power factor
compensation arrangement used in commercial power systems.
As one can observe, the scheme typically consists of one or
more breaker switched capacitor units along with an intelligent
control unit, current (CT) and voltage (VT) transformers, which
are connected at the low side of the supply transformer.
These banks often include three to nine capacitor units connected in three-phase grounded-wye, ungrounded-wye, or delta
configurations [5]. In practice, commercial installations employ
switched capacitor banks, instead of fixed banks, in which the
capacitors units are switched on and off automatically to compensate for changing load conditions (minimum condition up
to peak load) [17], [18]. Fig. 2 illustrates a typical kilovar demand over a 24 h period [5]. This curve can be determined
by a recording kilovar meter or calculated using kilowatt and
power factor measurements. The fixed banks satisfy the base
load requirements, and the switched banks compensate for the
inductive kilovar peak during the heavier load periods [5], [17].

Fig. 2. Application of switched and fixed capacitors for a time varying kvar
demand condition.

Fig. 3. Conventional strategy for power factor correction through switched capacitor banks.

In order calculate the capacitive kilovars necessary to correct

to a new, higher power factor, one must subtract the inductive
kvar of the corrected
power factor from the existing
power factor. The difference is the amount of capacitive kvar to be added to the system. The following formula is a
convenient way of doing this [5]:
(1)
where:
kW

is the system kilowatt load;

kvar

is the amount of capacitive kilovar to be added.

The capacitor switching control scheme illustrated in Fig. 1

is based on a local automatic controller. The control senses
voltage and current, and uses either these parameters directly
or a derived parameter like power factor to compare against a
threshold. Fig. 3 shows a flowchart that illustrates the conventional strategy of the automatic controller based on the measured
power factor.
According to the flowchart above, the following steps are performed for PF correction [4], [19]:

1564

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012

Fig. 4. Parallel resonance at a point of common coupling (PCC).

1) The power factor control is performed by controlling the

opening and closing of the capacitor switches based on the
measured power factor.
2) The control unit measures voltage (VT) and current (CT)
on the feeder side (as shown in Fig. 1) and the results of the
computed power factor is compared to the predetermined
target power factor setting.
3) Using measured and target power factors and the capacitor
information, the control unit determines if one or more capacitor banks need to be switched on or off to bring the actual power factor as close as possible to the targeted power
factor setting.
Based on the above considerations, automatic capacitor
controllers have been developed and marketed for commercial
power system designers and operators. While such controllers
work well for traditional passive loads such as motors, more
and more facility operators have reported frequent capacitor
failures or trips. As a result, reactive power compensation
cannot be achieved.
This problem is caused by the parallel resonance between
the capacitor and the upstream impedance. This resonance is
excited by the harmonic currents produced by modern facility
loads such as office electronics and variable frequency drives.
Industry, therefore, has a strong need for capacitor controllers
that can perform power factor correction on one hand and can
avoid harmonic resonance on the other. In addition, the controller shall not require additional sensors or inputs and can
retrofit current controllers with zero alteration to the existing
facility. In the next section, the resonance problem is explained
in detail.
III. HARMONIC RESONANCE PROBLEM
With any application of capacitor banks, there is always
the risk of resonance. This is due to the interaction of the
banks capacitance with the inductive reactance characteristics of the supply system. Harmonic currents at or near the
resonant frequency can create high harmonic voltages across
the high parallel impedance and the capacitor may not be able
to withstand the resonance voltage, leading to fuse blowing
or capacitor damage [5], [18]. In order to facilitate the description of the resonance problem, Fig. 4 is used to represent
a harmonic-producing commercial facility with a shunt PF
correction capacitor connected at the PCC with the distribution
system. In this figure, the impedance
and current source
represent the linear and nonlinear loads of the facilities,
respectively [20]. Assume that the supply system can be represented by a Thvenin impedance of
, where h

Fig. 5. Frequency response of the combined system and capacitor impedances.

is the harmonic order (or per-unit frequency normalized to the

fundamental frequency).
The total impedance
seen by the harmonic current
source can be determined as
(2)
of the supply system
The inductive reactance
impedance increases and the capacitive reactance
decreases as the frequency increases, or as the harmonic order
increases. At a given harmonic frequency in any system where
a capacitor exists, there will be a crossover point where the
inductive and capacitive reactances are equal
.
Consequently, the total impedance
approaches infinity
and a very high voltage harmonic may result if the commercial
facility harmonic current has a frequency close to
(3)
where
is the system short-circuit level and
is the
capacitor size. The above frequency is called the resonance frequency of the system. In this case, the resonant components
and
are in parallel. The resulting resonance is called parallel resonance. The parallel resonance phenomenon can also
be visualized from a frequency scan plot, as shown in Fig. 5.
This figure illustrates how both system and capacitor reactances
change with the frequency. At the resonance frequency both reactances are equal and total impedance seen from the capacitor
location
will tend to a very large value. It is extremely
unlikely that these two impedances are exactly identical, but
near resonance can be very damaging as well. For example,
consider a system fault level of 250 MVA and a capacitor bank
rating of 10.8 Mvar. Substituting these numbers on (3) yields
the following:

The parallel resonance order of 4.83 is too close to the 5th harmonic order and if any magnitude of 5th harmonic current flows
from the harmonic-producing loads into the power system at the
capacitor bus, the capacitor may not be able to withstand the resonance voltage, leading to fuse blowing or capacitor damage.

LIN et al.: AN INTELLIGENT CONTROL STRATEGY FOR POWER FACTOR COMPENSATION

TABLE I
CAPACITOR LOADING LIMITS ESTABLISHED
THE IEEE STANDARD 1036-1992

1565

IV. PROPOSED CONTROL ALGORITHM

BY

A practical (rule of thumb) way to find out whether parallel

resonance should be a concern is to use (4), which shows how
further away the resonance frequency
should be from any
dominant harmonic frequency
.
(4)
However, the condition given by (4) is not sufficient because
resonance frequency shift can occur due to capacitance deviation [21], for example. Therefore, the final condition to decide
if a certain combination of capacitors should be switched is to
verify if the stress levels on the capacitors bank meet the limits
defined in Table I.
When large levels of voltage and current harmonics are
present, the ratings are quite often exceeded, resulting in
failures. Therefore, the consideration of power capacitor installation should include harmonic resonance analysis at the
design stage. Several solutions can be employed for harmonic
resonance damping as follows [5]:
1) ungrounding grounded-wye capacitors;
2) changing capacitor bank sizes and/or locations;
3) adding a reactor to an existing capacitor bank;
4) adding a filter capacitor;
5) controlling the capacitor switching scheme to avoid
resonance.
The installation of filters can bring unacceptable additional
operational and capital costs to the PF correction scheme and,
furthermore, a detailed harmonic study must be conducted to
ensure that the application of the filters will not cause other side
effects on both the facility and the distribution power system,
such as parallel resonance at harmonic frequencies other than
the one targeted by the filter. We think that focusing on a more
intelligent algorithm to control the capacitor switching scheme
to achieve both power factor correction and resonance avoidance requirements is more of interest for consumers and utilities.
Employ an adaptive control to monitor the harmonic distortion and switch the capacitors to avoid resonance might be
appropriate for commercial loads where there are numerous
switched capacitors coming on and off line randomly [18]. Basically, the idea is to develop a controller that relies on common
low cost sensing devices and does not require additional measurements. As a result, the controller can be constructed as a
retrofitting device to replace existing power factor correction
controllers with little effort. Therefore, a new strategy for the
PF correction capacitors bank controller is proposed in this
paper and it is discussed in detail in the next section. It is
important that the controller is able to achieve both power
factor and resonance avoidance requirements under varying
demand conditions.

Based on the previous discussion, the problem to be solved

is to determine the number of capacitor units to be switched
that can yield the highest power factor for the facility without
causing excessive harmonic stress on the capacitors. Since there
are limited numbers of capacitor combinations, the simplest algorithm is to scan through these combinations and pick the best
candidate. This approach is doable but is not efficient. Another
extreme is to formulate the problem as an optimization problem.
Such an approach complicates the problem, it is not guaranteed
to converge and it might be time-consuming. More importantly,
they cannot be easily implemented into a microcontroller. In this
paper, a practical, efficient, and robust algorithm is proposed.
Easy implementation is one of the main considerations of the
algorithm.
It is important to note that the switching control algorithm
is only one of the components of the controller. The algorithm
needs the system impedance information as input. There is
also a need to detect if a capacitor is being overstressed due to
changing harmonic conditions. Therefore, the proposed controller actually has at least the following three major functions:
A. measurement of the system impedance;
B. detection of resonance condition;
C. determination of capacitor units.
The following subsections provide description of each one of
the above functions.
A. Measurement of the System Impedance
In the previous section, it was discussed that in order to detect
a resonance condition, it is necessary to determine the system
impedance. One important issue is that the system impedance is
not constant, but varies due to loading and topological changes
on the system. Therefore, the following issue must be solved,
how the harmonic resonance condition can be determined for a
time varying load demand and topology?
A number of impedance measurement methods have been
developed, which can be classified into two types: the transients-based methods and the steady-state-based methods. The
transients-based methods inject transient disturbances into
the system. The frequency-dependent network impedances
are extracted from voltage and current transients [14]. The
main problems associated with these methods are the need
for a high-speed data acquisition system and for the source
of disturbances. The steady-state-based methods use pre- and
postdisturbance steady-state waveforms [14][16]. Typical
disturbances are harmonic current injections produced by an
external source or switching of a network component. Since
there are no transients involved, the methods can only determine network impedances at harmonic frequencies. Since
there is no need for a high-speed data acquisition system, the
steady-state method can be implemented with many common,
low-cost power quality monitors and it relies on the common
voltage and current transformer sensors illustrated on Fig. 1.
The simplest form of the steady-state measurement method
involves the switching of a network component at the location
where the network impedance is to be measured. Assuming
that there is a shunt capacitor available for switching, the basic
idea of this method can be summarized as follows [14], [15]:

1566

Fig. 6. Relationship of the resonance frequency

switched shunt capacitor units.

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012

and the number of

1) Record the steady-state waveforms of the capacitor voltages and currents. If the capacitor is not connected, its currents are treated as zero.
2) Changes are then made to the status of the capacitor. For
example, a capacitor unit can be switched on or off to meet
the power factor requirement.
3) The postdisturbance steady-state voltage and current
waveforms are recorded.
4) Discrete Fourier transform (DFT) is applied to the pre- and
postdisturbance waveforms. For each harmonic, the following system equations can be developed:
(5)
(6)
where
and
are the predisturbance hth harmonic current and voltage, and
and
are the
post-disturbance hth harmonic current and voltage.
and
are the internal system voltage and system impedance,
respectively.
5) The system harmonic impedances can be determined from
the above two equations as follows:
(7)
The impedance does not include the switched capacitor
impedance. Practical implementation and field experiences
regarding this method to calculate
can be found on [16].
B. Detection of Resonance Condition
As mentioned before, from the system impedance and the existing capacitor impedance, the resonance frequency can be calculated through (3). For a certain system impedance (or system
fault level,
), the number of capacitor units
that
lead to a harmonic resonance frequency
equal or close
to the dominant harmonic frequencies. Fig. 6 illustrates, for a
particular system impedance, how the harmonic resonance frequency changes as more and more capacitor units are switched
on. For the figure below, the system is represented by a transformer of 1600 kVA with reactance of 6.0% and each capacitor

unit has a capacity of 50 kvar. From this figure, it is clear that the
resonance frequency can be shifted from a harmonic frequency
by changing the number of switched capacitor units.
A practical way to verify if
is too close to any harmonic
frequency
is to apply (4). In the example of Fig. 6, if 11
capacitors are switched ON, the resonance frequency
is too
close to the 7th harmonic order (point A in the figure), therefore, the bank should be increased or decreased. If two more
capacitor are switched (13 in total),
is around 6.4 (point B
in the figure), which is further away from 5th and 7th harmonic
orders. However, it is also necessary to evaluate for the current
combination of capacitor units if its loading conditions meet the
limits specified by Table I.
One can also observe from Fig. 6 that more than one combination of capacitor units can be considered to avoid resonance.
In the following subsections, it will be shown the criteria to select the most appropriate combination. In this paper, each combination refers to a particular number of capacitor units to be
switched on to the circuit.
C. Determination of Capacitor Units
The final step is to determine the number of capacitor units
that can be switched without violating power factor and resonance constraints. From the previous subsection, it is possible
to estimate, from the current system impedance, the combinations of capacitor units that can be switched so that the resonance frequency
is further away from the harmonic frequencies. This can be done through the following steps:
1) The system impedance
calculated from the last
capacitor switching is used as input.
2) Substituting (3) in (4) yields (8), from which it is possible
to determine the combinations of capacitors
that
can be switched. Normally, the dominant harmonic frequencies
are the odd harmonic orders from 3 to 29
(8)
3) Among the combinations found in step 2), it is possible
to determine which combinations (kvar) lead to a power
between utility lower
and upper
factor
limits. This verification can be done as follows:
(9)
4) From the combinations found in step 3), select the combination that lead to minimum switching relative to the current capacitor bank configuration.
5) Calculate the anticipated loading for this combination this
using the indices presented in Table I.
6) If loading indices meet the standard limits, switch the combination, otherwise discard this option from the combinations obtained in step 3) and go back to step 4) to select a
suboptimal solution.
The next subsection combines the previously discussed functionalities into a single flowchart.

LIN et al.: AN INTELLIGENT CONTROL STRATEGY FOR POWER FACTOR COMPENSATION

1567

TABLE II
PARAMETERS OF A TYPICAL DISTRIBUTION POWER TRANSFORMER

TABLE III
ACTIVE AND REACTIVE POWER DEMAND OF THE CASE STUDY FACILITY

Fig. 7. Flowchart of the proposed capacitor bank controller.

D. Proposed Controller Flowchart

Fig. 7 presents the flowchart of the proposed capacitor bank
controller combining the previously discussed functionalities
to achieve acceptable utility power factor level as well as harmonic resonance avoidance. The flowchart is composed of the
following steps:
1) Read voltage and current from VT and CT sensors as illustrated in Fig. 1 and switch on a capacitor unit.
2) Read postswitching voltage and current and calculate harmonic system impedance
from (7).
3) Determine the combinations of capacitor units that meet
the resonance constraint given by (8).
4) Read voltage/current and calculate power factor (PF).
5) If PF is within utility lower
and upper
limits go to step 6), otherwise go to step 7). Normally,
is around 0.92 (depending on each utility comis equal to 1.
pany) and
6) If the capacitor bank loading indices (defined in Table I)
are below standard limits go back to step 4), otherwise go
to step 7).
7) Using (9), determine the combinations of capacitor units
from those determined in Step 3) that leads to PF between
and
.
8) Select the combination that leads to minimum switching
relative to the current capacitor bank configuration.

9) If the bank loading indices (defined in Table I) are below

the standard limits switch the selected combination, otherwise go to step 11).
10) After the switching, if the capacitor bank loading indices
are below standard limits go back to step 2), otherwise go
to step 11).
11) Remove the combination selected in step 8) from those
determined in step 7). Go back to step 8).
One can observe from the flowchart of Fig. 7 that the proposed controller is simple and practical and has other advantage
as follows:
It does not require additional measurements relying on
common voltage and current transformer sensors.
The controller does not require the installation of reactors
and filters.
Furthermore, the controller not only checks if the harmonic
resonance frequency
is further away from any harmonic frequency, but the capacitor stress levels are also
verified to ensure the selection of the most appropriate
combination of capacitor units.
The controller takes into account the time-varying system
conditions to determine both the power factor and the resonance condition.
The proposed controller also preserves the objective of
conventional controllers, which is to achieve the highest
facility power factor.
In the following section, some case studies are conducted to
evaluate how the proposed controller performs.
V. CASE STUDIES
A case study is analyzed in this section to illustrate the
proposed control method. Suppose the parameters of a typical
power transformer feeding a commercial building as shown in
Table II.

1568

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012

TABLE IV
HARMONIC CURRENT LIMITS OF THE INVESTIGATED SYSTEM
ACCORDING TO IEEE 519-1992 [20]

Fig. 9. Harmonic resonance frequencies

tions of switched on capacitors (n).

Fig. 8. Equivalent circuit of the transformer and switched on shunt capacitors.

obtained for different combina-

TABLE V
COMBINATIONS OF SWITCHED ON CAPACITORS LEADING TO HARMONIC
RESONANCE FREQUENCY CLOSE TO TYPICAL HARMONIC ORDERS

From the parameters of the power transformer, the secondaryside short-circuit current at the PCC is

The facility is equipped with a PF correction scheme composed of units of shunt capacitor of 20 kVAR each one. The demand of active and reactive power from the facility is provided
in Table III. In this table, the maximum demand load current
(fundamental frequency) is

Therefore, the ratio of

to is approximately equal to 24.
According to the standard IEEE519-1992 [20], the harmonic
current limits corresponding to each harmonic order could be
calculated, shown in Table IV.
It is known that the system impedance mostly depends on the
impedance of the power transformer for LV distribution system.
According to the parameter of the power transformer and shunt
capacitors, the equivalent circuit of the transformer and shunt
capacitor could be drawn in Fig. 8, in which, the values are in
p.u. units and base value is 1000 kVA.
and
are the resistance and inductance of the power transformer, is the harmonic order, is the number of the switched on capacitors
and
are the resistance and capacitance of the combination
of switched on capacitors.
From Fig. 8, the resonance frequency corresponding to different combinations of switched on capacitors can be determined, as shown in Fig. 9.
Fig. 9 shows that when the number of switched on capacitors
is 6, 8, 12, 19, or 20, the resonance frequency does not meet
the constraint defined by (8), which means that the resonance
frequency is so close to the characteristic harmonics that the

Fig. 10. Total impedance

obtained for different harmonic orders
and combinations of switched on capacitors
.

capacitors could be damaged. Table V shows the values of the

resonance frequency for these combinations.
Taking the base impedance
as 0.23 ohms, the total
impedance
from Fig. 8 can be calculated for different
and , as shown in Fig. 10.
Table IV gives the current limits corresponding to each harmonic order in LV distribution system. If multiplying the current
limits and the total impedance
, the voltage limit in each
harmonic order can be determined. Therefore, the total RMS
working voltage of the capacitors can be expressed as
(10)
is the harmonic voltage
where is the fundamental voltage,
and is the maximum harmonic order. The maximum working
voltage of the capacitors for different combinations of switched
on capacitors considering the case study is shown in Fig. 11.

LIN et al.: AN INTELLIGENT CONTROL STRATEGY FOR POWER FACTOR COMPENSATION

1569

Fig. 13. Number of the switched on capacitors and corresponding power factors before and after compensation using proposed control strategy.

Fig. 11. Total working voltage of capacitor bank

tions of switched on capacitors (n).

for different combina-

B. PF Correction With the Proposed Control Strategy

Supposing the system harmonic impedance could be precisely measured by using the capacitor switching, for simplicity,
the system harmonic impedance is assumed to be

Fig. 12. Number of the switched on capacitors and corresponding power factor
before and after compensation using conventional control strategy.

It is observed that the working voltage of the capacitors exceed the 110% when the number of the switch-on capacitors is
equal to 6, 8, 12, or 20.

A. PF Correction With Traditional Control Strategy

Assume the required threshold PF range for the facility is
. According to the flowchart of the traditional
capacitor controller shown in Fig. 3, the required bank size necessary to meet this PF range can be calculated from (1). Fig. 12
shows the existing and corrected power factor with the traditional strategy and the number of switched on capacitor units
considering the demand profile provided in Table III.
It is obvious that, the number of the switched on capacitors
is equal to 6 for sample points 1 and 2 and 8 for sample points
3, 4, and 5. From Fig. 11, one can observe that these combinations cause the bank working voltage to exceed the 110% limit
defined in [5].

For the first sample point given in Table III, the number of
switched on capacitors
can be firstly determined as 6 to meet
. From the system impedance
, the
resonance frequency is calculated as
, which does
not respect the requirement defined by (8) and means that
cannot be selected as 6. Instead, if
, the new PF can be
estimated as
, which is acceptable. Similarly
for other cases, the corrected PF can be calculated for all the
sample points given in Table III, as shown in Fig. 13.
One can observe from both Figs. 13 and 11 that the new control strategy avoids the harmonic resonance problem, which, in
turn, it can prevent damage to the capacitor bank of the facility.
VI. CONCLUSION
This paper presents a new control strategy for power factor
compensation on distorted low voltage power systems. The proposed strategy can perform power factor correction without exciting harmonic resonance under varying demand conditions.
Practical and robust control algorithms are proposed for the purpose of easy implementation in a microcontroller. In addition,
the controller relies on common low cost sensing devices and
does not require additional hardware circuits. As a result, the
proposed controller can be constructed as a retrofitting device
to replace existing power factor correction controllers with little
effort and low cost. Analysis of representative case studies validates the proposed strategy and illustrates how the proposed
controller performs.
REFERENCES
[1] IEEE Recommended Practice for Electric Power Systems in Commercial Buildings, ANSI/IEEE Std. 241-1990, Gray Book.
[2] T. M. Blooming and D. J. Carnovale, Capacitor application issues,
IEEE Trans. Ind. Appl., vol. 44, no. 4, pp. 10131026, Jul. 2008.

1570

[3] A. Novitskiy and H. Schau, Energy saving effect due to the voltage reduction in industrial electrical networks, in Proc. Elec. Power Quality
Supply Rel. Conf. (PQ), Jun. 2010, pp. 6772.
[4] T. A. Short, Electric Power Distribution Handbook. Boca Raton, FL:
CRC, 2004.
[5] IEEE Guide for Application of Shunt Power Capacitors, IEEE Std.
1036-1992, 1992.
[6] J. Arrillaga and N. R. Watson, Power System Harmonics, 2nd ed.
Hoboken, NJ: Wiley, 2003.
[7] A. F. Zobaa, Capacitive compensation at nonsinusoidal buses based
on IEEE Std. 18-1992, IEEE Trans. Power Del., vol. 18, no. 4, pp.
15621563, Oct. 2003.
[8] M. M. Abdel Aziz, E. E. Abou Elzahab, and A. M. Ibrahim, Sizing of
capacitors to optimize the power factor at non-sinusoidal frequencies,
Elec. Power Syst. Res., vol. 64, no. 1, pp. 8185, Jan. 2003.
[9] R. H. Simpson, Misapplication of power capacitors in distribution systems with nonlinear loads-three case histories, IEEE Trans. Ind. Appl.,
vol. 41, no. 1, pp. 134143, Jan. 2005.
[10] L. Herman and I. Papic, Hybrid active filter for power factor correction and harmonics elimination in industrial networks, in Proc. IEEE
Electr. Power Energy Conf. (EPEC), Oct. 2011, pp. 302308.
[11] J.-C. Wu, H.-L. Jou, K.-D. Wu, and N. C. Shen, Power converter-based method for protecting three-phase power capacitor from
harmonic destruction, IEEE Trans. Power Del., vol. 19, no. 3, pp.
14341441, Jul. 2004.
[12] M. M. Abdel Aziz, E. E. Abou El-Zahab, A. M. Ibrahim, and A. F.
Zobaa, LC compensators for power factor correction of nonlinear
loads, IEEE Trans. Power Del., vol. 19, no. 1, pp. 331336, Feb.
2004.
[13] P. Jintakosonwit, S. Srianthumrong, and P. Jintakosonwit, Implementation and performance of an anti-resonance hybrid delta-connected capacitor bank for power factor correction, IEEE Trans. Power Del., vol.
22, no. 6, pp. 25432551, Nov. 2007.
[14] A. Robert and T. Deflandre, Guide for assessing the network harmonic
impedances, , CIGRE 36.05, Working Group CC02 Rep., Mar. 1993.
[15] A. de Oliveira, J. C. de Oliveira, J. W. Resende, and M. S. Miskulin,
Practical approaches for AC system harmonic impedance measurements, IEEE Trans. Power Del., vol. 6, pp. 17211726, Oct. 1991.
[16] W. Xu, E. E. Ahmed, X. Zhang, and X. Liu, Measurement of network
harmonic impedances: Practical implementation issues and their solutions, IEEE Trans. Power Del., vol. 17, no. 1, pp. 210216, Jan. 2002.
[17] J. C. Whitaker, AC Power Systems Handbook, 2nd ed. Boca Raton,
FL: CRC, 1999.
[18] R. C. Dugan, M. F. McGranaghan, and H. W. Beaty, Electrical Power
System Quality. New York: McGraw-Hill, 1996.
[19] R. Tinggren, Y. Hu, L. Tang, H. Mathews, R. Tyner, R. Heinemeyer, H.
Breder, and C. Cereda, Power factor controllerAn integrated power
quality device, in Proc. IEEE Transm. Power Distrib. Conf., 1999.

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 3, SEPTEMBER 2012

[20] Task force on harmonics modeling and simulation, modeling and simulation of the propagation of harmonics in electric power networks, part
I: Concepts, models, and simulation techniques, IEEE Trans. Power
Del., vol. 11, no. 1, pp. 452465, Jan. 1996.
[21] IEEE Standard for Shunt Power Capacitors, IEEE Std. 182002, 2002.
[22] IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems, IEEE Std 5191992, 1993.

Shunfu Lin (M12) received the B.S. degree in applied physics and the Ph.D.
degree in nuclear technology and application from the University of Science
and Technology of China in 2002 and 2007, respectively.
He worked for the Corporate Technology of Siemens Limited China as a Research Scientist in power monitoring and control of low-voltage distribution
system from July 2007 to September 2009. He was a Postdoctoral Fellow at the
Department of Electrical and Computer Engineering of University of Alberta,
Canada from October 2009 to October 2010. He is currently a Distinguished
Professor at the Shanghai University of Electric Power, China. His research interests include power quality and power measurement.

Diogo Salles (S04) received the B.Sc. and M.Sc. degrees in electrical engineering from the University of Campinas, Campinas, Brazil in 2006 and 2008,
respectively, where currently he is working toward the Ph.D. degree.
From 2010 to 2011, he was a visiting Ph.D. student at the University of Alberta, Edmonton, Canada. His research interests are power quality and analysis
of distribution systems.

Walmir Freitas (M02) received the Ph.D. degree in electrical engineering from
the University of Campinas, Campinas, Brazil in 2001.
He is currently an Associate Professor, University of Campinas. His areas of
research interest are analysis of distribution systems and distributed generation.

Wilsun Xu (M90SM95F05) received the Ph.D. degree from the University

of British Columbia, Vancouver, Canada, in 1989.
Currently, he is a Professor and a NSERC/iCORE Industrial Research Chair at
the University of Alberta, Edmonton, Canada. His research interests are power
quality and distributed generation.