This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL
1
Abstract —This paper proposes a coordinated voltage regulation scheme for on-load tap changer (OLTC) and renewable distributed generation (DG) units to provide a proper voltage regulation for active distribution systems. The main motivation of applying fuzzy logic is that it can deal with environments of imperfect information, and thus, it can reduce communication requirements. The proposed regulation scheme consists of three fuzzy-based control algorithms. The first control algorithm is proposed for the OLTC such that it can mitigate the effect of DG units on the voltage profile. The second control algorithm is proposed to provide a DG reactive power sharing, in order to relax the OLTC tap operation.
The third control algorithm aims to partially curtail DG active powers to restore a feasible solution from the OLTC prospective.
The proposed fuzzy algorithms have the advantage of providing proper voltage regulation with relaxed tap operation, utilizing only the estimated system minimum and maximum voltages. Moreover, it avoids numerical instability and convergence problems associated with centralized approaches, as it does not require to run an optimization algorithm. Real-time simulations are developed to show the effectiveness of the proposed fuzzy algorithms on a typical distribution network, using OPAL-RT real-time simulator.
Index Terms —Active distribution systems (ADS), distributed generation (DG), fuzzy control, on-load tap changer (OLTC), real-time simulator (RTS), voltage regulation.
I. I NTRODUCTION
T HE integration of renewable distributed generation (DG) units alters the distribution systems from their passive structures, with unidirectional power flow, toward active distribution systems (ADS), with multidirectional power flow [1],
[2]. While numerous benefits are associated with the change toward ADS, such transition represents many challenges [3].
Voltage regulation is considered as one of the main challenges that are accompanied with high penetration of renewable energy sources. The intermittent nature of renewable energy sources, such as wind and solar energy, can significantly change the system voltage profile and interact with the conventional
Manuscript received April 14, 2014; revised May 28, 2014; accepted June 7,
2014.
M. A. Azzouz and E. F. El-Saadany are with the Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON N2L 3G1,
Canada.
H. E. Farag is with the Department of Electrical and Computer Engineering,
York University, Toronto, ON M3J 1P3, Canada.
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/JSYST.2014.2330606
control of on-load tap changers (OLTCs) [4]–[6]. This interference may lead to overvoltage, undervoltage, increasing in the system losses, and abnormal wear of OLTC due to excessive taping actions.
Different control schemes are proposed in literature to overcome the effects of DG units on the voltage regulation. Traditionally, most of distribution network operators (DNOs) require all DG units connected to a distribution system to operate in a constant-power-factor control (PFC) mode [7]. The authors in
[8] proposed a local reactive power control approach for voltage rise mitigation in ADS. In this approach, each DG absorbs a reactive power to compensate the effect of its injected active power on the voltage rise. The authors in [9] suggested the use of droop-based active power curtailment for voltage rise prevention in radial low-voltage feeders. Two intelligent local controllers are proposed in [10] to mitigate the impacts of DG on the voltage profile in weak distribution networks. The first controller uses a set of rules to switch between PFC and voltage control modes, whereas the second controller is designed based on a fuzzy inference system that adjusts the reference setting of the power factor in response to the DG terminal voltage. A local voltage control scheme for multiple DG units in a distribution feeder is proposed in [11]. In [12], the authors introduced two local controllers to regulate the voltage profiles at buses, where wind power distributed generators are connected. The first method relies on the sensitivity analysis, and the other is based on a fuzzy reactive power compensation to support DG voltages. On the other hand, the authors in [13] focused on the voltage regulation from a dynamic point of view rather than the steady-state point of view. An adaptive PI control algorithm is proposed, based on local voltage variations and the nonactive power theory, to guarantee a satisfactory dynamic voltage response. Due to miscoordination between DG and OLTC local controllers, the following drawbacks are introduced: 1) high stress on OLTC, particularly in case of renewable power sources; and 2) the energy capture from DG units is not maximized.
Some of these drawbacks have been recently tackled in [14]–
[16] using coordinated control schemes. In [14], the authors suggested an agent-based algorithm for the DG reactive power dispatch to provide proper voltage regulation with less communication requirements, as compared with centralized approaches. However, the coordination between the DG reactive power support and the OLTC is not considered. Moreover, no solution is provided when the DG reactive power reaches
1932-8184 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL 2 its limit. A coordinated control between distributed energy storage systems (ESS) and OLTC is proposed in [15]. The proposed method relies on minimizing the reverse power flow during light loadings by activating ESS chargers. This method assumes that there is an ESS attached to each DG; however, this is not a common practice. In addition, it does not take into consideration the case when the ESS is fully charged. In [16], the authors presented a new voltage estimation methodology to estimate the maximum and minimum voltages for multifeeder distribution systems. Nonetheless, the OLTC is assumed to be the only voltage control device, an approach that can stress the
OLTC. In addition, relying solely on the OLTC can lead to an infeasible solution when the difference between the system maximum and minimum voltages exceeds the standard regulation band.
In [17], the authors applied an optimal power flow algorithm to show that both intelligent distributed and centralized voltage control can achieve similar results and obtain better performance than the PFC. In [18]–[20], the authors proposed optimal voltage regulation techniques utilizing the DG voltage support. The proposed algorithm in [18] reduces the computational complexity using a sequential convex programming, but it provides near optimal solution, as compared with a global solver. In [19], the distribution network is divided into subnetworks using ε decomposition to reduce the size and complexity of the optimization problem; however, the coordination with OLTC is not considered. In [20], the authors introduced an optimal coordination technique, which minimizes the tap operation utilizing the photovoltaic (PV) reactive power, based on a 24-h-ahead forecasting for both load and DG powers.
However, PV active power curtailment is not considered when
DG reactive powers are limited. Given the high uncertainty of the renewable power generation, conventional 24-h-ahead optimization approaches are highly associated with forecasting errors. Hence, the main drawbacks of the optimization-based voltage regulation schemes (centralized approaches) are in general: 1) they need real-time measurements of load and DG powers; 2) they require updating the system parameters with the change in its topology; and 3) they are accompanied with high complexity and convergence problems [21].
With high penetration of renewable energy resources, a proper online (real-time) control scheme is required to mitigate the challenges of voltage and reactive power control in ADS.
Thus, new expert-based (logic) distributed control schemes have been recently proposed in the literature [4], [16], [21].
Nonetheless, the logic-based techniques represent the controller inputs, e.g., voltage magnitudes, as crisp values, and they fall short in considering the uncertainties arise due to renewable energy resources. Furthermore, previous works do not consider the self-objectives of the devices participating in voltage regulation, e.g., OLTC needs to relax its tap operation and DG owners need to reduce their active power curtailments. Thus, the main contribution of this paper is to propose a fuzzy-based coordination between the OLTC and DG units to: 1) mitigate the interaction between the OLTC and DG units, which can lead to overvoltages and undervoltages; 2) reduce the excessive tap operation of the OLTC; and 3) avoid unnecessary DG active power curtailments.
The proposed fuzzy controllers have the advantage of reducing the communication burden compared with distributed and centralized techniques, as they rely only on the estimated system minimum and maximum voltages. In addition, they can map nonlinear, multivariable, and heuristic relations between their inputs and output. The proposed algorithms also avoid the numerical instability and convergence problems that are associated with other centralized control schemes, which run power flow algorithms in each time step, particularly in case of low X/R ratio. Finally, the proposed coordinated voltage regulation is validated in real-time using OPAL-RT real-time simulator (RTS).
The rest of this paper is organized as follows. Section II illustrates the need of applying fuzzy logic in the voltage regulation of ADS. The proposed fuzzy-based OLTC control is explained in Section III, whereas the proposed fuzzy-based DG voltage supports are discussed in Section IV. In Section V, the proposed coordination scheme between all fuzzy controllers is clarified.
Real-time simulations are provided in Section VI. Section VII concludes this paper and summarizes its main contributions.
IN V OLTAGE R EGULATION
According to Zadeh [22], fuzzy logic can be viewed as a tool to emulate human mental capability, because it can make rational decisions in an environment of imprecision, uncertainty, and incompleteness of information—in short, in an environment of imperfect information. Voltage regulation in ADS can be considered as an environment of imperfect information, because the proposed fuzzy controllers utilize only the system maximum and minimum voltages to determine the
DG reactive power support and active power curtailment. To prove the claim that the environment under study is of imperfect information, and thus, there is a need to apply fuzzy logic, the following analysis is conducted. The sensitivity matrix S can be calculated from the Jacobian matrix of the Newton power flow as follows: with
II. A
PPLICATION OF
F
UZZY
L
OGIC
V
Δ
Δ
θ
V r ( n )
=
S
θP
S
V P
S
θQ
S
V Q
= V o ( n )
+ S
V Q x
Q
Δ
Δ
P
Q
+ S
V P x
P
.
⎧
⎨ x
Q
= Q r
−
Q o
= x
DG
1
Q
, x
DG
2
Q
, . . . , x
DG
N g
Q
T
T x
P
= P r
− P o
= x
DG
1
P
, x
DG
2
P
, . . . , x
DG
N g
P
(1)
S
To calculate the DG reactive and active powers, which are needed to bring an initial voltage V o at a certain node n to a reference voltage V r
, the following linear relation can be used based on the sensitivity matrix S [19]:
(2) where ( P o
, Q o
) and ( P r
, Q r
) are DG active and reactive powers before and after voltage regulation, respectively; and N g is the total number of DG units. Suppose that the system maximum

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AZZOUZ et al.
: VOLTAGE REGULATION FOR NETWORKS INCORPORATING PENETRATION OF RENEWABLE SOURCES 3 voltage V voltage V sys max sys min happens at bus occurs at bus n
1 n
, whereas the system minimum
2
. The main objective of the
DG voltage support is determining the DG reactive and active powers that bring both V sys max and V sys min to the standard voltage limits V
Upper and V
Lower
, i.e., 1.05 and 0.95 pu, respectively.
Applying (2) at n
1 and n
2 leads to
V r ( n
1
)
V r ( n
2
)
= V
Upper
= V
Lower
= V o ( n
1
)
= V o ( n
2
)
+ S
V Q x
Q
+ S
V Q x
Q
+ S
V P x
P
+ S
V P x
P
.
(3)
As shown in (3), we have two equations in 2
×
N g unknowns, assuming that the system maximum and minimum voltages are estimated. Thus, this problem cannot be deterministically solved due to the lack of system information. Alternatively, such problem can be solved using a centralized control scheme (which runs an optimization algorithm), an approach that requires access to all nodes’ active and reactive powers and is susceptible to convergence problems. In terms of acquired information, the proposed fuzzy controllers require only the system minimum and maximum voltages to provide proper voltage regulation. For instance, the authors in [23] have proposed a fuzzy logic controller for the OLTC to have an adaptive nature, which can deal with the load uncertainty. However, the proposed fuzzy controllers do not consider the DG contribution to the voltage regulation problem, which adds to its uncertainty and complexity. The proposed fuzzy controllers also have the following advantages over conventional controllers, such as hysteresis or PI controllers.
1) They are multivariable controllers, as they accept two inputs (i.e., the system maximum and minimum voltages).
Accepting multi-inputs increases the controller degree of freedom and allows for emulating an adaptive reference.
This feature is not available with conventional controllers as they accept a single input, i.e., the error between a fixed reference and a regulated variable. For instance, a voltage violation may happen if the OLTC controller regulates the voltage at a fixed target point, because the voltage trend from the substation to the feeder terminal is not descending in the presence of DG units.
2) They can map nonlinear and heuristic relations between their inputs and output, a feature that cannot be provided by conventional controllers.
III. F UZZY -B ASED OLTC C ONTROL
The OLTC can vary the tap position n
( t ) voltage compensation) to
± N max from zero (no
(maximum voltage compensation). Typically, the process of tap changing involves two time delays: 1) a controller time delay T d
, which is intentionally introduced to avoid tap changing during fast voltage transients; and 2) a mechanical time delay T m due to the motor drive mechanism of the OLTC. The mechanical time delay T m has a constant value, which usually varies from 3 to 10 s, whereas the controller time delay T d commonly depends on the voltage error Δ V and controller dead band DB [24], [25]
T d
= τ
0
DB
|
Δ V
|
(4)
Fig. 1.
OLTC control. (a) Conventional controller. (b) Proposed FOC.
where τ
0 is a constant selected based on the tap changing mechanism.
Currently, most OLTCs employ line drop compensators
(LDCs), as shown in Fig. 1(a). The LDC measures the secondary voltage V
1 and current through the OLTC I
OLTC to estimate the voltage drop at a target point V k estimated value of V k
, i.e., V k
, regulating the
, by adaptively adjusting V
1
. However, the integration of DG units changes the voltage profile significantly and complicates the voltage regulation due to the following.
1) The voltage trend is not descending from the substation to the feeder terminal; thus, a fixed target point is not valid anymore.
2) The voltage estimation, based on local measurements, becomes worse in the presence of highly intermittent renewable sources, such as wind and PV.
Therefore, a local voltage estimation with fixed target point operation can lead to improper decisions of OLTCs, which may result in overvoltage, undervoltage, and excessive wear and tear of OLTCs. Here, a fuzzy-based OLTC controller (FOC) is proposed to mitigate the drawbacks of the conventional OLTC control. To provide proper estimation of V sys max and V sys min
, communication links are required. In this paper, the state estimation algorithm in [16] is adapted to estimate V sys max and V sys min
.
Fig. 1(b) shows the block diagram of the proposed OLTC controller. The proposed fuzzy control adapts the voltage error
Δ V such that V sys max and V sys min lie within the acceptable range, and hence, it can guarantee a proper voltage regulation at all buses. The proposed FOC imitates the behavior of DNOs, i.e., vary the OLTC tap setting in order to keep the system voltages within the standard limits. In other words, the proposed FOC emulates an adaptive reference behavior for the OLTC control.
The membership functions (MFs), which are assigned for V and V sys min
MFs, namely, “ V. Low ,” “Low,” and “ Normal ” for V
“ Normal min sys max
, are shown in Fig. 2. Each input is assigned three
,” “ High ,” “ V. High ” for V sys max sys
; and
, where letter “ V ” stands for very. The consequent part, Δ V , is assigned five singleton

4
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL
Fig. 2.
Input MFs of the proposed FOC.
TABLE I
R ULE B ASE OF THE P ROPOSED FOC
Fig. 3.
Proposed DG voltage support algorithms. (a) FQC. (b) FPC.
MFs, namely, “PL,” “P,” “Z,” “N,” and “NL,” where letters “P,”
“L,” “Z,” and “N” stand for positive, low, zero, and negative, respectively. The output singleton values are 2, 1, 0,
−
1, and
−
2 for “PL,” “P,” “Z,” “N,” and “NL,” respectively. The scaling factor ( K c
) of the FOC is tuned in order to normalize the fuzzy output with respect to the maximum expected deviation in Δ V .
The rule base of the FOC is illustrated in Table I, in which the “MIN” function is used for the fuzzy “AND” operator. The final crisp output of the controller is derived using the weighted average defuzzification method. The shaded rules in Table I represent the infeasible scenarios at which the OLTC fails to provide a proper action, and thus, it should hold its tap position.
IV. F UZZY -B ASED DG V OLTAGE S UPPORT
Typically, OLTCs are considered the main devices that are responsible for voltage regulation in distribution systems. The application of the OLTC in ADS introduces two challenges.
The first challenge is the excessive wear and tear of OLTCs, particularly when ADS has a high penetration of variable power sources. Second, OLTCs fail to provide proper tap settings when both V sys max and V sys min violate their specified limits, simultaneously. On the other hand, DG units are incorporated in the voltage regulation by: 1) reactive power supports; and 2) active power curtailments. Here, two fuzzy-based control algorithms are proposed for the DG reactive power support and active power curtailment.
A. Fuzzy-Based Reactive Power Control
Fig. 3(a) shows the block diagram of the proposed fuzzybased reactive power control (FQC), which is dedicated for all
DG units connected to a certain feeder. The FQC receives the feeder minimum V its output Δ V
F f min and maximum V f max voltages to generate
, which is then integrated to produce V
F
. The main reason of the integration is to avoid resetting DG reactive powers after recovering a voltage violation, which can lead to unnecessary tap operation. The FQC algorithm utilizes the same rule base of the FOC, as shown in Table I.
To achieve a proper reactive power sharing among all DG units at the same feeder, the reactive power reference Q
∗
DG
( i ) is determined via multiplying V
F by a voltage sensitivity factor
K
Q
, which is proportional to S
V Q ( i,j )
, where i is the DG local bus, and j is the bus at which V f min or V f max occurs. Voltage sensitivity analysis can determine the most effective nodes and amounts of DG reactive powers to support the grid voltage. A voltage sensitivity matrix S is calculated by solving load flow equations and determine the inverse of the Jacobian matrix [26].
In this paper, the modified Newton–Raphson load flow is used to calculate the sensitivity factors, since it has the following merits: 1) it can avoid the convergence problems associated with the conventional Newton–Raphson load flow when the distribution network has low X/R ratio; and 2) it can be applied in meshed networks [27]. The DG reactive power Q
∗
DG
( i ) is limited by the DG power factor and reactive power capability curves discussed in [28]. The reactive power capability curves set the DG reactive power limits based on the DG power rating and dc-link voltage. Hence, Q Limit
DG( i ) can be calculated as
Q
Limit
DG( i )
= min Q pf
DG( i )
, Q
S
DG( i )
, Q
V
DG( i ) with
⎧
⎪
⎨
Q pf
DG( i )
Q S
DG( i )
⎪
⎩ Q V
DG( i )
= P pf
DG( i ) tan cos
−
1
= S 2
DG( i )
− P 2
DG( i )
=
V max
DG( i )
V
DG( i )
X
( i ) pf
DG
( i )
2
−
P 2
DG( i )
−
V 2 g ( i )
X
( i )
(5)

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AZZOUZ et al.
: VOLTAGE REGULATION FOR NETWORKS INCORPORATING PENETRATION OF RENEWABLE SOURCES 5 where P
DG( i ) is the DG active power at bus i , S
DG( i ) is the
DG rating, pf
DG
( i ) is the DG power factor, V
DG
( i ) point of common coupling voltage, V max
DG
( i ) is the DG is the maximum
DG converter voltage, which depends on the dc-link voltage, and X
( i ) represents the total reactance of the DG interfacing transformer and filter.
When either V f min or V f max violates the standard limits, the
FQC is energized to increase or decrease Δ V
F
, such that the voltage violation problem can be mitigated. For instance, if a certain feeder suffers from undervoltage, the FQC generates a positive Δ V
F
. Hence, all DG units inject reactive power, based on their voltage sensitivities, in order to boost V f min to the standard lower limit and vice versa. The proposed FQC has the following advantages:
1) relieving the excessive OLTC operation;
2) reducing the possibility of having infeasible solutions, as it shrinks the gap between V sys min and V sys max
; and
3) increasing the reactive power capability that is required to mitigate the problem of voltage violation, where all DG units at a certain feeder participate in solving the problem, irrespective of the locations, at which the voltage violations occur.
It is noteworthy that, the proposed fuzzy algorithms are designed to deal with radial networks, which are the common topology of distribution networks. The proposed FOC is generic and it can be applied in both radial and meshed networks, as it is system configuration independent (i.e., at any state(s), a change in the substation voltage magnitude has the same effect on all downstream buses, regardless of their topology).
Nonetheless, the proposed FQC needs to be modified in case of meshed distribution networks, because the DG influence is not limited to its feeder but can propagate to other feeders.
Thus, the typical inputs of the proposed FQC, i.e., V
V f max
, should be replaced by V min networks.
sys and V sys max f min and in case of meshed
B. Fuzzy-Based Active Power Curtailment
When the difference between V sys max and V sys min is greater than the difference between the standard upper and lower voltages, i.e., V
Upper and V
Lower
, then the solution becomes infeasible from the OLTC prospective. The problem would be worse when the FQC is unable to provide the required reactive power due to the DG reactive power limitations. In such a case, the OLTC severely fluctuates and the solution can be only provided by two means: 1) DG active power curtailment to decrease V sys max
; or 2) load shedding to increase V sys min
. Due to the utility commitments, the second option is not proposed. In this paper, a fuzzy-based DG active power curtailment (FPC) is proposed to provide shared DG active power curtailments based on the DG participation in the overvoltage problem. As shown in Fig. 3(b), the proposed FPC has two inputs and one output. The FPC inputs are Δ V sys and Δ V
DG
( i ) max
, which are defined as
Δ V sys
= V sys max
− V sys min
Δ V
DG
( i ) max
= V
DG
( i )
− V
Upper
.
(6)
Fig. 4.
Input MFs of the proposed FPC.
TABLE II
R ULE B ASE OF THE P ROPOSED FPC
The FPC generates ϕ ( i ) , which is then rescaled based on a sensitivity factor K
P to get the DG active power curtailment factor Φ ( i ) . The sensitivity factor K
P is proportional to
S
V
V P ( i,j ) sys max
, where i is the DG local bus, and j is the bus at which occurs. Therefore, all DG units share the active power curtailments based on their contributions to the overvoltage problem. In addition, the second input, i.e., Δ V
DG
( i ) max a measure of the DG contribution to a large Δ V
DG
( i ) max
Δ V sys
, is used as
. For a certain DG, implies that such DG has a relatively more impact on Δ V sys , and thus, its power curtailment should be more and vice versa. Similarly, when Δ V sys is large, the power curtailment should be more and vice versa. It is worth noting that the active power curtailment is applied only on DG units, whose feeder maximum voltage is equal to V
As shown in Fig. 4, both Δ V sys and Δ V sys max
.
DG
( i ) max have three triangle MFs, namely, normal ( N ) , high ( H ) , and very high
( V H ) ; whereas ϕ ( i ) is assigned five singleton values, namely, unity ( U ) , High ( H ) , medium ( M ) , low ( L ) , and zero ( Z ) . The singleton values of the output MFs are 1, 0.75, 0.5, 0.25, and 0 for “U,” “H,” “M,” “L,” and “Z,” respectively. The FPC rule base is summarized in Table II. For a certain DG unit, Φ ( i ) multiplied by its active power reference P the updated reference P has normal Δ V
DG
( i ) max
DG
∗
( i ) ref
AND
DG
( i ) ref is to determine
. According to Table II, IF a DG
Δ V sys
THEN P
DG
( i )
A is unity, which implies no active power curtailment.
C. Application of ESS in Voltage Regulation
Typically, distributed ESS are interfaced through power electronic converters, similar to DG units [29], [30]. The

6
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL interfacing converters have two main tasks, i.e., regulating both active and reactive powers. Hence, ESS can be incorporated in the reactive-power voltage support by applying the proposed
FQC. Unlike renewable DG units, ESSs are dispatchable power sources and have bidirectional power flow (i.e., charging and discharging). During discharging, ESSs act as DG units; hence the proposed FPC can be applied, without any modification.
During the charging, ESSs are treated as loads. Thus, the ESS participates in the undervoltage rather than the overvoltage. In such a case, the proposed FPC needs to be slightly modified to fit the ESS. During the charging mode of operation, the second input of the proposed FPC, i.e., Δ V
DG
( i ) max by Δ V
ESS
( i ) min
, which is defined as
, should be replaced
Δ V
ESS
( i ) min
= V
Lower
−
V
ESS
( i )
(7) where V
ESS
( i ) is the ESS point of common coupling voltage.
It is noteworthy that the proposed FPC has the same inference system, i.e., rule base and input–output MFs, during ESS charging and discharging.
V. C OORDINATION S CHEME B ETWEEN THE P ROPOSED
F UZZY -B ASED A LGORITHMS
The coordination between the proposed fuzzy controllers is essential to provide an efficient operation of the OLTC with minimum DG active power curtailment. The flowchart of the proposed coordination scheme, which manages the three fuzzy controllers, is shown in Fig. 5. First, the maximum and minimum voltages for each feeder and for the entire distribution system are estimated. In case of voltage violation, the FQC is activated to solve the voltage violation problem by injecting or absorbing reactive power under prespecified reactive power limits. After activating either FQC or FPC, a time delay Δ t conv is introduced to ensure that all DG converters reach the desired reactive or active power references. This time delay depends on the settling times of the converter primary controllers, which can vary from 50 to 150 ms [31]. In this paper, Δ t conv is assumed to be 200 ms, whereas the total update time of the proposed coordination algorithm Δ t is 5 min. If the proposed
FQC cannot mitigate the voltage violation problem, the OLTC takes action by activating the FOC when a feasible solution exists. To guarantee a feasible solution using the FOC, the following condition has to be satisfied:
Δ V sys
VI. R EAL
≤
Δ V max
−
Δ a
-T IME S IMULATIONS
(8) where Δ a is the step change of the transformer turns ratio, and
Δ V max is the difference between V
Upper and V
Lower
.
Δ a is considered in (8) to allow a margin of change up or down for the
OLTC. If (8) is not satisfied (infeasible solution exists), the FPC is energized in order to decrease V sys max
Δ V sys
; resulting in decreasing to restore a feasible solution for the FOC.
The proposed fuzzy algorithms are modeled by the RTS of
OPAL-RT using the SimPowerSystems blockset and ARTEMiS
Fig. 5.
Proposed coordination cycle.
plug-in. The RTS provides a parallel computation, which allows distributing large and complex models over several processors to perform powerful computations with high accuracy and low-cost real-time execution. The real-time simulations are considered to prove the applicability of the fuzzy algorithms as prototype controllers, which is an important stage before practical implementations. The RTS is used to perform two main functions: rapid control prototyping (RCP) and hardwarein-the-loop (HIL) application. The RCP realization is used to implement the proposed fuzzy controllers in order to mimic actual voltage regulators. Compared with actual voltage regulators, RCP controllers are more flexible, easier to debug, and faster to implement. The HIL application is needed to test the proposed controllers, implemented as RCP controllers, when attached to a visual distribution network modeled in real time.
Fig. 6 shows the test network in an HIL application. The RTS lab consists of two processors (targets), each has 12 3.33-GHz dedicated cores to perform parallel computations. To achieve an
HIL realization, the network and DG models are implemented in Target 1. Each DG model is assigned to one core similar to the network model to achieve parallel computation. The fuzzy controllers are implemented in Target 2. Each DG voltage support, i.e., FQC and FPC, is assigned to one core similar to the FOC. Both targets exchange data in real time to test the fuzzy controllers as prototyping controllers. The sampling time used to realize the HIL is 100 μ s. For more details about OPAL-
RT and HIL applications, readers can refer to [32]–[35].
The distribution test system consists of two feeders, at 20 kV, namely, Feeder A and B. Feeder A has residential load profiles and six wind-based DG units, whereas Feeder B has different load types (residential, commercial, and industrial) and three

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AZZOUZ et al.
: VOLTAGE REGULATION FOR NETWORKS INCORPORATING PENETRATION OF RENEWABLE SOURCES 7
Fig. 6.
Distribution test network, implemented using OPAL-RT RTS in an
HIL application.
Fig. 7.
Typical daily load and generation profiles.
PV-based DG units. The typical daily load and generation power profiles are shown in Fig. 7 [36]–[38]. All DG units are assumed to be converter based, and hence, they have the capability of supplying reactive power. The total connected load at Feeder A is equal to 14.31 MW/4.13 Mvar, whereas Feeder
B has a total connected load equal to 13.88 MW/5.22 Mvar. A detailed description of the bus data and line parameters can be found in [36]. To test the robustness of the proposed algorithms, four scenarios are considered as follows.
Fig. 8.
OLTC responses under different control schemes, Case A. (a) V and V sys min
. (b) Tap position.
sys max
A. Comparison With Conventional Control
In this part, simulation studies are performed to compare between the following control schemes:
1) the conventional OLTC control (based on LDC);
2) the proposed FOC without incorporating FQC and FPC;
3) the proposed coordination algorithm without considering a power factor limitation;
4) the proposed coordination algorithm with 0.95 power factor limitation.
Fig. 8 shows the response of the OLTC under different control schemes, as previously discussed. Although, the conventional
OLTC control results in no excessive tap operation (6 taps/ day), the system voltages violate their specified limits at different operating conditions. The overvoltage problem is introduced during the peak wind power generation at Feeder A.
Contradictorily, the undervoltage happens during the peak loading condition at Feeder B. To tackle this voltage violation, the proposed FOC should be introduced. The proposed FOC, without incorporating the proposed FQC and FPC, can recover the system voltages, but with relatively excessive tap operation
(15 taps/day). To avoid such an excessive tap operation, the
DG fuzzy voltage support need to be integrated in the voltage regulation. The proposed coordination, with no restriction on the power factor limits, results in a relaxed tap operation (3 taps/ day). To detect the effectiveness of the proposed fuzzy algorithms, the DG power factor is limited to be within 0.95 lag or lead. Although the DG power factors are limited, the proposed coordination scheme can solve the voltage violation problem with reasonable number of taps (8 taps/day). Obviously, the number of taps increased because the reactive power capability is limited. In addition, the OLTC responds with extra two taps during the peak loading (approximately between 19:00 to 20:00), because the PV-based DG units at Feeder B have no active power, and thus, their reactive power contribution is null. It is worth noting that, the FPC is not activated in this scenario because the solution is feasible, i.e., (8) is satisfied in all operating conditions.

8
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL
Fig. 10.
Active power curtailment factors for all DG at Feeder A, Case B.
Fig. 9.
OLTC responses during an infeasible solution, Case B. (a) V
V sys min
. (b) Tap position.
sys max and
B. Active Power Curtailment Scenario
In this part, the FPC is examined, which is only activated when Condition (8) is not satisfied. To stimulate Condition (8),
DG A2, DG A4, and DG A5 are replaced by PV-based DG units, whereas DG B2 is replaced by a wind-based DG unit. All other system parameters remain as in the previous part. Fig. 9 shows comparative studies between the conventional OLTC control and proposed fuzzy algorithms at different incorporations, during an infeasible scenario. The conventional OLTC control results in a voltage violation. The system maximum and minimum voltages occur at the same time interval, i.e., approximately from 9:00 to 16:00, which leads to a violation of the OLTC feasibility condition. Activating the FOC, without incorporation of FQC and FPC, results in a hunting problem. In such a case, the FOC cannot solve the voltage violation problem alone and it must be either coordinated with the FQC and
FPC or deactivated. To detect the effectiveness of the proposed fuzzy algorithms, the DG power factors are limited to be within
0.95 lag or lead. Integrating the DG reactive power support
(FQC) with the FOC can solve the problem partially; however, the OLTC still has a fluctuating response, approximately from
12:00 to 13:00, due to the power factor limit implemented in the FQC. Finally, the overall coordinating scheme is examined.
The proposed FPC provides a solution for the infeasible case with a reasonable number of taps (6 taps/day).
It is worth noting that, the active power curtailment is activated only for the DG units at Feeder A, as they are the reason behind the system overvoltage. Fig. 10 shows the active power curtailments factors for all DG units at Feeder A. The FPC curtails active powers based on the DG participation in the overvoltage problem. The largest active power curtailments are made for DGA5 and DGA6 because: 1) they have the largest ratings, and 2) they are connected at the feeder terminals; hence, their contribution to the system overvoltage is significant. The results show that the maximum active power curtailments for
DGA5 and DGA6 are 18% and 13%, respectively, and occur around 12:00. At that time, DGA5 has higher active power
Fig. 11.
OLTC responses in a mesh network, Case C. (a) V sys max
Tap position.
and V sys min
. (b) injection than DGA6, so the FPC curtails more power from
DGA5 than from DGA6.These results show the effectiveness of the proposed FPC in identifying the curtailment factors based on the contribution of each DG in the overvoltage problem.
C. Meshed Network Scenario
In order to investigate the effectiveness of the proposed algorithms in meshed networks, the tie switches ( S1 and S2 ) are closed to form a meshed distribution network, see Fig. 6.
The DG power profiles are similar to Case B. Fig. 11 shows comparative studies between the conventional and proposed control algorithms at different incorporations, for a meshed network scenario. Again, the conventional OLTC control fails to provide a proper voltage regulation in ADS. Without coordination, both FOC cannot solve the problem and fluctuates, approximately between 12:00 and 13:00, since (8) is not satisfied. Alternatively, the proposed coordination algorithm solves the hunting problem with relaxed tap operation (5 taps/day). It is noteworthy that, the FOC and FQC can solve the problem without the need of active power curtailment, as compared with the radial configuration. From the aforementioned results, it can be concluded that the proposed coordination scheme can deal with meshed networks, if the FQC considers the system minimum and maximum voltages instead of the feeder minimum and maximum voltages, as previously discussed.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AZZOUZ et al.
: VOLTAGE REGULATION FOR NETWORKS INCORPORATING PENETRATION OF RENEWABLE SOURCES 9
ESS at B16, as V sys min occurs at B23. These results demonstrate the effectiveness of the proposed FPC in dealing with the ESS charging.
Fig. 12.
OLTC responses during ESS charging, Case D. (a) V sys max
(b) Tap position.
and V sys min
.
Fig. 13.
Active power curtailment factors during ESS charging, Case D.
(a) For DG at Feeder A. (b) For ESS at Feeder B.
VII. C ONCLUSION
The conventional control of OLTCs, which relies on a fixed target point, does not take into account the DG effect which complicates the voltage regulation due to reverse power flow and voltage estimation difficulties. In this paper, three fuzzybased voltage regulators are proposed in order to tackle the voltage violation problems, which are introduced by the high
DG penetration. In short, DG units start fixing the voltage violation by controlling their reactive powers. Then, if the problem still exits, due to DG reactive power limits, the proposed fuzzy
OLTC controller starts to solve the problem if the solution is feasible. In case of an infeasible solution, DG units curtail their active powers to restore a feasible solution from the OLTC prospective. The proposed fuzzy algorithms can also deal with
ESS. All DG and ESS units share their reactive- and activepower voltage supports according to their relative contributions to the problem. Therefore, the proposed coordination scheme can provide proper voltage regulation with relaxed tap operation and proportionate DG and ESS active power curtailments.
The proposed fizzy algorithms can be integrated as ancillary services within the DSP controllers of the voltage control devices. Nonetheless, communication links are necessary to acquire the system maximum and minimum voltages. Compared with distributed and centralized voltage regulation approaches, the proposed algorithms have a relatively less communication cost, because they rely only on the estimated system minimum and maximum voltages. The proposed coordination also mitigates the numerical instability and convergence problems that are associated with centralized approaches, which run the power flow algorithms in each time step; particularly in case of low X/R ratio. Real-time simulations are performed to show the effectiveness of the proposed algorithms using OPAL-RT.
The results demonstrate the success of the proposed fuzzy algorithms under different operating conditions and system configurations.
D. ESS Charging Scenario
The ESS can play a part in the undervoltage problem during the charging. In such a case, the proposed FPC acts as a smart charger to avoid the undervoltage problem. To examine the robustness of the proposed algorithms during the ESS charging, two 0.5-MW ESS are added at B16 and B23, respectively. All other system parameters remain as in Case B. The conventional control results in a voltage violation, in which the difference between V sys max and V sys min violates the OLTC feasibility constraint, as shown in Fig. 12. The application of the proposed FOC without introducing the FQC and FPC results in a hunting problem.
Merging all fuzzy controllers, using the proposed coordination scheme, lead to a proper voltage regulation with relaxed tap operation. Fig. 13 shows the active power curtailment factors for both DG units at Feeder A, which suffers from overvoltage, and
ESS at Feeder B, which suffers from undervoltage. The ESS at B23 has more active power curtailment compared with the
R EFERENCES
[1] A. P. Sakis Meliopoulos et al.
, “Smart grid technologies for autonomous operation and control,” IEEE Trans. Smart Grid , vol. 2, no. 1, pp. 1–10,
Mar. 2011.
[2] G. T. Heydt, “The next generation of power distribution systems,” IEEE
Trans. Smart Grid , vol. 1, no. 3, pp. 225–235, Dec. 2010.
[3] R. A. Walling, R. Saint, R. C. Dugan, J. Burke, and L. A. Kojovic,
“Summary of distributed resources impact on power delivery systems,”
IEEE Trans. Power Del.
, vol. 23, no. 3, pp. 1636–1644, Jul. 2008.
[4] H. E. Farag, E. F. El-Saadany, and R. Seethapathy, “A two ways communication-based distributed control for voltage regulation in smart distribution feeders,” IEEE Trans. Smart Grid , vol. 3, no. 1, pp. 271–281,
Mar. 2012.
[5] F. A. Viawan and D. Karlsson, “Voltage and reactive power control in systems with synchronous machine-based distributed generation,” IEEE
Trans. Power Del.
, vol. 23, no. 2, pp. 1079–1087, Apr. 2008.
[6] C.-L. Su, “Stochastic evaluation of voltages in distribution networks with distributed generation using detailed distribution operation models,” IEEE
Trans. Power Syst.
, vol. 25, no. 2, pp. 786–795, May 2010.
[7] Standard for Interconnecting Distributed Resources With Electric Power
Systems , IEEE Std. 1547-2003, Jul. 2003.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE SYSTEMS JOURNAL 10
[8] P. M. S. Carvalho, P. F. Correia, and L. A. F. Ferreira, “Distributed reactive power generation control for voltage rise mitigation in distribution networks,” IEEE Trans. Power Syst.
, vol. 23, no. 2, pp. 766–772,
May 2008.
[9] R. Tonkoski, L. A. C. Lopes, and T. H. M. El-Fouly, “Coordinated active power curtailment of grid connected PV inverters for overvoltage prevention,” IEEE Trans. Sustain. Energy , vol. 2, no. 2, pp. 139–147,
Apr. 2011.
[10] A. E. Kiprakis and A. R. Wallace, “Maximising energy capture from distributed generators in weak networks,” Proc. Inst. Elect. Eng.—Gener.
Transmiss. Distrib.
, vol. 151, no. 5, pp. 611–618, Sep. 2004.
[11] M. A. Kashem and G. Ledwich, “Multiple distributed generators for distribution feeder voltage support,” IEEE Trans. Energy Convers.
, vol. 20, no. 3, pp. 676–684, Sep. 2005.
[12] V. Calderaro, G. Conio, V. Galdi, and A. Piccolo, “Reactive power control for improving voltage profiles: A comparison between two decentralized approaches,” Elect. Power Syst. Res.
, vol. 83, no. 1, pp. 247–254,
Feb. 2012.
[13] D. T. Rizy and J. D. Kueck, “Adaptive voltage control with distributed energy resources: Algorithm, theoretical analysis, simulation, and field test verification,” IEEE Trans. Power Syst.
, vol. 25, no. 3, pp. 1638–1647,
Aug. 2010.
[14] M. E. Baran and I. M. El-Markabi, “A multiagent-based dispatching scheme for distributed generators for voltage support on distribution feeders,” IEEE Trans. Power Syst.
, vol. 22, no. 1, pp. 52–59, Feb. 2007.
[15] X. Liu, A. Aichhorn, L. Liu, and H. Li, “Coordinated control of distributed energy storage system with tap changer transformers for voltage rise mitigation under high photovoltaic penetration,” IEEE Trans. Smart Grid , vol. 3, no. 2, pp. 897–906, Jun. 2012.
[16] M. E. Elkhatib, R. El-Shatshat, and M. M. A. Salama, “Novel coordinated voltage control for smart distribution networks with DG,” IEEE Trans.
Smart Grid , vol. 2, no. 4, pp. 598–605, Dec. 2011.
[17] P. N. Vovos, A. E. Kiprakis, A. R. Wallace, and G. P. Harrison, “Centralized and distributed voltage control: Impact on distributed generation penetration,” IEEE Trans. Power Syst.
, vol. 22, no. 1, pp. 476–483,
Feb. 2007.
[18] S. Deshmukh, B. Natarajan, and A. Pahwa, “Voltage/VAR control in distribution networks via reactive power injection through distributed generators,” IEEE Trans. Smart Grid , vol. 3, no. 3, pp. 1226–1234,
Sep. 2012.
[19] L. Yu, D. Czarkowski, and F. de Leon, “Optimal distributed voltage regulation for secondary networks with DGs,” IEEE Trans. Smart Grid , vol. 3, no. 2, pp. 959–967, Jun. 2012.
[20] Y. P. Agalgaonkar, B. C. Pal, and R. A. Jabr, “Distribution voltage control considering the impact of PV generation on tap changers and autonomous regulators,” IEEE Trans. Power Syst.
, vol. 29, no. 1, pp. 182–192,
Jan. 2014.
[21] H. E. Z. Farag, “A novel cooperative protocol for distributed voltage control in active distribution systems,” IEEE Trans. Power Syst.
, vol. 28, no. 2, pp. 1645–1656, May 2013.
[22] L. A. Zadeh, “Is there a need for fuzzy logic?” Inf. Sci.
, vol. 178, no. 13, pp. 2751–2779, Jul. 2008.
[23] D. H. Spatti, I. N. da Silva, W. F. Usida, and R. A. Flauzino, “Realtime voltage regulation in power distribution system using fuzzy control,”
IEEE Trans. Power Del.
, vol. 25, no. 2, pp. 1112–1123, Apr. 2010.
[24] F. Milano, “Hybrid control model of under load tap changers,” IEEE
Trans. Power Del.
, vol. 26, no. 4, pp. 2837–2844, Oct. 2011.
[25] B. Kasztenny, E. Rosolowski, J. Izykowski, M. M. Saha, and B. Hillstrom,
“Fuzzy logic controller for on-load transformer tap changer,” IEEE Trans.
Power Del.
, vol. 13, no. 1, pp. 164–170, Jan. 1998.
[26] E. Demirok et al.
, “Local reactive power control methods for overvoltage prevention of distributed solar inverters in low-voltage grids,” IEEE J.
Photovoltaics , vol. 1, no. 2, pp. 174–182, Oct. 2011.
[27] C. S. Cheng, “A modified Newton method for radial distribution system power flow analysis,” IEEE Trans. Power Syst.
, vol. 12, no. 1, pp. 389–
397, Feb. 1997.
[28] N. R. Ullah, K. Bhattacharya, and T. Thiringer, “Wind farms as reactive power ancillary service providers—Technical and economic issues,”
IEEE Trans. Energy Convers.
, vol. 24, no. 3, pp. 661–672, Sep. 2009.
[29] P. Denholm, J. Jorgenson, M. Hummon, T. Jenkin, and D. Palchak, “The value of energy storage for grid applications,” Nat. Renewable Energy
Lab., Golden, CO, USA, NREL/TP-6A20-58465, pp. 1–37, May 2013.
[30] P. Grbovi´c, Ultra-Capacitors in Power Conversion Systems . Hoboken,
NJ, USA: Wiley, 2013.
[31] A. Yazdani and R. Iravani, Voltage-Sourced Converters in Power Systems .
Hoboken, NJ, USA: Wiley, 2010, p. 451.
[32] RT-LAB Version 10.4 User Guide , Opal-RT, Monréal, QC, Canada.
[33] P. Venne, J.-N. Paquin, and J. Bélanger, “The what, where and why of real-time simulation,” in Proc. IEEE PES Gen. Meet. Tuts.
, 2010, pp. 37–49.
[34] O. Cr˘aciun et al.
, “Hardware-in-the-loop simulation applied to protection devices testing,” Int. J. Elect. Power Energy Syst.
, vol. 54, pp. 55–64,
Jan. 2014.
[35] J.-N. Paquin, C. Dufour, and B. Jean, “A hardware-in-the-loop simulation platform for prototyping and testing of wind generator controllers,” in
Proc. CIGRÉ Canada, Conf. Power Syst.
, 2008, pp. 1–5.
[36] F. Bignucolo, R. Caldon, and V. Prandoni, “Radial MV networks voltage regulation with distribution management system coordinated controller,”
Elect. Power Syst. Res.
, vol. 78, no. 4, pp. 634–645, Apr. 2008.
[37] A. Sasitharanuwat, W. Rakwichian, N. Ketjoy, and S. Yammen, “Performance evaluation of a 10 kWp PV power system prototype for isolated building in Thailand,” Renew. Energy , vol. 32, no. 8, pp. 1288–1300,
Jul. 2007.
[38] J. A. Jardini, C. M. V. Tahan, M. R. Gouvea, S. U. Ahn, and
F. M. Figueiredo, “Daily load profiles for residential, commercial and industrial low voltage consumers,” IEEE Trans. Power Del.
, vol. 15, no. 1, pp. 375–380, Jan. 2000.
Maher Abdelkhalek Azzouz (S’10) was born in
Giza, Egypt, on 1986. He received the B.Sc. and
M.Sc. degrees (with honors) in electrical engineering from Cairo University, Giza, in 2008 and 2011, respectively. He is currently working toward the Ph.D.
degree in the Department of Electrical and Computer
Engineering, University of Waterloo, Waterloo, ON,
Canada.
His research interests include dynamics and control of power converters, distributed and renewable generation, and control of smart distribution systems.
Hany E. Farag (M’13) was born in Assiut, Egypt, on November 21, 1982. He received the B.Sc. (with honors) and M.Sc. degrees in electrical engineering from Assiut University, Assiut, in 2004 and 2007, respectively, and the Ph.D. degree in electrical engineering from the University of Waterloo, Waterloo,
ON, Canada, in 2013.
Currently, he is an Assistant Professor with the
Department of Electrical Engineering and Computer
Science, Lassonde School of Engineering, York University, Toronto, ON. His research interests include active distribution networks, integration of distributed and renewable energy resources, modeling, analysis, and design of microgrids, and applications of multiagent technologies in smart grids. His biography is listed in the 30th Pearl
Anniversary Edition of Marquis Who’s Who in the World in 2012.
Ehab F. El-Saadany (SM’05) was born in Cairo,
Egypt, on 1964. He received the B.Sc. and M.Sc. degrees in electrical engineering from Ain Shams University, Cairo, Egypt, in 1986 and 1990, respectively, and the Ph.D. degree in electrical engineering from the University of Waterloo, Waterloo, ON, Canada, in 1998.
Currently, he is a Professor with the Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada. His research interests include smart grids operation and control, power quality, distributed generation, power electronics, digital signal processing applications to power systems, and mechatronics.