Electric Power Components and Systems

ISSN: 1532-5008 (Print) 1532-5016 (Online) Journal homepage: www.tandfonline.com/journals/uemp20

Optimal Control of Active Power of Two Micro-

grids Interconnected with Two AC Tie-Lines

Diambomba Hyacinthe Tungadio, Ramesh C. Bansal & Mukwanga Willy Siti

To cite this article: Diambomba Hyacinthe Tungadio, Ramesh C. Bansal & Mukwanga
Willy Siti (2017) Optimal Control of Active Power of Two Micro-grids Interconnected
with Two AC Tie-Lines, Electric Power Components and Systems, 45:19, 2188-2199, DOI:
10.1080/15325008.2017.1384940

To link to this article: https://doi.org/10.1080/15325008.2017.1384940

Published online: 13 Feb 2018.

Submit your article to this journal

Article views: 396

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uemp20
Electric Power Components and Systems, 45(19):2188–2199, 2017
Copyright C Taylor & Francis Group, LLC
ISSN: 1532-5008 print / 1532-5016 online
DOI: 10.1080/15325008.2017.1384940

Optimal Control of Active Power of Two Micro-grids

Interconnected with Two AC Tie-Lines
Diambomba Hyacinthe Tungadio,1 Ramesh C. Bansal,1 and Mukwanga Willy Siti2
1
Department of Electrical, Electronics and Computer Engineering, University of Pretoria, Pretoria, Gauteng, South Africa
2
Department of Electrical Engineering, Tshwane University of Technology, Pretoria, Gauteng, South Africa

CONTENTS
Abstract—Active power balance and frequency control are impor-
1. Introduction tant tasks in the daily management of a power system. With the
2. Configuration of the System Studied integration of several microgrids in the modern power system, the
3. Problem Formulation and Modeling balancing of sources of energy has become a major concern for
electrical power engineers and researchers worldwide. New power
4. Results and Discussions systems require more flexibility in optimization and control design to
5. Conclusion ensure their ability to maintain the balance between generation and
References load of the system. The present paper discusses an optimal control
design for minimization of the power flow in tie-lines and frequency
deviations in the microgrid, which will lead to power balance
between the generation and load demand of the system. The system
being studied consists of two microgrids, each made up of a wind
farm, conventional power system (large hydro or thermal plant), pho-
tovoltaic system, battery energy storage system and the system load.
Optimal control theory is applied to control the power flow between
two microgrids. The Matlab environment is used for simulations.

1. INTRODUCTION
Active power control is one of the main requirements in the
daily management of any electrical systems. The aim of this
control is to monitor the variation of the active power and
to keep it within the limits prescribed by the standards [1],
[2]. The problem becomes more complex when dealing with
an interconnected system [3], [4]. In fact, the reliability and
continuity of service of two or more isolated power grids
(microgrids) can be achieved by their interconnection through
ties-lines. An interconnected power system can be considered
as being divided into control areas; all generating units are
assumed to form a coherent system [5]–[10]. In the steady
state power system demands of areas are satisfied at the nom-
Keywords: power system, load frequency control, optimal control, tie-line, inal frequency. The change of active power in one area is sus-
renewable energy sources tained by the increase in generation in all areas associated
Received 30 September 2016; accepted 27 August 2017
with a change in the tie-lines power flow and a reduction in
Address correspondence to Ramesh C. Bansal, Department of Electrical,
Electronics and Computer Engineering, University of Pretoria, Room 27, frequency [11]–[14]. As the power load demand varies ran-
Level 14, Engineering Building 1, Lynwood Campus, Pretoria, Gauteng domly, all areas frequency and the tie-lines power interchange
0002, South Africa. E-mail: [email protected]
also vary [15], [16]. The objectives of load frequency control
Color versions of one or more of the figures in the article can be found online
at www.tandfonline.com/uemp. (LFC) is to minimize the deviations in frequency and tie-line

2188
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2189

power flows which are taken as the control variables and to to the conventional method, it becomes more expensive to
ensure that the frequency variation is returned to zero. A basic build centralized control architecture. However as seen above
control system strategy for the normal mode is to operate in few authors did discusses on the grid interactive Wind and
such a way that it keeps the frequency at approximately nomi- PV-solar with the aims to optimize the frequency and the
nal value, maintaining the tie-line power flow out of each area renewable energy sources. Optimal control model was used
at the planned value. Each area should absorb its own load applied fmincon. Fmincon is an inbuilt function of Matlab
changes [17]–[19]. That is why the primary purpose of oper- for solving optimization problem. The variables are divided
ating the LFC is to keep the frequency changes during the into the control variables and the state variables, with a link
load changes uniform. Furthermore, the control model must to the dynamics of the networks. In real time power system
be sensitive against changes in the area [20]. operation and control, using the fmincon application makes it
Most of the research in the field of control of micro- easier to have the measurement of variables. The construction
grids is devoted to the use of conventional controllers such of the fmincon model requires only the measured variables,
as integral, proportional-integral (PI) or proportional-integral- reducing the complexity of the controllers.
derivative (PID) ones. The new tendency is the combination of In order to keep the system simple and to prevent cost relat-
the artificial intelligence algorithms particle swarm optimiza- ing to direct current converters, two alternating tie-lines are
tion (PSO), fuzzy logic (FL), genetic algorithm, etc. with the used. The control strategy of this work is based on the differ-
conventional controller to solve the frequency or active power ence in active power magnitude through these tie-lines. The
flow control issue in microgrids [21]–[25]. Intelligent fre- objective of this work is therefore the minimization of the dif-
quency control in an AC microgrid online PSO-based tuning ference inactive power flowing between the two tie-lines using
approach was presented in Ref. [26]. The research in Ref. [27] the optimal control approach. This paper presents the optimal
presents the design of optimal AGC regulators using an out- active power control regulator designs based on an optimal
put feedback control strategy for a multi-area interconnected control theory. The optimal active power control regulator is
power system. Li et al. [28] proposed frequency control in the designed for two interconnected microgrids composed of con-
microgrid power system combined with an electrolyser system ventional hydro and thermal plants, PV system, wind farm and
and fuzzy PI controller. The proposed control can be regarded two ac tie-lines. The investigation of the performance of the
as a means of power quality control to improve frequency fluc- power system dynamics has been studied using the response
tuations. Optimal fuzzy logic-based PID and self-tuning fuzzy curves of the disturbed areas (area 1, area 2 and both of them
PID controller was presented in researches in Refs. [29], [30]. simultaneously). The proposed active power regulator is found
Vigneysh and Kumarappan [31] have presented the control of to demonstrate its ability to bring the system state deviation as
PV-solid oxide fuel cell-battery energy storage based micro- per the desired ones in an effective manner. In practice the reg-
grid using an FL controller. FL based droop control for simul- ulator will improve the daily management of a modern electri-
taneous voltage and frequency regulation in an AC micro- cal network and will control the flow of energy from different
grid was discussed. Ahmadi et al. [32] have presented the energy sources to the main grid during periods of imbalance
design of intelligent control for stabilization of a microgrid in the tie-lines.
system using a modified general regression neural network. The rest of this paper is organized as follows: In
A Fuzzy gain scheduling controllers for automatic generation Section 2, the configuration of the studied system is explained.
of two-area interconnected electrical power systems was pre- The problem formulation and modelling are described in
sented in Ref. [33]. A novel hybrid local unimodal sampling– Section 3. Section 4 presents the system data, simulations
teaching learning based optimized fuzzy-PID controller for results and discussions. Finally, the conclusion of the work
load frequency control of multi-area power system was pre- is summarized in Section 5.
sented in Ref. [34]. Novel control scheme based on droop
characteristics for frequency control of microgrids working in
2. CONFIGURATION OF THE SYSTEM STUDIED
autonomous mode was presented by Khederzadeh and Maleki
[35]. The researchers in [36] have proposed a fractional order The system studied consists of microgrids, connected through
PID controller to solve LFC problem. Robust H for LFC two power tie-lines, as shown in Figure 1.
solution was presented Sondhi and Hote [37]. Each microgrid feeds its consumers and the tie-line allows
The weakness of most of the methods as discussed above electric power to flow between the areas. Information about
can be summarized by the conventional controller applied the local loads area is found in the tie-line power fluctuations.
to the nonlinear system, most of the time it is dependent Therefore, the tie-line power is sensed, and the resulting tie-
on the mathematical model of the plant. With reference line power is fed back into both areas. Each power area has
2190 Electric Power Components and Systems, Vol. 45 (2017), No. 19

The operational conditions become more favorable for each

electrical system, increasing their efficiency. The generating
units of each microgrid will not be overloaded by meeting
their power reserve limits. Connecting microgrids makes man-
agement more complex, despite the listed benefits. For sta-
ble operation, each network must be capable of supplying
and completely absorbing its own load and the tie-line power
flow must be kept minimum. Because of the varied produc-
tion and load demands, a microgrid presents high nonlinear-
ities, changing dynamics and uncertainties that may require
advanced intelligent control strategies, such as the optimal
approach used in this paper. The use of more effective strate-
gies would increase the performance of these systems.
The renewable energy sources such as wind and PV work
in turbulent conditions in an unforeseeable environment, and
microgrid control system must adjust to changes and thus the
effectiveness and reliability of microgrids depend heavily on
the chosen and applied control strategies. This paper focuses
on the regulation of active power and frequency in two tie-
FIGURE 1. Network model. lines between two microgrids. The system constraints i.e. state
of charge (SOC) of the battery, power output of the wind farm,
PV and SGs are taken as control variables. The limits of each
number of generators, which are closely coupled together so
generating unit and the power balance should be met during
as to form a coherent group. All generators or power sources
the control horizon. The tie-lines power flows are regarded as a
respond in unison to changes in the load. Such a coherent area
state variable and the control objective function is formulated
is called a control area, in which the frequency is assumed to
as the minimization of the difference of active power flow in
be the same throughout in static as well as dynamic situations.
two tie-lines.
It is assumed that each control area can be represented by a
synchronous generator (SG), wind farm, PV array and battery 3.2 Mathematical Modelling of the System
energy storage system (BESS). The active power from SG 1
In order to maintain stable power network operation, the total
and SG 2 are known (P1 = P2 = 150 MW); the wind power
energy generated must be adequately controlled and correctly
(from wind farm 1 and wind farm 2) and solar power (from PV
dispatched to meet the total power demand of the connected
1 and PV 2) are determined by their profile which varies dur-
loads. Based on Figure 1, the total power generated (PM1 and
ing the 24 hours period between the times [t0 , tf ], taken in this
PM2 ) for microgrids 1 and 2 respectively are given by the fol-
research as the control horizon. The maximum and minimum
lowing equations, all the terms are discretized:
values were divided by an N sample interval into sampling
time t in the control horizon. PM1 (k) = P1 (k) + P2 (k) + P3 (k) + P4 (k) − P5 (k) (1)
The load demands in two microgrids are taken under a pro-
where for microgrid 1: P1 (k) is the active power from the SG
file form. Each microgrid can supply a larger load than its
(hydro plant), P2 (k) the active power from the WF, P3 (t) the
maximum capacity using the power transfer through tie-lines.
active power from the PV array; P4 (k) and P5 (k) the power of
All active powers are in MW.
the BESS (charge and discharge mode).
PM2 = P8 (k) + P9 (k) + P10 (k) + P11 (k) − P12 (k) (2)
3. PROBLEM FORMULATION AND MODELING
where for microgrid 2: P8 (k) is the power from the SG (ther-
3.1. Problem Definition mal plant), P9 (k) the active power from the PV array, P10 (t)
The interconnection of electrical systems or microgrids the active power from the WF; P11 (k) and P12 (k) the power of
presents many benefits. It improves the reliability of sup- the BESS (charge and discharge mode).
ply and the peak load is less than the sum of the general All those powers are taken as control variables, P6 (k)and
peaks for each interconnected system. Using the bidirectional P7 (k) are considered as power flow in tie-lines (are taken as
power flow via tie-lines, all load demands can be managed. state variables), and the power deviations for each microgrid
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2191

(or area) are given by the following equations:

PM1 = PM1 (k) − PL1 (k) (3)
PM2 = PM2 (k) − PL2 (k) (4)
where PL1 (k) and PL2 (k) are respectively the power demand
(load) for microgrids 1 and 2. The two equations demonstrate
the difference between the power generated in the two areas
and their power demands
The frequency deviation of the system is calculated as fol-
lows for each microgrid:
PM1
ω1 = (5)
Ksys1
and
PM2
ω2 = (6)
Ksys2
where ω1 and ω2 are the frequency difference for micro-
grids 1 and 2 respectively, Ksys1 and Ksys2 is system frequency FIGURE 3. Flowchart of fmincon controller.
characteristic constants of the microgrids 1 and 2 respectively.
Tie-line is used to exchange power between control areas
to provide inter-area supporting abnormal conditions. If there The overall control system is given in Figure 3. The main
is a mismatch between the powers generated and demanded, a steps of the optimisation-control process are:
deviation in frequency for a particular area occurs. A model of r Control and state variables definition;
two areas interconnected with one tie-line is given by Figure 2. r Matrix Aineq and vector Bineq formulations (linear
Since a tie-line transports power in or out of an area, inequality constraint);
this fact must be accounted for the incremental power bal- r Determination of the linear equality constraint (matrix
ance equation of each area. The tie-line power deviation Aeq and vector Beq );
(PTL ) using microgrids frequency deviation and synchroniz- r Formulation of lower and upper bounds vectors;
ing power coefficient (PS ) is calculated as follows [15]: r Formulation of the state of charge matrix (Aeq1) and
  
vector Beq1;
PT L = PS ω1 dt − ω2 dt (7) r Starting point choice;
r Evaluations process:
++Objective function;
++Constraints;
r Iterations and decision process;
r End of process.

3.3 Introduction to the Proposed Model

To solve this problem, 12 control variables were considered
(all the active powers). The powers from wind farm and PV
array are modelled as a variable source, with a profile that
varies from zero to its maximum for 24 hours [38], [39]. Each
area feeds its own load; load 1 is supplied by powers from
wind farm 1, PV system 1, SG 1 and BESS 1 and load 2 is
fed by powers from wind farm 2, PV 2, SG 2 and BESS 2.
When an unbalance situation occurs, the power from one area
via tie-line can support the affected area. If power from the SG
(in area 1 or 2) is insufficient and powers from the wind farm
FIGURE 2. Modelling of one tie-line connection. (in area 1 or 2) and PV (in area 1 and 2) cannot meet the load,
2192 Electric Power Components and Systems, Vol. 45 (2017), No. 19

the power from the BESS (area 1 or 2) can be used to feed the Each energy source is constrained by minimum and maximum
load and to keep the system balanced. values as specified by Eq. (14). Pi (k) is the power from each
The optimal control approach is applied by controlling the energy source during the network operation.
frequency in the areas and the power flow in the tie-lines as
socmin ≤ soc (k) ≤ socmax (15)
a first objective function and the maximization of the use of
renewable energy resources (REN) as a second objective func- Eq. (15) represents the boundary constraint of the BESS
tion. The model is given by the following discretized cost 1 and BESS 2. Their SOC value must be kept less than the
function (Eq. (8) gives the first objective function): full capacity of the BESS socmax and larger than its minimum
capacity socmin .

N
f1 = min t (P6 (k) + P7 (k)) + B f . f (8)
0 ≤ Pi (k) ≤ Pimax (16)
k=1

where t is the sampling time; P6 (k) and P7 (k) are power Eq. (16) represents the power flow constraint for the safety
flows for tie-lines 1 and 2 respectively and Bff is the fre- of operations in the power network. The power flow from each
quency deviation’s constant. area cannot be negative and less than its allowable maximum
The second objective function is given by: value.

f2 = − max REN (9) P6min ≤ P6max , P7min ≤ P7max (17)

Taking into consideration the conflict of the two objective Eq. (14) represents the power generation limits of P2 (k) and
functions in the model, the problem is formulated as a multi- P10 (k), P3 (k) and P9 (k), P1 (k) and P8 (k), and Eq. (15) repre-
objective function given by the following equation: sents the limits of the BESS (BESS 1 and BESS 2).
This model has been applied using fmincon solver in Mat-
f = f1 + f2 (10) lab because of the non-linearity of the power systems. The
canonical form is given by the basic conditions for the opti-
3.3.1 Objective Function and Inequalities constraints. mization as follows: min fT x (Eq. 14)
Considering Eqs. (8) and (9), the objective function can be Subject to the following constraints
represented as follow: ⎧
⎨Aineq × X ≤ Bineq

n
A × X = Beq (18)
f = min [w1 t (P6 (k)) + w2 t (P7 (k))] ⎩ eq
Lb ≤ X ≤ Ub
i=1

n
where:
− max [P2 (k) + P3 (k) + P9 (k) + P10 (k)] + B f . f
i=1
r Bineq , Beq are vectors; Aineq , Aeq are matrices;
(11) r Lb  X  Ub lower and upper bounds of all given vari-
ables;
where w1 and w2 are the weight factors r Aineq × X  Bineq linear inequality constraint of the sys-
Subject to the following equalities constraint: tem derive from the dynamic of the batteries, given in
P1 (k) + P2 (k) + P3 (k) + P4 (k) − P5 (k) + P8 (k) + P9 (k) Eqs. (19) and (20);
r Aeq × X = Beq linear equality constraint based on the
+ P10 (k) + P11 (t ) − P12 (t ) = PL (12)
battery dynamics given in Equations (22) and (23) for
Eq. (12) represents the power balance of the two microgrids. BESS 1 and Eqs. (25) and (26) for BESS 2;

PL = PL1 (k) + PL2 (k) (13) The vector X is a binary integer vector represented by the
power flow from different components of the system. This
Subject to the following inequalities constraint: variables are responsible for the optimal process and is given
as follows: X = [x1 , x2 ,..., xN ]T
Pimin ≤ Pi (k) ≤ Pimax (14) Based on Eq. (12), the linear inequality constraints of the
system is found in matrix form as follows:
 
1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0
Aeq = (19)
0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2193

  These equations can be translated into the inequality con-

PL1
Beq = (20) straint matrix given by Eq. (25):
PL2
⎛ ⎞
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 −h 0 +h 0
⎜0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 −h −h +h +h ⎟
Aineq 2 = ⎜
⎝0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
⎟ (25)
+h 0 −h 0 ⎠
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 +h +h −h −h

The linear constraint is the same as for the BESS 1,

3.3.2 Battery Storage System State of Charge: Equalities
Constraints. The dynamics of the battery, charge and dis- Bineq 1 = Bineq 2 (26)
charge gives its SOC. For the BESS in area 1 the SOC is given
by the following equations:

⎧ 4. RESULTS AND DISCUSSIONS

⎪
⎪ soc (0) − h j
(P4 (k) − P5 (k)) ≤ soc max
⎪
⎪
k=1 The main conventional hydro and thermal generator running
⎪
⎨ ⇒ −h j
k=1 (P4 (k) − P5 (k)) ≤ socmax − soc (0) or synchronous generator in each area (SG1 and SG2) have
⎪
⎪
j
socmin ≤ soc (0) − h k=1 (P4 (k) − P5 (k)) the following rating power: P1 (k) = P8 (k) = 150 MW, dur-
⎪
⎪
⎪
⎩ ⇒ +h j (P (k) − P (k)) ≤ socmin − soc (0) ing the simulation process the optimal value of these active
k=1 4 5 power were found equal to 149.9 MW. Rated frequency is
(21) 50 Hz. SG1 and SG2 are conventional hydro and thermal
t
where h = ( Cn ), a coefficient taking into account the charging power plants. The characteristics of PV1, PV2, wind farm
and discharging of the battery. 1 and wind farm 2 and a group of loads are given and
Applying optimal control theory, the system of Eq. in (21) used under profile form. The data for hydro-electric power
can be translated into the following inequality constraints plant and thermal plant (steam turbine) can be found in
matrix for BESS 1 given by Eq. (22): Refs. [40], [41].
⎛ ⎞
0 0 0 0 0 0 −h 0 +h 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
⎜0 0 0 0 0 0 −h −h +h +h 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ⎟
Aineq 1 = ⎜
⎝0
⎟
⎠ (22)
0 0 0 0 0 +h 0 −h 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 +h +h −h −h 0 0 0 0 0 0 0 0 0 0 0 0 0 0

The linear vector constraint is represented by the following This research presents optimization-control using the fmin-
vector (Eq. (23)): con approach to control the active power flow and frequency in
two tie-lines of two microgrids (MG1 and MG2). The imple-
⎛ ⎞ mentation of the model was done in Matlab software. All the
socmax − soc (0)
⎜ socmax − soc (0) ⎟ sources in each microgrid supply at their maximum power.
Bineq 1 = ⎜ ⎟
⎝ soc (0) − socmin ⎠ (23) The system frequency deviations are zero before any distur-
soc (0) − socmin bances in the power system. Assuming a load change in area 1,
the system’s behaviour was observed. Observations were also
made after disturbing the system in area 2 by changing the
For the BESS in area 2 the state of charge is given by the load. The loads were used as inputs in the system under pro-
following equations: file form, since they vary constantly. The analyses of the vari-
⎧ ations of active power in the tie-lines and the frequency in the
⎪
⎪
j
soc (0) − h k=1 (P11 (k) − P12 (k)) ≤ socmax microgrids were made during 24 seconds, while the remain-
⎪
⎪
⎪
⎨ ⇒ −h j (P11 (k) − P12 (k)) ≤ socmax − soc (0) ing observations were made during 24 hours as control hori-
k=1
zon. From the profile form, the model gives the optimal MW
⎪
⎪ soc ≤ soc (0) − h k=1
min j
(P11 (k) − P12 (k))
⎪
⎪ values from each renewable energy sources and for the SG.
⎪
⎩ ⇒ +h j (P (k) − P (k)) ≤ socmin − soc (0)
k=1 11 12 This can be seen in Table 1, which reflects the optimal val-
(24) ues of all the active power after simulations. Area 1 pick up the
2194 Electric Power Components and Systems, Vol. 45 (2017), No. 19

Hours P2 (t) P3 (t) P4 (t) PL1 (t) P9 (t) P10 (t) P11 (t) PL2 (t)

1 43.31 0.00 6.35 191.22 21.79 0.00 8.84 178.69

2 43.32 0.00 12.71 190.23 16.78 0.00 11.82 176.94
3 43.52 0.00 9.81 190.42 13.80 0.00 8.93 168.64
4 43.42 0.00 14.68 191.33 13.79 0.00 10.72 173.87
5 43.14 0.00 18.14 193.05 13.78 0.00 12.72 175.67
6 43.78 0.08 19.65 206.43 13.76 0.09 17.81 172.05
7 42.62 7.74 9.12 211.26 13.77 2.18 8.36 172.06
8 44.41 10.54 17.28 215.12 19.78 5.90 19.94 197.26
9 41.63 11.55 14.47 204.09 13.77 5.02 16.29 184.97
10 41.70 14.54 18.40 206.14 17.77 11.14 19.42 174.10
11 40.79 16.44 19.87 208.13 18.82 18.43 17.22 183.99
12 40.58 17.35 16.08 213.91 14.81 26.62 10.65 200.99
13 41.87 19.60 11.38 227.75 13.71 19.62 3.16 191.02
14 42.62 19.78 12.09 213.31 13.74 19.49 14.58 185.00
15 43.26 20.04 18.60 221.78 13.99 21.40 19.78 204.49
16 45.72 17.53 11.07 204.53 14.32 17.71 10.04 191.03
17 50.25 21.88 18.50 213.03 15.13 22.62 19.83 206.39
18 55.55 0.01 19.14 215.95 20.16 0.00 18.78 177.64
19 59.03 0.00 19.23 212.53 36.17 0.00 19.09 189.98
20 59.95 0.00 19.70 209.21 47.31 0.00 16.78 189.81
21 59.85 0.00 12.73 207.02 46.10 0.00 12.04 177.02
22 59.62 0.00 11.00 201.57 42.86 0.00 11.50 175.03
23 59.04 0.00 12.56 198.95 36.49 0.00 15.18 189.20
24 56.59 0.00 0.62 197.11 22.94 0.00 0.075 170.02

TABLE 1. Optimal values of active power for microgrids 1 and 2.

load between 13:00 and 20:00 as shown by PL1 (t) which repre- occurring during the transmission and this value remains
sents the load profile of our control horizon, the change of the constant for 24 hours. This shows that area 1 has absorbed its
load contribute to the deviation of the frequency between the own power to keep the frequency deviation equal to zero. The
two areas. The optimisation by the renewable energy sources variable P2 (k) represents the wind farm output in area 2. There
which are P2 (t) and P3 (t), responds to the increase of the load
during the peak time to minimise the frequency deviation. The
same applied to area 2 when PL2 (t) increase between 15:00
and 17:00; once again the P9 (t) and P10 (t) respond to opti-
mise the frequency deviation. Figures 4a and 4b show the load
profile and the total generated power for areas 1 and 2, for
24 hours. The contribution of BESS 1 and BESS 2 in terms
of energy can be seen in Figures 5a and 5b. Figures 6c and 5d
show the SOC of BESS 1 and BESS 2. The frequency devia-
tions in microgrids 1 and 2 can be seen in Figures 7a, 7b, 7c
and 7d. The active power flow deviation in tie-lines 1 and 2
without active power flow control are given in Figures 8a and
8b. In Figure 9a to Figure 10d, the power flow deviation in
tie-lines is shown for three different load variation cases. All
the powers are in mega-watt (MW) and time in hour (h).
The snapshot of the two microgrids after 24 hours of opera-
tion is given in the Table 1. The optimal output of each sources
and the energy consumed by the loads are given in MW. The
variable P1 (k) represents the power flow from the main gener- FIGURE 4. Active power generation and load demand in
ating unit in area 1; the value 149.9 MW shows some losses microgrids 1 and 2.
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2195

FIGURE 7. Frequency deviation with disturbance in MG 1

FIGURE 5. Active power output and SOC of BESS1 and and MG 2.
BESS2.

The same situation is observed in area 2, where P8 (k) rep-

are changes in the morning and peak periods to keep area 1 in resents the active power generated by the conventional gener-
a stable state. The power P3 (k) represents the PV array output ation and P9 (k) and P10 (k) represent simultaneously the wind
in area 1 during the early hours (from 01:00 to 06:00) when the farm and the solar array output in area 2. The discharge of the
PV output is weak. The compensation to feed area 1 is taken battery is represented by the power P11 (k) and PL2 (k) is the
by P1 (k) and P2 (k). The power output of the battery in area power demand in area 2. The load variation in area 2 is shown
1 is represented by P4 (k), showing some change during the graphically in Figure 4b and its frequency deviation can be
24 hours taken as a control horizon. The power load demand found in Figures 7b and 7d.
in area 1 PL1 (k) changes with the need of consumers during the
24 hours. This can be seen in Figure 4a for the load variation
and Figures 7a and 7c for the frequency deviation for area 1.

FIGURE 8. Active power deviation in Tie-Lines with distur-

FIGURE 6. SOC of BESS 1 and BESS 2. bances in MG 1.
2196 Electric Power Components and Systems, Vol. 45 (2017), No. 19

Disturbance in area 1

PID Fmincon

Ts (s) OS (MW) Ts (s) OS (MW)

P_TIE1 13.11 10.62 9.20 8.29

P_TIE2 12.32 11.21 9.61 8.31
Disturbance in area 2

P_TIE1 13.54 10.71 11.01 8.33

P_TIE2 13.25 11.23 10.85 9.24
Disturbances in area 1 and 2

P_TIE1 17.11 15.61 15.49 10.34

P_TIE2 16.01 13.91 14.72 10.23

TABLE 2. Settling time (Ts) and maximum over-shoot (O.S) of the

system.

The active power flow deviation in the tie-lines 1 and 2

when disturbance occurs in microgrid 1 and the restoration of
FIGURE 9. Active power deviations in tie-line using PID con- the power flow balance after control are shown in Figures 7a
troller with disturbances in MG 1 and MG 2.
and 7b. Figures 7c and 7d give the frequency variation with
Considering area 1 f  0 and Ptie  0 that is the PID controller, and Figures 9a, 9b, 9c and 9d represent the
steady state. The frequency errors are returned to zero. Area simulation results obtained with the traditional PID controller.
1 picks up its own 209 MW as referred to in Table 1, load Figures 7c and 7d indicate the frequency variation during 300
increase, and area 2 returns to its original operating condi- s. It is difficult to see the reaction of the controller during that
tions, as shown in Table 1, that is, the condition before the time. This is the reason why the simulation time was extended
load increase. to 300 s. The return to the steady state of the system was
ensured after a relatively long time.
Figures 8a and 8b show the active power deviation in the
tie-lines when no control system is considered. These results
were obtained by introducing disturbances in MG 1 and MG
2 simultaneously. MG 1 and MG 2 were first loaded at their
maximum power and then a 1% overload was added in each
of them to disrupt the entire system. With the increase of the
load in the MG 1 only, the active power deviation in the tie-
lines is controlled in +/-3 s using the proposed method. This
can be seen from Figures 8c and 8d. This shows an improve-
ment in response time of the control system strategy used as
compared with what PID has given as control time. Generally,
short time response implies a good control process. The pro-
posed method has allowed to attenuate the active power in the
tie-lines with a good settling time and a good maximum over-
shoot, the summary of all three distances cases used can be
seen in Table 2.
The effectiveness of the control model (regulator) using
fmincon is shown in all the cases. This can be seen in
Figures 10a, 10b, 10c and 10d (when disturbances occur in
microgrids 1 and 2 respectively and both of them at the same
FIGURE 10. Active power deviations in tie-lines using fmin-
time). The reaction in the tie-lines when microgrid 2 was dis-
con controller with disturbances in MG 1 and MG 2 and 2
simultaneously. turbed can be seen in Figures 10a and 10b. The Figures 10c
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2197

and 10d give the reaction in the tie-lines when disturbances [4] M. J. Hossain, M. A. Mahmud, F. Milano, S. Bacha, and A.
occur in the two microgrids at the same time. For all the load Hably, “Design of robust distributed control for interconnected
variations that the systems have undergone, the balance of the microgrids,” IEEE Trans. Smart Grid, Vol. 7, no. 6, pp. 2724–
2735, December 2015.
system was reached after control. When the simulation results [5] D. E. Olivares, et al., “Trends in microgrids control,” IEEE
obtained using the traditional PID controller are compared, the Trans. Smart Grid, Vol. 5, no. 4, pp. 1905–1919, July 2014.
time response of fmincon is better. doi:10.1109/TSG.2013.2295514.
The settling time and the maximum over-shoot for the stud- [6] K. P. S. Parmar, S. Majhi, and D. P. Kothari, “Load frequency
ied system can be found in the above Table (Table 2) control of a realistic power system with multi-source power
generation,” Int. J. Electr. Power Energy Syst., Vol. 42, pp. 426–
The comparison between PID and fmincon according to
433, 2012.
the studied system and using the simulations parameters in [7] G. T. C. Sekhar, R. K. Sahu, A. K. Baliarsingh, and S. Panda,
[40] can be concluded as the fmincon controller shown a little “Load frequency control of power system under deregulated
improvement in time when compare with the PID controller. environment using optimal firefly algorithm,” Int. J. Electr.
Power Energy Syst., Vol. 74, pp. 195–211, 2016.
[8] D. Q. Hung, N. Mithulananthan, and R. C. Bansal, “Inte-
gration of PV and BES units in commercial distribu-
5. CONCLUSION
tion systems considering energy loss and voltage sta-
This paper has presented the active power control of two inter- bility,” Appl. Energy, Vol. 113, pp. 1162–1170, 2014.
connected microgrid systems minimising the difference in doi:10.1016/j.apenergy.2013.08.069.
[9] R. Majumder, A. Ghosh, G. Ledwich, and F. Zare, “Load fre-
active power flow between two ac tie-lines. The mathematical quency control for rural distributed generation,” Electr. Power
model of the proposed system was detailed and results were Compon. Syst., Vol. 38, no. 6, pp. 637–656, April 2010.
shown. Because of the complexity and multi-variable nature doi:10.1080/15325000903489652.
of power systems, a conventional control method did not pro- [10] C. Yuen, A. Oudalov, and A. Timbus, “The provision of
vide satisfactory solutions. The fmincon solver attempted to frequency control reserves from multiple microgrids,” IEEE
Trans. Industrial Electronics, Vol. 58, no. 1, Jan. 2011.
solve the issue, yielding promising results, with respect to
doi:10.1109/TIE.2010.2041139.
handling the robustness, reliability and nonlinearities asso- [11] M. Khederzadeh, H. Maleki, and V. Asgharian, “Frequency
ciated with power system control. The frequency oscillation control improvement of two adjacent microgrids in autonomous
and tie-lines power deviations in area one and area two were mode using back to back voltage-sourced converters” Int.
reduced. This is evident from a comparison of the results pre- J. Electr. Power Energy Syst., Vol. 74, pp. 126–133, 2016.
sented in the figures in the results and discussion section. doi:10.1016/j.ijepes.2015.07.002.
[12] P. M. Costa, and M. A. Matos, “Assessing the contribu-
The responses curves in these Figures show the deviations in tion of microgrids to the reliability of distribution net-
power system in respective areas and from their inspection it works,” Electr. Power Syst. Res., Vol. 79, pp. 382–389, 2009.
has been inferred that the proposed optimal active power con- doi:10.1016/j.epsr.2008.07.009.
trol regulator is effective in settling the change in the power [13] J. Zhao, and F. Dorfler, “Distributed control and optimisa-
system network to the required value in reasonably small time. tion in DC microgrids,” Automatica, Vol. 61, pp. 18–26, 2015.
doi:10.1016/j.automatica.2015.07.015.
Table 1 gives a numerical representation of the system after a
[14] N. Nikmehr, and S. N. Ravadanegh, “Reliability evaluation
24 hour control horizon. The MW transfer in the two micro- of multi-microgrids considering optimal operation of small
grids during off-peak and peak time can be seen. scale energy zones under load-generation uncertainties,” Int.
J. Electr. Power Energy Syst., Vol. 78, pp. 80–87, 2016.
doi:10.1016/j.ijepes.2015.11.094.
REFERENCES [15] A. H. Chowdhury, and M. D. A. Zuman, “Load frequency con-
trol of multi-microgrid using energy storage system,” in Proc.
[1] S. Doolla, T. S. Bhatti, and R. C. Bansal, “Load frequency con- IEEE 8th International Conference on Electrical and Computer
trol of an isolated small hydro power plant using multi pipe Engineering, Dhaka, Bangladesh, Dec. 20–22, 2014, pp. 548–
scheme,” Electr. Power Compon. Syst., Vol. 39, no. 1, pp. 46– 551.
63, 2011. [16] C. X. Dou, D. L. Liu, X. B. Jia, and F. Zhao, “Management and
[2] N. Kang, and Y. Liao, “Contingency studies of a real power sys- control for smart microgrid based on hybrid control theory,”
tem and violation mitigation measures,” Int. Rev. Model. Simul., Electr. Power Compon. Syst., Vol. 39, no. 8, pp. 813–832, May
Vol. 3, no. 5, pp. 882–887, 2010. 2011. doi:10.1080/15325008.2010.541414.
[3] M. Mehrasa, E. Poursemaeil, B. N. Jorgensen, and J. P. S. [17] M. Goyal, and A. Ghosh, “Microgrids interconnection
Catalao, “A control plan for the stable operation of microgrids to support mutually during any contingency,” Sustain-
during grid-connected and islanded modes,” Electr. Power Syst. able Energy Grids Networks, Vol. 6, pp. 100–108, 2016.
Res., Vol. 129, pp. 10–22, 2015. doi:10.1016/j.segan.2016.02.006.
2198 Electric Power Components and Systems, Vol. 45 (2017), No. 19

[18] P. Kundur, Power system stability and control. New York, USA: [32] S. Ahmadi, S. Shokoohi, and H. Bevrani, “A fuzzy logic-based
McGraw-Hill, 1994. droop control for simultaneous voltage and frequency regula-
[19] Q. Fu, et al., “Microgrids: architecture, controls, protection tion in an AC microgrid,” Int. J. Electr. Power Energy Syst.,
and demonstration,” Electr. Power Compon. Syst., Vol. 43, Vol. 64, pp. 148–155, 2015.
no. 12, pp. 1453–1465, July 2015. doi:10.1080/15325008. [33] Y. Arya, and N. Kumar, “Fuzzy gain scheduling controllers for
2015.1039098. automatic generation control of two-area interconnected elec-
[20] J. Yang, et al., “Load frequency control in isolated micro- trical power systems,” Electr. Power Compon. Syst., Vol. 44,
grids with electrical vehicles based on multivariable general- no. 7, pp. 737–751, 2016.
ized predictive theory,” Energies, vol. 8, pp. 2145–2164, 2015. [34] J. Fattahi, H. Schriemer, B. Bacque, R. Orr, and K. Hinzer,
doi:10.3390/en8032145. “High stability adaptive microgrid control method using fuzzy
[21] K. Jagatheesan, B. Anand, and M. Omar, “Design of logic,” Sustainable Cities Soc., Vol. 25, pp. 57–64, 2016.
proportional-integral-derivative controller using ant colony [35] M. Khederzadeh, and H. Maleki, “Frequency control of micro-
optimization technique in multi-area automatic generation con- grids in autonomous mode by a novel control scheme based on
trol,” Int. J. Electr. Eng. Inf., Vol. 7, no. 4, pp. 541–558, 2015. droop characteristics,” Electr. Power Compon. Syst., Vol. 41,
doi:10.15676/ijeei.2015.7.4.1. no. 1, pp. 16–30, 2013.
[22] C. S. A. Nandar, “Robust PI control of smart con- [36] S. Sondhi, and Y. V. Hote, “Fractional order PID controller
trollable load frequency stabilization of microgrid power for perturbed load frequency control using Kharitonov’s the-
system,” Renewable Energy, Vol. 56, pp. 16–23, 2013. orem,” Int. J. Electr. Power Energy Syst., Vol. 78, pp. 884–896,
doi:10.1016/j.renene.2012.10.032. 2016.
[23] R. C. Bansal, “Bibliography on the fuzzy set theory [37] Y. Son, et al., “Robust H load frequency control of delayed
applications to power systems (1994–2001),” IEEE multi-area power system with stochastic disturbances,” Neuro-
Trans. Power Syst., vol. 18, no. 4, pp. 1291–1299, 2003. computing, Vol. 193, pp. 58–67, 2016.
doi:10.1109/TPWRS.2003.818595. [38] R. D. Lopez, et al., “Multi-objective optimization minimiz-
[24] R. C. Bansal, “A bibliographical survey of evolutionary com- ing cost and life cycle emissions of stand-alone PV-wind-
putation applications in power systems (1994–2003),” Int. diesel systems with batteries storage,” Appl. Energy, Vol. 88,
J. Power Energy Syst., Vol. 26, no. 3, pp. 216–225, 2006. pp. 4033–4041, 2011.
doi:10.2316/Journal.203.2006.3.203-3364. [39] X. Li, D. Hui, and X. Lai, “Battery energy storage station
[25] F. Daneshfar, and H. Bevrani, “Multiobjective design (BESS)-based smoothing control of photovoltaic (PV) and
of load frequency control using algorithms,” Int. J. wind power generation fluctuations,” IEEE Trans. Sustain.
Electr. Power Energy Syst., Vol. 42, pp. 257–263, 2012. Energy, Vol. 4, no. 2, Apr. 2013.
doi:10.1016/j.ijepes.2012.04.024. [40] D. H. Tungadio, and R. C. Bansal, “Active power reserve esti-
[26] H. Bevrani, F. Habibi, P. Babahajyani, M. Watanabe, and mation of two interconnected microgrids,” Energy Procedia,
Y. Mitani, “Intelligent frequency control in AC microgrid: Vol. 105, pp. 3909–3914, 2017.
Online PSO-based fuzzy tuning approach,” IEEE Trans. [41] K. S. S. Ramakrishna, P. Sharma, and T. S. Bhatti, “Automatic
Smart Grid, Vol. 3, no. 4, pp. 1935–1944, Dec. 2012. generation control of interconnected power system with diverse
doi:10.1109/TSG.2012.2196806. sources of power generation,” Int. J. Eng. Sci. Technol., Vol. 2,
[27] G. Sharma, I. Nasiruddin, K. R. Niazi, and R. C. Bansal, no. 5, pp. 51–65, 2010.
“Optimal AGC of a multi-area power system with parallel
AC/DC tie lines using output vector feedback control strategy,”
Int. J. Electr. Power Energy Syst., Vol. 81, pp. 22–31, 2016.
doi:10.1016/j.ijepes.2016.02.007. BIOGRAPHIES
[28] X. Li, Y. J. Song, and S. B. Han, “Frequency control in micro-
grid power system combined with electrolyser system and Diambomba Hyacinthe Tungadio received the B.Eng.
fuzzy PI controller,” J. Power Sources, Vol. 180, pp. 468–475, degree in electrical engineering from University of Lubum-
2008. doi:10.1016/j.jpowsour.2008.01.092. bashi (UNILU), Democratic Republic of Congo, in 2003
[29] S. Pothiya, and I. Ngamroo, “Optimal fuzzy logic-
and his master degree in electrical engineering in 2014 from
based PID controller for load-frequency control includ-
ing superconducting magnetic energy storage units,” Tshwane University of Technology (TUT), Pretoria, South
Energy Convers. Manage., Vol. 49, pp. 2833–2838, 2008. Africa. He was with University of Lubumbashi as a lecturer
doi:10.1016/j.enconman.2008.03.010. of power system and electrical measurements from 2004
[30] E. Yesil, M. Guzelkaya, and I. Eksin, “Self-tuning fuzzy PID to 2009. He is currently a PhD student with University of
type load and frequency controller,” Energy Convers. Manage., Pretoria since 2015. His research interests include real-time
Vol. 45, pp. 377–390, 2004.
[31] T. Vigneysh, and N. Kumarappan, “Autonomous operation and
monitoring, power system state estimation, application of
control of photovoltaic/solid oxide fuel cell/battery energy stor- artificial intelligence techniques in power systems, energy
age based microgrid using fuzzy logic controller,” Int. J. Hydro- management, control system and optimal control applications
gen Energy, Vol. 41, pp. 1877–1891, 2016. in energy management.
 Tungadio et al.: Optimal Control of Active Power of Two Microgrids Interconnected with Two AC Tie-lines 2199

Ramesh C. Bansal has more than 25 years of teach- He has diversified research interests in the areas of Renewable
ing, research & industrial experience. Currently he is the Energy and Conventional Power Systems which includes
Professor and Group head (Power) in the Department of wind, PV, hybrid power systems, distributed generation, grid
Electrical, Electronic and Computer Engineering at the integration of renewable energy, power systems analysis (reac-
University of Pretoria, South Africa. In previous post- tive power/voltage control, stability, faults and protection),
ings he was with the University of Queensland, Australia, smart grid, FACTS and power quality.
Birla Institute of Technology and Science, Pilani, India;
the University of the South Pacific, Fiji; and Civil Con- Mukwanga Willy Siti received the B.Sc. and M.Sc. degrees
struction Wing, All India Radio. He has published over in electrical engineering from University of Pretoria (UP),
250 papers in journals and conferences. Prof. Bansal is an Pretoria, South Africa, in 2000 and 2002, respectively. He
Editor/Associate Editor of member many reputed journals has a Ph.D. from University of KwaZulu-Natal since 2015.
including IET-Renewable Power Generation, Technology He is with the Electrical Engineering Department, Tshwane
and Economics of Smart Grids and Sustainable Energy and University of Technology (TUT), Pretoria, South Africa, as a
Electric Power Components and Systems. He is a Fellow, lecturer since 2002. His research interests include power qual-
and CEngg IET-UK, Fellow Engineers Australia, Fellow ity, energy management, state estimation, optimal control and
Institution of Engineers (India) and Senior Member-IEEE. distribution system reconfiguration.