Received: 7 February 2018 Revised: 25 April 2018 Accepted: 6 June 2018

DOI: 10.1002/etep.2646

RESEARCH ARTICLE

Renewable energy source‐based multiarea AGC system with

integration of EV utilizing cascade controller considering
time delay

Arindita Saha | Lalit Chandra Saikia

Department of Electrical Engineering,

National Institute of Technology Silchar,
Summary
Silchar, India Automatic generation control of the unequal 3‐area system with diversified
sources like small hydrothermal units in area‐1 and area‐2 and split shaft gas
Correspondence
Arindita Saha, Department of Electrical turbine thermal units in area‐3 with nonlinearity of generation rate constraint
Engineering, National Institute of is explored. A cascade combination of integer order integral‐derivative with
Technology Silchar, Silchar, India.
filter (IDN) and fractional order proportional derivative (FOPD) is considered
Email: [email protected]
as a secondary controller (IDN‐FOPD). The controller gains and other param-
eters are optimized using whale optimization algorithm. The performance of
IDN‐FOPD controller has an upper hand when compared with some classical
controllers like integral, proportional‐integral, and proportional‐integral‐
derivative with filter (PIDN). Analyses reveal that the performance of a combi-
nation of IDN‐FOPD in area‐1 and PIDN controllers in other areas of system
dynamics provides comparable results with IDN‐FOPD controllers in all areas.
Thus, this combination of controllers is carried for further analyses. Even
eigenvalue analysis has been performed for analyzing system stability. The
impact of time delay on system dynamics is also studied. The impact of the
inclusion of renewable source like a solar thermal power plant in area‐1 and
electric vehicles in all areas provides better system dynamics in the presence
of nonlinearity generation rate constraint and time delay. Sensitivity analysis
reveals that controller gains and other parameters obtained at nominal
conditions are robust to changes in variable solar insolation and demands no
further reset.

KEYWORDS
electric vehicle, generation rate constraint, load frequency control, solar thermal power plant, whale
optimization algorithm

1 | INTRODUCTION

An efficient, economic, and reliable function of a multiarea interconnected power system claims that system fre-
quency and tie‐line power exchange between areas to be maintained at their nominal values. This can be obtained
by close control of both real and reactive powers which are generated through various controllable sources. Thus,
it is necessary to maintain a balance between total powers generated and demanded otherwise it may lead to unde-
sirable effects. This maintenance is served with the concept of ancillary service automatic generation control

Int Trans Electr Energ Syst. 2018;e2646. wileyonlinelibrary.com/journal/etep © 2018 John Wiley & Sons, Ltd. 1 of 22
https://doi.org/10.1002/etep.2646
2 of 22 SAHA AND SAIKIA

(AGC). Many kinds of literature are found in the field of AGC both in isolated as interconnected systems. Elgerd et al1
laid the concept of AGC for a multiarea thermal system. Nanda et al2 carried work for a 2‐area hydrothermal system
with proper utilization of nonlinearity generation rate constraint (GRC). Nanda et al3 performed analysis in an
unequal 3‐area thermal system considering GRC and reheat turbine. Gozde et al4 have analyzed a 2‐area thermal
system but considering another nonlinearity governor dead band (GDB). Bhatt et al5 studied multiarea AGC system
considering diversified sources like hydrothermal diesel. Saikia et al6 have carried work in an interconnected AGC
system of two equal areas as well as 3 and 5 unequal areas. Analyses are carried out considering single reheat turbine
and GRC. But because of certain drawbacks of carbon emissions, it is an urgent need to incorporate renewable
sources for generation. Recently, small hydropower plant (SHPP) is gaining importance because it is a clean source
of renewable energy as well as it is economical. Small hydropower plant has high efficiency, low investment, as well
as environment‐friendly. Dolla et al7 have applied SHPP in the field of dumping load management by controlling
water flow to turbines. Fan et al8 have highlighted the transfer function model of SHPP. Authors in Fan et al8 have
undergone maintenance of deviation in frequency and tie‐line power considering only SHPP units. Sarma et al9 have
emphasized the application of SHPP in the study of AGC of a 2‐area system with other sources of generation. Power
generation from gas turbine‐based units is very common because of their capability of a quick start, reliability, effi-
ciency, and need of less maintenance. There are mainly 2 types of the gas turbine (GT). One of them is single shaft
GT, and the other is split shaft GT (SSGT). Both of them have a major structural difference. In case of single shaft GT,
the generator and turbine are on the same shaft, but in SSGT, they are not on the same shaft; rather, generator is
connected to the turbine via a gearbox. In SSGT, the generator and turbine are mounted on 2 different shafts via a
gearbox. It uses 2 different turbines. One turbine is to drive the air compressor while the other generator via a gear-
box. The gearbox is integrated to reduce the speed. Thus, it is convenient to adopt a conventional generator without
the use of any power electronic interface. Shankar et al10 have introduced SSGT in the field of AGC for a 2‐area
hydrothermal SSGT system. The authors have considered a simple system, ie, without any kind of nonlinearity like
GRC and GDB. The solar energy and wind energy are the most easily available renewable sources of energy. Utiliza-
tion of these resources may lead to less use of the limited availability of fossil fuels and reduce the chances of global
warming. Das et al11 have incorporated solar as well as wind energy in an isolated system. Sharma et al12 have inte-
grated the use of solar thermal power plant (STPP) in the study of AGC. Electric vehicles (EV) are used by Debbarma
and Dutta13 in a deregulated environment of AGC. The EV consists of a battery charger which maintains a balance
between grid and load. Sometimes, a situation might occur when all of a sudden, all EVs get disconnected and come
with undesirable responses. To avoid such situation, a frequency band limit is considered between −10 and +10 mHz.
The EV gain is dependent upon the state of charge of EV. The discharging and charging capacity of EVs for this
analysis is considered within ±5 kW. Thus, such a combination of SHPP‐STPP‐thermal‐SSGT system with EV in
conventional AGC requires more studies.
Automatic generation control has 2 control loops namely primary and secondary control. The design of secondary
control of AGC is a recent trend of research. Various classical controllers are reported in the literature as integral (I),
proportional‐integral (PI), and proportional‐integral‐derivative with filter (PIDN). Many works considering these classi-
cal controllers are reported in past literature. Elgerd et al1 used basic I controller strategy for smooth maintenance of
frequency. Nanda et al2,3 have used PI and I controllers as secondary controllers respectively. Even higher degree of
freedom (DOF) integer order (IO) controllers, fractional order (FO) controller, and cascade controllers are gaining an
upper hand. It is observed that Sahu et al14 applied 2DOF‐PID controller for maintaining system dynamics of a 2‐area
thermal system with GDB. Again, Sahu et al15 proposed the use of a 2DOF‐PID controller for the 2‐area thermal system
and carried their analysis for the multisource system of thermal, hydro, and gas. Javad et al16 carried out work in AGC
using FO PID controller for a multiarea multisource system with consideration of nonlinearities GRC and GDB. Zamani
et al17 proposed the use of FO PID controller for mitigating system deviations for a 2‐area AGC system in the presence of
generation saturation limit. Ismayil et al18 applied FO PID controller for a single‐area AGC system with nonreheat tur-
bines. Recently, Saha et al19 have proposed a cascade combination of IO PIDN and FO PD controller (PIDN‐FOPD) in a
deregulated environment of AGC. Similarly, cascading IO integral‐derivative with filter (IDN) and FO proportional‐
derivative (PD) controllers can also be used whose performance is neither evaluated in conventional AGC system nor
deregulated AGC. Thus, this calls for further investigations.
Most of the work in AGC is done considering a single type of controllers in all areas1-6,9,10,12-20. A very few litera-
ture21 give the idea of using a combination of controllers in the study of AGC under conventional environment. Thus,
this can be explored in the present work of restructured AGC system by using the IDN‐FOPD controller in 1 particular
area and other IO controllers I, PI, and PIDN in other areas one at a time.
SAHA AND SAIKIA 3 of 22

A realistic system needs the inclusion of various nonlinearities like GRC, GDB, time delay, and boiler dynamics.
Most of the work in AGC seems to have GRC and GDB as nonlinearities. Very less literature shows the use of GRC
and time delay as nonlinearities together. Study of the effect of time delay on system dynamics for the system considered
in this present article has not been reported in previous literature. Thus, investigations need to be carried out with both
GRC and time delay.
The performance of the controller is dependent on its optimum values of gains and other parameters. The optimum
values can be obtained by considering various meta‐heuristic algorithms. In the past literature techniques like genetic
algorithm,5 bacteria foraging algorithm,6 firefly algorithm,9 quasioppositional harmony search algorithm,10 gray wolf
optimizer algorithm,12 flower pollination,13 teaching‐learning‐based algorithm,15 hybrid firefly and pattern search,20
craziness‐based particle swarm optimization,4 ant lion optimizer algorithm,22 etc. are used for obtaining the optimum
values. But many of them have the disadvantage of getting trapped in local optima as well as slow convergence. Mirjalili
et al23 proposed a whale optimization algorithm (WOA) which overrides all such disadvantages because it performs

FIGURE 1 The block diagram and TF of the unequal 3‐area system with SHPP thermal units in both area‐1 and area‐2 and SSGT thermal
in area‐3 along with STPP, EV, and time delay: A, block diagram of considered 3‐area system and B, TF model of considered system
4 of 22 SAHA AND SAIKIA

exploration and exploitation separately using 2 different set of equations. Thus, it shows reduced chances of getting
trapped in local optima as well as has high convergence. Recently, Saha et al19 have used WOA in the deregulated
environment, but its use in the conventional system is more to be done. Thus, utilization of WOA in conventional
AGC in such system combination needs studies.
Survey of past literature in AGC, it is observed that the controller gains and other parameters of the system are opti-
mized at nominal loading conditions. A step load perturbation (SLP) of 1% is used as a disturbance. But there may be
some abnormal conditions like a change in loading conditions, changes in parameters which will have an impact on
dynamic responses. To prove the robustness of optimum parameters obtained at nominal, sensitivity analysis is under-
taken. Sensitivity analysis for the proposed controller in this considered system of work is not reported in the previous
literature and hence needs more studies.
With regard to above highlights, the main objectives of the present work are as follows:

a. Development of a 3‐area conventional AGC system considering GRC, using SHPP thermal units in area‐1 and area‐
2 and SSGT thermal units in area‐3.
b. Simultaneous optimization of I, PI, PIDN, and IDN‐FOPD using WOA and comparison of dynamic responses.
c. Selection of best combination of controllers in the system and eigenvalue analysis.
d. Study of the effect of the presence of time delay on system dynamics for system mentioned in (a) using the best‐
obtained controller from (c).
e. The system mentioned in (d) is integrated with STPP, and its effect is studied using the best‐obtained controller
from (c).

FIGURE 2 Structure of proposed controller—IDN‐FOPD controller

TABLE 1 Optimum value of controller gains and other parameters under 1% SLP in area‐1

Controllers Gains/Parameters Area‐1 Area‐2 Area‐3

I KIi* 0.8547 0.1247 0.6761

PI KPi* 0.0002 0.0047 0.0094
KIi* 0.7327 0.1341 0.5220
PIDN KPi* 0.9213 0.3648 0.6325
KIi* 0.7528 0.9513 0.6121
KDi* 0.2529 0.4332 0.3996
Ni* 11.00 10.00 12.00
IDN‐FOPD KIi* 0.2203 0.2713 0.2112
KDi* 0.8127 0.6627 0.7227
Ni* 68.98 76.86 78.25
Kpi* 0.9281 0.2656 0.3154
Kdi* 0.7228 0.0741 0.9307
μi * 0.0120 0.0116 0.0126
SAHA AND SAIKIA 5 of 22

FIGURE 3 Analyzing dynamic responses vs time of WOA optimized I, PI, PIDN, and IDN‐FOPD controllers for system with SHPP
thermal in area‐1 and area‐2 and SSGT thermal in area‐3 at 1% SLP in area‐1 at nominal loading: A, frequency change in area‐1, B,
frequency change in area‐2, C, tie‐line power change in the line connecting area‐2 and area‐3, and D, tie‐line power change in the line
connecting area‐1 and area‐3
6 of 22 SAHA AND SAIKIA

f. The system mentioned in (e) is integrated with EV, and its effect is studied using the best‐obtained controller
from (c).
g. Sensitivity analysis is performed to check the robustness of controller gains and other parameters to changes in
solar insolation.

2 | SYSTEM INVESTIGATED

An unequal 3‐area multisource AGC system with area capacity ratio 1:3:4 is considered for the study. The system
comprises of SHPP thermal units in both area‐1 and area‐2 and SSGT thermal unit in area‐3. The system is later
incorporated into renewable energy source STPP in area‐1 and EVs in all areas. The system is made realistic with
the proper association of GRC of 3%/minute for thermal units and time delay. An SLP of 1% is considered as a dis-
turbance in area‐1. The system which comprises of SHPP thermal units in area‐1 and area‐2 and SSGT thermal in
area‐3 have area participation factors (pdfs) of apf11 = 0.8 and apf12 = 0.2 in area‐1, apf21 = 0.8 and apf22 = 0.2 in
area‐2, and apf31 = 0.7 and apf32 = 0.3 in area‐3. This set of apf is regarded as apfset‐1. The system comprising of
STPP‐SHPP thermal units in area‐1, SHPP thermal units in area‐2, and SSGT thermal units in area‐3 have apfs of
apf11 = 0.6, apf12 = 0.2, and apf13 = 0.2 in area‐1, apf21 = 0.8 and apf22 = 0.2 in area‐2, and apf31 = 0.7 and apf32 = 0.3
in area‐3. This set of apf is regarded as apfset‐2. The system consisting of STPP‐SHPP thermal and EV units in area‐1,
SHPP thermal and EV units in area‐2, and SSGT thermal and EV units in area‐3 have apfs of apf11 = 0.4, apf12 = 0.2,
apf13 = 0.2, and apf14 = 0.2 in area‐1; apf21 = 0.6, apf22 = 0.2, and apf23 = 0.2 in area‐2; and apf31 = 0.5, apf32 = 0.3,

TABLE 2 Peak overshoot (PO), peak undershoot (PU), and settling time (ST) of Figure 3

Controllers Measures Figure 3A Figure 3B Figure 3C Figure 3D

I PO (+ve) 0.0289 0.0264 0.0047 0.0056

PU (−ve) 0.0360 0.0278 0.0044 0.0097
ST (s) 150 144 146 146
PI PO (+ve) 0.0204 0.0207 0.0041 0.0047
PU (−ve) 0.0359 0.0279 0.0042 0.0098
ST (s) 120 104 138 113
PIDN PO (+ve) 0.0194 0.0192 0.0046 0.0041
PU (−ve) 0.0358 0.0278 0.0040 0.0097
ST (s) 110 91 80 98
IDN‐FOPD PO (+ve) 0.0114 0.0073 0.0007 0.0019
PU (−ve) 0.0356 0.0275 0.0039 0.0091
ST (s) 80 73 75 90

TABLE 3 Optimum values of controller gains and other parameters of IDN‐FOPD in area‐1 and I, PI, and PIDN in area‐2 and area‐3
independently

Optimum Values of Controller Gain and Other Parameters

Different Combination of Controllers
Combination‐1 IDN‐FOPD in Area‐1 and Combination‐2 IDN‐FOPD in Area‐1 and Combination‐3 IDN‐FOPD in Area‐1 and
I in Area‐2 and Area‐3 PI in Area‐2 and Area‐3 PIDN in Area‐2 and Area‐3

Area‐1: KI1* = 0.8876, KD1* = 0.2543, Area‐1: KI1* = 0.5296, KD1* = 0.7608, Area‐1: KI1* = 0.2423, KD1* = 0.7412,
N1* = 28.67, Kp1* = 0.8490, Kd1* = 0.3086, N1* = 95.00, Kp1* = 0.9109, Kd1* = 0.9522, N1* = 66.00, Kp1* = 0.1815, Kd1* = 0.8612,
μ1* = 0.0202 μ1* = 0.0195 μ1* = 0.0125
Area‐2: KI2* = 0.5843 Area‐2: KP2* = 0.0095, KI2* = 0.6075 Area‐2: KP2* = 0.0195, KI2* = 0.2503,
KD2* = 0.7531, N2* = 42.00
Area‐3: KI3* = 0.4136 Area‐3: KP3* = 0.0096, KI3* = 0.4504 Area‐3: KP3* = 0.0164, KI3* = 0.4392,
KD3* = 0.9631, N3* = 56.00
SAHA AND SAIKIA 7 of 22

FIGURE 4 Analyzing dynamic responses vs time of WOA optimized different combinations of controllers for system with SHPP thermal
in area‐1 and area‐2 and SSGT thermal in area‐3 at 1% SLP in area‐1 at nominal loading: A, frequency change in area‐1, B, frequency change
in area‐3, C, tie‐line power change in the line connecting area‐1 and area‐2, and D, tie‐line power change in the line connecting area‐1 and
area‐3
8 of 22 SAHA AND SAIKIA

and apf33 = 0.2 in area‐3. This set of apf is regarded as apfset‐3. The block diagram and transfer function (TF) of
the system integrated with renewable sources, nonlinearities, and EV is shown in Figure 1A, B respectively. The
nominal parameters for thermal, SHPP, SSGT, STPP, and EV are taken from Nanda et al,2 Sarma et al,9 Shankar
and Mukherjee,10 Sharma and Saikai,12 and Debbarma and Dutta,13 respectively. The controller gains and
other parameters are optimized utilizing WOA considering integral squared error as objective function as given by
Equation 1.
Tn

2 o
J ISE ¼ ∫ ðΔf i Þ2 þ ΔPtie i− j dt (1)
0

3 | I D N‐ F O P D C O N T R O L L ER

The proposed controller is cascade combination of an IO controller IDN‐FOPD in this conventional system of AGC. The
IDN part of the controller is given by Equation 2.

ðK Di N i Þs2 þ K Ii s þ K Ii N i
Ci ðsÞ ¼ (2)
s ðs þ N i Þ

It is known that fractional calculus is the base of FO controllers. The Riemann‐Liouville (R‐L) definition for
fractional derivative can be obtained from Equation 3.

FIGURE 5 Analyzing dynamic responses vs time for IDN‐FOPD controller in all areas and IDN‐FOPD in area‐1 and PIDN controllers in
other area‐2 and area‐3: A, frequency change in area‐1 and B, frequency change in area‐2
SAHA AND SAIKIA 9 of 22

t
1 dn
αDαt ¼ ∫ ðt−τ Þn−α−1 f ðτ Þ dτ (3)
Γðn−αÞdt n α

where n − 1 ≤ α < n, n is an integer, Γ(.) is the Euler gamma function, and αDαt is the fractional operator. The Laplace
transformation of above Equation 3 is given by Equation 4.

  n−1
L αDαt f ðt Þ ¼ sα F ðsÞ− ∑ sk αDα−k−1
t f ðt Þjt¼0 (4)
k¼0

Therefore, differential equations can even describe a dynamic system behavior under the assumptions of zero initial
conditions. This may lead to a transfer function with the FO of s. There may be an infinite number of poles and zeros.
Oustaloup et al24 have proposed a transfer function which approximates both FO derivative and integrator within the
range of frequency [ωl, ωh] using the recursive distribution of poles and zeros given by Equation 5. Here, this frequency
range is considered as [0.01, 50].

TABLE 4 Eigenvalues and minimum damping ratio of the system with different secondary controllers

Minimum Damping
Controllers Eigenvalues Ratio (ξ)

I −0.3333 + 0.0000i, −12.8860 + 0.0000i, −12.8944 + 0.0000i, −12.9090 + 0.0000i, −0.1682 + 3.3252i, 0.0505
−0.1682 − 3.3252i, −0.1585 + 3.1233i, −0.1585 − 3.1233i, −2.4345 + 0.0000i, −2.3506 + 0.0000i,
−1.5617 + 0.7594i, −1.5617 − 0.7594i, −0.4881 + 1.0759i, −0.4881 − 1.0759i, −2.0000 + 0.0000i,
−1.9999 + 0.0000i, −0.2322 + 0.1838i, −0.2322 − 0.1838i, −0.2398 + 0.0151i, −0.2398 − 0.0151i,
−0.1601 + 0.0148i, −0.1601 − 0.0148i, −0.0899 + 0.0764i, −0.0899 − 0.0764i, 0.0000 + 0.0000i,
−0.0347 + 0.0000i, −0.0322 + 0.0000i, 0.0000 + 0.0000i, 0.0000 + 0.0000i, 0.0000 + 0.0000i
PI −0.3333 + 0.0000i, −12.9119 + 0.0000i, −12.8883 + 0.0000i, −12.8978 + 0.0000i, −0.1768 + 3.3335i, 0.0530
−0.1768 − 3.3335i, −0.1746 + 3.1307i, −0.1746 − 3.1307i, −2.4440 + 0.0000i, −2.3604 + 0.0000i,
−1.5418 + 0.8462i, −1.5418 − 0.8462i, −0.5206 + 1.0173i, −0.5206 − 1.0173i, −1.9999 + 0.0000i,
−2.0000 + 0.0000i, −0.1918 + 0.1711i, −0.1918 − 0.1711i, −0.2386 + 0.0000i, −0.2252 + 0.0000i,
−0.1563 + 0.0229i, −0.1563 − 0.0229i, −0.0905 + 0.0820i, −0.0905 − 0.0820i, −0.0347 + 0.0000i,
−0.0322 + 0.0000i, −0.0000 + 0.0000i
PIDN −0.3333 + 0.0000i, −16.0210 + 0.0000i, −15.2824 + 0.0000i, −14.9384 + 0.0000i, −8.5912 + 0.0000i, 0.0970
−7.7886 + 0.0000i, −0.3918 + 4.0178i, −0.3918 − 4.0178i, −0.6109 + 3.7952i, −0.6109 − 3.7952i,
−4.2853 + 1.2267i, −4.2853 − 1.2267i, −2.6592 + 0.0000i, −0.6259 + 0.8915i, −0.6259 − 0.8915i,
−2.2242 + 0.0000i, −1.7492 + 0.0000i, −2.0000 + 0.0001i, −2.0000 − 0.0001i, −0.2222 + 0.1675i,
−0.2222 − 0.1675i, −0.1705 + 0.1272i, −0.1705 − 0.1272i, −0.2318 + 0.0041i, −0.2318 − 0.0041i,
−0.1696 + 0.0208i, −0.1696 − 0.0208i, −0.0336 + 0.0000i, −0.0344 + 0.0000i, 0.0000 + 0.0000i
IDN‐FOPD −0.3333 + 0.0000i, −77.4014 + 0.0000i, −76.9047 + 0.0000i, −66.8244 + 0.0000i, −55.2605 + 0.0000i, 0.2899
−52.6746 + 0.0000i, −50.2338 + 0.0000i, −34.4326 + 0.0000i, −34.7620 + 0.0000i,
−34.7281 + 0.0000i, −15.9710 + 0.0000i, −15.7362 + 0.0000i, −15.6338 + 0.0000i,
−11.8881 + 0.0000i, −6.0736 + 5.6606i, −6.0736 − 5.6606i, −6.9402 + 1.9404i, −6.9402 − 1.9404i,
−7.4209 + 0.0000i, −6.7723 + 0.0000i, −4.1842 + 3.5407i, −4.1842 − 3.5407i, −1.0217 + 3.3725i,
−1.0217 − 3.3725i, −3.5609 + 0.0000i, −3.4098 + 0.0000i, −3.2169 + 0.0000i, −2.3426 + 0.0000i,
−1.4906 + 0.4185i, −1.4906 − 0.4185i, −2.0000 + 0.0000i, −1.9999 + 0.0000i, −1.5775 + 0.0000i,
−1.4474 + 0.0599i, −1.4474 − 0.0599i, −0.4219 + 0.6225i, −0.4219 − 0.6225i, −0.7440 + 0.0193i,
−0.7440 − 0.0193i, −0.7255 + 0.0000i, −0.3459 + 0.0055i, −0.3459 − 0.0055i, −0.3343 + 0.0000i,
−0.1084 + 0.1230i, −0.1084 − 0.1230i, −0.2166 + 0.0000i, −0.2059 + 0.0000i, −0.0764 + 0.0661i,
−0.0764 − 0.0661i, −0.1289 + 0.0147i, −0.1289 − 0.0147i, −0.1566 + 0.0000i, −0.1539 + 0.0000i,
−0.1499 + 0.0000i, −0.0676 + 0.0000i, −0.0699 + 0.0000i, −0.0696 + 0.0000i, −0.0350 + 0.0000i,
−0.0294 + 0.0000i, −0.0334 + 0.0000i, −0.0327 + 0.0000i, −0.0316 + 0.0000i, −0.0145 + 0.0000i,
−0.0150 + 0.0000i, −0.0149 + 0.0000i, −0.0000 + 0.0000i, −50.0000 + 0.0000i, −50.0000 + 0.0000i,
−50.0000 + 0.0000i
a
Bold signify best values.
10 of 22 SAHA AND SAIKIA

 
M 1þ s
=ωZ;n
sα ¼ K ∏   (5)
n¼1 1 þ s
=ωp;n

Thus, in Equation 5, if it is considered that “K” which is adjusted gain is 1 then gain is zero db for 1 rad/second
frequency, “M” is the predefined number of poles and zeros, and range of frequencies for poles and zeros is given by
Equations 6 to 10.
pffiffiffi
ωZ;l ¼ ωl n (6)

ωp;n ¼ ωZ;n ε; n ¼ 1; …; M (7)

pffiffiffi
ωZ;nþ1 ¼ ωp;n η; n ¼ 1; :…; M−1 (8)

ε ¼ ðωh =ωl Þ =M
v
(9)

TABLE 5 Eigenvalues and minimum damping ratio of the system with a different combination of secondary controllers

Minimum
Damping Ratio
Controller Combination Eigenvalues (ξ)

IDN‐FOPD in area‐1 and I in −0.3333 + 0.0000i, −49.8747 + 0.0000i, −33.9824 + 0.0000i, −29.6793 + 0.0000i, 0.0978
area‐2 and area‐3 −15.6609 + 0.0000i, −10.9748 + 0.0000i, −12.9064 + 0.0000i, −12.8945 + 0.0000i,
−7.2241 + 0.0000i, −0.4955 + 3.4504i, −0.4955 − 3.4504i, −0.2613 + 2.6587i,
−0.2613 − 2.6587i, −3.3301 + 0.0000i, −1.9219 + 0.7365i, −1.9219 − 0.7365i,
−0.3702 + 1.1176i, −0.3702 − 1.1176i, −2.1760 + 0.1083i, −2.1760 − 0.1083i,
−1.5362 + 0.0000i, −1.9995 + 0.0000i, −1.9998 + 0.0000i, −0.7080 + 0.0000i,
−0.3268 + 0.0000i, −0.2287 + 0.1366i, −0.2287 − 0.1366i, −0.1531 + 0.1447i,
−0.1531 − 0.1447i, −0.2523 + 0.0000i, −0.2261 + 0.0000i, −0.1602 + 0.0397i,
−0.1602 − 0.0397i, −0.1504 + 0.0000i, −0.0693 + 0.0000i, −0.0147 + 0.0000i,
−0.0344 + 0.0000i, −0.0339 + 0.0000i, −0.0320 + 0.0000i, 0.0000 + 0.0000i,
−50.0000 + 0.0000i, 0.0000 + 0.0000i, 0.0000 + 0.0000i
IDN‐FOPD in area‐1 and PI in −0.3333 + 0.0000i, −94.9219 + 0.0000i, −51.7340 + 0.0000i, −34.0322 + 0.0000i, 0.0543
area‐2 and area‐3 −15.6621 + 0.0000i, −12.9060 + 0.0000i, −12.8938 + 0.0000i, −4.8116 + 5.2632i,
−4.8116 − 5.2632i, −7.1599 + 0.0000i, −0.1722 + 3.1696i, −0.1722 − 3.1696i,
−1.6956 + 2.1929i, −1.6956 − 2.1929i, −3.2268 + 0.0960i, −3.2268 − 0.0960i,
−2.3692 + 0.0000i, −0.4102 + 0.8542i, −0.4102 − 0.8542i, −1.5271 + 0.0000i,
−2.0000 + 0.0000i, −2.0000 + 0.0000i, −1.1041 + 0.0000i, −0.7162 + 0.0000i,
−0.2312 + 0.1638i, −0.2312 − 0.1638i, −0.3315 + 0.0000i, −0.1743 + 0.1606i,
−0.1743 − 0.1606i, −0.2433 + 0.0000i, −0.2343 + 0.0000i, −0.1672 + 0.0288i,
−0.1672 − 0.0288i, −0.1503 + 0.0000i, −0.0693 + 0.0000i, −0.0344 + 0.0000i,
−0.0336 + 0.0000i, −0.0319 + 0.0000i, −0.0147 + 0.0000i, 0.0000 + 0.0000i,
−50.0000 + 0.0000i
IDN‐FOPD in area‐1 and PIDN −0.3333 + 0.0000i, −64.1538 + 0.0000i, −53.2947 + 0.0000i, −56.8266 + 0.0000i, 0.3907
in area‐2 and area‐3 −43.0514 + 0.0000i, −33.9704 + 0.0000i, −15.6517 + 0.0000i, −9.2082 + 0.0000i,
−8.6673 + 0.0000i, −6.9762 + 3.7094i, −6.9762 − 3.7094i, −7.2411 + 0.0000i,
−1.7909 + 4.2191i, −1.7909 − 4.2191i, −2.0000 + 3.6650i, −2.0000 − 3.6650i,
−3.3284 + 0.0000i, −1.8248 + 0.0000i, −1.5936 + 0.1181i, −1.5936 − 0.1181i,
−1.5205 + 0.0000i, −2.0003 + 0.0000i, −2.0001 + 0.0000i, −0.3957 + 0.6335i,
−0.3957 − 0.6335i, −0.7081 + 0.0000i, −0.3264 + 0.0000i, −0.1105 + 0.1426i,
−0.1105 − 0.1426i, −0.2425 + 0.0000i, −0.2120 + 0.0000i, −0.0873 + 0.1045i,
−0.0873 − 0.1045i, −0.1542 + 0.0250i, −0.1542 − 0.0250i, −0.1502 + 0.0000i,
−0.0692 + 0.0000i, −0.0147 + 0.0000i, −0.0352 + 0.0000i, −0.0325 + 0.0000i,
−0.0319 + 0.0000i, −0.0000 + 0.0000i, −50.0000 + 0.0000i

Bold signify best values.

SAHA AND SAIKIA 11 of 22

ð1−vÞ
=
η ¼ ðωn =ωl Þ=M (10)

Therefore, the non‐IO of derivative μ can be any real number, and the transfer function for FOPD is given by
Equation 11.

F i ðsÞ ¼ K pi þ K di sμi (11)

Therefore, the ultimate output signal from IDN‐FOPD controller is given by Equation 12.


ðK Di N i Þs2 þ K Ii s þ K Ii N i
μi

U i ðs Þ ¼ × K pi þ K di s ×ACE i (12)
s ðs þ N i Þ

The IDN‐FOPD controller gains and parameters of Equation 12 are optimized using WOA to get minimum value of
cost function JISE given in Equation 1 subject to the constraint given in Equation 13.

K Ii min ≤K Ii ≤K Ii max
K Di min ≤K Di ≤K Di max
N i min ≤N i ≤N i max
(13)
K p i min ≤K pi ≤K p i max
K di min ≤K di ≤K di max
μi min ≤μi ≤μi max

4 | WOA TECHNIQUE

Mirjalili et al23 proposed a meta‐heuristic algorithm namely WOA based on the foraging behavior of humpback whales.
Whale optimization algorithm maintains a balance between exploitation and exploration which helps in finding global
optima. Whale optimization algorithm has a specialty compared to other algorithms as it performs 2 phases, exploration

FIGURE 6 Convergence characteristics of FA, ALO, PSO, and WOA

TABLE 6 Optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains and parameters with time delay

Areas Optimum Values of Gains and Parameters

Area‐1 (with IDN‐FOPD controller) KI1* = 0.4288 KD1* = 0.7235 N1* = 72 Kp1* = 0.8975 Kd1* = 0.1535 μ1* = 0.0111
Area‐2 (with PIDN controller) KP2* = 0.0026 KI2* = 0.6752 KD2* = 0.4030 N2* = 65
Area‐3 (with PIDN controller) KP3* = 0.0006 KI3* = 0.6033 KD3* = 0.4022 N3* = 58.41
12 of 22 SAHA AND SAIKIA

and exploitation, separately using 2 different equations. In the first half of iterations, WOA shows high exploration
where whales are allowed to move randomly, and in the next half, it shows high exploitation and convergence where
they rapidly relocate themselves around or in spiral‐shaped form toward the best‐obtained location so far. Thus, it
shows reduced chances of getting trapped in local optima as well as has high convergence. The detailed explanation
of WOA is given in Mirjalili and Lewis.23 Whale optimization algorithm is used for optimizations of controller gains

FIGURE 7 Analyzing dynamic responses vs time for IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 with and without time delay: A,
frequency change in area‐1, B, frequency change in area‐2, and C, tie‐line power change in the line connecting area‐1 and area‐2
SAHA AND SAIKIA 13 of 22

and other parameters. The input variables are a number of search agents, maximum iterations, and ! a which reduces
from 2 to 0 during iterations. The secondary controllers I, PI, PIDN, and IDN‐FOPD are used one at a time, and their
gains and other parameters are optimized. The step‐by‐step method for input parameter tuning is as follows:

1 Initialize the input variables: number of search agents (sa) and a number of iterations (noi). The input variable ! a
reduces from 2 to 0 during iterations.
2 At first, “sa” is kept fixed at some value and “noi” is varied. For each simulation, the value of JISE is noted. The value
of noi corresponding to the minimum value of JISE is selected.
3 After fixing noi, again sa value is changed and value of JISE is noted. The value of sa corresponding to the minimum
value of JISE is selected.
4 Thus, after multiple simulations, the values of sa and noi are fixed.

In this present work, sa = 50 and noi = 100. The flow chart for easy understanding of WOA is given in Saha and
Saika.19

5 | RESULTS AND ANALYS IS

5.1 | Performance comparison of I, PI, PIDN, and IDN‐FOPD controllers

The system considered for analysis comprises of SHPP thermal units in both area‐1 and area‐2, and SSGT thermal in
area‐3 is considered for the study. This system is associated with controllers I, PI, PIDN, and IDN‐FOPD each indepen-
dently. The system dynamics are obtained considering 1% SLP as a disturbance in area‐1. Simulations are done using
MATLAB R2015a tool. Simultaneous optimization of controller gains and other parameters are done using WOA tech-
nique. ISE is used as a performance index. The optimized values of each controller gains and other parameters are given
in Table 1. The dynamic responses are obtained with these optimized values and compared in Figure 3. Only 4
responses are shown in Figure 3. The entities of these shown responses like peak overshoot (PO), peak undershoots
(PU), and settling time (ST) are given in Table 2. Critical observation of Figure 3 and Table 2 gives clear idea that
IDN‐FOPD controller performs better than other abovementioned controllers in PO, PU, and ST. Thus, the IDN‐FOPD
controller is efficient enough to carry forward for further analysis.

5.2 | Utilization of IDN‐FOPD controller in area‐1 and I, PI, and PIDN in area‐2 and area‐3
each combination one at a time
It is observed from Figure 3 that using only I, PI, and PIDN in all areas provides an oscillatory responses showing much
peak deviations and higher settling time. This condition of high peak deviations and settling time can be further ana-
lyzed considering the best controller IDN‐FOPD obtained from Section 5.1 in area‐1 and each other classical controllers
I, PI, and PIDN in area‐2 and 3 each one at a time. The controller gains and other parameters for these conditions (a)
IDN‐FOPD in area‐1 and I in area‐2 and 3, (b) IDN‐FOPD in area‐1 and PI in area‐2 and area‐3, (c) IDN‐FOPD in area‐1
and PIDN in area‐2 and area‐3 are optimized using WOA technique. The optimum values are given in Table 3, and the
dynamic responses corresponding to these obtained optimum values are shown in Figure 4. The responses shown in
Figure 4 depicts that the best combination is IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 showing that this com-
bination is having settling time and much lesser peak deviations in all responses.

TABLE 7 Optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains and other parameters with STPP in
area‐1

Areas Optimum Values of Gains and Parameters

Area‐1 (with IDN‐FOPD controller) KI1* = 0.4324 KD1* = 0.1300 N1* = 26 Kp1* = 0.0942 Kd1* = 0.7964 μ1* = 0.1292
Area‐2 (with PIDN controller) KP2* = 0.0155 KI2* = 0.2163 KD2* = 0.1273 N2* = 29
Area‐3 (with PIDN controller) KP3* = 0.0111 KI3* = 0.1592 KD3* = 0.1412 N3* = 29
14 of 22 SAHA AND SAIKIA

FIGURE 8 Analyzing dynamic responses vs time for IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 with and without STPP in area‐1:
A, frequency change in area‐1, B, frequency change in area‐2, C, frequency change in area‐3, and D, tie‐line power change in the line
connecting area‐2 and area‐3
SAHA AND SAIKIA 15 of 22

5.3 | Dynamic response comparison of IDN‐FOPD controllers in all areas and IDN‐FOPD
in area‐1 and PIDN in area‐2 and area‐3
The dynamic responses corresponding to the best‐obtained combination of IDN‐FOPD in area‐1 and PIDN in area‐2 and
area‐3 obtained from Section 5.2 is compared with responses considering IDN‐FOPD controllers in all areas. The
dynamic responses are compared in Figure 5. It is inferred from Figure 5 that using IDN‐FOPD in area‐1 and PIDN
in area‐2 and area‐3 shows comparable results with responses using IDN‐FOPD controllers in all areas. Only 2 dynamic
responses are provided. Because this combination has less tuning parameters compared to IDN‐FOPD controllers in all
areas, further analysis is carried out using IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3.

5.4 | Eigenvalue analysis

In this subsection, eigenvalue analysis is performed for stability analysis of the system under controlled manner, ie, in
the presence of secondary controllers. The considered system analysis comprises of SHPP thermal units in both area‐1
and area‐2, and SSGT thermal in area‐3. The system is provided with different secondary controllers I, PI, PIDN, and
IDN‐FOPD independently, and eigenvalue analysis is performed. In this case, the same controller is used in all areas.
The eigenvalues are provided in Table 4 (column 2). Analysis of the values says that using I, PI, and PIDN as secondary
controllers, there is a possibility where a condition may arise where the system may become unstable as there are few
values which are having zero real part which makes the system marginally stable. When the system is provided with
IDN‐FOPD as a secondary controller, then all eigenvalues are having a negative real part, so the system is completely
stable. Even the value of minimum damping ratio (ξ) corresponding to IDN‐FOPD controller (ξ = 0.2899) (Table 4
column 3 row 5) is much more than the values obtained with other controllers. Thus, the system with IDN‐FOPD
controller reaches stable conditions faster than others.
From Table 4, it is observed that the system with IDN‐FOPD controller is showing better stability and highest value
of ξ. Now, the same considered system is provided with a different combination of controller, and stability analysis is
performed. The different combinations are used one at a time, and the combinations are (a) IDN‐FOPD in area‐1
and I in area‐2 and area‐3, (b) IDN‐FOPD in area‐1 and PI in area‐2 and area‐3, and (c) IDN‐FOPD in area‐1 and PIDN
in area‐2 and area‐3. The eigenvalues and value of ξ are provided in Table 5. In Table 5, it is observed that the system
with a combination of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 is showing better stability with all negative
real part of eigenvalues and higher value of ξ.
The system with a combination of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 is having ξ = 0.3907 which is
higher than ξ = 0.2899 for IDN‐FOPD in all areas. So, the system with a combination of IDN‐FOPD in area‐1 and PIDN
in area‐2 and area‐3 has a higher possibility of reaching stability than IDN‐FOPD in all areas. Therefore, further
analyses are done using the IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 combinations.

5.5 | Performance of WOA with other meta‐heuristic algorithms

The system comprising of SHPP thermal units in both area‐1 and area‐2 and SSGT thermal in area‐3 with a combination
of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 as secondary controllers are considered for the study. The con-
sidered system with WOA optimized such combination of controllers is compared with other algorithms like FA, ALO,
and PSO. The optimum values of gains and other parameters using different algorithms are not provided. The conver-
gence characteristic comparison of FA, ALO, PSO, and WOA is shown in Figure 6. From Figure 6, it is inferred that the
value of JISE corresponding to WOA is lesser compared to other considered algorithms. It is also observed that WOA
converges faster compared to FA, ALO, and PSO in the considered system.

TABLE 8 Optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains and other parameters with STPP in
area‐1 and EVs in all areas

Areas Optimum Values of Gains and Parameters

Area‐1 (with IDN‐FOPD controller) KI1* = 0.9731 KD1* = 0.4621 N1* = 26 Kp1* = 0.1476 Kd1* = 0.0001 μ1* = 0.0211
Area‐2 (with PIDN controller) KP2* = 0.1478 KI2* = 0.8625 KD2* = 0.1150 N2* = 35
Area‐3 (with PIDN controller) KP3* = 0.2455 KI3* = 0.9714 KD3* = 0.1558 N3* = 22
16 of 22 SAHA AND SAIKIA

FIGURE 9 Analyzing dynamic responses vs time for IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 in the presence of STPP in area‐1
with and without EVs in all areas: A, frequency change in area‐1, B, frequency change in area‐2, C, tie‐line power change in the line
connecting area‐2 and area‐3, and D, tie‐line power change in the line connecting area‐1 and area‐3
SAHA AND SAIKIA 17 of 22

5.6 | System dynamics in the presence of time delay with IDN‐FOPD controller in area‐1
and PIDN in area‐2 and area‐3
In the particular subsection, the effect of time delay (τd) is considered for the study. The dynamic responses with
τd = 0.1 second are considered for analysis. The IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 are considered
as secondary controllers, and optimum values of gains and other parameters are obtained with the help of WOA tech-
nique. The obtained optimum values are given in Table 6. The dynamic responses obtained with time delay are com-
pared with responses without time delay to examine the effect of time delay on system dynamics as shown in
Figure 7. From Figure 7, it is seen that system dynamics degrades in the presence of time delay with a higher value
of settling time and peak deviations. But, from the realistic point of view, inclusion of time delay is necessary. Thus,
further analyses are carried out with consideration of time delay and IDN‐FOPD in area‐1 and PIDN in area‐2 and
area‐3.

5.7 | System dynamics in the presence of STPP with IDN‐FOPD in area‐1 and PIDN in
area‐2 and area‐3
The system mentioned in Section 5.6 which consists of SHPP thermal units in area‐1 and area‐2 and SSGT thermal in
area‐3 with time delay in each area is integrated with renewable source solar as STPP in area‐1. The best‐obtained con-
troller combination IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 is considered for analysis. The IDN‐FOPD in
area‐1 and PIDN in area‐2 and area‐3 controller gains and other parameters are optimized using WOA, and optimum
values are given in Table 7. The dynamic responses corresponding to these optimum values are compared with
responses without STPP in Figure 8. Only 4 dynamic responses are shown to reflect the effectiveness of incorporation
of STPP. On keen observation of Figure 8, it is seen that incorporation of STPP leads to responses with reduced oscil-
lations and settling time. It is seen that when STPP is incorporated, the dynamic responses have settling time in lesser
range than the system in the absence of STPP using IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controllers.
Thus, the system dynamics are improved.

5.8 | System dynamics in the presence of EVs with IDN‐FOPD in area‐1 and PIDN in area‐
2 and area‐3
In this subsection, aggregates of EVs are integrated into each area of the system considered in Section 5.7 with STPP.
The IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller combination is used for analysis of the effect of
the presence of EVs for a system with STPP. The IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains
and other parameters are optimized using WOA, and optimum values are given in Table 8. The dynamic responses cor-
responding to these optimum values are compared with responses for a system with STPP in Figure 9. Only 4 dynamic
responses are shown to reflect the effectiveness of incorporation of EVs. On keen observation of Figure 9, it is seen that
incorporation of EVs in the system leads to responses with reduced peak deviations and fast settling time. It is seen that
when EVs are incorporated, the dynamic responses have to settle in the range of 20 to 35 seconds, providing a noticeable
improvement in system dynamics.

5.9 | Variable solar insolation sensitivity analysis

Sensitivity analysis of proposed controller combination IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 is per-
formed for a system with both STPP and EVs. It is done to check the robustness of controller gains and parameters

TABLE 9 Optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains and other parameters for variable solar
insolation

Areas Optimum Values of Gains and Parameters

Area‐1 (with IDN‐FOPD controller) KI1* = 0.8646 KD1* = 0.1395 N1* = 15 Kp1* = 0.1900 Kd1* = 0.0002 μ1* = 0.0182
Area‐2 (with PIDN controller) KP2* = 0.1170 KI2* = 0.9555 KD2* = 0.1601 N2* = 18
Area‐3 (with PIDN controller) KP3* = 0.3366 KI3* = 0.8144 KD3* = 0.1635 N3* = 18
18 of 22 SAHA AND SAIKIA

FIGURE 10 Analysis vs time of WOA optimized IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller with time delay for system
with STPP‐SHPP‐thermal‐EV in area‐1, SHPP‐thermal‐EV in area‐2, and SSGT‐thermal‐EV in area‐3 for variable solar insolation: A, variable
solar insolation, B, frequency change in area‐2, C, frequency change in area‐3, and D, tie‐line power change in the line connecting area‐1 and
area‐2
SAHA AND SAIKIA 19 of 22

obtained at nominal conditions. A good controller is the one which can handle any worst conditions without any need
of reset because an interconnected system will have disturbances. The robustness of controller gains and other param-
eters can be checked for variable solar insolation. The controller gains and other parameters are optimized using WOA.
The optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller gains and parameters are given in
Table 9. The dynamic responses corresponding to these optimum values are compared with responses with nominal
values at the changed condition of variable solar insolation. The dynamic response comparisons are shown in
Figure 10. Figure 10 interprets that optimum values of IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller
gains and other parameters are robust to any such kind of changes and further resetting is not required.

6 | CONCLUSION

An analysis of an unequal 3‐area multisource system comprising of SHPP thermal in area‐1 and area‐2 and SSGT ther-
mal in area‐3 is done in this article. Appropriate GRC is also considered as nonlinearity. The controller gains and other
parameters are optimized using WOA technique. It is seen that IDN‐FOPD controller outperforms other classical con-
trollers I, PI, and PIDN. Performance comparison is done to study the best combination of controllers like IDN‐FOPD in
area‐1 and either I, PI, or PIDN in other areas. It is inferred that the combination of IDN‐FOPD in area‐1 and PIDN in
area‐2 and area‐3 gives better combination. The performance comparison of the best combination with IDN‐PIDN con-
trollers in all areas shows that the responses are more or less similar. Thus, further analyses are done using IDN‐FOPD
in area‐1 and PIDN in area‐2 and area‐3 controller combinations. Eigenvalue analysis is performed to infer about the
stability of the system. Investigation of convergence curve shows that WOA has faster convergence and less value of cost
function compared to other algorithms. The analysis is done to study the effect of inclusion of time delay. A system with
the inclusion of STPP in area‐1 shows improvement of dynamic responses with reduced oscillations and lesser settling
time. For further stabilization of system dynamics, EVs are included in all areas. The inclusion of EVs along with STPP
in the system provides much‐improved responses with reduced peak deviations and reduced settling time. Sensitivity
analysis reveals that the optimum values of controller gains and parameters obtained at nominal conditions are robust
to any changes in solar insolation, and no further resetting is not required. Thus, it is suggested to use such a
combination of renewable source STPP with EVs using IDN‐FOPD in area‐1 and PIDN in area‐2 and area‐3 controller
combination for improving system dynamics considering nonlinearities of GRC and time delay.

L I S T O F A B B R E V I A T I O N S A N D S Y M BO L S

ACE area control error

AGC automatic generation control
aij –(Pri/Prj)
ALFC automatic load frequency control
apf area participation factor
Bi frequency bias coefficient of area i
Di ΔPDi/Δ f i (pu MW/Hz)
Dtur turbine damping in pu of SSGT
EV electric vehicle
f nominal frequency, Hz
FO fractional order
GDB governor dead band
GRC generation rate constraint
Hi inertia constant of area i, s
I integral
i subscript refers to area i (i = 1, 2, 3) for two area
IO integer order
ISE integral squared error
Kdi derivative gain of FOPD controllers for area i
KDi derivative gain of PIDN and IDN controller for area i
20 of 22 SAHA AND SAIKIA

KEVi gain of EV of area i

KIi integral gain of I, PI, PIDN, and IDN controllers for area i
Kpi 1/Di (Hz/pu MW)
Kpi proportional gain of FOPD controller for area i
KPi proportional gain of PI and PIDN controllers for area i
Kri reheat turbine gain for thermal power plant of area i
Ksi solar field gain of STPP of area i
KT temperature loop gain of SSGT, s
Lmax load limit of SSGT
N number of search agents
Ni derivative filter coefficient of PIDN for area i
PI proportional‐integral controller
pi π
PIDN proportional‐integral derivative with controller
Pri rated power of area i, MW
Ri governor speed regulation parameter of area i (Hz/pu MW)
SHPP small hydro power plant
SLP step load perturbation
SSGT split shaft gas turbine
STPP solar thermal power plant
T simulation time, s
T1 fuel time lag constant of fuel opening valve block of SSGT, s
T2 fuel time lag constant of fuel system block of SSGT, s
t2i reset time constant of SHPP of area i, s
T3 load limiter time constant of the exhaust temperature block of SSGT, s
TEVi time constant of EV of area i, s
Tgi steam governor time constant for thermal power plant of area i, s
Tgsi steam governor time constant for STPP of area i, s
T ij synchronizing coefficients
Tpi (2 × Hi)/( f i × Di), s
Tri reheat turbine time constant for thermal power plant of area i, s
tRi transient droop time constant of SHPP of area i, s
TSHgi governor time constant of SHPP of area i, s
TSHi nominal starting time of water of area i, s
Tsi solar field time constant of STPP of area i, s
Tti turbine time constant for thermal power plant of area i, s
Ttsi steam turbine time constant for STPP for area i, s
WOA whale optimization algorithm
* superscript denotes optimum value
Δfi deviation in frequency of area i, Hz
ΔPDi incremental load change in area i (pu MW)
ΔPgi incremental generation change in area i (pu MW)
ΔPtie i‐j incremental change in tie‐line power in the line connecting area i and area j (i ≠ j)
βi area frequency response characteristics of area i (=Di + 1/Ri)
μi order of differentiator of FOPD controllers in area i

ORCID
Arindita Saha http://orcid.org/0000-0002-4927-1341
SAHA AND SAIKIA 21 of 22

R EF E RE N C E S
1. Elgerd OI, Fosha CE. Optimum megawatt‐frequency control of multiarea electric energy systems. IEEE Trans Power App Syst.
1970;89(4):556‐563.
2. Nanda J, Mangla A, Suri S. Some new findings on automatic generation control of an interconnected hydrothermal system with conven-
tional controllers. IEEE Transactions on Energy Conversion. 2006;21(1):187‐194.
3. Nanda J, Mishra S, Saikia LC. Maiden application of bacterial foraging based optimization technique in multiarea automatic generation
control. IEEE Trans Power Syst. 2009;24(2):602‐609.
4. Gozde H, Cengiz Taplamacioglu M. Automatic generation control application with craziness based particle swarm optimization in a
thermal power system. Int J Electr Power Energy Syst. 2011;33(1):8‐16.
5. Bhatt P, Roy R, Ghoshal SP. GA/particle swarm intelligence based optimization of two specific varieties of controller devices applied to
two‐area multi‐units automatic generation control. Int J Electr Power Energy Syst. 2010;32(4):299‐310.
6. Saikia LC, Nanda J, Mishra S. Performance comparison of several classical controllers in AGC for multi‐area interconnected thermal
system. Int J Electr Power Energy Syst. 2011;33(3):394‐401.
7. Doolla S, Bhatti TS. Load frequency control of an isolated small‐hydro power plant with reduced dump load. IEEE Trans Power Syst.
2006;21(4):1912‐1919.
8. Fan RX, Zhao J, Pan B, Chen N, Wang T, Ma H. Automatic generation control of three‐area small hydro system based on fuzzy IDD
control. International Conference on Power System Technology (POWERCON)2014: 2522‐2528.
9. Sarma U, Saikia LC, Sana A, Dash P. Load frequency control of a multi‐area, multi‐source system using firefly algorithm optimized
F2DOFIDD controller. IEEE Region 10 Conference (TENCON), Singapore 2016:1475‐1479.
10. Shankar G, Mukherjee V. Quasi oppositional harmony search algorithm based controller tuning for load frequency control of multi‐
source multi‐area power system. Int J Electr Power Energy Syst. 2016;75:289‐302.
11. Das DC, Roy AK, Sinha N. GA based frequency controller for solar thermal–diesel–wind hybrid energy generation/energy storage system.
Int J Electr Power Energy Syst. 2012;43(1):262‐279.
12. Sharma Y, Saikia LC. Automatic generation control of a multi‐area ST‐thermal power system using Grey wolf optimizer algorithm based
classical controllers. Int J Electr Power Energy Syst. 2015;73:853‐862.
13. Debbarma S, Dutta A. Utilizing electric vehicles for LFC in restructured power systems using fractional order controller. IEEE Trans
Smart Grid. 2017;8(6):2554‐2564.
14. Sahu RK, Panda S, Rout UK. DE optimized parallel 2‐DOF PID controller for load frequency control of power system with governor dead‐
band nonlinearity. Int J Electr Power Energy Syst. 2013;49:19‐33.
15. Sahu RK, Panda S, Rout UK, Sahoo DK. Teaching learning based optimization algorithm for automatic generation control of power
system using 2‐DOF PID controller. Int J Electr Power Energy Syst. 2016;77:287‐301.
16. Javad M, Kazem Z, Mehrdad TH. Applying fractional order PID to design TCSC‐based damping controller in coordination with automatic
generation control of interconnected multi‐source power system. Eng Sci Technol Int J. 2017;20(1):1‐17.
17. Zamani A, Barakati SM, Yousofi‐Darmian S. Design of a fractional order PID controller using GBMO algorithm for load–frequency
control with governor saturation consideration. ISA Trans. 2016;64:56‐66.
18. Ismayil C, Sreerama RK, Sindhu TK. Automatic generation control of single area thermal power system with fractional order PID (PIλDμ)
controllers. IFAC Proc Vol. 2014;47(1):552‐557.
19. Saha A, Saikia LC. Utilization of ultra‐capacitor in load frequency control under restructured STPP‐thermal power systems using WOA
optimized PIDN‐FOPD controller. IET Gener Transm Distrib. 2017;11(13):3318‐3331. https://doi.org/10.1049/iet‐gtd.2017.0083
20. Sahu RK, Panda S, Padhan S. A hybrid firefly algorithm and pattern search technique for automatic generation control of multi area
power systems. Int J Electr Power Energy Syst. 2015;64:9‐23.
21. Saikia LC, Chowdhury A, Shakya N, Shukla S, Soni P K. AGC of a multi area gas‐thermal system using firefly optimized IDF controller.
2013 Annual IEEE India Conference (INDICON). 2013 Mumbai. https://doi.org/10.1109/INDCON.2013.6725998:1‐6
22. More R, Saikia LC, Sinha N. Automatic generation control of a multi‐area system using ant lion optimizer algorithm based PID plus
second order derivative controller. Int J Electr Power Energy Syst. 2016;80:52‐63.
23. Mirjalili S, Lewis A. The whale optimization algorithm. Adv Eng Softw. 2016;95:51‐67.
24. Oustaloup A, Mathieu B, Lanusse p. The CRONE control of resonant plants: application to a flexible transmission. Eur J Control.
1995;1(2):113‐121.

How to cite this article: Saha A, Saikia LC. Renewable energy source‐based multiarea AGC system with
integration of EV utilizing cascade controller considering time delay. Int Trans Electr Energ Syst. 2018;e2646.
https://doi.org/10.1002/etep.2646
22 of 22 SAHA AND SAIKIA

A P P EN D I X

Nominal parameters of system: f = 60 Hz; loading = 50%; Kpi = 120 Hz/p.u. MW; Tpi = 20 seconds; T12 = 0.086 p.u.
MW/rad; Hi = 5 seconds; Di = 8.33 × 10−3 p.u. MW/Hz; Bi = βi = 0.425 p.u. MW/Hz; Ri = 2.4 Hz/p.u. MW
Thermal unit: Tgi = 0.08 second, Tti = 0.3 second, Tri = 10 seconds, Kri = 0.5
SHPP: t2i = 0.496 second, tRi = 5 seconds, TSHgi = 36.5 seconds, TSHi = 1 second
SSGT: T1 = T2 = 1.5 seconds, T3 = 3 seconds, Lmax = 1, KT = 1, FOVmin = −0.02, FOVmax = 1, Dtur = 0 p.u.
STPP: Ks = 1.8, Ts = 1.8 seconds, Tgs = 1.0 second, Tts = 3.0 seconds
EV: KEVi = 1, TEVi = 1 second.