IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO.

12, DECEMBER 2023 7431

EV Hosting Capacity Enhancement in a Community

Microgrid Through Dynamic Price
Optimization-Based Demand Response
Md Juel Rana , Forhad Zaman , Tapabrata Ray , Senior Member, IEEE, and Ruhul Sarker , Member, IEEE

Abstract—Community microgrids, as an emerging technology, Variables

offer resiliency in operation for smart grids. Microgrids are
seeing an increased penetration of eco-friendly electric vehicles
πt Dynamic price function.
(EVs) in recent years. However, the uncontrolled charging of πt,h
DR Dynamic price for selected user h for DR.
EVs can easily overwhelm such electric networks. In this work, Et /Etev
batt Energy level of in-house battery/EV battery.
we propose an efficient demand response (DR) scheme based Pbatt
t Charging/discharging power of battery installed
on dynamic pricing to enhance the capacity of the microgrid to at the house.
securely host a large number of EVs. A hierarchical two-level
optimization framework is introduced to realize the DR scheme. Pev
t Charging/discharging power of EV.
At the upper level, the dynamic prices for the participating users PMG
t Power exchanged by household with microgrid.
PV
in DR are optimized while at the lower level, each user optimizes Pt roof_top Power from roof-top PV.
its energy consumption based on the price signal from the upper
level. An evolutionary algorithm and a mixed-integer linear pro-
gramming model is employed to solve the upper and lower level I. I NTRODUCTION
problems, respectively. Energy scheduling problems of the users OMMUNITY microgrids are an indispensable part of
are solved in a distributed manner which adds to the scalabil-
ity of the approach. The proposed DR scheme is tested on a
C the future smart grids that promote efficient utilization
of local distributed power generation. In recent years, envi-
microgrid system adopted from the IEEE European low-voltage
distribution network. Numerical experiments confirm the effec- ronmental concerns have prompted an increased adoption of
tiveness of the proposed DR scheme compared to the benchmark electric vehicles (EVs). EVs are high energy-consuming appli-
pricing policies from the literature. ances when compared to other residential loads. Uncontrolled
Index Terms—Demand response (DR), dynamic pricing, elec- charging of EVs not only raises the energy cost of the house-
tric vehicle (EV) hosting capacity, evolutionary algorithm (EA), hold but also creates network issues that negatively affect
microgrid. the secure operation of the microgrid. These network issues
mainly include voltage deviation from statutory limits, over-
N OMENCLATURE loading of the distribution lines and transformers [1]. These
Sets and Parameters factors limit the EV hosting capacity which is defined as the
δ Conversion loss for charging/discharging. maximum number of EVs that can be hosted by the elec-
B Set of buses of the test electric network. tric network without negatively affecting its normal operating
L Set of lines of the test electric network. conditions [2]. Hence, coordinated control of EV charging is
N Set of residential customers. of utmost importance for increased EV hosting capacity and
∇t time-step. reliable operation of the microgrid.
batt /Eev
Einit Initial energy level of in-house battery/EV
init
battery. A. Motivation
ev
Etarget Target energy level for EV battery. Although demand response (DR) is envisioned as an effec-
Hselected Selected users for DR. tive tool to aid the secure operation of the electric grid, its
T Time of simulation. success depends on its ability to motivate users’ participation.
tdeadline Deadline time for EV charging. One of the forms of DR is the incentive-based scheme such as
tplug_in Plug-in time for EV. direct load control (DLC) [3], where utility companies directly
control (on/off) the user’s appliances by providing agreed
Manuscript received 8 June 2022; accepted 31 July 2022. Date of pub-
lication 31 August 2022; date of current version 22 November 2023. This financial incentives. The main deterrent for users to enroll in
work was supported by the Australian Research Council under Discovery such schemes stems from the fear of compromising comfort
Grant DP190101271. This article was recommended by Associate Editor level. Another form is the price-based DR, where the price is
S. Senatore. (Corresponding author: Md Juel Rana.)
The authors are with the School of Engineering and Information designed to motivate users to consume less during high-price
Technology, University of New South Wales, Canberra, ACT 2610, instants and vice versa [4]. This form of DR is voluntary;
Australia (e-mail: [email protected]). thus, users are in control of their comfort level. The most
Color versions of one or more figures in this article are available at
https://doi.org/10.1109/TCYB.2022.3196651. common price-based DR adopts time-of-use (ToU) [5] pric-
Digital Object Identifier 10.1109/TCYB.2022.3196651 ing in which different prices are charged at different periods.
2168-2267 
c 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7432 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

However, the ToU fails to reflect the short-term variation in to their amount of energy consumption even if the microgrid
the supply–demand balance as they are set months in advance. has ample generation. Other works focusing on distributed
The dynamic pricing which is set a day or less in advance can dynamic pricing-based DR have also been reported in [15],
effectively match the supply–demand variations. The dynamic [16], and [17]. In [18], DR based on the mechanism of the
pricing is also conducive to the integration of renewables as energy storage system (ESS) sharing between operator and
well as supports wider adoption of EVs [6]. Nevertheless, high users is presented. Bilevel programming was utilized to model
price volatility from the pure form of dynamic pricing makes the interaction between the SO and the users. Both price-
it less attractive for residential customers. Therefore, in this based and incentive-based DRs are modeled for a microgrid
work, we focus on a DR scheme based on effective dynamic in [19]. As part of the price-based DR, tariff of ToU and
pricing to enhance the EV hosting capacity of the microgrid. real-time pricing (RTP) are optimized using particle swarm
optimization (PSO) for maximizing social welfare. In [20], a
bilevel optimization framework is proposed aiming to optimize
B. Related Works the transactive price signals for DR in a distribution network.
In recent years, many studies have focused on the price- In the upper level, the utility company computes the price sig-
based DR models. In [7], dynamic pricing was designed nals for DR aggregators, whereas, in the lower level, energy
for scheduling flexible loads in the distribution networks. consumptions are modified by DR aggregators in response to
The loads were scheduled using centralized optimization via price signals. A price-based residential DR is proposed in [21]
DR aggregators. A centralized DR management system is where reinforcement learning and fuzzy reasoning were uti-
proposed in [8] to maximize the social welfare of the com- lized to implement the scheme. An event-triggered mechanism
munity. In the proposed system, DR aggregators facilitate the is used in [22] for the hybrid energy system operation consid-
interaction between the utility and the end users. Alternating ering price-based DR. The multiagent system and consensus
direction method of multipliers (ADMM), a iteration-based algorithm are used for the proposed DR.
algorithm, is used to solve the DR model. A DR scheme to mit- However, in most of the works except in [7] unified dynamic
igate distribution network congestion caused by EVs and heat price is offered to all users without differentiating among
pumps is proposed in [9]. In this work, the distribution system the type of energy consumers. In reality, the users with
operator (SO) implements dynamic tariffs and daily power- energy-intensive flexible loads such as EVs are most likely
based network tariffs to manage congestion through a central- to pose negative impacts on the electric network; conse-
ized coordinated home energy management system (HEMS). quently, they should pay more than those with fixed/regular
The problem is modeled in a general algebraic modeling energy consumers. This may lead to cross-subsidies among
system (GAMS) and solved by CPLEX. Nevertheless, the the energy-intensive customers with EVs and those with reg-
direct control of EVs by utility considered in this work ular usage. Furthermore, in the pure form of dynamic pricing
may affect end-user preferences. Furthermore, the centralized (such as in [13]), peak rebounding of loads (especially with
optimization-based DR scheme suffers from scalability issues EVs) may still occur without coordination among the users.
with the increasing number of customers. To overcome this
problem, DR models with distributed optimization have been
reported. In these models, scheduling of appliances in the con- C. Contributions of This Article
sumer premises is done locally. In [10], a distributed dynamic Based on the aforementioned discussion, this work introduces
pricing policy (D2P) was proposed for the management of EV an efficient dynamic-pricing-driven DR for a microgrid to
charging within a multimicrogrid environment. The dynamic support high penetration of EVs. An effective dynamic price
pricing is determined based on the supply–demand imbalance, function is proposed based on the supply–demand imbalance
where EVs from home microgrids see lower prices than for- in the system. In the proposed dynamic price function, for
eign microgrids. However, the possibility of simultaneous EV fairness, the users with regular residential consumption remain
charging resulting in peak loads in the microgrids have been unaffected by the price spikes caused by the energy-intensive
ignored. Dynamic pricing-based EV load management in the consumers with EVs at all instants. On the other hand, any price
distribution network is also proposed in [11]. A DR frame- reductions at the instants with ample renewable generation in the
work based on reinforcement learning is proposed in [12] microgrid are enjoyed by all users. In such a scheme, renewable
to optimize the dynamic price for the end users. However, generation is promoted by encouraging users to consume more
the scalability of the framework remains questionable as it at those periods. Each user optimizes their energy profile locally
was tested with only three customers. Real-time dynamic in response to price functions and reports to the SO. The SO
pricing was proposed in [13] to reduce peak loads through assesses the optimized net load and triggers a special DR event
the management of EV charging and discharging. A decen- if there are any network issues. An optimization framework
tralized cloud-based architecture was used for the proposed conforming to the bilevel programming is proposed to realize
scheme. However, the proposed price function only accounts the DR scheme. At the upper level, coordinated dynamic prices
for the price variation above the reference value ignoring any are optimized for a set of users chosen through a fair selection
price decrease during abundant supply from renewable gener- strategy with a view to improve the net load profile of the
ations. In [14], a distributed DR algorithm-based pricing was system. At the lower level, all selected users for DR optimize
proposed to manage plug-in hybrid EVs within the microgrid. their profiles in a distributed fashion. The interaction between
In such a pricing model, the users are charged proportionally the upper and lower level progresses iteratively until all the

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 RANA et al.: EV HOSTING CAPACITY ENHANCEMENT IN A COMMUNITY MICROGRID 7433

network issues are eliminated. The contributions of this work

can be summarized as follows.
1) The proposed dynamic price function prevents cross-
subsidies among users.
2) A DR model based on coordinated dynamic price
optimization is proposed. The optimization of user’s
energy scheduling is realized in a distributed fash-
ion offering scalability (i.e., can accommodate a large
number of customers).
3) The participants for DR are selected following a fair
selection strategy and the motivation of guaranteed
financial gains attracts users’ participation.
4) Numerical experiments show that the proposed DR
scheme results in an improved EV hosting capacity and
better financial gains for the consumers compared to
other schemes.

D. Organization of This Article Fig. 1. Schematic of the system and conceptual framework of the DR model.

The remainder of this article is organized as follows.

Section II presents the overview of the system and the math- energy profile in the cloud environment. The net load of the
ematical formulation of the problem. The proposed solution system is then derived from the optimized energy profiles in
approach is elaborated in Section III. Section IV discusses the edge cloud to be sent to the core cloud. In the core cloud,
and analyzes the results obtained from numerical experiments. the SO performs power-flow analysis with the net load and
Finally, the conclusions are summarized in Section V. calls for DR if there is any possible network issues. Once DR
is activated the SO selects potential customers to participate in
II. S YSTEM M ODELING AND P ROBLEM F ORMULATION DR and optimizes day-ahead dynamic prices for them through
iterative interactions between SO and the selected HEOs until
This section presents the overview of the system and the network issues are resolved. The process of iterative inter-
details of the underlying mathematical model. actions creates large communication traffic which needs to
be managed efficiently for a successful implementation of
A. System Overview DR. The important factor for establishing efficient cloud-based
The schematic of a residential microgrid connected to the communication is to maintain higher quality of service (QoS),
utility grid is presented in Fig. 1. The microgrid meets the which depends on the two major factors such as the band-
customer energy demand from its own distributed generations width and communication delays. Implementation details of
(renewable and nonrenewables) as well as power imported cloud computing is out of scope of this article and readers are
from the upper level utility grid. It is assumed that some of the referred to [23] and [24] for further details.
households have their own generations from installed rooftop
photovoltaic (PV) and battery systems. Different levels of EV C. Dynamic Price Function
penetration are considered based on the availability of EVs in
Based on the forecasted baseline load profile of the
a house. Since the majority of the EV owners (about 83% [2])
microgrid, the SO determines the day-ahead dynamic price
prefer to charge at home, only home charging is considered.
(πt ) of electricity based on the supply–demand imbalance
The charging demand of EVs is considered as a flexible load
(λt ) according to (1)–(3). The variation of πt around the
as it can be shifted/modified. Each household is assumed to be
reference/base price πb reflects the actual generation capac-
equipped with HEMS. The HEMS comprises a local controller
ity/supply (St ) and system-wide energy demand (Dt ). To avoid
(LC) and a home energy optimizer (HEO). The LC implements
continuous variation in the price (high volatility is a con-
the optimized schedule of the home appliances and the EVs
cern) with supply–demand imbalance, the price is increased
obtained from HEO.
above πb only when the demand-to-supply ratio (Rt ) exceeds
a certain threshold (Rth ). On the contrary, the price reduc-
B. Cloud Computing Platform for Distributed Optimization tion limited by π min occurs only when demand is less than
Each household optimizes its energy usage in a distributed supply and renewable generations (Pren t ) in the microgrid
manner based on the price signals received from the SO. exceeds a certain threshold (Pren
th ). To establish fairness, the
For this purpose, a cloud computing platform consisting of increased prices at any instant are imposed on the users who
the edge cloud and core cloud are considered. The edge are responsible for the price surge, that is, users who are
clouds are closer to the end users that reduces end-to-end consuming more energy (e.g., by charging EVs) than the max-
latency [23]. The advantages of the cloud computing for DR imum threshold for individual residential consumers. At the
implementation lies in its reliability, scalability, and flexible instants of usual energy consumption (when electric network
characteristics. HEO of the individual houses optimizes the is not congested/has enough supply), users receive the resulted

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7434 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

t
regular dynamic prices. The price reductions due to high
Etev = Einit
ev
+ s (1 − δ)∇t, (charging)
pev
(11)
renewable generation are enjoyed by all users
s=tplug_in s (1 + δ)∇t, (discharging)
pev
 ev
Emin ≤ Etev ≤ Emax
ev
(12)
Dno_EV − St , without EVs
λt = t
(1) Etevdeadline = Etarget .
ev
(13)
Dwith_EV
t − St , with EVs
⎧ no_EV The power balance constraint in the house is enforced by
⎨ Dt , without EVs PVroof_top
Rt = St (2) PMG + Pt = Pbatt + Pev
t + Pt
Hnon_EV
. (14)
⎩ Dt
with_EV
t t
St , with EVs
⎧ Now, the optimization problem for HEO can be written in a
⎨ πb + aλ2t +bλt ,  if λt > 0, Rt > Rth
compact form as follows:
πt = max {πb − aλ2t + bλt }, π min , if λt < 0, Pren
t > Pth
ren
⎩
πb , else min f (xHEO ) = CH ∀H ∈ N (15)
xHEO
(3) Subject to: (6)−(14).
where a and b are both positive constants and their values are The vector of the decision variables (xHEO ) consists of the
set at the discretion of the SO. active power of the available energy consuming devices as well
One can observe that the above dynamic price function can as the power exchanged with the microgrid for the respective
avoid cross-subsidies among customers by offering different house.
prices for energy-intensive (EV charging) customers and gen-
eral users (users without EV charging). At the same time, it E. Proposed Optimization Framework for DR
facilitates renewable generations by giving price incentives in
certain periods. It is important to note that with Rth = 1, the Each user solves the optimization problem (15) based on
πt turns into the pure form of the dynamic price that varies the price signal (πt ) received from the SO and reports back
with any variation in λt . their energy demand from the MG. The SO performs power-
flow studies of the MG network and activates the DR program
if there are any potential network issues. The DR program
D. Home Energy Optimizer
is based on an optimization framework resembling bilevel
The individual houses schedule their energy usage using programming as discussed in the following.
the HEO to minimize the total cost (CH ). The decision vari- 1) Upper Level Problem: In the upper level problem, the
ables for HEO are the active powers of the energy-consuming SO selects the customers with EV charging demand for DR
devices available in a particular house as well as the power based on a fair selection strategy which will be discussed
exchanged (PMG t ) with the microgrid. The house pays the in the next section. Then, for each of the selected users,
microgrid as per πt when PMG t > 0 and gets paid according dynamic price optimization is performed to minimize the peak-
to feed-in-tariff (FiT) if PMG
t < 0. to-average ratio (PAR) of the system net load that resolves all
The cost of energy for a house is determined by the network issues. The decision variables of this optimization
T problem (Xupper ) are the dynamic prices for the selected users
CH = Ct (4) within their preferred window of EV charging as shown in (16)
t=1  DR
Xupper = πt,1 . . . πt,1
DR
. . . πt,h
DR
. . . πt,h
DR
∀h ∈ Hselected ⊂ N.
Ct = t πt ,
PMG if PMG
t ≥0
(5)
Pt FiT, if PMG
MG < 0. (16)
t
The operation of the battery in any battery-installed house is The price optimization is subjected to the constraint (17) to
governed by the following constraints: guarantee financial gain of the DR participants. It computes
the cost of EV charging with the updated price πt,h DR which

max ≤ Pt
− Pbatt batt
≤ Pbatt
max (6) needs to be less or equal than that of usual price πt . The
t network constraints for statutory voltage limits as well as line
Etbatt = Einit
batt
+ s (1 − δ)∇t,
pbatt (charging)
(7) and transformer loading are also considered as presented in
s (1 + δ)∇t,
pbatt (discharging)
s=1 the following:
batt
Emin ≤ Etbatt ≤ Emax
batt
. (8)
Chev (πt,h
DR
) ≤ Chev (πt ) (17)
For houses with EVs, it is considered that vehicle-to-house pu pu pu
Vmin ≤ ≤ Vmax ∀b ∈ B
Vb,t (18)
(V2H) energy transfer is possible by discharging EV battery to
minimize energy costs. The charging/discharging of the EV is Loadingline
l,t ≤ lmax ∀l ∈ L
Loadingline (19)
performed within the user’s preference defined by plug-in time transformer
Loadingt ≤ Loadingtransformer
max . (20)
(tplug_in ), deadline of charging (tdeadline ) and required energy
ev ). EV charging/discharging is characterized by the Now, the optimization problem for the upper level can be
level (Etarget
represented as follows:
following constraints:
min f (xupper ) = PAR (21)
tplug_in ≤ ∀t ≤ tdeadline (9) xupper

−Pev
max ≤ Pt ≤ Pmax
ev ev
(10) Subject to: (17)−(20).

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 RANA et al.: EV HOSTING CAPACITY ENHANCEMENT IN A COMMUNITY MICROGRID 7435

2) Lower Level Problem: The lower level optimization Algorithm 1 Proposed Optimization Framework for Dynamic
problem shown in (22) remains similar to (15) and is solved Price-Based DR
simultaneously for individual users. However, only selected Require: Price (πt ), user preferences for EV charging, num-
user/houses is considered in this case ber of generations (Ng ), population size (Np ) and other EA
parameters
min = f (xHEO ) = Ch ∀h ∈ Hselected ⊂ N (22) Output: Dynamic prices (πt,hDR ) for users participating in DR
xHEO
Subject to: (6)−(14). 1: Get system netload by solving (15). {Initial generation}
2: Perform powerflow with system netload
The formulation of problem (22) is a mixed-integer nonlin- 3: Identify time indexes of network issues (NIt ) if any
ear programming (MINLP) model that contains both continu- 4: NIt = {tVoltage_violation tTranformer_overloading tLine_overloading }
ous and integer decision variables. This nonlinear model can ∀t ∈ T
be converted to an equivalent mixed-integer linear program- 5: if NIt = ∅ then
ming (MILP) model using a linearization approach. In this 6: Set minimum number of users (nDR user ) to call for DR
7: while NIt = ∅ do
research, the MILP approach is chosen for its ability to off-
8: Select users according to Algorithm 2
set the imprecision emanating from linearization with benefits
9: Encode Xupper for selected users as in (16)
of faster computation and global optima as compared to the 10: Set generation count g = 0
MINLP approach. The steps of linearization using standard Initialize (Pinit
11: g ). {Initial set of solutions}
techniques have been presented in the Appendix. Evaluate (Pinit
12: g ) by solving lower level problem of
The outcomes of upper and lower levels are dependent on (22)
each other. In simple words, the dynamic prices from the upper 13: for 1 : Ng do
level affects the energy usage of the users in the lower level 14: Select parent solutions Pg from current population
and responses of the users to price signals in the lower level 15: Generate child Cg from Pg via crossover and
modifies the PAR of the net load in the upper level. mutation
16: Evaluate Cg
17: Sort Tx = Pg ∪ Cg
III. S OLUTION A PPROACH
18: Pg+1 ←− Tx (taking top Np solutions)
In this section, the solution approach is detailed. To solve 19: g = g + 1.
the upper level problem, an evolutionary algorithm (EA) is 20: end for
employed. On the other hand, MILP is used to solve the 21: Perform powerflow with the netload obrained from
lower level problem. The pseudocode of the proposed solu- best solution
tion approach is presented in Algorithm 1. First, the users 22: Update NIt
report their optimized energy profile by solving (15) based 23: Increase nDRuser
on the dynamic price πt . Then, the SO performs power-flow 24: end while
analysis using the resulted net load of the system and stores 25: Broadcast final πt,h DR

the time indexes of any consequent network issues (lines 1–4). 26: else
The network issue indicates any occurrence of statutory bus 27: No additional DR action is needed
voltage limit violation as well as transformer and distribution 28: end if
lines overloading. For any sort of network issue, the proposed
DR scheme is activated and solved (lines 5–25). Otherwise,
Algorithm 2 Strategy for Selecting Users for DR
the scheduled energy profiles based on πt are followed by
respective LCs in the user’s premises. The proposed scheme Require: minimum number of users (nDR user ), time indexes of
comprises of three main steps: 1) selection of the potential network issues (NIt ), optimized energy profile of the houses
users to invite for DR; 2) solution of the upper level problem (Et,H ) with respect to πt , historical DR compliance rate (CRH )
via EA; and 3) distributed solution of lower level problems of the users
(HEOs) using MILP. These steps are discussed succinctly as Output: List of selected users for DR (Hselected )
follows. 1: Enumerate energy consumption (ENIt ,H ) of all users at
the instants of network issues
2: Rank users based on the merit of ENIt ,H and CRH
A. Strategy for Selecting Users for DR 3: List user-id as Fi from non-dominated fronts of ranked
The strategy for selecting users to participate in DR is users (in ascending order of fronts)
presented in Algorithm 2. At the instants where there are 4: Hselected ←− Fi (1 to nDR
user )
network issues, the energy consumption of all users are cal-
culated. Also, an archive of the historical DR compliance rate
is maintained for all users. The compliance rate indicates how
well a user complies with the self-optimized energy profile and compliance rate following the principle of nondominated
reported to the SO while participating in the DR program. sorting (line 2 in Algorithm 2). This ensures the fairness of the
Failure to follow the reported energy profile during actual con- selection process and increases the probability of success rate
sumption decreases the compliance rate of the user. Then, all for acquiring the required DR. Finally, from the top-ranked
the users are ranked based on the merit of energy consumption users, the required numbers of participants are selected for DR.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7436 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

TABLE I
B. EA for Upper Level Optimization S YSTEM PARAMETERS
An EA is used to optimize the dynamic prices for the users
participating in DR. EA solves the upper level problem (21)
where the solutions are encoded as in (16). In the begin-
ning, a set of initial solutions are randomly generated for the
encoded decision variables; which are then evaluated by solv-
ing lower level problems. From the current population, parent
solutions (Pg ) are selected which are used to generate child
solutions (Cg ) employing crossover and mutation operation. In
this work, the well-known simulated binary crossover (SBX)
and polynomial mutation [25] are considered. Binary tourna-
ment [26] strategy is followed to select parents that increases 114 residential customers distributed throughout the network.
the likelihood of choosing elite parents for child generation. About 30% [28] of the household is considered to have installed
The evaluated child solutions are then combined with the rooftop PV with 9% [29] having battery system that conforms
parent solution and the combined solutions are ranked. For with Australian statistics. The real rooftop PV production data
ranking, the population is divided into feasible and infeasi- of Brisbane, Australia, with a range of PV panel capacities are
ble solutions. The infeasible solutions are ranked based on the taken from [30]. The hourly load data for the households are
amount of constraint violation (CV) with the preference to the taken from [31]. The statutory voltage limits of the electric
lower value of CV. The feasible solutions are ranked based network are set as +10%/−6% as per Australian standard AS
on the minimum value of the objective function. The over- 61000.3.100 [32]. The real EV trial data are extracted from [33]
all ranking process adheres to the feasible first strategy where to model the home EV charging behavior. To make it more
feasible solutions obtain a higher rank above the infeasible realistic, three popular EV models are considered and uni-
ones. The parent solutions for the next generation are selected formly assigned among the houses. These include Nissan Leaf
from the top-ranked solutions. The mechanism of passing best of battery size of 24 and 40 kWh, as well as Tesla Model 3 with
solutions in the subsequent generations facilitates continu- a battery capacity of 50 kWh. Standard type-2 EV chargers for
ous improvement in the objective value, which expedites the homes with charging power 3.6–11 kW are considered. Other
convergence of the algorithm. This process is repeated until parameters of the case study are presented in Table I.
stopping criteria is met or the number of generation reaches
its maximum value.
B. Simulation Settings
All the numerical simulations are performed in the
C. MILP for Lower Level Optimization
MATLAB (R2021a) environment on a computer equipped with
The lower level problem (22) deals with the energy pro- 3.20-GHz Intel Core i7 processor and 16-GB of RAM. The
file optimization of all the selected users for DR in response OpenDSS engine (an electric power distribution system sim-
to the decision variables of the upper level problem. In other ulator) [34] is utilized for power flow of the test microgrid
words, the solutions of the upper level problem are evalu- network using MATLAB COM interfacing. The EA is config-
ated by solving lower level problems of individual houses in ured with the parameter values of 30, 50, 1, and 0.1 for Np , Ng ,
a distributed manner. The problem of cost minimization for crossover probability, and mutation probability, respectively.
individual users’ energy consumption is formulated using the The MILP formulation of the lower level problem is solved
MILP approach which can be solved optimally using available using the Gurobi9.12 solver with the default settings.
solvers.
C. Benchmarks
IV. N UMERICAL E XPERIMENTS The performance of the proposed user coordinated dynamic
In this section, the proposed DR strategy is tested on a price-based DR (DPDR-C) policy is compared with other
residential microgrid adopted from a real low-voltage (LV) pricing policies, such as D2P [10], decentralized energy man-
distribution network. agement with dynamic pricing (DP-DEM) [13], and ToU
pricing [5]. Although ToU pricing is not fully dynamic, it has
A. System Data been considered due to its wide adoption in electricity pricing
markets. In addition, dynamic pricing without user coordina-
For the test microgrid network, the original IEEE European
tion (DPDR-NC) is considered to study the impacts of user
LV test feeder [27] is reduced in dimension spatially and
coordinated pricing and uncoordinated pricing on the success
electrically while keeping intact the system parameters. The
of the DR program.
test microgrid is connected to the main utility grid through
an 11/0.416-kV distribution transformer at the substation. The
microgrid also contains four distributed generators, namely: D. Results and Discussions
1) PV; 2) wind-turbine (WT); 3) fuel-cell (FC); and 4) micro- To evaluate the performance of the proposed dynamic
turbine (MT), with the maximum generation capacity of 40, price-based DR policy, following three simulation cases are
35, 30, and 30 kW, respectively. The microgrid consists of considered.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 RANA et al.: EV HOSTING CAPACITY ENHANCEMENT IN A COMMUNITY MICROGRID 7437

1) Case I: This is considered as the baseline case where

customers use energy as per their habit without paying any
attention to the price signals.
2) Case II: In this case, the customers optimize their energy
usage without any coordination among them by solving (15)
based on the received dynamic price signal from the SO. This
simulation case is called “DPDR-NC” throughout the text.
3) Case III: In this case, the SO calls for DR if there
is any network issue from the uncoordinated energy usage
in case II. A minimum number of customers are called for
DR and coordinated dynamic prices are optimized for each (a)
selected customer to ensure normal operation of the network
as discussed in Section II-E. This case is called “DPDR-C.”
All the cases are simulated over a wide range of EV
penetration levels as discussed in the following.
1) Case I (Baseline): In the baseline case, the users do not
care about price signals and use energy as per their instincts.
The EVs start charging as soon as they are plugged in. The
net load in the microgrid emanating from the individual user’s
baseline profile for different penetration of EVs is shown in
Fig. 2(a). It reveals that the net load is minimum at the mid-
day (12 P. M .) due to the user’s self-power generation from the
rooftop PV system. The users feed back their excess energy
to the microgrid network. On the other hand, peak load occurs
in the evening (around 8 P. M .) when people come back home (b)
and plug-in their EVs. The usual peak load (with 0% EV)
also coincides with this peak. As time progresses, the load
decreases gradually. The power flow of the test network with
the baseline load profile provides the transformer loading, line
loading, and minimum bus voltages in per unit as shown in
Fig. 2(b)–(d), respectively. It can be seen that the transformer
is overloaded at 35% of EV penetration while voltage viola-
tion and line overloading occurs at 35% (voltage falls below
the minimum limit at phase C) and 60% of EV penetration,
respectively. All these network issues occur at peak hours.
Transformer and voltage violation appear as the bottleneck for
higher EV penetration due to the simultaneous uncoordinated
charging of EVs in the baseline profile.
2) Case II (DPDR-NC): In case II, the customers optimize (c)
their energy usage through HEOs based on the electricity
prices from the dynamic price function. The users need to
set their preferences ahead once by logging in to their user
accounts and HEOs determine the optimal schedules of energy
usage that minimizes the cost. The dynamic electricity prices
for the users with and without EV charging are shown in
Fig. 3. At any instant of time, the users whose energy con-
sumption profile exceeds the maximum threshold for non-EV
users will be charged with the prices as shown in Fig. 3(a).
For other instants, they will be charged with regular dynamic
prices resulting from non-EV users’ demand as shown in
Fig. 3(b). For fairness, the higher energy prices during peak
(d)
loads due to the increased penetration of EVs are imposed
on the EV users only while keeping other users unaffected. It Fig. 2. Results for case I. (a) Net load. (b) Transformer loading. (c) Line
can be seen that at the instants with high renewable generation loading. (d) Bus voltages, top (phase A)–bottom (phase C).
(e.g., around noon), the prices are reduced dynamically which
encourages users to consume more. The lower price is enjoyed EV charging demand from the peak of the baseline profile to
by all irrespective of their types/consumption profiles. The the end of scheduling horizon causing peak rebounds. This is
uncoordinated dynamic prices prompt the users to shift their reflected in the loading of the transformer as shown in Fig. 4.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7438 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

Fig. 3. Dynamic price for users (a) with EV charging demand (b) without (a)
EV charging demand.

(b)

Fig. 4. Transformer overloading due to uncoordinated EV charging in case II.

It is found that transformer overloading and voltage violation

occurs at 60% EV penetration. Although the EV penetration
is increased from the baseline profile, it is limited to 59% due
to uncoordinated optimization across users.
3) Case III (DPDR-C): In this case, the SO calls for
the DR program if any potential network issue is detected
due to the uncoordinated energy optimization by the users
based on the dynamic prices. The DR program is formulated (c)
through the proposed optimization framework discussed in
Section II-E and implemented via Algorithm 1. The partic-
ipants for the DR program are selected as per Algorithm 2.
For 60% EV penetration, nondominated fronts of users based
on their shiftable energy and compliance rate are presented
in Fig. 5(a). The DR participants are selected from the first
front and then subsequent fronts are considered if required.
The coordinated dynamic price optimization takes into account
the consumption of all selected users to minimize the PAR
of the system load. The net load of the system, loading of
the transformer, and bus voltages are shown in Fig. 5(b)–
(d). The variation of coordinated dynamic prices and its (d)
impact on the EV charging demand are portrayed in Fig. 6.
Fig. 5. Results for case III. (a) For 60% EV selection of users in DR (number
Fig. 6(b) shows the optimized dynamic prices of three ran- inside markers denotes user-id). (b) Net load. (c) Transformer loading. (d) Bus
domly selected DR participants within their preferred time voltages, top (phase A)–bottom (phase C).
window for EV charging. Only three users were selected for
simplicity and properly visualizing the variations of prices
with respect to each other. It is interesting to notice that the evenly distributed EV charging demand presented in Fig. 6(c)
optimized prices for all selected users vary in such a man- as opposed to the uncoordinated one presented in Fig. 6(a). It
ner that can influence the scheduling of HEOs to avoid the reveals that DR with the proposed coordinated dynamic pric-
simultaneous charging of EVs. The impacts of the coordi- ing optimization avoids all network issues even for 100% EV
nated and optimized dynamic pricing are evident from the penetration.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 RANA et al.: EV HOSTING CAPACITY ENHANCEMENT IN A COMMUNITY MICROGRID 7439

(a) similar to D2P and DP-DEM. In the case of ToU, a larger off-
peak duration allows 79% EV penetration. With the proposed
DPDR-C, hosting capacity can be reached to 100% which ben-
efits the SO by delaying investment on network upgrades. The
(b) improved EV hosting capacity with the proposed DR frame-
work is realized by the reduced PAR of the net load compared
to other schemes as shown in Fig. 7(a). This enhancement
(c) of hosting capacity corroborates the efficacy of the proposed
approach.
b) Economic performance: Economic gain of the SO and
the customers are the main factors determining the success of
Fig. 6. For 60% EV penetration: (a) user’s EV charging profile with DPDR- the DR programs. Alongside the improved network operation,
NC, (b) dynamic price with DRDR-C, and (c) user’s EV charging profile with the SO is interested in maximizing revenue from the enhanced
DPDR-C.
EV hosting capacity. On the other hand, the customers are
TABLE II inherently greedy, that is, they want to minimize energy bills
I MPACTS OF D IFFERENT P RICING S CHEME ON THE EV H OSTING without affecting their comforts. The revenue gathered by the
C APACITY OF THE N ETWORK AND R EVENUE OF THE M ICROGRID
SO for different level of EV penetration is presented in the
last column of Table II. It can be seen that the revenue of SO
is increased with the proposed DR model in use due to the
improved EV hosting capacity. However, the revenue does not
increase at the same rate as of EV penetration. This is due to
the increased cost of DR service incurred through incentivized
prices for the DR participants. The more the number of cus-
tomers are allowed to participate in DR, the higher will be
the cost of DR service. The increased number of DR partic-
ipants in turn will facilitate peak power management. In our
model, the SO has the freedom to decide the tradeoff between
peak modification and cost incurred through DR services by
4) Performance Evaluation: The DR programs aid the SO varying the maximum number of participants to be included in
to maintain the reliable operation of the power network. At DR program. This directly affects the peak power modification
the same time, users’ participation in DR solely depends on achieved and cost of DR service.
the incentives/benefits they receive. Therefore, from both SO As discussed earlier, the customers can only be motivated to
and customer perspectives, the performance of the proposed participate in DR if they receive appropriate financial incen-
DR policy is evaluated against other schemes in this section. tives. In the proposed model, all the DR participants enjoy
The convergence analysis of the proposed algorithm is also the savings in energy bills. The average energy bills of the
presented. DR participants with different EV penetration are presented in
a) Performance in terms of electric network operation: Fig. 7(b). It shows that the proposed DR model offers lower
Unmanaged high penetration of EVs poses significant secu- energy bill for the DR participants compared to other dynamic
rity issues to the electric network. From the perspective of the pricing schemes. This indicates that the proposed approach
SO, alongside system stability, higher EV hosting capacity of can offer appropriate incentives to the DR participants. These
the network is also important from a revenue standpoint. On findings confirm the efficacy of the proposed DR model
the other hand, the improved EV hosting capacity and sta- in creating win-win situation for both the SO and the end
ble operation stem from the reduced PAR of the net load. users.
The impacts of different DR policies on the operating condi- c) Performance of the algorithm: To ensure the conver-
tions of the test network and consequent EV hosting capacity gence of the algorithm, elitism strategy is employed in the
are presented in Table II. It can be seen that the EV host- underlying EA. In this strategy, the best solution is carried
ing capacity is limited to 34% for the baseline profile as well forward as members of the next generation. The convergence
as using D2P and DP-DEM. For both D2P and DP-DEM, the of the algorithm is assessed by calculating the change in objec-
continuously varying price function depending on the demand– tive function (value of PAR for net load) between consecutive
supply imbalance forces the users to shift their EV charging iterations and the algorithm is assumed to be converged if
to the end of the scheduling window. This results in simul- the objective value remains unchanged for 15 iterations. The
taneous charging of many EVs causing network issues that convergence plot of the proposed algorithm shown in Fig. 8
result in limited hosting capacity. In contrast, the proposed reflects the improvement of the objective function in each
dynamic price function prevents oscillations in price signals iteration. In addition, a statistical analysis was carried out
up to a certain power threshold. This results in EV charg- with the objective values obtained from 11 independent trials
ing distributed over the scheduling horizon giving a hosting of the algorithm. The standard deviation of objective values
capacity of 59% with DPDR-NC. It is worth mentioning that was found to be very small (0.0097) which further indicates
DPDR-NC with Rth = 1 varies continuously that performs consistency in performance.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7440 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

revenue of the microgrid is also increased with the proposed

DR compared to other pricing policies. Further economic gain
of the microgrid with the proposed method can be perceived by
delaying the investment for the upgradation of electric network
to host increased EV penetration. From the consumer perspec-
tive, the DR participants obtained better financial savings with
proposed DPDR-C than those obtained from D2P, DP-DEM,
and DPDR-NC. This is reflected in the minimum average cost
of households enrolled with the proposed DR model than those
(a) with other dynamic pricing policies. These findings supported
the effectiveness of the proposed DR scheme.
In our future work, we will consider experimental validation
of the proposed DR model within a cloud computing platform.
The incorporation of peer-to-peer (P2P) energy trading strat-
egy among the users is another possibility. The optimization
framework that can support a large number of users within the
P2P scheme will also be investigated for improved operation
of the microgrid.

A PPENDIX
(b) Major equations governing the MILP formulation are given
as follows:
Fig. 7. Comparison of PAR and energy bills for the users. (a) PAR of net T
load. (b) Average cost of energy for the houses participated in DR. Ct = t πt xt − Pt yt FiT
PMG DR MG
(23)
t=1
xt + yt ≤ 1 (24)
T  
Etbatt = Einit
batt
+ pbatt
s (1 − δ)cbatt
t − (1 + δ)dtbatt ) t (25)
t=1
cbatt
t + dtbatt ≤1 (26)
T

Etev = Einit
ev
+ s (1 − δ)ct − (1 + δ)dt ) t
pev ev ev
(27)
t=1
t + dt ≤ 1
cev ev
(28)
xt , yt , cbatt
t , dt , ct , dt ∈ {0, 1}
batt ev ev
(29)
where PMG , pbatt , and pev are the continuous variables between
Fig. 8. For 60% EV penetration convergence plot of the algorithm. zero and their respective maximum value. The associated
binary variables are presented by xt , yt , cbatt
t , dt , ct , and
batt ev
ev
dt . The power import and export of the house from/to the
V. C ONCLUSION
In this work, a DR scheme based on dynamic pricing was microgrid are controlled by xt and yt , respectively. Charging
proposed for a microgrid to aid a higher penetration of EVs. and discharging of the household battery system and EV
First, an efficient dynamic price function was designed that battery are regulated by corresponding ct and dt , respectively.
can prevent cross-subsidies among customers. Second, a DR However, in (23), (25), and (27) the product of continuous
framework (i.e., DPDR-C) based on a two-level optimization and binary variables introduces nonlinearity in the problem.
formulation was proposed to be initiated by SO when deemed This can be resolved by introducing additional variables for
necessary. the products [35]. These additional variables for (23) can be
Numerical experiments showed that the designed dynamic defined as follows:
price function can effectively capture the load-generation z1t = PMG
t xt (30)
imbalance and promotes renewable generation while motivat-
z2t = PMG
t yt (31)
ing users to shift their peak loads. It can extend EV hosting
capacity to 59% compared to the other pure form of dynamic where z1t
and z2t
correspond to the power import and export
pricing, such as D2P and DP-DEM. The proposed DPDR-C power, respectively. These additional variables brings forth the
further enhances the EV hosting capacity to 100% at an following linearization constraints [35]:
expense of financial incentives provided to the DR partici-
pants. The improved EV hosting capacity results from the z1t,n ≥ 0 (32)
reduced PAR of the netload with the proposed DR model. The z1t ≤ PMG
t,max xt (33)

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 RANA et al.: EV HOSTING CAPACITY ENHANCEMENT IN A COMMUNITY MICROGRID 7441

z1t ≥ PMG
t − max 0, PMG
t,max (1 − xt ) (34) [18] B. Li, Q. Yang, L. Duan, and Y. Sun, “Operator-as-a-consumer:
A novel energy storage sharing approach under demand
z1t ≤ PMG
t + max 0, PMG
t,max (1 − xt ) (35) charge,” IEEE Trans. Cybern., early access, Aug. 16, 2021,
doi: 10.1109/TCYB.2021.3088221.
z2t ≥ 0 (36) [19] M. Nikzad and A. Samimi, “Integration of designing price-based demand
response models into a stochastic bi-level scheduling of multiple energy
z2t ≤ PMG
t,max yt (37) carrier microgrids considering energy storage systems,” Appl. Energy,
vol. 282, Jan. 2021, Art. no. 116163.
z2t ≥ PMG
t − max 0, PMG
t,max (1 − yt ) (38) [20] C. Feng, Z. Li, M. Shahidehpour, F. Wen, and Q. Li, “Stackelberg game
based transactive pricing for optimal demand response in power distri-
bution systems,” Int. J. Electr. Power Energy Syst., vol. 118, Jun. 2020,
z2t ≤ PMG
t t,max (1 − yt )
+ max 0, PMG (39) Art. no. 105764.
[21] F. Alfaverh, M. Denaï, and Y. Sun, “Demand response strategy based
0 ≤ PMG
t ≤ PMG
t,max . (40) on reinforcement learning and fuzzy reasoning for home energy man-
agement,” IEEE Access, vol. 8, pp. 39310–39321, 2020.
In a similar fashion, (25) and (27) can also be linearized. As [22] H. Zhang, D. Yue, C. Dou, K. Li, and X. Xie, “Event-triggered
a result, the problem of the HEO is completely transformed Multiagent optimization for two-layered model of hybrid energy system
into an MILP form. with price bidding-based demand response,” IEEE Trans. Cybern.,
vol. 51, no. 4, pp. 2068–2079, Apr. 2021.
[23] M. H. Yaghmaee, A. Leon-Garcia, and M. Moghaddassian, “On the
R EFERENCES performance of distributed and cloud-based demand response in smart
grid,” IEEE Trans. Smart Grid, vol. 9, no. 5, pp. 5403–5417, Sep. 2018.
[1] J. Zhao, J. Wang, Z. Xu, C. Wang, C. Wan, and C. Chen, “Distribution [24] A. Montazerolghaem and M. H. Yaghmaee, “Demand response applica-
network electric vehicle hosting capacity maximization: A chargeable tion as a service: An SDN-based management framework,” IEEE Trans.
region optimization model,” IEEE Trans. Power Syst., vol. 32, no. 5, Smart Grid, vol. 13, no. 3, pp. 1952–1966, May 2022.
pp. 4119–4130, Sep. 2017. [25] K. Deb and R. B. Agrawal, “Simulated binary crossover for continuous
[2] J. Z. William J. Nacmanson, and L. Ochoa, “Network modelling and search space,” Complex Syst., vol. 9, no. 2, pp. 115–148, 1995.
EV impact assessment,” Dept. Elect. Electron. Eng., Univ. Melbourne, [26] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist
Melbourne, VIC, Australia, Rep. UoM-ENA-C4NET-EV-Integration- multiobjective genetic algorithm: NSGA-II,” IEEE Trans. Evol. Comput.,
M6-v03, 2021. vol. 6, no. 2, pp. 182–197, Apr. 2002.
[3] N. Ruiz, I. Cobelo, and J. Oyarzabal, “A direct load control model for [27] “IEEE PES Distribution System Analysis Subcommittee’s Distribution
virtual power plant management,” IEEE Trans. Power Syst., vol. 24, Test Feeders.” Accessed: Dec. 15, 2021. [Online]. Available: https://
no. 2, pp. 959–966, May 2009. cmte.ieee.org/pes-testfeeders/resources/
[4] M. J. Rana, K. H. Rahi, T. Ray, and R. Sarker, “An efficient optimization [28] “Australian Photovoltaic Installations.” Accessed: Apr. 10, 2021.
approach for flexibility provisioning in community microgrids with an [Online]. Available: https://pv-map.apvi.org.au/historical
incentive-based demand response scheme,” Sustain. Cities Soc., vol. 74, [29] “Australian Battery Market 2021.” Accessed: Jul. 14, 2021. [Online].
Nov. 2021, Art. no. 103218. Available: https://www.sunwiz.com.au/wp-content/uploads/2021/03/
[5] E. Celebi and J. D. Fuller, “Time-of-use pricing in electricity markets SunWiz-Australian-Battery-Market-2021-Purchasers_Redacted.pdf
under different market structures,” IEEE Trans. Power Syst., vol. 27, [30] “Rooftop PV Output.” Accessed: Jun. 20, 2021. [Online]. Available:
no. 3, pp. 1170–1181, Aug. 2012. https://pvoutput.org/live.jsp
[6] S. Borenstein, “Effective and equitable adoption of opt-in residen- [31] M. R. Islam, H. Lu, J. Hossain, M. R. Islam, and L. Li, “Multiobjective
tial dynamic electricity pricing,” Rev. Ind. Org., vol. 42, pp. 127–160, optimization technique for mitigating unbalance and improving volt-
May 2012. age considering higher penetration of electric vehicles and distributed
[7] D. T. Nguyen, H. T. Nguyen, and L. B. Le, “Dynamic pricing design generation,” IEEE Syst. J., vol. 14, no. 3, pp. 3676–3686, Sep. 2020.
for demand response integration in power distribution networks,” IEEE [32] (Essential Services Commission, Melbourne, VIC, Australia). Electricity
Trans. Power Syst., vol. 31, no. 5, pp. 3457–3472, Sep. 2016. Distribution Code, 11th ed. (Apr. 2020). [Online]. Available:
[8] X. Kou et al., “A scalable and distributed algorithm for managing resi- https://www.esc.vic.gov.au/sites/default/files/documents/Electricity
dential demand response programs using alternating direction method [33] J. Quiros-Tortos, L. Ochoa, and T. Butler, “How electric vehicles and
of multipliers (ADMM),” IEEE Trans. Smart Grid, vol. 11, no. 6, the grid work together: Lessons learned from one of the largest electric
pp. 4871–4882, Nov. 2020. vehicle trials in the world,” IEEE Power Energy Mag., vol. 16, no. 6,
[9] M. A. F. Ghazvini et al., “Congestion management in active distribution pp. 64–76, Dec. 2018.
networks through demand response implementation,” Sustain. Energy, [34] “Simulation Tool—Opendss.” Electric Power Research
Grids Netw., vol. 17, Mar. 2019, Art. no. 100185. Institute. Accessed: Jan. 2, 2022. [Online]. Available:
[10] S. Misra, S. Bera, and T. Ojha, “D2P: Distributed dynamic pricing poli- https://smartgrid.epri.com/SimulationTool.aspx
cyin smart grid for PHEVs management,” IEEE Trans. Parallel Distrib. [35] L. Liberti. Mathematical Progamming. Ecole Polytechnique.
Syst., vol. 26, no. 3, pp. 702–712, Mar. 2015. 2018. [Online]. Available: https://www.lix.polytechnique.fr/∼liberti/
[11] W. Kong, H. Ye, N. Wei, D. Xing, and W. Chen, “Dynamic pricing teaching/dix/inf580-15/mathprog.pdf
based EV load management in distribution network,” Energy Rep., vol. 8,
pp. 798–805, Aug. 2022.
[12] R. Lu, S. H. Hong, and X. Zhang, “A dynamic pricing demand response
algorithm for smart grid: Reinforcement learning approach,” Appl.
Energy, vol. 220, pp. 220–230, Jun. 2018.
[13] D. A. Chekired, L. Khoukhi, and H. T. Mouftah, “Decentralized cloud-
SDN architecture in smart grid: A dynamic pricing model,” IEEE Trans. Md Juel Rana received the B.Sc. degree in elec-
Ind. Informat., vol. 14, no. 3, pp. 1220–1231, Mar. 2018. trical and electronic engineering from the Rajshahi
[14] Z. Fan, “A distributed demand response algorithm and its application to University of Engineering and Technology, Rajshahi,
PHEV charging in smart grids,” IEEE Trans. Smart Grid, vol. 3, no. 3, Bangladesh, in 2011, and the M.Sc. degree in elec-
pp. 1280–1290, Sep. 2012. trical engineering from the King Fahd University
[15] L. Jia and L. Tong, “Dynamic pricing and distributed energy manage- of Petroleum and Minerals, Dhahran, Saudi Arabia,
ment for demand response,” IEEE Trans. Smart Grid, vol. 7, no. 2, in 2017. He is currently pursuing the Ph.D. degree
pp. 1128–1136, Mar. 2016. with the School of Engineering and Information
[16] S. Mhanna, A. C. Chapman, and G. Verbič, “A faithful and tractable Technology, University of New South Wales,
distributed mechanism for residential electricity pricing,” IEEE Trans. Canberra, ACT, Australia.
Power Syst., vol. 33, no. 4, pp. 4238–4252, Jul. 2018. His current research interests include power
[17] A. Safdarian, M. Fotuhi-Firuzabad, and M. Lehtonen, “Optimal residen- system operation and control, energy management of microgrids, demand
tial load management in smart grids: A decentralized framework,” IEEE response, management of EVs in the distribution networks, and evolutionary
Trans. Smart Grid, vol. 7, no. 4, pp. 1836–1845, Jul. 2016. algorithms.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.
 7442 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 53, NO. 12, DECEMBER 2023

Forhad Zaman received the Ph.D. degree from the Ruhul Sarker (Member, IEEE) received the Ph.D.
University of New South Wales (UNSW), Canberra, degree from Dalhousie University, Halifax, NS,
ACT, Australia, in 2017. Canada, in 1992.
He worked as a Research Fellow with UNSW He is currently a Professor with the School of
from 2017 to 2019, where he is currently an Engineering and Information Technology, and the
Adjunct Lecturer with the School of Engineering Director of the Faculty Postgraduate Research,
and Information Technology. His research interests University of New South Wales, Canberra,
include power system operation and control, active ACT, Australia. He has authored a book titled
bidding strategy in the electricity market, evolu- Optimization Modeling: A Practical Approach.
tionary algorithms, multiobjective optimization, han- His current research interests include evolutionary
dling real-life uncertainty, and game theory. optimization and applied operations research.
Dr. Sarker is a member of INFORMS.

Tapabrata Ray (Senior Member, IEEE) received

the B.Tech. (Hons.), M.Tech., and Ph.D. degrees
from the Indian Institute of Technology Kharagpur,
Kharagpur, India, in 1990, 1993, and 1997, respec-
tively.
He is a Professor with the School of Engineering
and Information Technology, University of New
South Wales, Canberra, ACT, Australia, where he
leads the Multidisciplinary Design Optimization
Group. His current research interests include surro-
gate assisted/multifidelity optimization, multi/many-
objective optimization, robust design, and decision making.

Authorized licensed use limited to: National Institute of Technology - Jamshedpur. Downloaded on February 21,2025 at 05:05:48 UTC from IEEE Xplore. Restrictions apply.