Research ProposalPDF Available

Efficient Utilization of Energy Employing Meta-heuristic Techniques with the Incorporation of Green Energy Resources in Smart Cities Signature of Student

Authors:

Abstract and Figures

In the recent era, the energy management problem considering peak load is solved using demand side rather than supply side management. However, various electricity consumers have different interests and willingness when considering peak load management. In this regard, efficient energy management solutions are required where the priority of an appliance is considered according to user's interest. Shifting of the load based on priorities will be beneficial for consumers during time slots where electricity prices are high and also to a utility to control the peak load demand. However, it is difficult for consumers to respond manually to demand response incentives like time of use (ToU) and real-time pricing (RTP) signals due to the lack of interest, busy schedules or their unwillingness. Therefore, in order to take full benefits of such incentives, the need for energy management controller (EMC) capable of making smart decisions in response to the price signals, without jeopardizing user comfort is need of the hour. The researchers have been focused on designing autonomous EMC for energy management based on shifting the residential load from on-peak to off-peak hours without considering appliances' priorities and threshold limits. This causes in the creation of rebound peaks, that is also a risk to the grid sustainability. When consumers shift the operation load of their smart home appliances to off-peak hours, it also results in an increased peak-to-average ratio (PAR). Demand side management (DSM) in smart grid (SG) authorizes consumers to make informed decisions regarding their energy consumption pattern and helps the utility in reducing the peak load demand during an energy stress time. This results in reduced carbon emission, consumer electricity cost, and increased grid sustainability. Most of the existing DSM techniques ignore priority defined by consumers. In this work, we present a DSM strategy based on the load shifting technique, which considers shifting of various energy cycles of an appliance according to the consumer defined priority. The proposed day-ahead load shifting technique is mathematically formulated and mapped as multiple knapsack problem (MKP). This reduces the rebound-effect caused by load shifting to off-peak time slots and also minimize the PAR. The autonomous EMC proposed embeds three meta-heuristics optimization techniques; genetic algorithm, enhanced differential evolution, and binary particle swarm optimization along with optimal stopping rule (OSR), which is used for solving the load shifting problem. Simulations are carried out using three different appliances and the preliminary results validate that the proposed DSM strategy successfully shifts the appliances' operational time to off-peak time slots, which consequently leads to substantial electricity cost savings in reasonable waiting time, and also help in reducing the peak load demand from the SG through knapsack capacity limit. In addition, we calculate the feasible regions to show the relationship between cost, energy consumption, and delay. The simulation section presents the initial results that we have got so far. Next, we will extend our work by considering multiple appliances and homes along with hybridization of meta-heuristics schemes for better exploratory and exploitive search space. The integration of green energy resources obtained from the natural resources like sun, wind etc. will also be considered in the proposed model. We will also evaluate microgrids energy trading using fogs in various regions of the smart city to minimize energy losses as compared to the main grid.
Content may be subject to copyright.
COMSATS Institute of Information Technology,
Islamabad Campus
Synopsis For the degree of M.S/M.Phil. Ph.D.
PART-1
Name of Student Asif Khan
Department Department of Computer Science
Registration No.
FA15-PCS-004
Date of Thesis
Registration
Fall 2016
Name of
(i) Research Supervisor
(ii) Co-Supervisor
Research Area
(i)
Dr. Nadeem Javaid
(ii) Dr. Mariam Akbar
Artificial Intelligence
Members of Supervisory Committee
1. Dr. Nadeem Javaid
2. Dr. Mariam Akbar
3. Dr. Sohail Asghar
4.
Title of Research Proposal Efficient Utilization of Energy Employing Meta-heuristic
Techniques with the Incorporation of Green Energy
Resources in Smart Cities
Signature of Student:
Summary of the Research
In the recent era, the energy management problem considering peak load is solved using
demand side rather than supply side management. However, various electricity
consumers have different interests and willingness when considering peak load
management. In this regard, efficient energy management solutions are required where
the priority of an appliance is considered according to user’s interest. Shifting of the load
based on priorities will be beneficial for consumers during time slots where electricity
prices are high and also to a utility to control the peak load demand. However, it is
difficult for consumers to respond manually to demand response incentives like time of
use (ToU) and real-time pricing (RTP) signals due to the lack of interest, busy schedules
or their unwillingness. Therefore, in order to take full benefits of such incentives, the
need for energy management controller (EMC) capable of making smart decisions in
response to the price signals, without jeopardizing user comfort is need of the hour.
The researchers have been focused on designing autonomous EMC for energy
management based on shifting the residential load from on-peak to off-peak hours
without considering appliances’ priorities and threshold limits. This causes in the creation
of rebound peaks, that is also a risk to the grid sustainability. When consumers shift the
operation load of their smart home appliances to off-peak hours, it also results in an
increased peak-to-average ratio (PAR).
Demand side management (DSM) in smart grid (SG) authorizes consumers to make
informed decisions regarding their energy consumption pattern and helps the utility in
reducing the peak load demand during an energy stress time. This results in reduced
carbon emission, consumer electricity cost, and increased grid sustainability. Most of the
existing DSM techniques ignore priority defined by consumers. In this work, we present a
DSM strategy based on the load shifting technique, which considers shifting of various
energy cycles of an appliance according to the consumer defined priority. The proposed
day-ahead load shifting technique is mathematically formulated and mapped as multiple
knapsack problem (MKP). This reduces the rebound-effect caused by load shifting to off-
peak time slots and also minimize the PAR. The autonomous EMC proposed embeds
three meta-heuristics optimization techniques; genetic algorithm, enhanced differential
evolution, and binary particle swarm optimization along with optimal stopping rule
(OSR), which is used for solving the load shifting problem. Simulations are carried out
using three different appliances and the preliminary results validate that the proposed
DSM strategy successfully shifts the appliances’ operational time to off-peak time slots,
which consequently leads to substantial electricity cost savings in reasonable waiting
time, and also help in reducing the peak load demand from the SG through knapsack
capacity limit. In addition, we calculate the feasible regions to show the relationship
between cost, energy consumption, and delay. The simulation section presents the initial
results that we have got so far.
Next, we will extend our work by considering multiple appliances and homes along with
hybridization of meta-heuristics schemes for better exploratory and exploitive search
space. The integration of green energy resources obtained from the natural resources like
sun, wind etc. will also be considered in the proposed model. We will also evaluate
microgrids energy trading using fogs in various regions of the smart city to minimize
energy losses as compared to the main grid.
1 Introduction
Smart grid (SG) uses new and advanced technologies including intelligent hardware, autonomous
controllers, robust software and other resources for the management of data along with two-way
communication between consumers and power utilities to deliver energy in a reliable and effi-
cient way. The main objectives of the SG are to improve the reliability, efficiency, and safety
of the entire system [1]. According to National Institute of Standards and Technology (NIST),
Framework and Roadmap for SG Interoperability Standards give the definition as, a combination
of computing and communication services and integration with power infrastructure. Two-way
communication flow of energy and control capabilities will open an array of new applications
and functionalities that go well beyond smart meters for businesses and homes [2].
Demand side management (DSM) also called energy demand management allows con-
sumers to change their demand for energy so that electricity usage is minimized during on-peak
hours. Demand response (DR) programs are in the form of financial incentives or other time-
based rates which provide an ample opportunity for consumers in shifting or reducing appli-
ances load to off-peak hours. Thus, DR has been considered the most reliable and cost-effective
solution to reduce the peak demand and smooth the demand curve, under system stress. DR
offers motivation in the form of price signals including time-of-use (ToU), critical peak pricing
(CPP), peak load pricing (PLP), real-time pricing (RTP), day-ahead pricing (DAP) or day-ahead
RTP (DA-RTP). The limitation associated with ToU is that scheduling has to be performed on
a day-ahead basis while RTP requires continuous real-time communication between the utility
and consumers, which may cause network congestion and data loss problems. Therefore, an
alternate solution proposed is the DA-RTP, where predicted real-time prices are announced to
customers beforehand and consumers are billed on this day-ahead price [3].
Electricity cost minimization using appliance scheduling is one of the most difficult tasks
due to the intermittent nature of renewable energy sources (RESs), randomness in electricity
price and unknown user behaviors. Integration of RESs in SG cause intermittent generation
of electricity [4]. To deal with this stochastic phenomenon, authors in [5] have considered the
scheduling problem as optimal stopping problem and proposed optimal stopping rule (OSR) as
a solution. The authors in these papers have only focused on multiple appliance scheduling and
ignored energy scheduling of dominant cycles of an appliance.
The load scheduling problem of appliances is considered as an optimization problem and
solved using various techniques in the literature. Appliance load shifting is done through task or
energy scheduling. In task scheduling appliances are switched on/off, while in energy schedul-
ing power consumption of appliances are reduced and their length of operational time (LoT)
extended when the system is under stress [6]. In [7], the authors have proposed compress de-
lay scenario, where the operational time of appliances is expanded by decreasing their power
consumption for a finite and infinite number of the appliances using recursive approach for
peak demand calculation. In these papers, a large number of appliances have been considered,
however, consumer-defined appliance priorities have not been addressed.
Authors in [8], solved the appliances load scheduling problem by using mixed integer non-
linear programming (MINLP). The home energy management (HEM) architecture proposed in
[9], has also modeled the energy management problem as MINLP. The technique gives an ac-
curate and effective solution, however, also requires a large amount of computational time. As
the size of the problem increases it becomes difficult to handle the constraints and parameters in
linear programming.
A very important issue in SG is user comfort which is often neglected in uni-objective cost
minimization problems. Some studies reveal that consumers wish to reduce their costs, however,
do not want to compromise on their comfort. The scheduling of appliance results in the delay in
the operational time of appliance, which causes inconvenience for the consumers. Muralitharan
et al. have used a multi-objective evolutionary algorithm to reduce electricity cost and user
waiting time [10]. Authors have identified the trade-off between waiting time and electricity
cost. Thus, the schemes used for load scheduling of appliances may also consider waiting time.
Due to the aforementioned limitations, we have considered appliance priorities in schedul-
ing and formulated (mapped) the optimization problem using multiple knapsacks in order to
mitigate the rebound-effect which will create peaks in the off-peak time slots. For prelimi-
nary results, we adopted a DSM strategy (load shifting) using three meta-heuristic techniques
enhanced differential evolution (EDE), genetic algorithm (GA), and binary particle swarm op-
timization (BPSO) and compared our results with other optimization technique OSR. We found
that all proposed schemes successfully shift various energy cycles of an appliance during sys-
tem stress time to reduce the consumer cost and discomfort based on priorities. The simulations
are conducted using two scenarios. In scenario-1, we take individual or atomic appliances to
schedule their energy cycles. In scenario-2, aggregate appliances effect is considered in two
different cases: a) no load capacity limit and b) knapsack load capacity limit. The results reveal
substantial electricity cost savings when no load capacity limit is considered, however, results
in peak-to-average ratio (PAR) creation. The rebound-peaks created during off-peak hours are
then minimized by applying the knapsack capacity limit. Moreover, the monthly and yearly cost
savings of appliances are demonstrated along with user discomfort.
2 Objectives and impact of research
The objectives of this work are to present an optimal solution to minimize the consumer elec-
tricity cost along with maintaining user comfort. Appliance priorities will be considered which
shifts the device load as per willingness of the consumer. We will also consider knapsack ca-
pacity limit in order to mitigate the rebound-effect where the peak is created in the off-peak
time-slots. To make the consumer more active in DR programs and energy trading with utility,
we also incorporate green energy resources (GERs) obtained from natural resources like wind
and sun. Further, energy storage system (ESS) in the form of batteries and plug-in-hybrid elec-
tric vehicles (PHEVs) will be also considered in our model. The use of GERs has a significant
effect on the environment as other electricity generating resources like fossil fuels, coal etc. pol-
lute the nature by causing carbon dioxide (CO2) emissions. Thus, the objectives are summarized
as below:
Reduce the consumer electricity cost while maintaining user comfort level.
Provide an opportunity to the electricity consumers to set the appliance priorities accord-
ing to their wish and willingness.
The knapsack capacity limit will be applied to maintain PAR, which balances the demand
and supply of electricity. This effect also decreases the re-bound peaks and increases the
grid sustainability.
Incorporate GERs and ESS at microgrid local generation levels so that consumers take
part in energy trading programs with the utility and sell their surplus energy back to the
grid. The integration of GERs and PHEVs will significantly reduce the CO2emissions
and play a key role in providing a clean environment.
This work proposes meta-heuristic methods which provide an approximate solution near
to the optimal with less computational effort as compared with mathematical techniques.
These meta-heuristic techniques are more suited for stochastic environment.
3 Scope of work
The preliminary results are obtained using a single home with three appliances, clothes dryer,
dishwasher, and refrigerator. These are considered suitable appliances for DR because their duty
cycles are short, also require high energy consumption and peak demand. The HEM system
uses low and high priority of an appliance to shift the load according to consumer willingness.
Optimization schemes like GA, EDE, BPSO, and OSR are applied to schedule the consumer
load from on-peak to off-peak time slots. We will further extend our work and the scope is
limited to the following aspects:
We will use meta-heuristic algorithms and also hybridize to get better exploratory and
exploitative search results.
Use smart appliances in scenarios with the worst cases like using operational time interval
(OTI) of one minute will also be considered.
We will extend our work by proposing an EMC that will consider a larger number of the
appliance data set.
Multiple DR pricing schemes like ToU and CPP will be considered.
Multiple homes/buildings will be considered in our model.
Microgrid local generation and integration of GERs will be taken in this work.
ESS will be taken into account where batteries, PHEVs will be charged in off-peak time-
slots and utilized back in the on-peak period.
A fog-induced microgrids energy harvesting solution for smart city (SC) will also be
contemplated.
4 Related work
Consumers make intelligent decisions for better and improved consumption patterns via DR.
Various optimization techniques have been opted in studies to achieve cost minimization as an
objective for a large number of consumers. Like Derakhshan et al. scheduled appliances of four
residential houses in Tehran city of Iran using different types of DR price schemes [11]. They
proposed two algorithms called teaching and learning based optimization (TLBO) and shuffled
frog leaping (SFL) for scheduling appliances in the SG. The objective was to reduce the con-
sumer cost. The results showed that DR programs (DRPs) successfully shift load to minimize
the electricity cost. Mhanna et al. proposed large-scale households DR aggregation distributed
algorithm for a huge number of appliances based on the non-convex DR decomposition [12].
The method was applied and tested on 2560 households, with average 10 devices resulted in
near-optimal solutions. Similarly, Logenthiran et al. used DSM controller to schedule load for
2604 devices present in residential, 808 devices in commercial and 109 devices in industrial
areas, respectively, using DAP strategy [13]. The load shifting strategy was opted to reduce the
cost and PAR using an evolutionary algorithm (EA). The results bring the load curve close to
the objective load curve in order to reduce electricity cost and PAR. The limitation consists of
applying the proposed technique on real-time scenarios.
Real time scenarios use RTP signals which require continues data transmission between
utility and consumers. In [5], Rasheed et al. used OSR technique to reduce cost and maximize
user comfort in the real-time pricing environment. The users are categorized as active, passive
and semi-autonomous depending upon the available resources of renewable energy and storage
system. The autonomous home architecture is proposed using three algorithms first come first
serve (FCFS), modified FCFS (MFCFS) and priority enables early deadline first (PEEDF). The
simulation results show that cost is minimized by first come first serve algorithm however, the
LoT does not meet when compared with other two algorithms. Further, the PEEDF algorithm
has reduced cost and also achieved maximum user comfort as compared to MFCFS algorithm.
In some studies, the appliance scheduling problem is formulated as an optimization prob-
lem and solved using linear programming. Shirazi et al. proposed a HEM system, which is
integrated with distributed energy resources (DERs) along with electrical and thermal appliance
scheduling (HEMDAS) [9]. The home appliances are categorized as controllable electrical, ther-
mal and uncontrollable appliances. Energy management minimization problem is formulated as
MINLP. The multi-objective function includes cost and user comfort performance parameters.
The results show that HEMDAS architecture has reduced the cost and also achieved a feasible
solution to optimal energy management problem. In [8], Moon et al. proposed electricity load
scheduling algorithm to satisfy user demands within budget limits through manipulating energy
consumption of appliances by using ToU pricing strategy. The MINLP problem was solved by
the use of generalized Bender’s decomposition approach. The results show optimal electricity
load scheduling and consumption of many appliances in a residence was managed in a unified
way. In [14], Ma et al. classified appliances into two different categories and named as the flex-
ible starting time and flexible power devices. The energy consumption of flexible starting time
may be shifted to off-peak time slots, while the power of flexible appliances is reduced during
system stress period. This effect causes consumer discomfort. The load scheduling problem is
modeled as a convex optimization problem. The authors used weighted sum approach in multi-
objective function to prioritize cost and discomfort and results are discussed in three different
scenarios. The model ignores the integration of RESs and ESS, which has a significant effect on
cost reduction and providing a clean environment. In addition, these formal approaches require
huge computational time and complexity arises with the increase in the size of a problem.
To overcome the limitation of linear programming some studies have proposed population-
based optimization algorithms, which are widely applied in HEM controller (HEMC) to reduce
the peak load demand along with electricity cost. Differential evolution (DE) algorithm was
proposed by Storn et al. in 1995 [15] which was enhanced by Arafa et al. [16]. The EDE was
proposed to enhance the accuracy and convergence speed of DE. In order to simplify tuning
process control parameters in EDE were reduced to two as compared with DE which has three
parameters. The simulation results of EDE on various functions performed well as compared
with DE. In [17], the author’s used wind-driven optimization (WDO) algorithm and min-max
regret-based knapsack algorithm to reduce electricity cost, peaks and waiting time using ToU
pricing signal. The simulation results were compared with particle swarm optimization (PSO)
and WDO algorithms. The WDO was found efficient. Rahim et al. designed HEMC based on
three heuristic algorithms including GA, ant colony optimization (ACO), and BPSO to reduce
consumer cost, PAR and user discomfort [18]. The problem is formulated as multiple knapsacks
and a combined model of inclined block rate and ToU pricing tariff is implemented. The ESS
along with RESs are also considered in the model. The simulations results show that GA-based
HEMC performed well as compared to ACO and BPSO.
Hybrid meta-heuristic schemes are widely applied to have better results. Manzoor et al.
[19] combined GA and TLBO and proposed teacher learning genetic optimization (TLGO) al-
gorithm which has comparable results with LP at the reduced complexity and computational
efforts. Six appliances were categorized into time-flexible and power-flexible categories. Multi-
objective function considering electricity cost and user discomfort is used to solve the load
scheduling problem. Javaid et al. [20] combined EDE and TLBO and proposed enhanced
differential teaching-learning algorithm (EDTLA) to minimize electricity cost while also main-
taining a desired user comfort level. RESs and battery storage system (BSS) are also considered
in the model to further reduce the consumer cost. The simulation results show high performance
of EDTLA as compared to EDE and TLBO. In [21], the authors merged GA with BPSO and
proposed genetic binary particle swarm optimization (GBPSO) to minimize consumer bill con-
sidering single and multiple home scenarios using RTP signals. The simulation results show that
GBPSO based HEM controller performed better and reduced 36% cost and 34% PAR. In [22],
authors hybridize GA and particle swarm optimization and proposed hybrid GA-PSO (HGPO)
to reduce consumer electricity cost. RES and ESS were considered in the model. Results show
HGPO outperforms other heuristic algorithms and significantly reduces the bill by 25.12% and
PAR by 24.88% respectively.
One of the goals of energy management is to reduce CO2emissions for providing a clean
environment. Morales et al. [23] used ToU pricing signal to shift load curves using polynomial
functions in Galapagos Islands belonging to Ecuador country. Different strategies were ap-
plied to annulate the rebound-effect by considering peak load reduction and shifting energy in
a day. In the aggregate residential demand, Muratori and Rizzoni found that simple DR pricing
schemes might create pronounced rebound peaks. The authors used dynamic programming to
find a global solution and proposed an energy management framework based on multi-ToU and
multi-CPP to deal with the rebound-peaks by synchronizing the individual residential demands
along with PHEVs consideration in a decentralized manner. In multi-ToU, electricity consumers
are placed into different groups, where each group sees different ToU prices. Simulation results
serve as a tool for energy policy solutions where different electricity price structures may be
applied to developing effective and robust DRPs [24]. Thus, HEM system may be designed to
reduce the rebound-peaks.
Researchers have proposed recursive formulas for the calculation of peak demand under
different power demand scheduling scenarios. In [6], Vardakas et al. have used recursive ap-
proach for finding peak demand under compressed, delay and postponement request scenarios
and compared it with non-scheduled default scenario using RTP scheme for an infinite number
of appliances in the residential home management system. User participation in energy manage-
ment program was also considered along with RES integration. The simulations result shows
satisfactory results in terms of accuracy while analytical models calculate peak demand in very
small computational time. The limitation of the paper includes power consumption overestima-
tion due to an infinite number of appliances assumption. To overcome the overestimation the
authors in [7] proposed four scenarios for a finite number of appliances. The customer partici-
pation in HEM is also considered in order to find social welfare. The analytical results produce
low timely results which are essential for near real-time decisions in HEM system.
The multiobjective function was formulated for DSM to achieve multiple goals including
cost minimization and user comfort maximization. In home appliance scheduling, there is a
trade-off between electricity cost and user comfort. Ogunjuyigbe et al. in [25], proposed load
satisfaction algorithm to maximize user comfort at minimum cost. The simulation results show
that the proposed algorithm has achieved minimum cost and maximum user satisfaction on three
different budget scenarios. Sensitivity analysis was also carried out on different user budgets and
it was found that user’s satisfaction increase with the increase in the user’s budget. Muralitharan
et al. used a multiobjective evolutionary algorithm to reduce consumer cost and waiting time
in SG [10]. The authors have applied threshold policy in order to avoid peak and balance load.
The penalty in form of additional charges has been incorporated in their proposed model if
consumers exceed price threshold limits. The simulation results minimize both electricity cost
and appliance waiting time.
The shifting of appliance operation based on the priorities defined by consumer’s viewpoint
is an essential factor to be considered in HEM schemes. Rastegar et al. addressed the issue
by considering the value of lost load to indicate the appliance operational priority based on
consumer’s perspective. Pricing tariffs including ToU and IBR are used and results show the
lower cost for the IBR [26]. In [27], the authors assigned static and infinite priorities to shiftable
and non-shiftable appliances. These static priorities of the appliance are time-independent and
selected on some criteria like welfare, emergent usage or electricity cost. The results reduced
the electricity cost with different threshold values. Therefore, it is vital to consider the priorities
of appliance defined by consumers in HEM systems.
The traditional power grid generates electricity from centralized sources, including hydro-
electric or coal plants. With the advent of SG and new technologies, the idea of producing
electricity generation through RESs close to the consumption points have emerged at the end of
the 20th century. Microgrids consists of small-scale electricity generation and storage systems
installed near to demand, resulting in high efficiency and transmission reductions, emergency
and system stress situations [28]. Microgrids have advantages of integrating GERs to fulfill
local consumer demands. Microgrids work in grid and islanded modes [29].
The proliferation of Information and Communication Technologies (ICTs), Internet of Things
(IoT) and cloud technologies has brought a tremendous change in the traditional cities. The
concept of SC is now becoming a reality and thus requiring new demands for data and energy
management. Fog computing has emerged which provides distributed data management and
aggregation at the edge of the network for improved throughput. Nazmudeen et al. [30] showed
through analyses that fog computing framework is highly beneficial for smart meter data in the
SG systems. It also results in improved bandwidth capacities of power line communication.
Similarly, a solution providing data storage and processing by fog computing was proposed
for advanced metering infrastructure (AMI) [31]. The solution provided was efficient, robust,
low-cost and also supported plug-and-play.
The SMs installed in SG produce a huge amount of data that is hard to store, process and
analyze with cloud computing [32]. The authors proposed a fog computing environment, where,
big data is preprocessed and computed before transmitting it to the cloud servers. The three-tier
model consists of SG, fog, and cloud layers. The fog acts as a bridge between SG and cloud
layers. This integration results in reduced latency and increased the privacy of data. Scalability
is a vital issue in cloud computing [33]. Therefore, the authors proposed energy management-as-
a-service over fog computing platform to achieve data privacy, scalability, interoperability and
interactivity among heterogeneous devices. In [34], the authors proposed fog-to-cloud (F2C)
layered approach with a goal to bring different heterogeneous fog/cloud layers into a hierarchical
architecture and establish a real need for coordinated management of the two computing system
models. In [35], the authors provided an optimization framework for IoT based energy efficient
and harvesting solution for SC. In addition, they proposed the use of lightweight protocols,
DSM for appliance scheduling, low-power transceivers along with cloud-based approach.
SC are very large distributed systems and complex. It integrates ICTs and IoT for enriching
government and other services for citizen welfare. The challenges associated with SC appli-
cations are mobility, real-time support, and heterogeneity. In [36], the authors used service-
oriented-middleware (SOM) using the fog and cloud of things (CoT) as a solution. However,
energy consumption of IoT devices is still lacking. In [37], the authors used microgrids and
fog computing to reduce the energy consumption of IoT devices. The results show less energy
consumption of IoT applications due to aforementioned integration. Therefore, fog computing,
providing solutions at the edge of the network is gaining popularity due to features like locality,
proximity, low latency, and privacy may also be considered in the energy management of SC.
The energy and task scheduling problem of home appliances is considered as an optimiza-
tion problem and various studies have mapped this problem to single and multiple knapsack
problems [38, 45]. The single knapsack problem (SKP) is associated with a knapsack having an
upper capacity limit which has to be filled up with items having value and weight such that the
total weight of inserted items does not exceed the knapsack capacity limit while the objective is
to maximize the profit [49]. Multiple knapsack problem (MKP) is a combinatorial optimization
problem consisting of multiple knapsacks and items where the goal is to find a subset of items
results in the maximum profit without exceeding capacity limit [50].
The summary of related work is shown in Table 1 and 2.
4.1 Analysis of literature
The work carried out in references [8, 9, 11, 12, 13, 18], lack prioritizing the operation of
shiftable appliances based on consumer’s viewpoint. HEM schemes enabled with appliance
priorities defined by consumer willingness is essential. In [26, 27], the authors assigned priority
Table 1 Heuristic algorithms based scheduling (N/A stands for Not Applicable)
Techniques
used Objectives
Time-
based
DR
schemes
Demand/user
classifica-
tion
Features Limitations
TLBO and
SFL [11] Reduce cost
ToU,
RTP, and
CPP
Shiftable,
sheddable
and non-
sheddable
loads
TLBO provides
more optimized
results than
SFL
Delay, user
comfort and
PAR are
ignored
Distributed
algorithm
[12]
Robust
algorithm N/A N/A
Consider large
number of
households i.e.
2560, Fast
convergence
and scalibility
Individual
appliance cost
saving is
ignored
EA [13]
Minimize
cost and
peak load
DAP Controllable
devices only
Residential,
commercial and
industrial areas
are considered
with large
number of
appliances
UC and delay
are not
considered,
individual
appliance
scheduling and
real time
scenarios were
ignored
OSR,
distributed
and
centralized
scheduling
algorithms
[51]
Cost and
average
delay mini-
mization
RTP
Considers
peak
demand
cycles of
appliances
Real-time
scenario
implemented,
Considers
appliance peak
demand duty
cycles
PAR and large
number of
appliances are
ignored
FCFS,
MFCFS,
PEEDF, and
OSR [5]
Minimize
cost and
energy
consumption
through
RESs
RTP
Active,
passive, and
semi-
autonomous
users classi-
fication
Real-time
scenario
implemented,
Integration of
RESs and
storage system
is considered
Individual
appliance
scheduling
along with
monthly and
yearly cost
savings are
ignored
MINLP [9]
Minimize
cost and
maximize
user comfort
RTP and
natural
gas fixed
price
Controllable
electrical,
thermal and
uncontrol-
lable
appliances
Reduced cost
and integrated
RESs, storage,
combined heat
and power unit.
Uncertainty is
considered
Increased
computational
time
Table 2 Heuristic algorithms based scheduling
Techniques
used Objectives
Time-
based
DR
schemes
Demand/user
classifica-
tion
Features Limitations
ACO, BPSO
and GA [18]
Minimize
cost, PAR,
energy
consumption
and
maximize
user comfort
ToU and
IBR
Passive,
semi-active
and active
users, fixed,
shiftable and
elastic
appliances
Integration of
RESs and
storage, cost
reduction,
minimizing
PAR and user
satisfaction is
considered
Peak duty
cycles of
individual
appliance is
ignored along
with monthly
savings
WDO and
min-max
regret-based
knapsack
algorithm
[17]
Minimize
energy con-
sumption,
cost, peak
and waiting
time
ToU
User
dependent,
interactive
schedulable
and un-
schedulable
WDO
performed well
when compared
with PSO
Scheduling of
peak duty
cycles of an
appliance is
ignored along
with the
monthly cost
Recursive
formulas [6]
Peak
demand
calculation
under four
different
scenarios
RTP
Compress,
postpone-
ment and
no-
participation
appliances
categoritza-
tion
User
participation
and RESs
integration, Fast
calculation of
peak demand
using analytical
model
Power
consumption
overestimation
due to infinite
number of
appliances
assumption
Recursion
[7]
Peak
demand
calculation
RTP
Compress,
postpone-
ment and
no-
participation
User
participation
and RESs
integration,
Social welfare
calculation,
Fast calculation
of peak demand
using analytical
model
Considers only
finite set of
appliances
Load
satisfaction
algorithm
[25]
Minimize
electricity
cost and
maximize
user comfort
ToU
Categorization
based on
different
sections of
the user
house
Multiobjective
funciton, Three
different budget
scenarios
implemented
Ignores PAR
and monthly
cost savings.
Multiobjective
EA [10]
Minimize
cost and
delay
ToU
Permanent
and
schedulable
devices
Threshold
policy along
with penalty
has been
considered
Dominant
energy
scheduling of
an appliance is
ignored
to the appliances. In [51], Yi et al. have used OSR to shift the operation of the appliance
from on-peak to off-peak hours based on priorities in order to save electricity cost using RTP
scheme. The authors have applied a threshold policy and considered waiting and electricity
cost in their objective function. The simulation results show a significant performance of OSR
when compared with linear optimization technique in terms of minimizing cost and waiting
time. However, the authors [26, 51] have not addressed the PAR, which is an essential factor
to control the peak load demand in the SG. Therefore, load capacity limit is vital to control the
rebound peaks which are again created in the off-peak hours. Further, the RTP signal used in
[51] has its own limitations and increases network bandwidth. Thus, DA-RTP is considered
effective in order to minimize and control the network traffic.
In the references [5, 6, 9, 13, 18] the work carried out do not take actual load profiles of
appliances. The actual load profile is replaced by the maximum or average load profile of
appliances. As with respect to time load profile of appliance varies thus, scheduling without
considering the actual load profile may not always give a feasible solution. In [52], actual load
profile of appliances considering different energy cycles is considered along with sequential ex-
ecution constraint without operational interruption. The optimization problem was solved using
mixed-integer programming technique. However, the user’s comfort has not been addressed in
[52]. User’s comfort is a relative term which also varies among consumers and situations. In
literature, many authors have associated the user’s comfort with appliance delay occurred in
shiftable appliances, while others relate it to thermal comfort. On the DSM, user’s comfort is a
vital characteristic that may also be obeyed with the cost minimization problem.
Moreover, mathematical techniques used in [8, 9, 14, 52], are highly computational and
their complexity increases exponentially with the increase in size of the problem. Furthermore,
these deterministic approaches are formulated with known parameters. Stochastic and dynamic
programming find an optimal solution, however, has its own disadvantages and suffers from the
curse of dimensionality problem [54]. In SG, the scheduling of home appliances is dependent
on consumers’ willingness and intermittent nature of RESs, which adds unknown variables to
the environment. Therefore, computational intelligence-based techniques like GA, BPSO, EDE
etc. can possess some inherent randomness and provide a solution to the uncertain problems.
Further, integration of RESs as used in works [9, 18, 22, 52] may also be considered in HEM
schemes to reduce the peaks and CO2emissions.
The authors [30, 31, 32], presented fog computing framework for the distributed data ag-
gregation, which reduced the size of data sent to the cloud. However, these work focused only
on distributed data aggregation and do not consider energy management at microgrids level. In
[33], the authors considered scalability issues in cloud and proposed energy management-as-a-
service over fog platform. In the simulations, a home and microgrid level prototypes were con-
sidered. The authors achieved interoperability, flexibility and data privacy features required for
energy management, however, the limitation consists of considering a single microgrid which
is enabled with a transformer and three homes. In [37], the authors reduced the energy of IoT
devices by considering microgrid and fog computing. The fog consumed less energy for IoT ap-
plications requiring low computational efforts on data. However, the microgrid was only limited
to a residence.
5 Problem statement
Based on the above-given analysis of literature, some limitations have been identified because
of which we are motivated to propose a HEM system enabled with meta-heuristic schemes that
are used to manage the energy consumption, which ultimately reduces the electricity cost, PAR
and user discomfort [13, 18]. In addition to all aforesaid performance parameters, our objective
is to consider the consumer’s lifestyle by assigning priorities through which they can schedule
appliances as per their requirements. Further, rebound peaks generated during off-peak hours
are also dangerous and may harm the grid stability if not handled with care [24]. Due to this fact,
the utility encourages consumers to reduce their load during on-peak hours by incentivizing in
terms of DRPs. In order to minimize the rebound peaks, it is vital for EMC to consider threshold
and load limits in the design. Moreover, actual load profiles are replaced by maximum or the
average load of devices which may not give accurate results when compared with the real load
profile, which takes different energy cycles of an appliance [51, 52]. Thus, the actual load profile
of appliances may be taken into account. The increased demand of electricity has caused energy
scarcity problems, where electricity demand is higher as compared to its supply. Therefore, a
model that integrates RESs and ESS is also essential to minimize the aforesaid issue.
In the literature, user comfort is associated with thermal comfort [39] or related to the ap-
pliances’ delay, cost savings and return on investment parameters [14, 40, 41]. Ogunjuyigbe
et al. considered the user’s satisfaction from a different perspective which is based on three
satisfaction postulations [42]. However, none of the referenced work has considered the user
comfort which may be derived from time and device based varying priorities. Further, the con-
sumer’s budget limitation is also one of the prominent constraints to the electricity usage mostly
neglected in the literature as their primary focus is on cost minimization. We are motivated to
present a new concept of absolute user comfort which is being derived from time and device
based priority values assigned to the appliances by a typical consumer. The challenge required
here is to schedule those appliances in different time-slots of a day that may yield maximum
user’s comfort based on the budget constraint in a dynamic environment.
Beside the HEM system and new user’s comfort derivation based on dynamic priorities, we
extend the work to SCs. The SCs are very large distributed systems and also complex by nature.
The concept of SC incorporating smart grids, microgrids, and PHEVs have emerged to manage
the energy in an efficient way. However, the microgrids installed at a different location of SC will
receive an exponential energy demand from a number of devices, including smart meters, IoT
devices, AMI, PHEVs etc. Further, the penetration of PHEVs is likely to affect the reliability of
the SG by consuming a massive amount of energy [43, 44]. Moreover, the energy demand of the
SC is exponentially rising due to the high proliferation of smart appliances and IoT devices. The
sheer amount of data collected from the PHEVs, smart appliances and SMs fall into the category
of big data [30]. Thus, it becomes very problematic to schedule all the entities, including PHEVs
and building devices from a centralized location while considering the minimization of network
cost and latency requirements. Therefore, the emerging technologies, including fog computing
is considered to resolve issues pertaining to big data problems and its scalability at distributed
level [33]. Without usage of emerging trends, PHEVs and other appliances may get starved
and their operations are being delayed, which may not be accepted by the consumers. So, the
deployment and energy trading of PHEVs at microgrids across different regions of the SC using
the fog computing technology to reduce the transmission latency and energy losses is the need
of the day.
6 Problem formulation and proposed solution
The appliance scheduling problem is formulated as an optimization problem using MKP. The
problem formulation of three appliances in terms of load categorization, PAR, threshold, and an
objective function with constraints are discussed below:
6.1 Load categorization
Let Anrepresents a set of three appliances consisting of clothes dryer Acd , dishwasher Adw ,
and refrigerator Are f . Thus, An={Acd ,Adw ,Aref }. Appliances are scheduled in 24 hours time
horizon τ ε TT={τ1,τ2,τ3, ...., τ24}. In the following subsections, we discuss the energy
consumption and cost calculation for appliances.
6.1.1 Clothes dryer
The clothes dryer operation pattern is similar to the dishwasher as shown in Fig. 2. Let P
cd
represent the power rating of clothes dryer. Then total energy consumption per day of clothes
dryer (εcd ) is represented by the following equation:
εcd =
T
τ=1
P
cd ×χ(τ)(1)
The clothes dryer per day total cost ζD
cd in time interval Tis calculated through formula:
ζD
cd =
T
τ=1
P
cd ×χ(τ)×ρ(τ)(2)
Similarly, clothes dryer monthly ζM
cd and yearly ζY
cd costs are given as:
ζM
cd =
M
d=1
ζD
cd (d)M=30 (3)
ζY
cd =
Y
y=1
ζM
cd (y)Y=12 (4)
Here, we minimize the hourly cost of a clothes dryer, which results in the overall cost re-
duction. The hourly cost of clothes dryer ζH
cd is given by the formula:
ζH
cd =P
cd ×χ(τ)×ρ(τ)τ={1,2, ...., T}(5)
In the above equations and including rest of the work, χ(τ)=[0,1]is a boolean integer
variable which shows appliance status. χ(τ) = 1, if appliance is ON in time slot τand otherwise.
Electric price is represented by ρ(τ)which in this case is DA-RTP signal.
6.1.2 Dishwasher
The dishwasher usage is similar to clothes dryer pattern which also spikes after breakfast and
dinner, coinciding peak hours Fig. 2. Let P
dw denotes the power rating of a dishwasher. The total
energy consumption per day of the dishwasher (εdw) is represented by the following equation:
εdw =
T
τ=1
P
dw ×χ(τ)(6)
The dishwasher per day total cost ζD
dw in Tis calculated through formula:
ζD
dw =
T
τ=1
P
dw ×χ(τ)×ρ(τ)(7)
Similarly, monthly and yearly cost of the dishwasher is given in Eq. 8 and 9.
ζM
dw =
M
d=1
ζD
dw (d)M=30 (8)
ζY
dw =
Y
y=1
ζM
dw (y)Y=12 (9)
In this work, we minimize the hourly cost of the dishwasher, which results in the overall
cost reduction. The hourly cost of dishwasher ζH
dw is given in Eq. 10.
ζH
dw =P
dw ×χ(τ)×ρ(τ)τ={1,2, ...., T}(10)
6.1.3 Refrigerator
The refrigerator has different energy pattern when compared to dishwasher and clothes dryer.
Let P
re f shows the power rating of the refrigerator. The total energy consumption per day of a
refrigerator (εre f ) is represented by the following equation:
εre f =
T
τ=1
P
re f ×χ(τ)(11)
The refrigerator per day total cost ζD
re f in Tis calculated through formula:
ζD
re f =
T
τ=1
P
re f ×χ(τ)×ρ(τ)(12)
Similarly, refrigerator monthly ζM
re f and yearly ζY
re f costs are given as:
ζM
re f =
M
d=1
ζD
re f (d)M=30 (13)
ζY
re f =
Y
y=1
ζM
re f (y)Y=12 (14)
We minimize the hourly cost of the refrigerator, which results in the overall cost reduction.
The hourly cost of refrigerator ζH
re f is given by the formula:
ζH
re f =P
re f ×χ(τ)×ρ(τ)τ={1,2, ...., T}(15)
The total energy consumed and total cost per day are given by equations 16 and 17:
εAn =εcd +εdw +εre f (16)
ζAn =ζD
cd +ζD
dw +ζD
re f (17)
6.2 Peak-to-average ratio
In order to maintain a balance between supply and demand It is essential to reduce PAR. PAR
is defined as the ratio of the peak load and average load consumed by the consumer in 24 hours
time slots. The mathematical equation for PAR is given in Eq. 18.
PAR =max(ετ
n)
1
TT
τ=1An
n(ετ
n)T=24 (18)
6.3 Threshold
The threshold of an appliance is dependent on the priority µset by the consumer and energy
consumption εand is calculated by the equation given below [51].
Z=r2(ρpρo)µτ
ε+ρo(19)
Here, ρpand ρorepresents the maximum and minimum electric price values at time slot τwhen
the appliance is using energy consumption ε.µrepresents priority also called time factor of
an appliance which is set by the consumer. We have implemented the above threshold equation
in proposed HEMC. The threshold value is provided to HEMC which checks and compares it
with current EP received from the utility. If EP is less than the threshold, HEMC turns on the
appliance [51].
6.4 Delay
In order to save cost, appliances are shifted from on-peak to off-peak hours, which causes delay.
Thus, delay is caused by the appliance deviation from its normal operational scheduling pattern.
The user comfort is associated with the appliance’s delay. A higher delay will cause more user
discomfort. The average delay is calculated by formula:
AvgDelay =abs(Unsch.Sch.)
LOT (20)
where Unsch. is the location of unscheduled appliance with ON status, while Sch. depicts
the location of scheduled appliance obtained via HEMC. The abs represents absolute function
and LOT is total length of operational time.
6.5 Knapsack problem formulation
In this section, we formulate home appliance scheduling problem as an optimization problem
and map it to a MKP. The SKP is a combinatorial problem which consists of one knapsack with
a specific capacity limit. Many objects with different weights and values need to be assigned to
knapsack such that the profit is maximized considering the capacity limit. On the other hand,
MKP is a resource allocation problem which consists of mresources or knapsacks and a set of
nobjects. Like SKP, each object in MKP is associated with certain weight and value. Each
resource or knapsack jhas a capacity limit C j, which depicts the maximum weight that can be
supported. The objective of MKP is to find a combination of objects that can be packed within
the knapsacks so that total profit or net value of objects in all knapsacks is maximized [45].
Following are assumptions made to map scheduling problem using MKP:
The objects in MKP are considered equal to the number of appliances.
The weight of each object is represented by the energy of the appliance consumed in each
time slot.
The object value in a specific time slot is specified by the cost of power consumption of
appliance in that time slot.
The capacity of jnumber of knapsacks represents the threshold of power limit in each
time slot. In this work, we assume this limit is fixed and equal to maximum non-scheduled
aggregate load.
The binary variable χ[1,0]which shows ON or OFF status of the appliance.
These assumptions provide consumer to participate in energy management schemes to re-
duce their electricity cost and delay. In this work, we proposed HEMC based on optimization
schemes EDE, GA, BPSO and OSR which successfully shifts load from on-peak to off-peak
hours.
6.6 Objective function
The cost minimization objective function is given as:
min
T
τ=1
An
n=1
ζD
n(τ)(20)
Here χn(τ)is boolean integer [0,1], which shows appliance nstatus On/Off in time slot τ.ζD
n
represents daily electricity cost of the appliance nin time slot τ. This cost is given by equation
given below:
ζD
n=
T
τ=1
An
n=1
P
n×χn(τ)×ρn(τ)(21a)
subject to:
T
τ=1
An
n=1
P
n×χn(τ)γcap (21b)
T
τ=1
An
n=1
χNsch.
n(τ) =
T
τ=1
An
n=1
χSch.
n(τ)(21c)
εNsch.
An =εSch.
An (21d)
ζNsch.
An >ζSch.
An (21e)
ζMNsch.
An >ζMSch.
An (21f)
ζYNsch.
An >ζYSch.
An (21g)
PAR Γmax (21h)
Eq. 20 shows objective function to minimize the electricity cost. The daily electricity cost is
represented in Eq. 21a. The Eq. 21b-21h shows constraints of objective function. Eq. 21b
ensures total load demand of appliances should be less than or equal to the grid capacity (γcap ).
Constraint Eq. 21c represents that LoT of all appliances before and after scheduling must be
equal. The constraint in Eq. 21c also guarantees that total energy consumed by appliances
before and after scheduling must also be equal is given in Eq. 21d. Constraints Eq. 21e, Eq.
21f, and Eq. 21g shows daily, monthly and yearly costs of nonscheduled must be greater than
scheduled costs respectively. Eq. 21h states that the PAR value must be less than or equal to
the knapsack capacity limit which is Γmax. In our scenario Γmax will be equal to maximum
non-scheduled peak load.
6.7 Proposed solution
We apply four optimization schemes EDE, GA, BPSO, and OSR to solve the MKP. The meta-
heuristic schemes EDE, GA and BPSO are similar in nature because of their population gen-
eration based search methods. Deterministic and probabilistic rules are applied to improve the
new generation of populations during each iteration. Metaheuristic algorithms make few or
no assumptions about the problem being optimized. This show that metaheuristics are very
general and use large population sizes that travel semi-randomly within a search space to find
global solutions [46, 47]. Metaheuristic algorithms are not fundamentally limited by restric-
tive assumptions of search space (assumptions concerning continuity, unimodality, existence of
derivatives, etc.) [48]. In the following subsections, we shortly discuss the proposed schemes.
6.7.1 Enhanced differential evolution
The EDE is an enhanced version of DE. The control parameters of DE are population size POP,
mutation factor MF , and crossover rate CR.POP parameter affects the ability to search the
space. The MF controls the convergence speed while CR is relevant to the number of changes in
population. The limitation of DE is low accuracy and slow convergence rate, which is improved
in EDE. In EDE the control parameters are reduced to two (POP and MF ).
The modification in EDE is done at a stage of generating trial vectors CR. Five groups of
trial vectors are generated in each iteration. The first three trial vectors are obtained by taking
three distinct CR values i.e. 0.3, 0.6 and 0.9. The fourth trial vector aims to speed up the
convergence rate while the last trial vector increases the population diversity. The equations for
generating five groups of trial vectors are given in [16]:
ui=vii f (rand (d)0.3,
xii f (rand (d)>0.3(22)
ui=vii f (rand (d)0.6,
xii f (rand (d)>0.6(23)
ui=vii f (rand (d)0.9,
xii f (rand (d)>0.9(24)
ui=rand(d).xi(25)
ui=rand(d).vi+ (1rand (j)).xi(26)
Here, d=1, uirepresents the five trial vectors, viis mutant vector and xiis the target vector. A
fitness function is used to evaluate generated trial vectors. The trial vector with minimum fitness
function value is considered.
6.7.2 Genetic algorithm
One of the metaheuristic optimization algorithm is GA, which is inspired by the biogenetic pro-
cess of living organisms. Here new genes are formed which carry properties and characteristics
of their parents. In GA, we generate a random population of chromosomes. Each of these chro-
mosomes represents a candidate solution to the problem. In our case, binary GA is implemented
in which each bit of chromosome is associated with on/off status of the appliance. The number
of bits in the chromosome is equal to different energy cycles of the appliance. A fitness function
based on the objective function evaluates each chromosome of the initial population. The best-
fitted value is chosen and recorded as current best value. Based on these current best solutions,
a new stream of a population is generated by using steps of crossover and mutation.
Despite other, two important control parameters in GA are crossover and mutation. In the
crossover step binary strings are crossover from the two parents selected randomly or using
another selection method i.e. tournament etc. The probability of crossover rate value specifies
the convergence rate. A high value of the crossover will result in faster convergence at the cost
of accuracy. The best crossover rate specified in the literature is given as:
pc=0.9 (27)
To create randomness in population so that it may not stuck in local, mutation function is ap-
plied. In mutation, one or more genes in a chromosome are changed. The probability of muta-
tion is very low and is calculated by:
pm=1pc(28)
Finally, new population based on crossover and mutation is generated. This population is
again evaluated by fitness function and compared with the previous population to find the global
best optimal solution.
6.7.3 Binary particle swarm optimization
Particle swarm optimization (PSO) was first proposed in 1995 [53] is used to find optimal so-
lutions to problems continuous by nature. However, this algorithm cannot be directly applied
to discrete problems. The authors extended PSO to BPSO in 1997 to solve discrete nature of
problems. It is based on the concept of swarm intelligence. In swarm intelligence, the emer-
gent behavior occurs when agents interact locally with their environment and results in coherent
global patterns. It is a nature-inspired swarm intelligence optimization technique which mimics
the social behavior of a flock of birds, fishes, bees or ants. The individuals start the search for
food in random directions and reach a food source by sharing information. This phenomenon
is also called the social concept of PSO [54]. The individuals are represented by the particles
that make a swarm, moving around the search space for finding the optimal solution. In BPSO,
the initial population is generated randomly in the form of position matrix. Each bit position
in the matrix represents the state of the appliance. Each row in the position matrix represents
a candidate solution to the problem. The initial velocity of individuals is generated by given
formula:
vi=vmax ×2×(rand(swarm,n)0.5)(29)
The particles move freely in the search space and are evaluated through fitness function and
updated via velocity equation given as [55].
vi(t) = w.vi(t1) + φ1.rand(1).pbest xi(t1)(30)
+φ2.rand(1).gbest xi(t1)
In above equation, φ1and φ2are acceleration constants which controls the movement of particle
towards pbest and gbest positions. In our case both acceleration constants have equal value of
2. wis a weighted factor and is calculated by using formula.
w=wi+wf(wi×k)
nitra (31)
Where wiand wfhave values 2 and 0.4 respectively, and nitra represents the total number of
generations while kis the loop index value. Since the velocity values of the particles are real
numbers which are mapped between 0 and 1 by using a function called sigmoid. The sigmoid
function is given by the equation:
sig(i,j) = 1
1+expvi(t)(32)
The position matrix is updated by comparing sigmoid function values with random function
values using the formula:
xf=1 sig(i,j)<rand(1)
0 otherwise.(33)
Moreover, the fitness of above position matrix is calculated and compared with old fitness
values. The minimum fitness value is chosen for the final position matrix for the next generation.
The whole process is repeated until a stopping criterion is met. Finally, from the final position
matrix gbest is selected based on the fitness function.
6.7.4 Optimal stopping rule
In the mathematics, OSR is related associated with the problem of choosing right time based on
sequentially observed random variables to make a decision or take a particular action, in order
to minimize cost or maximize an expected reward. The following two objects define stopping
rule problems:
• A sequence of random variables, X1,X2,.......,Xnwhose joint distribution is assumed
known.
A sequence of real-valued reward functions, y0,y1(x1),y2(x1,x2), ......, y(x1,x2, ...., xn)
The OSR is to observe the sequence X1,X2,... for each n=1,2, ..., and stop at a particular instant
to receive the know reward yn(x1, ...., xn)or may continue and observe Xn+1. If no observations
X1=x1,X2=x2,...Xn=xnis taken, a constant amount y0is received or if one never stop, will
receive y(x1,x2, ..).
The threshold value Zis calculated by eq. 19. If an appliance iis postponed to next
available hours, two additional cost may arise. The first is the electricity cost which may vary at
the next time slot. The second is waiting cost of an appliance. Thus, these costs are calculated
by eq. 34 given below [51].
κi=εZ
i+ρo
2ρ(t)+µ(i)
ρpρo
Z
iρo
(34)
Here, ρpand ρorepresents the maximum and minimum electric price values at time slot τ
when the appliance iis using energy consumption ε.µirepresents the priority of an appliance
which is set by the consumer. The objective is to switch on the appliance where total cost is
minimum.
7 Proposed system models
In this work, we propose two different models. One model is considered for home while the
other model is proposed for SC.
7.1 Smart home model
The proposed system model consists of home area network, AMI, HEMC, and appliances en-
ergy consumption pattern is shown in Fig. 1.
Smart
Clothes
dryer
Smart TV
Smart
Refrigerator
PC
Mobile
Juicer/ grinder
Fan
Smart Home
Smart
Dishwasher
Energy Management
Controller (EMC)
Wi-Fi/
ZigBee
routers
Utility network
NAN
Smart meter
Smart meter
Smart meter
RTP signals
Appliances
usage data
HAN
AMI
Smart plug
Figure 1. Proposed system model for smart home
7.1.1 Home area network
The physical and digital entities could be linked together to form a giant IoT network. Building
automation, SG, healthcare, and many other fields have incorporated IoT paradigm [56]. Many
appliance companies including Panasonic, LG, Samsung, and Whirlpool have invested in smart
appliances’ production which supports automatic DR [51]. The SG networks include home
area network (HAN) which supports wired and wireless technologies including ZigBee, WiFi,
WiMAX, etc. [57]. Smart appliances and IoT have the capability to schedule themselves to
off-peak hours based on the instructions received from the HEMC. The HEMC decisions are
based on the pricing signals received from the utility through the smart meter and consumer
defined appliance priorities. The HAN provides a platform that enables communication among
appliances, smart meter, and consumers.
The proliferation of IoT devices and widespread low-cost wireless technologies such as
Wi-Fi, ZigBee etc., has enabled to accelerate the deployment of HAN. Smart appliances are
supported with wireless and network interfaces through which they can communicate with con-
sumers or HEMC. The HEMC gets DA-RTP signal information from the utility through smart
meters via the Internet. The smart appliances also get input information from consumers through
various applications installed on their smartphones or laptops. The communication among con-
sumers, appliances, and smart meter could be achieved using wireless protocols. The intelli-
gence embedded in smart appliances enables them to operate in energy saving modes or delay
its operation till DR prices drop below a certain threshold value. The examples include smart
dishwasher to stop its operation from on-peak hours to off-peak hours to reduce cost in system
stress time. Similarly, a smart refrigerator can delay it defrost and ice-making cycles to off-peak
periods.
7.1.2 Advance metering infrastructure
Automatic meter reading (AMR) enables the utility to read consumers units from the meters
in an automated manner. The unidirectional communication flow from meters to utility does
not provide maximum benefits to both. In order to overcome this deficiency AMI has been
proposed which has the facility of bidirectional communication. AMI supports smart meters
which receive DA-RTP price signals from the utility. The DA-RTP signal is sent by the utility
to HEMC through neighbourhood area network (NAN). Based on the DA-RTP prices signal
HEMC makes intelligent decision to shift load from on-peak to off-peak hours considering
appliance priorities.
The consumers change their electric appliances usage pattern from its normal consumption
in response to the DR signals received from the utility. Thus, DR programs may fall into incen-
tive and price based categories. In our model, we only focus on the later. In order to save elec-
tricity cost, residential consumers schedule appliances in time slots where price is low. For this
purpose, DR requires the availability of AMI and smart meters which enables the bi-directional
flow of information between utility and consumers.
7.1.3 Home energy management controller
The HEMC installed in residence connects AMI and HAN to enable bidirectional communica-
tion between the two subdomains. The utility sends DA-RTP signals to all smart meters installed
in the residential area through NAN. The HEMC is connected to all IoT devices, smart appli-
ances through HAN. The HEMC also contains information of power ratings of appliances, op-
erational time, status, consumer-defined comfort level, threshold and the priorities information.
Based on the information, HEMC makes an intelligent decision using proposed optimization
techniques GA, EDE, BPSO and OSR to schedule load to off-peak hours. Thus based on the
appliance priorities consumer electricity cost is minimized.
Smart appliances like clothes dryer, refrigerator and IoT devices have DR signals informa-
tion and can operate in less energy saving modes when prices are high or they can delay their
operation until electricity prices get below a threshold level. Other legacy devices like light-
ings, fans etc. which do not have intelligence and communication abilities could be controlled
and managed via smart plug installed in HAN. Smart plugs are intelligent devices enabled with
communication capabilities to remotely on/off the appliances.
2 4 6 8 10 12 14 16 18 20 22 24
Hour of the day
0
0.02
0.04
0.06
0.08
0.1
0.12
Normalized hourly energy usage
Clothes dryer
Dishwasher
Refrigerator
Figure 2. Appliance energy profile data [58]
7.1.4 Appliances energy consumption pattern
Appliances’ energy consumption patterns are essential to be provided to HEMC for load schedul-
ing and energy management. Each appliance installed in the home has unique energy usage pro-
file. We consider three appliances: clothes dryer, dishwasher, and refrigerator for scheduling.
Their usage profile for 24-hours time slots is depicted in Fig. 2. The dishwasher is often used
after breakfast and dinner coinciding with on-peak hours. The DA-RTP signal is plotted in Fig.
4. The on-peak hour is 9 and off-peak hours are 21 and 24. We assume the DA-RTP remains
fixed during a month. It changes within limits range for the remaining months. The time slot for
scheduling is set to one hour and the duty cycles of appliances are usually less than an hour. It
means once the appliance starts, the price is constant during that time slot. For appliance having
duty cycles greater than an hour, task decomposition has been applied. In multi-stage home
appliances, there exists a dominant stage which consumes the most energy. The dishwasher
can be divided into stages like the main wash, final rinse, and heated dry. The clothes dryer
consists of only one cycle. We consider ice-making and defrosts cycles of the refrigerator for
scheduling, similar to [51]. Thus, based on the discussion above, we may consider the following
assumptions:
A single home is considered.
The home is equipped with multiple appliances however, we assume three different ap-
pliances for scheduling.
7.2 Smart city model
The SC model consists of various regions containing microgrids connected to fog and cloud is
shown in Fig. 3.
7.2.1 Smart city regions
The SC is divided into n-regions. Each region has some specific area consisting of a smart build-
ing (SB), smart community (SC), commercial area (CA) and industrial area (IA). We consider
Wind Turbine
Photovoltaic
Generator
Electric Vehicle
Smart Home
Commercial Area
Smart Buildings
Industrial Area
Smart Community
Battery Storage
SMs Region-1
Fog Server-N
Cloud
Clouds
Communication
Fogs Communication
Smart City Divided into N-Regions
Containing N-Microgrids
Fog Server-1
Utility
Microgrid-1
Region-1
Wind Turbine
Photovoltaic
Generator
Electric Vehicle
Microgrid-N
Smart Home
Commercial Area
Smart Buildings
Industrial Area
Smart Community
Region-N
SMs Region-N
Main Grid
Microgrids Energy Trading
Figure 3. Proposed system model for smart city
various smart appliances and IoT devices which are installed in each region and also connected
to smart meters (SMs). IoT devices and SMs produce a tremendous amount of data packets
that are hard to store, process and analyze by cloud computing [32]. According to Cisco statis-
tics, IoT devices that are connected to the Internet by 2020 will be more than 50 billions [59].
Therefore, we propose a distributed model equipped with fog servers to minimize the load on
clouds.
7.2.2 Microgrids
Each SC region has a microgrid which generates its own electricity through various GERs like
a wind turbine, photovoltaic etc. Heavy generator installed in the region is used as a backup,
while BSS along with electric vehicle (EV) is used for storing energy. The energy generation
close to its consumption has many advantages like fewer power losses and low carbon emission.
The microgrid installed in a region may have three different types of status including:
Balance energy status: Where demand and supply of electricity are equal.
Surplus energy status: Where electricity demand is less than supply.
Deficit energy status: A situation, where demand of electricity is more as compared to its
generation.
All microgrid statuses are conveyed to a respective fog server installed in the region. Instead
of arranging electricity from the main grid the fog may also communicate to nearby fogs to
resolve the energy problems. Thus, energy trading is only performed at microgrids level. This
leads to low data latency, reduced energy losses and also increase the overall system throughput.
7.2.3 Fog servers
Cloud computing offers a solution to big data generated by SMs and IoT however, has limitation
to low-latency applications like substation supervisory control and data acquisition (SCADA)
and inter-substation communication where time constraints range from milliseconds to a few
seconds [60]. As the distance between IoT and cloud server increases, it results in lower perfor-
mance, reliability, and security. Second, the services offered to low-latency SG applications by
the cloud has no guarantees. Thus, to cope with the cloud computing limitations, fog computing
offers facilities like computation, storage, and other network services, typically located at the
edge of the network [61]. Therefore, we propose a fog computing environment, where, big data
is preprocessed and computed before transmitting it to the cloud servers. In the proposed model
fog servers are connected to only private clouds. Thus, deployment and energy trading of micro
grids in different regions of the SC using fog computing to reduce the transmission latency and
energy losses is need of the day.
7.2.4 Clouds
The utility is connected to the public cloud using wired/wireless backhaul network to share DR
signals to manage the load and provide incentives to consumers to reduce their electricity cost.
The main grid information also resides in public cloud. The private clouds containing fog data
are connected to the public cloud. However, it is noted that private cloud only shares public data
which may be required for only billing purpose by the utility along with microgrid status reports.
Two private clouds may share their public data through the cloud to cloud (C2C) communication
link.
The next section discusses simulation results based on a smart home model in two scenarios.
Performance parameters, feasible region, and trade-off among performance parameters are also
discussed.
8 Simulations and discussion
Three different smart appliances clothes dryer, dishwasher, and refrigerator have been consid-
ered for scheduling. The appliance energy consumption profile is given in the Table 3 [51] and
usage profile in Fig. 2. Due to stochastic nature of metaheuristic techniques, we have plotted
mean values of GA, EDE, and BPSO after ten iterations. This section is divided into four differ-
ent subsections: comprising of performance parameters definition, simulations and discussion,
feasible region, and trade-off. We consider two different scenarios. In scenario-1, we apply high
and low µvalues to each atomic or individual appliance and calculate the corresponding cost
and delay parameters. In scenario-2, all the three appliances are aggregated to show the perfor-
mance of meta-heuristic schemes and OSR. Here, we also apply a knapsack capacity limit and
discuss all the four performance parameters accordingly.
Table 3 Energy profile of appliances
Appliance Total
energy
(kWh)
Cycle during
(hrs)
Peak power
(kW)
Average
power (kW)
Time fac-
tor (µ)
Clothes
dryer
3.0 0.75 6.0 3.0 0.001,
0.13
Dishwasher 1.4 1.75 1.18 0.8 0.001,
0.017
Refrigerator 2.4 24 0.574 0.089 0.0033,
0.0089
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hours)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Cost ($/kWh)
DA-RTP prices
Figure 4. DA-RTP signal
8.1 Performance parameters definition
We have considered four performance parameters cost, delay, energy consumption and PAR.
These parameters are discussed below:
8.1.1 Cost
Electricity cost is defined as the total power consumed by the consumer during a time slot
multiplied by the electricity price (EP) signals received from the utility. The day is divided into
hourly time slots. The objective is to reduce the electricity cost of a consumer by scheduling the
peak energy consumption cycles of the appliance to off-peak time slots.
8.1.2 Delay
The scheduling of appliances from on-peak to off-peak hours causes delay. Delay is the differ-
ence between scheduled time slots and arrival time slots of an appliance. The delay is inversely
proportional to cost and also causes user discomfort.
8.1.3 Energy consumption
Each appliance consumes energy in order to operate. Total energy, peak power and average
power consumed by appliances are given in Table 3.
8.1.4 Peak-to-average ratio
The ratio of peak load to its average load in time interval τis PAR, which is used to maintain
the demand and supply relationship. It is considered as an essential factor for both the utility
and the consumers. No unit is associated with the PAR since it is a ratio.
8.2 Scenario-1 atomic appliances
In scenario-1, we discuss the cost and delay parameters of three appliances: clothes dryer,
dishwasher, and refrigerator and compare the performance of meta-heuristic and OSR schemes.
We consider average monthly and yearly cost of the appliance with respect to time factor or
priority (µ) values. The µvalue is consumer defined variable and has a direct effect on the
cost and delay. The higher µvalue of an appliance shows that consumer wants maximum user
comfort, with lower appliance delay and is willing to pay more cost.
8.2.1 Clothes dryer
We observe that clothes dryer is a suitable appliance for DR due to the facts that its duty cycles
are short and requires high energy consumption and high peak demand. The cost and delay
Table 4 Comparison of cost and delay: clothes dryer
Scheduling
Technique
Time
Factor
(µ)
Average
Cost ($)
Difference Percentage
decrement in
Cost
Average
Delay
(hrs)
Non-
scheduled
- 153.32 - - -
EDE 0.001 55.10 98.22 64.06 % 7.62
0.13 102.57 50.75 33.10 % 2.38
GA 0.001 60.83 92.49 60.32 % 7.42
0.13 104.21 49.11 32.03 % 3.09
BPSO 0.001 55.16 98.16 64.02 % 7.62
0.13 103.21 50.11 32.68 % 3.40
OSR 0.001 45.38 107.94 70.40 % 6.57
0.13 99.15 54.17 35.33 % 2.69
parameters comparison of the clothes dryer is summarized in table 4. Four different optimization
schemes EDE, GA, BPSO, and OSR have been implemented in HEMC to reduce the electricity
cost. A high µvalue will lead to shorter appliance delay. The non-scheduled yearly load cost
incurred by clothes dryer is $153.32. When the µvalue of the clothes dryer is low (0.001) EDE,
GA, BPSO, and OSR have reduced the cost by 64.06%, 60.32%, 64.02 and 70.40% with an
average delay of 7.62, 7.42, 7.62 and 6.57 hours respectively. These high savings are the result
of a difference between high on-peak and low off-peak price signals communicated by the utility
to the consumer via AMI network. When the µvalue of the clothes dryer is high (0.13), EDE,
GA, BPSO, and OSR have reduced the cost by 33.10%, 32.03%, 32.68% and 35.33%. The
average delay incurred by EDE, GA, BPSO, and OSR is 2.38, 3.09, 3.40 and 2.69 hours. When
the µvalue is set low or high, EDE and BPSO perform well as compared to GA while OSR has
achieved highest savings with least amount of delay. Thus, different µvalues of clothes dryer
set by the consumer have direct effect on average cost and delay. The average cost increases
with the increase in µof appliance resulting in low operational delay.
1 2 3 4 5 6 7 8 9 10 11 12
Time (Month)
2
4
6
8
10
12
14
16
18
Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.13
GA mu=0.001
GA mu=0.13
BPSO mu=0.001
BPSO mu=0.13
OSR mu=0.001
OSR mu=0.13
Figure 5. Average monthly cost of clothes dryer
We also notice that there is a trade-off between cost and delay with respect to time factor (µ).
When clothes dryer’s µvalue is high, the consumer has to bear maximum cost with least amount
of delay. Fig. 5 shows the average monthly cost of the clothes dryer. The non-scheduled monthly
cost curve is the base case which is compared to our four proposed optimization schemes. We
noticed that optimization schemes perform well and significantly reduce the average monthly
cost with respect to two different priority values defined by the consumer. In Fig. 5, we also
notice that electricity prices have been constantly increasing for the months 5 to 7, which also
result in consumers bill to pay more for the aforementioned duration.
0
Cases
0
20
40
60
80
100
120
140
160
Average Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.13
GA mu=0.001
GA mu=0.13
BPSO mu=0.001
BPSO mu=0.13
OSR mu=0.001
OSR mu=0.13
Figure 6. Average cost of clothes dryer
Fig. 6 and Fig. 7 show corresponding yearly cost savings and the average waiting delay
of the clothes dryer. Fig. 6 shows that all proposed schemes have reduced the yearly costs
as compared to non-scheduled load cost which is $153.32. When the µvalue of the clothes
dryer is low, EDE, GA, BPSO, and OSR based HEMC have reduced the yearly costs up to
$55.10, $60.83, $55.16 and $45.38 with an average daily delay of 7.62, 7.42, 7.62 and 6.57
hours, respectively. The performance of GA based HEMC is not good in terms of reducing cost
however, it has achieved a high comfort level with less delay when compared to EDE and BPSO
in low µvalue, as shown in Fig. 7. On the other hand, OSR beats the meta-heuristic schemes
0
Cases
0
1
2
3
4
5
6
7
8
Time (Hours)
EDE mu=0.001
EDE mu=0.13
GA mu=0.001
GA mu=0.13
BPSO mu=0.001
BPSO mu=0.13
OSR mu=0.001
OSR mu=0.13
Figure 7. Average delay of clothes dryer
when µis low and high in terms of consumer cost reduction. Considering a high µvalue, EDE
beats the OSR in terms of delay 2.38 hours which is less than 2.69 hours. However, OSR has
also reduced the cost at the expense of high delay as compared to EDE.
When the µof the clothes dryer is high (0.13) the consumer has to bear maximum cost
with least delays as shown in Fig. 6 and Fig. 7, respectively. When the µof the clothes dryer
is high, EDE, GA, BPSO, and OSR based HEMC have the yearly costs of $102.57, $104.21,
$103.21 and $99.15 with an average delay of 2.38, 3.09, 3.40 and 2.69 hours, respectively. Here,
we notice that EDE has performed very well in achieving maximum user comfort with a delay
of 2.38 hours as shown in Fig. 7 and also achieved a high-cost saving of $102.57 when µis
high with respect to GA and BPSO based HEMC. Here GA performed well by maximizing
user comfort with waiting delay of 7.42 hours when compared to EDE and BPSO. In addition,
OSR has achieved the highest cost savings when compared to meta-heuristic schemes. It also
performs well in minimizing average delay when µof the appliance is low. In a nutshell, we
conclude that all our proposed optimization techniques along with OSR have reduced the cost
well as compared to the cost of the non-scheduled load. Fig. 5 and Fig. 6 satisfy our constraints
mentioned in Eq. 21f and Eq. 21g which also show monthly and yearly costs of the non-
scheduled load must be greater than the cost of scheduled load respectively. The µvalue set by
the consumer for an appliance has a direct effect on cost and delay parameters.
0
Cases
0
0.5
1
1.5
2
2.5
3
Total Load (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.13
GA mu=0.001
GA mu=0.13
BPSO mu=0.001
BPSO mu=0.13
OSR mu=0.001
OSR mu=0.13
Figure 8. Total load of clothes dryer
Total energy consumption of clothes dryer in a day are shown in Fig. 8. Clothes dryer’s total
energy is 3.0 kWh as given in the Table 3. Fig. 8 shows that all our proposed schemes equally
run the load of 3 kWh. This figure validates the constraints given in Eq. 21c and Eq. 21d, by
showing that LoT of an appliance and its energy consumption before and after scheduling is
equal.
8.2.2 Dishwasher
Smart dishwasher is also considered as a suitable appliance for DR because its duty cycles are
short and it requires high energy consumption. The simulation results of dishwasher have been
summarized in Table 5. When no HEMC is used the yearly energy consumption cost of the
dishwasher is $143.10. When the µvalue is low (0.001), EDE, GA, BPSO, and OSR based
HEMC have reduced the cost by 48.83%, 47.45%, 48.83% and 55.85% bearing an average
daily delay of 5.00, 4.45, 5.00 and 5.02 hours respectively. When the µvalue is high (0.017),
EDE, GA, BPSO, and OSR based HEMC have reduced the cost by 33.86%, 32.36%, 33.52%
and 28.62% bearing an average daily delay of 0.75, 1.22, 1.18 and 0.72 hours, respectively.
From Table 5, it is obvious that EDE and BPSO HEMC performed equally well in terms of
cost and delay reduction with low µvalue, while GA-based HEMC’s cost is high, however, its
delay is less. When µis set high EDE performs well in reducing both performance parameters
with respect to GA and BPSO. When meta-heuristic schemes are compared with OSR, we find
that OSR reduces cost when µis low. When µis high, meta-heuristic perform better in cost
reduction as compared to the OSR, however, OSR bears the minimum delay and gives maximum
user comfort. This also shows the tradeoff relationship between cost and delay parameters.
Table 5 Comparison of cost and delay: dishwasher
Scheduling
Technique
Time
Factor
(µ)
Average
Cost ($)
Difference Percentage
decrement in
Cost
Average
Delay
(hrs)
Non-
scheduled
- 143.10 - - -
EDE 0.001 73.22 69.88 48.83 % 5.00
0.017 94.64 48.46 33.86 % 0.75
GA 0.001 75.20 67.90 47.45 % 4.45
0.017 96.79 46.31 32.36% 1.22
BPSO 0.001 73.23 69.87 48.83 % 5.00
0.017 95.13 47.97 33.52 % 1.18
OSR 0.001 63.18 79.92 55.85 % 5.02
0.017 102.14 40.96 28.62 % 0.72
1 2 3 4 5 6 7 8 9 10 11 12
Time (Month)
2
4
6
8
10
12
14
16
18
Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.017
GA mu=0.001
GA mu=0.017
BPSO mu=0.001
BPSO mu=0.017
OSR mu=0.001
OSR mu=0.17
Figure 9. Average monthly cost of dishwasher
The average monthly and yearly energy consumption’s cost of the dishwasher is shown in
Fig. 9 and Fig. 10, respectively. We notice that proposed schemes including EDE, GA, BPSO,
and OSR have reduced the cost significantly as compared to non-scheduled load cost. Fig. 10
depicts the annual cost of EDE, GA, BPSO and OSR based HEMCs are $73.22, $75.20, $73.23
and $63.18, bearing delay of 5.00, 4.45, 5.00 and 5.02 hours respectively, when µvalue is set
low (0.001) value. This effect of delay has been shown in Fig. 11. The EDE and BPSO have
decreased the cost and delay by the same percentage and performed well as compared to GA
based HEMC when the µvalue is set low. However, GA has achieved higher user comfort.
In this case, OSR has achieved the highest cost savings with respect to meta-heuristic schemes
when µis low at the same delay as compared to EDE and BPSO. All meta-heuristic perform
well as compared to OSR in terms of cost savings when the µvalue is high however, OSR
achieves maximum user comfort by delaying appliance operation by only 0.72 hours. This is
also due to the fact that OSR cost considers both electricity cost and appliance waiting cost as
given in eq. 34.
0
Cases
0
50
100
150
Average Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.017
GA mu=0.001
GA mu=0.017
BPSO mu=0.001
BPSO mu=0.017
OSR mu=0.001
OSR mu=0.17
Figure 10. Average cost of dishwasher
0
Cases
0
1
2
3
4
5
6
Time (Hours)
EDE mu=0.001
EDE mu=0.017
GA mu=0.001
GA mu=0.017
BPSO mu=0.001
BPSO mu=0.017
OSR mu=0.001
OSR mu=0.17
Figure 11. Average delay of dishwasher
In short, we have applied minimum and maximum µvalues to show its effects on average
cost and delay. As the µof dishwasher increases consumer’s electricity cost will also increase
with least amount of delay. So, maximum user comfort is achieved when µis set to highest
value. In this situation, HEMC will run the appliance on the same pattern as given in non-
scheduled load scenario and will not reduce any cost. However, when µof the appliance is
decreased by the consumer, then appliance has to wait for low peak time slots, in order to
reduce the electricity cost. The Fig. 12 shows that all schemes equally run the load of 1.4 kWh
before and after scheduling.
8.2.3 Refrigerator
Table. 6 shows the comparison of cost and delay of the refrigerator. We found that proposed
optimization schemes EDE, GA, and BPSO along with OSR have reduced the cost when com-
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Total Load (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.017
GA mu=0.001
GA mu=0.017
BPSO mu=0.001
BPSO mu=0.017
OSR mu=0.001
OSR mu=0.17
Figure 12. Total load of dishwasher
pared to non-scheduled load’s cost with respect low and high µvalues defined by the consumer.
When the µof the refrigerator is set to a low value of (0.0033), EDE, GA, BPSO, and OSR have
reduced the cost by 9.42%, 8.18%, 9.29% and 13.55% with an average delay of 11.88, 11.25,
11.88 and 10.38 hours, respectively. When µof the refrigerator is set to the higher value of
(0.0089), EDE, GA, BPSO, and OSR have reduced cost by 4.17% , 4.17%, 4.07% and 8.18%
with the same delay of 5.50 hours for meta-heuristic schemes and 5.38 hours delay for OSR
based HEMC. The maximum cost and delay are shown in the Table 6. The OSR scheme has
significantly decreased the cost and delay values, as compared to other meta-heuristic schemes.
Table 6 Comparison of cost and delay: refrigerator
Scheduling
Technique
Time
Factor
(µ)
Average
Cost ($)
Difference Percentage
decrement in
Cost
Average
Delay
(hrs)
Non-
scheduled
- 122.65 - - -
EDE 0.0033 111.10 11.55 9.42 % 11.88
0.0089 117.54 5.11 4.17 % 5.50
GA 0.0033 112.62 10.03 8.18 % 11.25
0.0089 117.54 5.11 4.17 % 5.50
BPSO 0.0033 111.26 11.39 9.29 % 11.88
0.0089 117.66 4.99 4.07 % 5.50
OSR 0.0033 106.03 16.62 13.55 % 10.38
0.0089 112.61 10.04 8.18 % 5.38
Fig. 13 and Fig. 14 show average monthly and yearly cost savings of the refrigerator after
scheduling. The non-scheduled yearly load cost incurred by the refrigerator is $122.65. When
µof the refrigerator is set to a low value, EDE, GA, BPSO and OSR consumers have to pay a
yearly cost of $111.10, $112.62, $111.26 and $106.03. When the µvalue of the refrigerator is
set high, EDE and GA consumers pay same electricity cost of $117.54, while BPSO and OSR
consumers bear the cost of $117.66 and $112.61, respectively. While comparing meta-heuristic,
EDE performed better in reducing cost as compared to GA and BPSO when µis low. Both EDE
and GA performed equally and better than BPSO when µis set to high value. In this case, OSR
beats all meta-heuristic schemes in terms of cost and delay parameters when µof the appliance
is low or high. Fig. 13 shows that prices are high and constantly increasing from 5 to 7 months,
while these are lowest during the months from 9 to 12.
We also notice that the cost of the refrigerator has not been significantly decreased as com-
1 2 3 4 5 6 7 8 9 10 11 12
Time (Month)
6
7
8
9
10
11
12
13
14
15
Cost (dollar)
Non-schedule
EDE mu=0.0033
EDE mu=0.0089
GA mu=0.0033
GA mu=0.0089
BPSO mu=0.0033
BPSO mu=0.0089
OSR mu=0.0033
OSR mu=0.0089
Figure 13. Average monthly cost of refrigerator
0
Cases
0
20
40
60
80
100
120
140
Cost (dollar)
Non-schedule
EDE mu=0.0033
EDE mu=0.0089
GA mu=0.0033
GA mu=0.0089
BPSO mu=0.0033
BPSO mu=0.0089
OSR mu=0.0033
OSR mu=0.0089
Figure 14. Average cost of refrigerator
pared to clothes dryer and dishwasher because only the ice-making and defrost phases of the
refrigerator have been considered in scheduling. Both ice-making and defrost phases of the re-
frigerator account only small portion of its operation time. Due to this fact, the average cost
of the refrigerator has not been significantly reduced as compared with clothes dryer and dish-
washer.
0
Time (Day)
0
2
4
6
8
10
12
Time (Hours)
EDE mu=0.0033
EDE mu=0.0089
GA mu=0.0033
GA mu=0.0089
BPSO mu=0.0033
BPSO mu=0.0089
OSR mu=0.0033
OSR mu=0.0089
Figure 15. Average delay of refrigerator
The average waiting delay of the refrigerator is shown in Fig. 15, which depicts that with
low µvalue, optimization schemes delay the operation by more than 10.00 hours. When the
µvalue is set high all meta-heuristic schemes have an equal delay of 5.50 hours, while OSR
achieved the highest user comfort level and has a delay of only 5.38 hours. The total load
of the refrigerator is depicted in Fig. 16, which shows that LoT along with the total energy
consumption of 2.4 kWh in a day, before and after scheduling is complete.
0
0
0.5
1
1.5
2
2.5
Total Load (kWh)
Non-schedule
EDE mu=0.0033
EDE mu=0.0089
GA mu=0.0033
GA mu=0.0089
BPSO mu=0.0033
BPSO mu=0.0089
OSR mu=0.0033
OSR mu=0.0089
Figure 16. Total load of refrigerator
In the next subsection, we discuss the simulation results by aggregating all the three appli-
ances and also apply knapsack capacity constraint limit to reduce the PAR.
8.3 Scenario-2 aggregate appliances
In scenario-2, all the three appliances are considered in scheduling and their combined effect is
shown using four performance parameters. Here, we also consider the knapsack capacity limit
to maintain the peak load. As already stated that knapsack capacity limit value is equal to the
maximum non-scheduled load value in each time slot. This will yield a balanced load in each
hour causing PAR value reduction. Another objective achieved through the knapsack capacity
limit is high user comfort which is achieved at the expense of high cost. This effect is due to the
trade-off between cost and delay, which will be discussed in performance trade-off section.
8.3.1 Aggregate appliances without knapsack capacity limit
Table. 7 summarizes the results of optimization algorithms. When the µvalue is set low (0.001),
EDE, GA, BPSO and OSR schemes have reduced the cost by 66.09%, 66.07%, 65.97% and
66.89% with an average delay of 10.63 hours each for meta-heuristic schemes and 9.38 hours
for OSR. Here, OSR beats all other schemes in terms of cost reduction and delay. When µis set
to a high value (0.105), EDE, GA, BPSO and OSR schemes have reduced the cost by 42.84%,
42.52%, 42.82% and 43.13% with an average delay of 3.50, 4.13, 3.50 and 2.75 hours, respec-
tively. Again OSR has better performance in terms of cost and delays reduction with respect to
meta-heuristic schemes. Thus, results show that meta-heuristic schemes provide approximate
solution near to optimal and have comparable performance with respect to OSR. The perfor-
mance of EDE is slightly better than GA and BPSO schemes in terms of reducing cost when µ
value is set low.
Table 7 Comparison of cost and delay: aggregate appliances
Scheduling
Technique
Time
Factor
(µ)
Average
Cost ($)
Difference Percentage
Decrement in
Cost
Average
Delay
(hrs)
Non-
scheduled
- 419.07 - - -
EDE 0.001 142.12 276.96 66.09 % 10.63
0.105 239.54 179.53 42.84 % 3.50
GA 0.001 142.18 276.89 66.07 % 10.63
0.105 240.89 178.18 42.52 % 4.13
BPSO 0.001 142.62 276.45 65.97 % 10.63
0.105 239.62 179.45 42.82 % 3.50
OSR 0.001 138.75 280.32 66.89 % 9.38
0.105 238.34 180.73 43.13 % 2.75
Monthly and yearly cost of appliances are shown in Fig. 17 and Fig. 18, respectively. The
average delay associated with appliances is depicted in Fig. 19. The monthly cost is high from
5 to 8 months. The results show that meta-heuristic schemes have comparable and near-optimal
performance as compared to the OSR scheme. When comparing meta-heuristic schemes, we
find that EDE performance is better in terms of reducing cost from $419.07 to $142.12 with the
same delay of 10.63 hours, when the µvalue is set low. When the µvalue is set high, EDE and
BPSO schemes have almost equal performance. The performance of GA based HEMC is worst
in terms of user comfort, causing an average delay of 4.13 hours. From Fig. 18 and Fig. 19,
it is obvious that OSR scheme has performed best in terms of cost and delay reduction when
compared to proposed metaheuristic schemes. The metaheuristic schemes also have achieved
very near optimal solution, when compared with the OSR scheme.
Daily energy consumption and total energy consumption of appliances are shown in Fig.
20 and Fig. 21, respectively. The non-scheduled load peak created in the on-peak hour is 3.79
kWh. when the appliance µis set to a low value, peaks are created in the off-peak hours 21
and 24, respectively. All the meta-heuristic schemes along with OSR create a high peak of
3.92 kWh which is greater than the non-scheduled load peak. This is due to the rebound-effect
when no load capacity limit is considered, peaks are again created in the off-peak hours. We
will tackle this rebound-effect problem in next sub-section where knapsack capacity is being
applied. When µis set to a high value, appliances cycles are scheduled with less delay and no
peaks are created in the off-peak hours. The scheduled load peak created by all schemes is 3.09
kWh when the µvalue is high and it is less than the maximum non-scheduled load peak. In Fig.
21, the total load of 6.8 kWh before and after scheduling is equal and this satisfies that the LoT
1 2 3 4 5 6 7 8 9 10 11 12
Time (Month)
5
10
15
20
25
30
35
40
45
50
Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 17. Average monthly cost of appliances
0
Cases
0
50
100
150
200
250
300
350
400
450
Average Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 18. Average cost of appliances
0
Cases
0
2
4
6
8
10
12
Time (Hours)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
Figure 19. Average delay of appliances
of all appliances is complete.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hours)
0
0.5
1
1.5
2
2.5
3
3.5
4
Energy consumption (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 20. Energy consumption of appliances
0
Cases
0
1
2
3
4
5
6
7
Total Load (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 21. Total load of appliances
The PAR of the aggregate appliances is depicted in Fig. 22. The non-scheduled load has
created a PAR value of 13.37. All meta-heuristic schemes along with OSR create PAR values of
13.84 and 10.90, when the µvalue of appliances is set low and high, respectively. Thus, peaks
created in the off-peak hours has increased the PAR value with respect to the non-scheduled
PAR value. Based on the discussion above, it is stated that, when the µvalue of appliances is
set low, it causes rebound-effect by creating rebound-peaks in the off-peak hours. To resolve
this issue, we opted knapsack capacity limit which is discussed next.
8.3.2 Aggregate appliances with knapsack capacity limit
The simulation results of aggregate appliances with knapsack capacity limit have been sum-
marized in the Table 8. The knapsack capacity limit in this scenario is equal to maximum
non-scheduled load. This capacity limit is being applied to each time slot in order to mitigate
the rebound-peaks created in the off-peak time slots. The yearly cost of a non-scheduled load is
same $419.07 when no HEMC is used. When µis set to low (0.001) value, EDE, GA, BPSO and
OSR based HEMC have reduced the cost by 60.85%, 60.83%, 60.66% and 65.03% bearing an
average appliance delay of 10.13 hours for all meta-heuristic schemes and 9.88 hours for OSR,
respectively. When the µvalue is set high (0.105), EDE, GA, BPSO, and OSR based HEMCs
have reduced the cost by 42.84%, 42.50%, 42.83% and 43.13% bearing an average daily delay
of 3.50, 4.00, 3.50 and 2.75 hours, respectively. From the Table 8 values, it is obvious that
OSR based HEMC performed best in terms of reducing cost and delay in both µvalues. While
EDE based HEMC achieves the highest cost savings of 60.85% when compared to other two
0
Cases
0
2
4
6
8
10
12
14
PAR
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 22. PAR of appliances
meta-heuristic schemes GA and BPSO in case of low µvalue. The performance of GA based
HEMC is worst in terms of cost and delay reduction with respect to all other schemes when the
µvalue is set high.
Table 8 Comparison of cost and delay: aggregate appliances (knapsack)
Scheduling
Technique
Time
Factor
(µ)
Average
Cost ($)
Difference Percentage
Decrement in
Cost
Average
Delay
(hrs)
Non-
scheduled
- 419.07 - - -
EDE 0.001 164.08 254.99 60.85 % 10.13
0.105 239.54 179.53 42.84 % 3.50
GA 0.001 164.14 254.93 60.83 % 10.13
0.105 240.98 178.09 42.50 % 4.00
BPSO 0.001 164.86 254.21 60.66 % 10.13
0.105 239.60 179.47 42.83 % 3.50
OSR 0.001 146.56 272.51 65.03 % 9.88
0.105 238.34 180.73 43.13 % 2.75
The monthly and yearly cost of appliances with knapsack capacity constraint is shown in
Fig. 23 and Fig. 24, respectively. The average delay associated with appliances is plotted in
Fig. 25. All proposed optimization schemes reduced the monthly cost, as compared to the
base case which is non-scheduled load’s cost. As shown in Fig. 24, the annual cost beard by
EDE, GA, BPSO, and OSR based HEMC are $164.08, $164.14, $164.86 and $146.56, bearing
delay of 10.13 for all meta-heuristic schemes and 9.88 hours for OSR based HEMC when the
µvalue is set low (0.001). The average delay plot is depicted in Fig. 25. Here, we notice
that, after implementing the knapsack capacity limit the monthly and yearly cost of all schemes
is increased, while its delay decreased in case of low µvalue, as compared to the previous
scenario where no knapsack capacity limit was considered. Thus, the knapsack capacity limit
balances the load between supply and demand and reduces the peak formation caused by the
rebound-effect in off-peak time slots.
Daily energy consumption and total energy consumption of appliances with knapsack ca-
pacity limit are shown in Fig. 26 and Fig. 27, respectively. In Fig. 26, the peak created by the
non-scheduled load is 3.79 kWh. With low and high µvalues, all optimization schemes create
a lower peak of 3.09 kWh in the off-peak time slots. The Fig. 27 shows that all optimization
schemes equally run the load of 6.8 kWh before and after scheduling.
The PAR of appliances is presented in Fig. 28. The non-scheduled PAR value is 13.37.
1 2 3 4 5 6 7 8 9 10 11 12
Time (Month)
5
10
15
20
25
30
35
40
45
50
Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 23. Average monthly cost of appliances (knapsack)
0
Cases
0
50
100
150
200
250
300
350
400
450
Average Cost (dollar)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 24. Average cost of appliances (knapsack)
0
Cases
0
2
4
6
8
10
12
Time (Hours)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
Figure 25. Average delay of appliances (knapsack)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hours)
0
0.5
1
1.5
2
2.5
3
3.5
4
Energy consumption (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 26. Energy consumption of appliances (knapsack)
0
Cases
0
1
2
3
4
5
6
7
Total Load (kWh)
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 27. Total load of appliances (knapsack)
Table 9 Possible cases - clothes dryer
Case Load
(kWh)
Electricity
price
($/kWh)
Cost ($)
Min. load, Min. ρ0.0890 0.0154 0.0014
Min. load, Max. ρ0.0890 0.1359 0.0121
Max. load, Min. ρ3.7890 0.0154 0.0584
Max. load, Max. ρ3.7890 0.1359 0.5148
When µis low or high, all schemes have equal PAR of value 10.90, which is less than non-
scheduled PAR value. Thus, by incorporating knapsack capacity limit in each hour results in
18.47% reduced PAR value, which significantly plays a vital role in the grid sustainability.
0
Cases
0
2
4
6
8
10
12
14
PAR
Non-schedule
EDE mu=0.001
EDE mu=0.105
GA mu=0.001
GA mu=0.105
BPSO mu=0.001
BPSO mu=0.105
OSR mu=0.001
OSR mu=0.105
Figure 28. PAR of appliances (knapsack)
8.3.3 Feasible region
Feasible region (FR) in an optimization problem is a set of all possible values and points that
satisfy the problem constraints. FR of aggregate appliances is discussed with respect to cost,
energy consumption, and delay. The DA-RTP (ρ) signals range is given as (0.0154 - 0.1359)
$/kWh. The hourly power consumption load range of appliances is (0.0890 - 3.7890) kWh.
Thus all the possible cases with respect to the DA-RTP range as shown in Table. 9.
The cost of FR obtained must be lower than or equal to maximum non-scheduled load
hourly cost of $0.4055. Thus, based upon these values we define our constraints where the cost
of scheduled load should be less than or equal to maximum non-scheduled load cost in each
time slot. FR constraints are given as:
C1: 0.0014 ζH
cd 0.4055
C2:ζD
cd <0.6495
C3: 0.0890 εH
cd 3.7890
Constraint C1shows the hourly cost of clothes dryer may lie in the given range. The HEMC
scheduled load cost should be less than or equal to maximum non-scheduled load hourly cost
$0.4055 at any time slot. Similarly, constraint C2shows that daily cost of clothes dryer should
be less than daily non-scheduled load cost which is $0.6495. C3shows the hourly energy con-
sumption load range of appliance.
Based on these constraints the FR of clothes dryer showing cost and energy relation is shown
in Fig. 29. We have pointed total region by four points (P
1,P
2,P
3,P
4). These points are cal-
culated using the minimum and maximum load multiplied by minimum and maximum electric
prices communicated by the utility as given in Table. 9. The maximum cost per hour of the
non-scheduled load is $0.4055. Thus the cost of scheduled load should be less than or equal to
non-scheduled load cost at each time slot. The line connecting P
5and P
6depicts that electricity
consumers can minimize their consumption cost by consuming more energy during off-peak
time slots. Therefore, the FR covered by points (P
1,P
2,P
5,P
6and P
4) shows that cost will be
feasible and it will reduce the overall electricity cost. We also notice that at the points P
5and
P
6loads are different, however, the cost is same. This phenomenon shows that high load can be
scheduled at off-peak hours resulting in low electricity cost.
0 0.5 1 1.5 2 2.5 3 3.5 4
Power consumption (kWh)
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Cost per hour ($)
P1(0.0890, 0.0014)
P2(0.0890, 0.0121)
P3(3.7890, 0.5148)
P4(3.7890, 0.0584)
P5(3.0000, 0.4055)
P6(3.7890, 0.4055)
Figure 29. FR cost-energy of aggregate appliances
The FR for cost and delay relationship of aggregate appliance scenario is depicted in Fig. 30.
The point P
1represents no delay with a maximum cost of $419.07, in case of the non-scheduled
load. The point P
2and P
3show the scenario without knapsack capacity limit with two costs
at the minimum and maximum delay point as given in Table 7. The last constraint shown in
points P
4and P
5depict that the maximum delay of the appliances when knapsack capacity limit
is implemented cannot be greater than the delay of 9.88 hours. This is the maximum limit and
beyond this limit will create high PAR value and also result in rebound-peaks creation in the
off-peak hours. We represent delay by λ. Thus, FR constraints are given as:
C1: 0 λD
cd 10.63
C2: 146.56 ζy
cd 419.07
C3:λD
cd 9.88
Constraint C1shows the daily delay of the clothes dryer with minimum and maximum range.
Constraint C2shows the yearly cost range of clothes dryer while the constraint C3shows that
daily delay should be less than or equal to 9.88 hours to maintain PAR value and reduce the
peak load. These constraints are implemented and shown in Fig. 30 which depicts yearly cost
of appliances with respect to daily average delay. The surrounded region by (P
1,P
2,P
4,P
5)
is the feasible region which illustrates that delay of appliances cannot exceed over 9.88 hours
which will create high PAR values. Note that at the point P
4and P
5the delay is same, however,
the costs are different, which shows that incorporating knapsack capacity limit will increase the
cost as compared to no limit. The costs calculated at these points are $146.56 and $162.02,
respectively. The Fig. 30 also shows an inverse relationship between cost and delay. The
increase in the amount of delay will reduce the cost accordingly.
8.3.4 Performance parameters trade-off
From the above discussions, it is obvious to see the trade-off between µ, cost and delay. We
see that priority (µ) value set by the consumer is directly proportional to cost and inversely
proportional to delay. From the simulation and discussions, it is obvious that high µvalue
increases consumer’s cost and decreases appliances’ delay. There is also a trade-off between
0 2 4 6 8 10
Delay (hours)
100
150
200
250
300
350
400
450
Cost (dollar)
P1(0, 419.07)
P2(2.75, 238.44)
P3(10.63, 142.62)
P4(9.88, 146.56)
P5(9.88, 162.02)
Figure 30. FR cost-delay of aggregate appliances
cost and delay. The increase in average cost means appliances minimum delay will be incurred.
Thus, if the consumer wants to save more cost the µof the appliance must be set with smaller
value so that delay is maximized and the appliance is scheduled during off-peak hours. This
cost can only be saved at the expense of user comfort which is the delay. On the other hand, if
the consumer wants to maximize its comfort than appliance µhas to be set to a higher value.
In this situation, the consumer achieves maximum comfort and pays more cost. However, we
also noticed that appliance types also have an effect on cost savings. We may save more on
delaying appliances like clothes dryer and dishwasher. As from simulations section, it is clear
that delaying refrigerator duty cycles may not contribute to major cost savings.
References
[1] Tuballa, M. L., & Abundo, M. L. (2016). A review of the development of Smart Grid
technologies. Renewable and Sustainable Energy Reviews, 59, 710-725.
[2] Framework, N. I. S. T. (2012). Roadmap for Smart Grid Interoperability Standards Release
2.0 [NIST SP-1108R2].
[3] Vardakas, J. S., Zorba, N., & Verikoukis, C. V. (2015). A survey on demand response
programs in smart grids: Pricing methods and optimization algorithms. IEEE Communi-
cations Surveys & Tutorials, 17(1), 152-178.
[4] Hossain, M. S., Madlool, N. A., Rahim, N. A., Selvaraj, J., Pandey, A. K., & Khan, A. F.
(2016). Role of smart grid in renewable energy: An overview. Renewable and Sustainable
Energy Reviews, 60, 1168-1184.
[5] Rasheed, M. B., Javaid, N., Ahmad, A., Awais, M., Khan, Z. A., Qasim, U., & Alrajeh, N.
(2016). Priority and delay constrained demand side management in realtime price envi-
ronment with renewable energy source. International Journal of Energy Research, 40(14),
2002-2021.
[6] Vardakas, J. S., Zorba, N., & Verikoukis, C. V. (2015). Performance evaluation of power
demand scheduling scenarios in a smart grid environment. Applied Energy, 142, 164-178.
[7] Vardakas, J. S., Zorba, N., & Verikoukis, C. V. (2016). Power demand control scenarios
for smart grid applications with finite number of appliances. Applied energy, 162, 83-98.
[8] Moon, S., & Lee, J. W. (2016). Multi-residential demand response scheduling with multi-
class appliances in smart grid. IEEE Transactions on Smart Grid.
[9] Shirazi, E., & Jadid, S. (2015). Optimal residential appliance scheduling under dynamic
pricing scheme via HEMDAS. Energy and Buildings, 93, 40-49.
[10] Muralitharan, K., Sakthivel, R., & Shi, Y. (2016). Multiobjective optimization technique
for demand side management with load balancing approach in smart grid. Neurocomput-
ing, 177, 110-119.
[11] Derakhshan, G., Shayanfar, H. A., & Kazemi, A. (2016). The optimization of demand
response programs in smart grids. Energy Policy, 94, 295-306.
[12] Mhanna, S., Chapman, A. C., & Verbic, G. (2016). A fast distributed algorithm for large-
scale demand response aggregation. IEEE Transactions on Smart Grid, 7(4), 2094-2107.
[13] Logenthiran, T., Srinivasan, D., & Shun, T. Z. (2012). Demand side management in smart
grid using heuristic optimization. IEEE transactions on smart grid, 3(3), 1244-1252.
[14] Ma, K., Yao, T., Yang, J., & Guan, X. (2016). Residential power scheduling for demand
response in smart grid. International Journal of Electrical Power & Energy Systems, 78,
320-325.
[15] Storn, R., & Price, K. (1997). Differential evolutiona simple and efficient heuristic for
global optimization over continuous spaces. Journal of global optimization, 11(4), 341-
359.
[16] Arafa, M., Sallam, E. A., & Fahmy, M. M. (2014, May). An enhanced differential evolution
optimization algorithm. In Digital Information and Communication Technology and it’s
Applications (DICTAP), 2014 Fourth International Conference on (pp. 216-225). IEEE.
[17] Rasheed, M. B., Javaid, N., Ahmad, A., Khan, Z. A., Qasim, U., & Alrajeh, N. (2015).
An efficient power scheduling scheme for residential load management in smart homes.
Applied Sciences, 5(4), 1134-1163.
[18] Rahim, S., Javaid, N., Ahmad, A., Khan, S. A., Khan, Z. A., Alrajeh, N., & Qasim,
U. (2016). Exploiting heuristic algorithms to efficiently utilize energy management con-
trollers with renewable energy sources. Energy and Buildings, 129, 452-470.
[19] Manzoor, A., Javaid, N., Ullah, I., Abdul, W., Almogren, A., & Alamri, A. (2017). An
intelligent hybrid heuristic scheme for smart metering based demand side management in
smart homes. Energies, 10(9), 1258.
[20] Javaid, N., Hussain, S. M., Ullah, I., Noor, M. A., Abdul, W., Almogren, A., & Alamri,
A. (2017). Demand Side Management in Nearly Zero Energy Buildings Using Heuristic
Optimizations. Energies, 10(8), 1131.
[21] Javaid, N., Naseem, M., Rasheed, M. B., Mahmood, D., Khan, S. A., Alrajeh, N., & Iqbal,
Z. (2017). A new heuristically optimized Home Energy Management controller for smart
grid. Sustainable Cities and Society, 34, 211-227.
[22] Ahmad, A., Khan, A., Javaid, N., Hussain, H. M., Abdul, W., Almogren, A., ... & Azim
Niaz, I. (2017). An optimized home energy management system with integrated renewable
energy and storage resources. Energies, 10(4), 549.
[23] Morales, D. X., Besanger, Y., Sami, S., & Bel, C. A. (2017). Assessment of the impact of
intelligent DSM methods in the Galapagos Islands toward a Smart Grid. Electric Power
Systems Research, 146, 308-320.
[24] Muratori, M., & Rizzoni, G. (2016). Residential demand response: Dynamic energy man-
agement and time-varying electricity pricing. IEEE Transactions on Power systems, 31(2),
1108-1117.
[25] Ogunjuyigbe, A. S. O., Ayodele, T. R., & Akinola, O. A. (2017). User satisfaction-induced
demand side load management in residential buildings with user budget constraint. Ap-
plied Energy, 187, 352-366.
[26] Rastegar, M., Fotuhi-Firuzabad, M., & Zareipour, H. (2016). Home energy management
incorporating operational priority of appliances. International Journal of Electrical Power
& Energy Systems, 74, 286-292.
[27] Azar, A. G., & Jacobsen, R. H. (2016). Appliance Scheduling Optimization for Demand
Response. International Journal on Advances in Intelligent Systems, 9(1 & 2), 50-64.
[28] Bakar, N. N. A., Hassan, M. Y., Sulaima, M. F., Naim Mohd Nasir, M., & Khamis, A.
(2017). Microgrid and load shedding scheme during islanded mode: A review. Renewable
and Sustainable Energy Reviews, 71, 161-169.
[29] Kaur, A., Kaushal, J., & Basak, P. (2016). A review on microgrid central controller. Re-
newable and Sustainable Energy Reviews, 55, 338-345.
[30] Nazmudeen, M. S. H., Wan, A. T., & Buhari, S. M. (2016, September). Improved through-
put for power line communication (plc) for smart meters using fog computing based data
aggregation approach. In Smart Cities Conference (ISC2), 2016 IEEE International (pp.
1-4). IEEE.
[31] Yan, Y., & Su, W. (2016, May). A fog computing solution for advanced metering in-
frastructure. In Transmission and Distribution Conference and Exposition (T&D), 2016
IEEE/PES (pp. 1-4). IEEE.
[32] Okay, F. Y., & Ozdemir, S. (2016, May). A fog computing based smart grid model. In
Networks, Computers and Communications (ISNCC), 2016 International Symposium on
(pp. 1-6). IEEE.
[33] Al Faruque, M. A., & Vatanparvar, K. (2016). Energy management-as-a-service over fog
computing platform. IEEE internet of things journal, 3(2), 161-169.
[34] Masip-Bruin, X., Marín-Tordera, E., Tashakor, G., Jukan, A., & Ren, G. J. (2016). Foggy
clouds and cloudy fogs: a real need for coordinated management of fog-to-cloud comput-
ing systems. IEEE Wireless Communications, 23(5), 120-128.
[35] Ejaz, W., Naeem, M., Shahid, A., Anpalagan, A., & Jo, M. (2017). Efficient energy man-
agement for the internet of things in smart cities. IEEE Communications Magazine, 55(1),
84-91.
[36] Mohamed, N., Al-Jaroodi, J., Jawhar, I., Lazarova-Molnar, S., & Mahmoud, S. (2017).
SmartCityWare: A Service-Oriented Middleware for Cloud and Fog Enabled Smart City
Services. Ieee Access, 5, 17576-17588.
[37] Jalali, F., Vishwanath, A., de Hoog, J., & Suits, F. (2016, November). Interconnecting Fog
computing and microgrids for greening IoT. In Innovative Smart Grid Technologies-Asia
(ISGT-Asia), 2016 IEEE (pp. 693-698). IEEE.
[38] Kumaraguruparan, N., Sivaramakrishnan, H., & Sapatnekar, S. S. (2012, January). Res-
idential task scheduling under dynamic pricing using the multiple knapsack method. In
Innovative Smart Grid Technologies (ISGT), 2012 IEEE PES (pp. 1-6). IEEE.
[39] Al Hasib, A., Nikitin, N., & Natvig, L. (2015, April). Cost-comfort balancing in a smart
residential building with bidirectional energy trading. In Sustainable Internet and ICT for
Sustainability (SustainIT), 2015 (pp. 1-6). IEEE.
[40] Mahmood, D., Javaid, N., Alrajeh, N., Khan, Z. A., Qasim, U., Ahmed, I., & Ilahi, M.
(2016). Realistic scheduling mechanism for smart homes. Energies, 9(3), 202.
[41] Mahmood, D., Javaid, N., Ahmed, S., Ahmed, I., Niaz, I. A., Abdul, W., & Ghouzali, S.
(2017). Orchestrating an effective formulation to investigate the impact of EMSs (energy
management systems) for residential units prior to installation. Energies, 10(3), 335.
[42] Ogunjuyigbe, A. S. O., Ayodele, T. R., & Akinola, O. A. (2017). User satisfaction-induced
demand side load management in residential buildings with user budget constraint. Ap-
plied Energy, 187, 352-366.
[43] Chekired, D. A., Khoukhi, L., & Mouftah, H. T. (2017). Decentralized Cloud-SDN Ar-
chitecture in Smart Grid: A Dynamic Pricing Model. IEEE Transactions on Industrial
Informatics.
[44] Chekired, D. A., & Khoukhi, L. (2017). Smart Grid Solution for Charging and Discharging
Services Based on Cloud Computing Scheduling. IEEE Transactions on Industrial Infor-
matics, 13(6), 3312-3321.
[45] Khameis, A., Rashed, S., Abou-Elnour, A., & Tarique, M. (2015). Zigbee based optimal
scheduling system for home appliances in the United Arab Emirates. Network Protocols
and Algorithms, 7(2), 60-80.
[46] Beaudin, M., & Zareipour, H. (2017). Home Energy Management Systems: A Review
of Modelling and Complexity. In Energy Solutions to Combat Global Warming (pp. 753-
793). Springer, Cham.
[47] Luke, S. (2009). Essentials of metaheuristics (Vol. 113). Raleigh: Lulu.
[48] Cárdenas, J. J., Garcia, A., Romeral, J. L., & Andrade, F. (2009, November). A genetic al-
gorithm approach to optimization of power peaks in an automated warehouse. In Industrial
Electronics, 2009. IECON’09. 35th Annual Conference of IEEE (pp. 3297-3302). IEEE.
[49] GüNtzer, M. M., & Jungnickel, D. (2000). Approximate minimization algorithms for the
0/1 knapsack and subset-sum problem. Operations Research Letters, 26(2), 55-66.
[50] Chih, M. (2015). Self-adaptive check and repair operator-based particle swarm optimiza-
tion for the multidimensional knapsack problem. Applied Soft Computing, 26, 378-389.
[51] Yi, P., Dong, X., Iwayemi, A., Zhou, C., & Li, S. (2013). Real-time opportunistic schedul-
ing for residential demand response. IEEE Transactions on smart grid, 4(1), 227-234.
[52] Qayyum, F. A., Naeem, M., Khwaja, A. S., Anpalagan, A., Guan, L., & Venkatesh, B.
(2015). Appliance scheduling optimization in smart home networks. IEEE Access, 3,
2176-2190.
[53] Kennedy, J. (2011). Particle swarm optimization. In Encyclopedia of machine learning (pp.
760-766). Springer US.
[54] Del Valle, Y., Venayagamoorthy, G. K., Mohagheghi, S., Hernandez, J. C., & Harley, R.
G. (2008). Particle swarm optimization: basic concepts, variants and applications in power
systems. IEEE Transactions on evolutionary computation, 12(2), 171-195.
[55] Joy, J., Rajeev, S., & Narayanan, V. (2016). Particle swarm optimization for resource
Constrained-project scheduling problem with varying resource levels. Procedia Technol-
ogy, 25, 948-954.
[56] Haider, H. T., See, O. H., & Elmenreich, W. (2016). A review of residential demand re-
sponse of smart grid. Renewable and Sustainable Energy Reviews, 59, 166-178.
[57] Gao, J., Xiao, Y., Liu, J., Liang, W., & Chen, C. P. (2012). A survey of communica-
tion/networking in smart grids. Future Generation Computer Systems, 28(2), 391-404.
[58] Iwayemi, A., Yi, P., Dong, X., & Zhou, C. (2011). Knowing when to act: An optimal
stopping method for smart grid demand response. Ieee Network, 25(5).
[59] Evans, D. (2011). The internet of things: How the next evolution of the internet is changing
everything. CISCO white paper, 1(2011), 1-11.
[60] Ho, Q. D., Gao, Y., & Le-Ngoc, T. (2013). Challenges and research opportunities in wire-
less communication networks for smart grid. IEEE Wireless Communications, 20(3), 89-
95.
[61] Bonomi, F., Milito, R., Zhu, J., & Addepalli, S. (2012, August). Fog computing and its
role in the internet of things. In Proceedings of the first edition of the MCC workshop on
Mobile cloud computing (pp. 13-16). ACM.
Tentative Time Table
Sr No. Activity Date
1 Background study and detailed literature re-
view
Completed
2 Formulation of problem and proposing solu-
tion
Completed
3 Analysis and dissemination of results November-December 2017
4 Comparative analysis of existing techniques January-March 2018
5 Thesis Writing April-May 2018
PART II
Recommendation by the Research Supervisor
Name Dr. Nadeem Javaid Signature___________________Date____________
Recommendation by the Research Co-Supervisor
Name Dr. Mariam Akbar ______Signature___________________ Date___________
Signed by Supervisory Committee
Name of Committee member
Designation
Signature & Date
Dr. Nadeem Javaid
Assistant Professor
Dr. Mariam Akbar
Assistant Professor
Dr. Sohail Asghar
Professor
Approved by Departmental Advisory Committee
Certified that the synopsis has been seen by members of DAC and considered it suitable
for putting up to BASAR.
Secretary
Departmental Advisory Committee
Name:__________________________________ Signature:_____________________
Date: _________________
Chairman/HoD ____________________________
Signature: _____________________________
Date: _____________________________
PART III
Dean, Faculty of Information Sciences & Technology
_________________Approved for placement before BASAR.
_________________Not Approved on the basis of following reasons
Signature ____________________________ Date__________________
Secretary BASAR
_________________Approved from BASAR.
_________________Not Approved on the basis of the following reasons
Signature ____________________________ Date__________________
Dean, Faculty of Information Sciences & Technology
________________________________________________________________________
________________________________________________________________________
Signature:_______________________________ Date__________________
Please provide the list of courses studied
1. Advanced Topics in Social Networks
2. Special Topics in Simulation and Modeling
3. Special Topics in Artificial Intelligence
4. Advanced Topics in Wireless Networks
5. Advanced Topics in Simulation and Modeling
6. Special Topics in Performance Evaluation of Networks
ResearchGate has not been able to resolve any citations for this publication.
Technical Report
Full-text available
Section 1305 of the Energy Independence and Security Act (EISA) of 2007 (Pub. L. 110–140) directs NIST ‘‘to coordinate the development of a framework that includes protocols and model standards for information management to achieve interoperability of smart grid devices and systems.’’ To meet these statutory goals, in January 2010, NIST published the NIST Framework and Roadmap for Smart Grid Interoperability Standards, Release 1.0 (Release 1.0), and in February 2012, NIST published the NIST Framework and Roadmap for Smart Grid Interoperability Standards, Release 2.0 (Release 2.0), which updated the material discussed in Release 1.0. This document is Release 3.0 and builds upon the work in previous releases with an update on the progress since Release 2.0; a description of the Smart Grid Interoperability Panel (SGIP); updated architecture, cybersecurity, and testing and certification chapters; and a new chapter on cross-cutting issues and future directions. Since the release of the last edition of the NIST Framework and Roadmap for Smart Grid Interoperability Standards (Release 2.0), advances in smart grid infrastructure have been implemented. Examples include the widespread deployment of wireless-communication power meters, the availability of customer energy usage data through the Green Button initiative, remote sensing for determining real-time transmission and distribution status, and protocols for electric vehicle charging, supported by standards development across the entire smart grid arena. This release updates NIST’s ongoing efforts to facilitate and coordinate smart grid interoperability standards development and smart grid-related measurement science and technology, including the evolving and continuing NIST relationship with the SGIP.
Article
Full-text available
Smart grid is an emerging technology which is considered to be an ultimate solution to meet the increasing power demand challenges. Modern communication technologies have enabled the successful implementation of smart grid (SG), which aims at provision of demand side management mechanisms (DSM), such as demand response (DR). In this paper, we propose a hybrid technique named as teacher learning genetic optimization (TLGO) by combining genetic algorithm (GA) with teacher learning based optimization (TLBO) algorithm for residential load scheduling, assuming that electric prices are announced on a day-ahead basis. User discomfort is one of the key aspects which must be addressed along with cost minimization. The major focus of this work is to minimize consumer electricity bill at minimum user discomfort. Load scheduling is formulated as an optimization problem and an optimal schedule is achieved by solving the minimization problem. We also investigated the effect of power-flexible appliances on consumers’ bill. Furthermore, a relationship among power consumption, cost and user discomfort is also demonstrated by feasible region. Simulation results validate that our proposed technique performs better in terms of cost reduction and user discomfort minimization, and is able to obtain the desired trade-off between consumer electricity bill and user discomfort.
Article
Full-text available
Smart cities are becoming a reality. Various aspects of modern cities are being automated and integrated with information and communication technologies (ICT) to achieve higher functionality, optimized resources utilization and management and improved quality of life for the residents. Smart cities rely heavily on utilizing various software, hardware and communication technologies to improve the operations in areas ike healthcare, transportation, energy, education, logistics and many others, while reducing costs and resources consumption. One of the promising technologies to support such efforts is the Cloud of Things (CoT). CoT provides a platform for linking the cyber parts of a smart city that are executed on the cloud with the physical parts of the smart city including residents, vehicles, power grids, buildings, water networks, hospitals and other resources. Another useful technology is Fog Computing, which extends the traditional Cloud Computing paradigm to the edge of the network to enable localized and real time support for operating enhanced smart city services. However, proper integration and efficient utilization of CoT and Fog Computing is not an easy task. The paper discusses how the service-oriented middleware (SOM) approach can help resolve some of the challenges of developing and operating smart city services using CoT and Fog Computing. We propose a SOM called SmartCityWare for effective integration and utilization of CoT and Fog Computing. SmartCityWare abstracts services and components involved in smart city applications as services accessible through the service-oriented model. This enhances integration and allows for flexible inclusion and utilization of the various services needed in a smart city application. In addition, we discuss the implementation and experimental issues of SmartCityWare and demonstrate its use through examples of smart city applications.
Article
Full-text available
Traditional power grid and its demand-side management (DSM) techniques are centralized and mainly focus on industrial consumers. The ignorance of residential and commercial sectors in DSM activities degrades the overall performance of a conventional grid. Therefore, the concept of DSM and demand response (DR) via residential sector makes the smart grid (SG) superior over the traditional grid. In this context, this paper proposes an optimized home energy management system (OHEMS) that not only facilitates the integration of renewable energy source (RES) and energy storage system (ESS) but also incorporates the residential sector into DSM activities. The proposed OHEMS minimizes the electricity bill by scheduling the household appliances and ESS in response to the dynamic pricing of electricity market. First, the constrained optimization problem is mathematically formulated by using multiple knapsack problems, and then solved by using the heuristic algorithms; genetic algorithm (GA), binary particle swarm optimization (BPSO), wind driven optimization (WDO), bacterial foraging optimization (BFO) and hybrid GA-PSO (HGPO) algorithms. The performance of the proposed scheme and heuristic algorithms is evaluated via MATLAB simulations. Results illustrate that the integration of RES and ESS reduces the electricity bill and peak-to-average ratio (PAR) by 19.94% and 21.55% respectively. Moreover, the HGPO algorithm based home energy management system outperforms the other heuristic algorithms, and further reduces the bill by 25.12% and PAR by 24.88%.
Article
Full-text available
Demand Response (DR) programs under the umbrella of Demand Side Management (DSM) tend to involve end users in optimizing their Power Consumption (PC) patterns and offer financial incentives to shift the load at “low-priced” hours. However, users have their own preferences of anticipating the amount of consumed electricity. While installing an Energy Management System (EMS), the user must be assured that this investment gives optimum comfort of bill savings, as well as appliance utility considering Time of Use (ToU). Moreover, there is a difference between desired load distribution and optimally-scheduled load across a 24-h time frame for lowering electricity bills. This difference in load usage timings, if it is beyond the tolerance level of a user, increases frustration. The comfort level is a highly variable phenomenon. An EMS giving optimum comfort to one user may not be able to provide the same level of satisfaction to another who has different preferences regarding electricity bill savings or appliance utility. Under such a diversity of human behaviors, it is difficult to select an EMS for an individual user. In this work, a numeric performance metric,“User Comfort Level (UCL)”isformulatedonthebasisofuserpreferencesoncostsaving,toleranceindelayregardinguse of an appliance and return of investment. The proposed framework (UCL) allows the user to select an EMS optimally that suits his.her preferences well by anticipating electricity bill reduction, tolerable delay in ToU of the appliance and return on investment. Furthermore, an extended literature analysis is conducted demonstrating generic strategies of EMSs. Five major building blocks are discussed and a comparative analysis is presented on the basis of the proposed performance metric.
Article
Smart grids (SG) energy management system and Electric Vehicle (EV) has gained considerable reputation in recent years. This has been enabled by the high growth of EVs on roads; however, this may lead to a significant impact on the power grids. In order to keep EVs far from causing peaks in power demand and to manage building energy during the day, it is important to perform an intelligent scheduling for EVs charging and discharging service and buildings areas by including different metrics, such as real time price and demandsupply curve. In this paper, we propose a real-time dynamic pricing model for EVs charging and discharging service and building energy management, in order to reduce the peak loads. Our proposed approach uses a decentralized cloud computing architecture based on Software Define Networking (SDN) technology and Network Function Virtualization (NFV). We aim to schedule user’s requests in a real-time way and to supervise communications between micro-grids controllers, smart grid and user entities (i.e., EVs, Electric Vehicles Public Supply Stations (EVPSS), Advance Metering Infrastructure (AMI), smart meters, etc.). We formulate the problem as a linear optimization problem for EV and a global optimization problem for all micro grids. We solve the problems using different decentralized decision algorithms. To the best of our knowledge, this is the first paper that proposes a pricing model based on decentralized cloud-SDN architecture in order to solve all the aforementioned issues. The extensive simulations and comparisons with related works proved that our proposed pricing model optimizes the energy load during peak hours, maximises EVs utility and maintains the micro grid stability. The simulation is based on real electric load of the city of Toronto.
Article
Recently, Home Energy Management (HEM) controllers have been widely used for residential load management in a smart grid. Generally, residential load management aims {to reduce the electricity bills and also curtail the Peak-to-Average Ratio (PAR)}. In this paper, we design a HEM controller on the {basis} of four heuristic algorithms: Bacterial Foraging Optimization Algorithm (BFOA), Genetic Algorithm (GA), Binary Particle Swarm Optimization (BPSO), and Wind Driven Optimization (WDO). Moreover, we proposed {a} hybrid algorithm which is Genetic BPSO (GBPSO). All the selected algorithms are tested with the consideration of essential home appliances in Real Time Pricing (RTP) environment. Simulation results show that each algorithm in the HEM controller reduces the electricity cost and curtails the PAR. GA based HEM controller performs relatively better in term of PAR reduction; it curtails approximately $34\%$ PAR. Similarly, BPSO based HEM controller performs relatively better in term of cost reduction {, as} it reduces approximately $36\%$ cost. Moreover, GBPSO based HEM controller performs better than the other algorithms based HEM controllers in terms of both cost reduction and PAR curtailment.
Article
Smart Grid (SG) technology represents an unprecedented opportunity to transfer the energy industry into a new era of reliability, availability, and efficiency that will contribute to our economic and environmental health. On the other hand, the emergence of Electric Vehicles (EVs) promises to yield multiple benefits to both power and transportation industry sectors, but it is also likely to affect the SG reliability, by consuming massive energy. Nevertheless, the plug-in of EVs at public supply stations must be controlled and scheduled in order to reduce the peak load. This paper considers the problem of plug-in EVs at public supply stations (EVPSS). A new communication architecture for smart grid and cloud services is introduced. Scheduling algorithms are proposed in order to attribute priority levels and optimize the waiting time to plug-in at each EVPSS. To the best of our knowledge, this is one of the first papers investigating the aforementioned issues using new network architecture for smart grid based on cloud computing. We evaluate our approach via extensive simulations and compare it with two other recently proposed works, based on real supply energy scenario in Toronto. Simulation results demonstrate the effectiveness of the proposed approach when considering real EVs charging-discharging loads at peak-hours periods.
Article
The Galapagos Islands are a fragile ecosystem, which belongs to Ecuador country. The supply of energy is through of combined technologies such as thermal, photovoltaic and wind. The government of Ecuador is fostering the policy zero CO2 emissions within the islands. In order to achieve this objective, some initiatives are carried out like replacing the conventional vehicles and the gas stoves by electrical ones. Taking into account this framework, it becomes vital to assess the impact of these new policies on the distribution network; for sure, this will change significantly the operation and control of Galapagos distribution grid. In addition, once determined the impact it is worth proposing smart strategies in order to deal with the load growth due to new services. Thus, this paper proposes an adapted and intelligent method for performing Demand Side Management in the Galapagos low voltage network, the proposal executes Demand Response in the presence of time of use pricing. An ingenious approach, which consists in modifying load curves by means of polynomial functions is presented; several scenarios with different strategies are analyzed, the main results show as the proposal strategy respect all the limits stablished by the regulator and as the rebound effect is annulated. Real field information has been used in the presented models.