2014 IEEE 28th International Conference on Advanced Information Networking and Applications

Peak Load Scheduling in Smart Grid

Communication Environment
M. A. Khan1, N. Javaid1, M. Arif1 , S. Saud1, U. Qasim2, Z. A. Khan3
1 COMSATS Institute of Information Technology, Islamabad, Pakistan
2 University of Alberta, Alberta, Canada
3 Internetworking Program, FE, Dalhousie University, Halifax, Canada

AbstractIn this paper, we consider an Energy Con-
sumption Scheduler (ECS) unit inside a smart meter. The
function of ECS is to flatten the peaks in the load curve
of residential type users. The ECS dually communicates
with the user as well as power utility in smart grid real
time communication environment. Each user has a specific
power capacity limit for their appliances operation . The
ECS keeps check and balance condition whenever a user
exceeds its power capacity limit. Also the most urgent ap-
pliance operations are given priority based on their internal
preemption state. By adopting ECS unit, the total peaks in
the load curve are reduced up to 33.3%. Also the users
payment to the power utility becomes less due to efficiently
managing their appliance operations. Additionally, the cost
of per unit energy generation is also reduced by avoiding the
peak power plants. Matlab/Simulink is used as a simulation
tool for the realization of this model.
Index TermsSmart grid, demand side management,
load scheduling.
I. I NTRODUCTION
HE traditional power grid covers the areas of power
T generation, transmission, distribution and control.
The generated power is transferred from a small number
of generation sites to a large number of users in a one
way fashion. In contrast, the smart grid (also called an
intelligent grid or future grid) ensures two way flow of
power and communication signals between power utility
and end users. A smart grid is an integration of the
modern technologies and subsystems. From a technical
point of view, a smart grid can be visualized as a three
major subsystems: smart infrastructure subsystem, smart
management subsystem, and smart protection subsystem.
The smart infrastructure subsystem consists of energy
related information and communication systems. The
smart management subsystem includes management and
control services. Whereas, smart protection subsystem
consists of failure protection, security, reliability and
privacy [1]. In USA special attributes used for smart
grid. For example, it is self-healing for power disturbance
events, provides power quality as needed in 21St cen-
tury, accommodates the option of Distributed Generation
(DG) and storage, active participation by consumers in
Demand Response (DR) and protects itself both from
physical and cyber-attacks. Additionally, optimization of
asset utilization, and more space for new technologies
and economic improvements. According to European
Commission (EC) report the smart grid must be reliable,
economical, accessible and flexible. Nonprofit Joint US-
China Cooperation on clean energy defines the smart grid
as, a power transmission and distribution system that in-
tegrates the elements of conventional and modern power
engineering, communication, information and monitoring
technology. It also uses sophisticated sensing for better
performance and advanced services to customers [2].
The gradual replacement of traditional meters by IC
card prepayment meters is also an important aspects of
smart grid. Also, standards are developed for security
techniques (for example, Q/GDW365 in China). The
extensive analysis of Q/GDW365 is presented in [24].
Energy efficiency is one of the most critical issues
for existing as well future smart grid. With the advent
of new types of loads such as Plug-in Hybrid Electric
Vehicles (PHEVs), and the involvement of DG has
badly affected the energy efficiency. Different types of
techniques are used to improve energy efficiency, i.e.,
economic dispatch, unit commitment, reducing power
transmission & distribution losses, reactive power plant
management, load forecasting and proficient attempts
to distribute the generated power [18]. In smart grid,
Demand side management (DSM) is one of the most
attractive areas having a significant impact on energy
efficiency [19]. DSM is the planning, implementation and
monitoring of those utility activities designed to influence
the customer use of electricity in ways that will produce
desired changes in the utilities load shape, i.e., changes
in the time pattern and magnitude of a utilitys load
[3], [4]. The need for DSM was first realized in 1970s
[5]. The first attempt towards DSM is the Direct Load
Control (DLC). However, in smart grid, smart pricing is
an alternate to DLC [6], [7].
The main focus of this paper is, to flatten the peaks
generated in the load curve of a residential type users.
The real time communication between power utility and
users occur through a smart meter having ECS unit
installed inside them. The ECS has the capability to

1550-445X/14 $31.00 2014 IEEE 1025

DOI 10.1109/AINA.2014.135
manage the peaks. The operation of appliances can be
accepted or rejected, based upon appliance operation
urgency and also the power capacity limit.
The remaining paper is structured as follows. Related
work and motivation are presented in Section II. In
Section III, evaluation of the system model consisting
of, load categorization and smart appliance modeling are
carried out. Simulation studies are presented in Section
IV. In Section V, the paper is concluded.
II. R ELATED W ORK A ND M OTIVATION
In smart grid, DSM has promising effects on energy ef-
ficiency (i.e., providing same services using less energy).
Different energy consumption scheduling techniques are
introduced. A three layered autonomous DSM model is
presented in [7]. This model consists of Admission Con-
troller (AC), Load Balancer (LB), and Demand Response
(DR) with Load Forecaster (LF) modules. Loads are cat-
egorized based upon their operational characteristics and
user requirements. Whenever a user starts an appliance
operation, the request is generated by appliance interface
for AC to process it further. The AC checks the available
capacity. If the capacity is available and also off-peak
hours are running, then AC accepts an appliance request
and initiates an appliance operation. In case, when an
appliance operation exceeds an available capacity, then
the request is rejected and forwarded to LB. The LB then
solves an optimization problem and assigns a future time
slot for that specific appliance to start later. DR and LF
is the upper layer and communicates with the smart grid
interface for real time information. However, DR and LF
modules are not considered. Also, when the number of
appliances request exceeds from a certain level, then the
proposed three layered autonomous DSM model cannot
maintain the available capacity limit. The results obtained
are satisfying. The flaws identified are the motivation of
our work.
A game-theoretic approach is used by Mohsenian-
Rad A., et al. [8], for residential energy consumption
scheduling. The pricing mechanism introduced is based
on convex and increasing cost function. The Authors
presented a distributed algorithm for optimization prob-
lem. This work is considered as a reference for DSM
techniques. An alternative mechanism known as Vickrey-
Clarke-Grove (VCG) mechanism is proposed in [9].
The aim of this mechanism is to achieve efficiency,
nonnegative transfer (i.e., payments from power utility
to user) and truthfulness among users. The proposed
VCG mechanism encourage users to shift their load from
peak hours to off-peak hours. Besides from obtaining
social welfare, the utility also gets benefit by getting a
reduced average load shape curve. To achieve effective
DSM model, pricing scheme also plays an important
role. Weighted average filter based price prediction and
1026
combining inclining block rates with real time pricing are
discussed in [10]. Price predictor and energy scheduler
are considered within a smart meter. Overall results of
this work is satisfying, however the problem of fairness is
not explored. Whereas in [11], a heuristic optimization
based DSM is adopted. This model is based on Evo-
lutionary Algorithm (EA). Also a new pricing scheme
known as day-ahead load shifting pricing scheme is
used. The model is simulated for all types of end user
including: residential, commercial, and industrial type.
In this model, the authors show that the waiting time
of appliances is inversely proportional to the delay of
scheduling algorithm. Simulation results are efficient in
terms of PAR reduction and total energy cost minimiza-
tion. However, the fairness issue among users is not
addressed in their proposed work.
The above discussed DSM techniques have considered
users with a fixed load curve. However, load uncertainty
is a big challenge especially in modeling a pricing
mechanism. This important issue is considered in [12].
The adopted DSM model forecasts the load curve of the
user from the previous knowledge of their energy usage.
Real Time Pricing (RTP) and Inclining Block Rates
(IBR) are combined for effective pricing mechanism.
The mechanism proposed in this model is multi-stage,
i.e., information related with appliances revealed after
time. The model presented is performing optimization
based scheduling. Simulation results show efficiency by
reducing total PAR, total energy cost and also fairness.
However, the issue of reducing waiting time of appli-
ances is not addressed. Two most effective and related
ideas, i.e., fairness and optimality are addressed in [13].
The authors presents an alternative pricing scheme by
combing RTP with hour by hour billing mechanism.
Analytical case study is evaluated, which shows that the
proposed scheme is 73% more efficient in fairness and is
inversely related to optimality. Backtracking algorithm,
based on artificial intelligence technique for ECS in
DSM is proposed [14]. In this technique, RTP billing
mechanism is adopted for residential type users. The is-
sues, PAR reduction and execution time of algorithm are
achieved effectively. However, all other issues regarding
fairness, coverage area and waiting time of appliances are
not considered. An alternatively, incentives are provided
for users to encourage them to efficiently schedule their
load [15]. Comparison of different type of DSM tech-
niques is performed in [20]-[24]. This is practically true
if the utility companies provide information about price
some hours ahead of usage time. For the communication
of smart appliances with ECS, special consideration has
been taken in [17]. Zigbee based communication is
proposed for remote power controlling at home and at
transmission side. In this paper, the authors discussed
the effective role of Zigbee in DLC for residential users.
 III. S YSTEM M ODEL fixed deadline. The MRL is the users prime appliances
In this section, the detailed description of load catego- acting as a constant load, which directly communicates
with smart meter, i.e., bypassing the ECS.
rization on residential level as well as smart appliances
modelling are presented. Moreover, we assume that each
residential user is equipped with a smart meter consisting B. Modelling of Smart Appliances
of ECS unit. In this work, each appliance is considered as a Finite
State Machine (FSM) having six different states (as
A. Load Categorization shown in fig. 1): off state (1), initiate state (2), run state
(3), interrupt state (4), complete state (5), and sleep state
In the power sector, users are divided into three main (6).
categories: residential, industrial and commercial type. In More specifically, the status of a smart appliance varies
this paper, we focus only on residential type users (i.e., among:
homes, offices, etc.) due to their high energy consump-
OFF state (1): Appliance is not yet started, and
tion profile. Residential users may have different types
hence consumes no power.
of loads (e.g., Air Conditioner (AC), refrigerator/fridge,
INITIATE state (2): SWITCH ON command re-
water heater and washing machine). For simplification,
ceived, appliance is ready to run. Power consump-
we further divide residential type load into three subcat-
tion is minimum.
egories based on their internal characteristics. This sub
RUN state (3): The START command received
categorization is flexible to prioritize the users choice.
from user or from thermostat, and so rated power
The users are capable to make changes in their ECS unit
consumption.
through a user interface, according to their own choice.
INTERRUPT state (4): User wants to interrupt
This sub categorization is given below.
appliance during operation, so HALT command
1. Must-Run Load (MRL) (e.g., lighting, television,
asserted and the appliance absorbs minimum power
laptop charger, etc.): This type of load consists of
(e.g., thermostat).
users prime appliances. The operation of such appliances
COMPLETE state (5): Appliance operation com-
cannot be delayed. Moreover, these appliances are not
pleted, jump to initiate (2), sleep (6) or off (1) state.
considered by the ECS for efficient energy consumption
SLEEP state (6): An auxiliary status for power
scheduling.
saving option.
2. Shiftable Load (SHL) (e.g., dishwasher, washing
machine, Plug-in Hybrid Electric Vehicles (PHEVs),
etc.): SHL includes all those appliances which have a IV. S IMULATION S TUDIES
specific time period for their operation (i.e., deadline). In this section, we discuss the setup of simulation as
Their operation can be delayed based on the remaining well as simulation results.
time for their operation. However, if the remaining time
is equal to or less than their deadline, then the operation A. Simulation Setup
of SHLs cannot be interrupted any more.
For simulation purpose, all seven types of appli-
3. Steady load (SL) (e.g. refrigerator, water heater,
ances are modeled using Matlab/Simulink with Stateflow
house heating, etc.): In this sub category, the loads are
toolbox. The power consumption for each appliance is
running steadily, according to their internal temperature
assumed to be 20 watts. The lower and upper limits
control system. Upper and lower temperature limits are
of temperature for the water heater is 20 C and 60 C
defined by thermostat for their operation. Appliance
respectively. Similarly, for house heating and refrigerator
running at these limits is said to be in a comfort zone.
the temperature ranges are 22 C - 24 C and 2 C - 5 C
As shown in fig. 1, in comfort zone the appliance goes
respectively. However, the accuracy, power consumption,
to complete state (5) and then immediately to initiate
and constraints of a practical model of these appliances
state (2) and so consumes no power. Additionally, from
may be different.
the complete state (5), the user also changes the state of
The time span for simulation is set to 30 minutes. The
appliance to off (1) or sleep state (6).
water heater, house heating, and refrigerator are modeled
Seven different types of electric appliances are con-
using the combination of one dimensional Fouriers law
sidered for the detailed simulation study of this model.
in its integral form for homogeneous material (1) and
These appliances include water heater, house heating,
1-D specific heat equation (2) given in [16].
refrigerator, electric iron, microwave oven, washing ma-
chine, and baseline load. The operation of the first four 
appliances (i.e., SLs) is controlled by thermostat, while Qexch
= K T.dA (1)
microwave oven and washing machine are SHLs having t s

1027
Fig. 1. FSM for smart appliances modeling.

TABLE I
L IST OF T YPICAL H OUSEHOLD A PPLIANCES AND T HEIR C HARACTERISTICS

Appliance Power (W) Temp. Range ( C) Parameters

Water Heater 1380-2300 20-60 Status, Preemption, Ab-
sorption, Temperature
Refrigerator 3100-6400 2-5 Status, Preemption, Ab-
sorption, Temperature
House Heating 1000-5000 22-24 Status, Preemption, Ab-
sorption, Temperature
Microwave Oven 600-1500 25-10 Status, Preemption, Ab-
sorption, Temperature
Washing Machine 300-500 Energy and water con-
sumption, Spin-drying ef-
ficiency class
Clothes Iron 1000-1800 90-230 Status, Preemption, Ab-
sorption, Temperature
Baseline Load Variable Lighting, Entertainment,
Office work

Fig. 2. Implementation of refrigerator FSM in Stateflow toolbox.

1028
Fig. 3. Characteristic curves of house heating.

Fig. 4. Characteristic curves of refrigerator.

Fig. 5. Combination of overall loads in Simulink.

1029
Fig. 6. Total load curve along with individual load curves without power capacity limit.

Fig. 7. Total load curve along with individual load curves with power capacity limit of 80 watts.

m, c are the mass and specific heat capacity of air

respectively.
Qtotal = mcT (2)
where,
The electric iron is modeled upon the principles of
T = Texternal Troom (3) thermostat. Upper and lower threshold limits of temper-
ature are defined for their operation. The burst loads,
microwave oven and washing machine have a fixed
Qtotal = Qexch + h Ph (4)
deadline. They can be interrupted during their operational
Texternal is the external temperature, time if they have enough time remaining to complete
Ph is the heater power, their operation and hence a preemptive state of zero.
h is the heating system thermal efficiency, However, when their preemption state becomes one (i.e.,
K is room global thermal conductivity, active), then burst loads cannot be delayed anymore due
Q is the heat transferred and to their internal characteristic.

1030
B. Simulation Results
Fig. 2, shows the refrigerator FSM implementation
in Stateflow toolbox. Only three status of refrigerator
is considered for the simulation: initiate state (2), run
state (3) and complete state (5). The characteristic curves
of refrigerator and house heating are presented among
SLs in fig. 3. The x-axis shows time in minutes and
the y-axis represents status, preemption, absorption and
temperature values respectively. In case of refrigerator,
the status shifts between initiate (2) and run state (3).
The preemption value shows that a task of an appliance
can be interrupted or not. The power consumption of
appliances is represented by an absorption curve. The
maximum power consumption that an appliance is used
for its operation is assumed 20 watts. The lowermost
curve shows the temperature of refrigerator in degree
centigrade. The comfort zone for refrigerator is 2 C -
5 C. In case of house heating shown in fig. 4, the status
is varying between initiate (2) and run state (3). However,
at 21 minute of simulation, the status jumps to complete
state (5) because the comfort zone (i.e., 22 C - 24 C)
has been reached.
The combination of overall loads in Simulink is given
in fig. 5. Fig. 6 shows the overall load curve and the indi-
vidual curve of each appliance. The highest peak load is
observed during the fifth minute of simulation, reaching
up to 120 watts. In this simulation, no scheduler is used
to manage the load. Requests coming from all types of
appliances are accepted and turned on without checking
any condition. As a result, the peak is generated in the
overall load curve, which is undesirable. The simulation
results shown in fig. 7, demonstrate the behavior of the
scheduler by maintaining the peaks in the load curve
within certain limits. The same set of appliances are
considered again. In this simulation, the scheduler is set
to limit the total load curve within 80 watts power limit.
As shown in fig. 6, the limit of 80 watts is only exceeded
between 3 to 10 minutes of simulation. So, the scheduler
is working on these time limits only to manage the peaks
of the overall load curve. In this paper, the peak load is
reduced up to 33.3%. So, the users payment to the utility
is also reduced by the same amount.
V. C ONCLUSION
The work presented in this paper is based upon
online power scheduling in smart grid communication
environment. The ECS installed inside the smart meter,
is responsible for efficient power scheduling of resi-
dential type users. The users equipped with ECS, have
a power consumption limit depending upon their load
requirement. Whenever an appliance starts an operation,
it initially sends an admission request to ECS. The ECS
concern the power utility through smart grid interface for
the power capacity limit of that specific user. If there is
1031
enough power capacity limit, the appliance is allowed
to operate. Otherwise, the appliance operation request is
deferred. Also, the ECS checks the preemption value of
each appliance. If an appliance has a preemption value of
1, it is given priority. The less urgent appliance operation
having a preemption value of 0, is replaced by the more
urgent appliance operation having a preemption value of
1. By applying the proposed ECS mechanism, simulation
results show that the peak load is reduced by 33.3%.
R EFERENCES
[1] Xi Fang, Satyajayant Misra, Guoliang Xue, and Dejun Yang,
Smart GridThe New and Improved Power Grid: A Survey
IEEE Communications Surveys & Tutorials (ComST), vol. 14,
no. 4, pp. 944980, Fourth Quarter 2012.
[2] S. M. Amin and B. F. Wollenberg, Toward A Smart Grid:
Power Delivery for the 21st Century, IEEE Power and Energy
Magazine, vol. 3, no. 5, pp. 3441, Sept. Oct. 2005.
[3] E. Santacana, G. Rackliffe, Tnag Le, and Feng Xiaoming, Get-
ting Smart, IEEE Power and Energy Magazine, vol. 8, no. 2, pp.
4148, MarchApril 2010.
[4] Li Fangxing, QIAO Wei, Sun Hongbin, Wan Hui, Wang Jianhui,
Xia Yan, Xu Zhao, and Zhang Pei, Smart Transmission Grid:
Vision and Framework, IEEE Transactions on Smart Grid, vol.
1, no. 2, pp. 168177, Sept. 2010.
[5] C. W. Gellings, The Concept of Demand-Side Management for
Electric Utilities, Proceedings of the IEEE, vol. 73, no. 10, pp.
14681470, Oct. 1985.
[6] M. Alizadeh, Li Xiao, Wang Zhifang, A. Scaglione, and R.
Melton, Demand-Side Management in the Smart Grid: Informa-
tion Processing for the Power Switch, IEEE Signal Processing
Magazine, vol. 29, no. 5, pp. 5567, Sept. 2012.
[7] G. T. Costanzo, Zhu Guchuan, M. F. Anjos, and G. Savard,
A System Architecture for Autonomous Demand Side Load
Management in Smart Buildings, IEEE Transactions on Smart
Grid, vol. 3, no. 4, pp. 21572165, Dec. 2012.
[8] A. H. Mohsenian-Rad, V. W. S. Wong, J. Jatskevich, R. Schober,
and A. Leon-Garcia, Autonomous Demand-Side Management
Based on Game-Theoretic EnergyConsumption Scheduling for
the Future Smart Grid, IEEE Transactions on Smart Grid, vol.
1, no. 3, pp. 320331, Dec. 2010.
[9] P. Samadi, A. H. Mohsenian-Rad, R. Schober, and V. W. S. Wong,
Advanced Demand Side Management for the Future Smart Grid
Using Mechanism Design, IEEE Transactions on Smart Grid,
vol. 3, no. 3, pp. 11701180, Sept. 2012.
[10] A. H. Mohsenian-Rad and A. Leon-Garcia, Optimal Residential
Load Control With Price Prediction in Real-Time Electricity
Pricing Environments, IEEE Transactions on Smart Grid, vol.
1, no. 2, pp. 120133, Sept. 2010.
[11] T. Logenthiran, D. Srinivasan, and Shun Tan Zong, Demand Side
Management in Smart Grid Using Heuristic Optimization, IEEE
Transactions on Smart Grid, vol. 3, no. 3, pp. 12441252, Sept.
2012.
[12] P. Samadi, A. H. Mohsenian-Rad,V. W. S. Wong, and R. Schober,
Tackling the Load Uncertainty Challenges for Energy Consump-
tion Scheduling in Smart Grid, IEEE Transactions on Smart
Grid, vol. 4, no. 2, pp. 10071016, June 2013.
[13] Z. Baharlouei, M. Hashemi, H. Narimani, and A. H. Mohsenian-
Rad, Achieving Optimality and Fairness in Autonomous Demand
Response: Benchmarks and Billing Mechanisms, IEEE Transac-
tions on Smart Grid, Vol. 4, no. 2, pp. 968975, June 2013.
[14] Lee. Junghoon, Kim Hye-Jin, Park. Gyung-Leen, and Kang.
Mikyung, Energy Consumption Scheduler for Demand Response
Systems in the Smart Grid, Institute of Information Science (IIS),
Journal of Information Science and Engineering, vol. 28, no. 5,
pp. 955969, March 2012.
[15] A. H. Mohsenian-Rad, V. W. S. Wong, J. Jatskevich, and R.
Schober, Optimal and Autonomous Incentive-Based Energy
Consumption Scheduling Algorithm for Smart Grid, IEEE Power
and Energy Society (PES) Innovative Smart Grid Technologies
Conference (ISGT), pp. 16, 1921 Jan. 2010.
[16] J.P. Holman, Heat Transfer, 10th ed. McGraw-Hill New York,
2009.
[17] N. Javaid, A. Sharif, A. Mahmood, S. Ahmad, U. Qasim, and
Z. A. Khan, Monitoring and Controlling Power Using Zigbee
Communications, Seventh International Conference on Broad-
band, Wireless Computing, Communication and Applications
(BWCCA12), 2012.
[18] A. Mahmood, N. Javaid, A. Zafar, R. A. Riaz, S. Ahmed,
and S. Razzaq, Pakistans Overall Energy Potential Assessment,
Comparison of LNG, TAPI and IPI Gas Projects, accepted in
Elseviers Renewable & Sustainable Energy Reviews Journal, Vol.
31, pp. 182193, March 2014.
[19] M. Anas, N. Javaid, A. Mahmood, M. S. Raza, U. Qasim, and Z.
A. Khan, Minimizing Electricity Theft Using Smart Meters in
AMI, Seventh International Conference on P2P, Parallel, Grid,
Cloud and Internet Computing (3PGCIC) Nov. 2012.
[20] M. N. Ullah, N. Javaid, I. Khan, A. Mahmood, and M. U. Fa-
rooq, Residential Energy Consumption Controlling Techniques
to Enable Autonomous Demand Side Management in Future
Smart Grid Communications, 8th IEEE International Conference
on Broadband and Wireless Computing, Communication and
Applications (BWCCA13), 2013.
[21] I. Khan, N. Javaid, M. N. Ullah, A. Mahmood, and M. U. Farooq,,
A Survey of Home Energy Management Systems in Future
Smart Grid Communications, 8th IEEE International Conference
on Broadband and Wireless Computing, Communication and
Applications (BWCCA13), 2013.
[22] A. Mahmood, H. Fakhar, S. H. Ahmed, and N. Javaid, Home
Energy Management in Smart Grid, The International Industrial
and Information Systems Conference (IIISC), 2014.
[23] M. A. Khan, A. Mahmood, M. Arif, M. N. Ullah, I. Khan, M.
S. Hassan, A. Zafar, and Z. A. Khan Improvement in Perturb
and Observe Method for Maximum Power Point Tracking of PV
Panel, J. Basic. Appl. Sci. Res., vol. 3, no. 9, pp. 456466, Sept.
2013.
[24] Jinfeng Yang, Rong Cheng, Wenming Liu, Yong Xiao, and
Fangguo Zhang Cryptoanalysis and improvement of smart pre-
payment meter protocol in standard Q/GDW 365, Int. J. Grid
and Utility Computing, vol. 4, no. 1, pp. 4046, June 2013.

1032