Load Shifting Technique on 24Hour Basis for a Smart-Grid to Reduce Cost and Peak Demand using Particle Swarm Optimization

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 10 | Oct -2017 www.irjet.net p-ISSN: 2395-0072
Load Shifting Technique on 24Hour Basis for a Smart-Grid to Reduce
Cost and Peak Demand Using Particle Swarm Optimization
K. Siva Sai Kumar1, Dr. M. Gopichand Naik2
1PG student, Dept. of Electrical Engineering, Andhra University College of Engineering (A), Andhra Pradesh, India
2Professor, Dept. of Electrical Engineering, Andhra University College of Engineering (A), Andhra Pradesh, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In present days, Load shifting is one of the
techniques used in Demand Side Management(DSM). The
increase in energy consumption in present scenario of power
distribution system introduces many demand side
management strategies to achieve desired changes in utility’s
load shape. This paper presents a demand side management
strategy based on day-ahead load shifting technique, by
Shifting the loads from peak hours to off peak hours. Particle
swarm optimization (PSO) algorithm has been modified for
the DSM problem and implemented in three area loads of
smart grid i.e. residential, commercial and industrial. The
Objective load curve is chosen inversely proportional to the
wholesale electricity price. The performance of PSO algorithm
is compared to the GA.
The simulation results show that the proposed load shifting
technique reduces overall operational cost and the peak load
demand of the smart grid.
Key Words: Demand side management, PSO algorithm,
Controllable loads, Load scheduling, Peak to average
value.
1.INTRODUCTION
For a variety of reasons, Electricity consumption is in rapid
growth around the world. LargemarketssuchasChina,India
and Brazil are gaining in global importance. And the more
their economies grow, the greater their hunger for
electricity. More people means a higher demand for
electricity. Every sector is expanding and more and more
electrically-powered devices are being used. To reduce
carbon emissions, fossil fuels are being replaced by new
technologies which also increases higher electricity
consumption.
To meet this demand for electricity, more and more
emphasis has been placedon improving energy efficiency for
both economic and environmental reasons. However, this
improvement is limited due to the existing infrastructure.
Hence much attention has been shifted to the consumerside
instead of altering the generation side of the power systems.
Generally, the wholesale electricity system adjusts to
changing demand by dispatching additional or less
generation. However, during peak periods, the additional
generation is usually supplied by less efficient ("peaking")
sources. New technologies are introduced to handle these
higher electricity demands in order to make the system
sustainable, efficient and reliable. Demandsidemanagement
[1]-[3] strategies generally focus on the peak load demand,
increasing the efficiency and reducing cost in load
scheduling. These strategies are not new – the idea has been
explored since the 1970s. More recently, there are different
approaches and algorithms being studied for demand side
management. Energy demand management activities
attempt to bring the electricity demand and supply closer to
a perceived optimum, and help give electricity end users
benefits for reducing their demand. With the help of recent
innovations, the integrated approach to DSM is becoming
increasingly common. Some focuses on reducing the
generation capacity byreducingthe peak loaddemand,some
on smoothening the load profile and bringing it close to the
desired load curve, and some on scheduling of the loads
according to the time intervals where the power is most
available and preferably the cheapest as well.
IDSM automatically sends signalstoend-usesystemstoshed
load depending on system conditions.Thisallowsforprecise
tuning of demand to ensure that it match with the supply at
all times, reduces capital expenditures for the utility.
In general, adjustments to demand can occur in different
ways: through consumer responses to attractive price
signals, such as permanent differential rates for eveningand
day times.
1.1 Particle Swarm Optimization
Particle swarm optimization (PSO) is a population based
optimization technique developed by Dr.
Eberhart and Dr. Kennedy in 1995 [4], inspired by social
behavior of bird flocking or fish schooling.
In PSO algorithm, the system to be optimized is initialized
with some random solutions and searches for optima by
updating generations. However, unlike Genetic Algorithms
(GA), PSO has no evolution operators such as crossover and
mutation [5]. In PSO, the potential solutions, calledparticles,
fly through the problem space by following the current
optimum particles.
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1180

PSO simulates the behaviors of bird flocking. Suppose the
following scenario: a group of birds are randomly searching
for the food in an area. If there is only one piece of food in
the area being searched, all the birds do not know where the
food is. But they will know how far the food is in each
iteration. So the best strategy to find the food is to follow the
bird which is nearest to the food.
PSO learned from above scenario and used it to solve the
optimization problems. Here each single solution is a "bird"
as in bird flock and call it as "particle". All of these particles
have fitness values which are calculated by the fitness
function to be optimized, and have velocities which direct
the flying of the particles. The particles fly through the
problem space by followingthecurrentoptimumparticles as
bird’s scenario.
In every iteration, each particle is updated by two "best"
values, the first one is the best solution (fitness) it has
achieved so far and this value is called pbest. Another "best"
value that is tracked by the PSO is the best value,obtained so
far by any particle in the population. This best value is called
gbest (global best). When a particle takes part of the
population as its topological neighbors, the best value is a
local best.
After finding these pbest and gbest, the particle updates its
velocity and positions with following equations
(1) v = v + c1 * rand * (pbest - prsnt) + c2 * rand * (gbest -
prsnt)
(2) prsnt = prsnt + v
where prsnt represents the current particle (solution), v is
the particle velocity, rand represents random number
between (0,1); c1, c2 are learning factors. usually
c1 = c2 = 2.
If the sum of accelerations would cause the velocity to
exceed Vmax, which is a parameter specified by the user.
Then the velocity is limited to Vmax.
Flow chart of PSO algorithm
1.2 Load Shifting Technique
Load shifting is one of the techniques used in DSM. It
involves moving the consumption of shift-able loads during
peak hours to off-peak hours of that days [6]. It doesn’t
reduce net quantity of energy consumed in an electricity.
To understand the reason behind load shifting, one needs to
realize that efficiency of electricity generation varies with
the load demand. Duringpeak hours,additional generationis
to be supplied by less efficient generation stations available
(peak load plants). This involves in increasing the
operational cost, and the system may not be sustain these
peak loads. As the demand in now days is inrapidgrowth we
cannot simply alter theinfrastructureofthesystem.To make
the system reliable, efficient and sustainable, load side
management attracts the researchers. Because of different
load generations at any given times there are different costs
associated with electricity generated. This means there
exists substantial savings in generation costs if some part of
load during peak hours could be moved around in time. This
is where load shifting comes into play.
This shifting can happen in different ways. For example,
having variable electricity price and thus encourage
consumption in specific hours, remote controlling specific
appliances etc.

Fig 1: Load shifting
2. PROPOSED STRATEGY
a) Problem formulation
The proposed demand side management strategy schedules
the connection moments of each shift-able device in the
system in a way that brings the load consumption curve as
close as to the objective load consumption curve. Proposed
load shifting technique is mathematically formulated as
follows.
Minimize: f1 max(loadf)
Minimize: f2
Where Load(h) is hourly load, Ch is the cost of electricity at
hour h in ct/Kwh, H no. of hours a day. i.e. we are minimizing
both peak load demand and overall cost. This can be
formulated by choosing an objective load curve, and then
optimizing the forecasted load curve to the objective load
curve subjected to the constraints.
Minimize: fh [|RLMh|-|∆Loadh|]
Where RLM is the reduced load margin, ∆Loadh is calculated
load that is to be1 either connected or disconnected at each
hour h.
RLMh= Forecast(h)-Objective(h)
If RLM is >0 , load is to be disconnected. i.e.
∆Loadh =Disconnect(h)
If RLM is <0 , load is to be connected. i.e. ∆Loadh =Connect(h)
In this paper, our main objective isreducing thecost,thiscan
be achieved by choosing the load curve inversely
proportional to wholesale cost of electricity at each time
step.
Objective(h)=(Pavg/Pmax) *( )/Ph
Mathematical expressions:
Disconnect(h)= ) +
( )
The above equation calculates the total power consumption
by the X number of devices at hth hour in the first part. If the
device is having more than one operating hours then with
the same quantity of devices, power consumption is
calculated at successive hours, i.e. h+1, h+2… (h+l-1) in the
second part of the equation. Here Xkh and Xk(h+l) is the
quantity of devices of type k assumed tobeoperatingathour
h and h+l, P1k is the power consumption of device k in the
first hour of operation and Pk(1+j) is the power
consumption of the same device at subsequent hours.
Connect(h)= ) +
( )
Connect(h) is similar to the expression of disconnect(h).
This minimization problem is subjected to following
constraints
1) The number of devices shifted away from a time step
cannot be more than the number of devices available for
control at the time step.
2) The no of devices that are to be shifted is always positive
at any time instant.
Xkh>0 for all k,h
b) Controllable Loads:
Firstly, DSM controller fetches the user inputs for the pre-
scheduled loads. For each load, the user will input the
starting time (i.e. the earliest time the load can run), the
deadline (i.e. the latest time by when the load must finish
running), the load duration and the ideal time for the load to
run. Furthermore, the loads are classified into following
three classes.
Class 1 – Uncontrollable Loads: These are fixed loads, those
have highest priority and must be run at ideal times
specified by the users No load shifting will be performed for
class 1 loads. ex: Tv, Fan etc.
Class 2 – Controllable but Uninterruptable Loads: These are
the loads which can perform their work at any instant of
time but operation should not interrupted after starting. By
considering starting time (Ti), max time (Fi), working
duration (Li) , min time (Si) and power demand (Pi), the
proposed algorithm will find the optimal slot within
minimum and maximum time which incurs the lowest cost.
Example for class 2: Washing machine, Dryer etc.

Class 3 – Controllable and Interruptible Loads:
In this case, the schedule of the loads will still in between the
starting time (Ti) and the Maxtime (Fi). However, the user
can interrupt the operation of load to work in 2 or more
different slots, results in giving a large number of possible
solutions.
To solve this nature of problem, binary particle swarm
optimization(BPSO) algorithm is introduced here. In BPSO
proposed, a particle or a solution is represented by a row of
vector with n variables, which in thiscasewill bethe number
of hours in a day. Thus, for each load, as solution is
represented by the equation S= [s1, s2 ,s3 …..,s23 ,s24]
Where, each variable si can be either ‘0’, if the load is not
connected or ‘1’, if the load is connected. Thefitnessfunction
in this algorithm is defined to serve the objective of the
demand side management approach – tosavecost.Hence for
each load i with a schedule, S, the fitness function can be
formulated as below.
Cost= )
Where, S(t)i is the status of load i at time slot t, Pi is the
power demanded by load i, r(t) is the electricity rate at time
slot h.
Here the three areas in a smart grid are considered with
different types of controllable devices:
1) Residential Area: There are total 2604 controllable
devices comprising 14 different devices as mentioned in[7].
2) Commercial Area: Therearetotal 808 controllabledevices
of 8 different types with their consumption, starting time
and operating duration as mentioned in[7].
3) Industrial Area: There are total 109 controllable devices
of 6 different types available as mentioned in[7].
TABLE I: Forecasted load demands andwholesaleprices
Time
(hr.)
Wholesale
Price
(cents/kWh
)
Hourly Forecasted
Load(kWh)
Residential Commercial Industrial
1 12.00 540.9 661.5 1170.5
2 9.19 593.8 892.4 1560.1
3 12.27 593.6 1181.0 1274.9
4 20.69 594.1 1293.0 1372.3
5 26.82 558.8 1257.4 680.1
6 27.35 545.6 1257.4 898.6
7 13.81 535.4 1139.8 898.6
8 17.31 529.6 1318.6 842.4
9 16.42 513.8 1338.4 1145
10 9.83 866.4 1301.7 706.7
11 8.63 1085.6 1446.0 917.0
12 8.87 1196.6 1246.1 809.7
13 8.35 1228.3 1298.7 863.6
14 16.44 1117.3 1096.7 964.9
15 16.19 911.2 923.5 970.1
16 8.87 545.4 577.2 1022.7
17 8.65 395.3 404.0 974.0
18 8.11 331.9 375.2 876.6
19 8.25 364.7 375.2 827.9
20 8.10 348.8 404.0 730.5
21 8.14 269.6 432.9 730.5
22 8.13 269.6 432.9 779.2
23 8.34 412.3 432.9 1120.1
24 9.35 539.1 663.8 1509.7
3. SIMULATION STUDIES AND RESULTS
a) Residential Area
Simulation usingPSOalgorithmis performedona residential
area load mentioned in [7]. Both thepeak loaddemand,Peak
to average(PAR) value and cost are reduced, and results are
better compared to GA[8]. Peak load, PAR are reduced by
23.25% shown in Fig2 while cost is reduced by 7.22%.
Fig 2: DSM results for Residential area load

TABLE II: CALCULATED DATA FOR RESIDENTIAL AREA
LOAD
Parameters Without
DSM
With DSM % reduction
GA[8] PSO GA[8] PSO
Peak
load(Kw)
1363.6 1114.4 1046.5 18.3 23.25
Cost 2302.90 2188.3 2136.43 4.97 7.22
PAR ratio 1.85 1.51 1.42 18.3 23.24
b) Commercial Area
Simulation using PSO algorithm is performed on a
Commercial area load mentioned in [7]. Both the peak load
demand, Peak to average(PAR) value and cost are reduced,
and results are better compared to GA[8].Peak load,PARare
reduced by 17.61% shown in Fig3 while cost is reduced by
8.71%.
Fig 3: DSM results for Commercial area load
TABLE III: CALCULATED DATA FOR COMMERCIAL
AREA LOAD
Parameter Without
DSM
GA[8] PSO GA[8] PSO
Peak
load(Kw)
1818.2 1485.2 1497.9 18.3 17.61
Cost 3626.66 3424.3 3310.7 5.8 8.71
PAR ratio 1.7 1.39 1.4 18.23 17.61
c) Industrial Area
Simulation using PSO algorithm is performed ona Industrial
area load mentioned in [7]. Both thepeak loaddemand,Peak
to average(PAR) value and cost are reduced, and results are
better compared to GA[8]. Peak load, PAR are reduced by
8.24% shown in Fig4 while cost is reduced by 18.13%.
Fig 4: DSM results for Industrial area load
TABLE IV: CALCULATED DATA FOR INDUSTRIAL AREA
LOAD
Parameter Without
DSM
GA[8] PSO GA[8] PSO
Peak
load(Kw)
2727.3 2343.6 2502.4 14.2 8.24
Cost 5712.1 5141.6 4675.9 10.0 18.13
PAR ratio 1.617 1.38 1.483 14.65 8.28
4. CONCLUSION
Load shifting technique proposed in this paper reduces the
peak load demand , so that the entire power system is
sustainable. It also reduces the overall operational cost by
improving the load factor. PSOgivesbetterresultscompared
to GA and simulation has been carried out on a smart grid
consists of residential, commercial and industrial loads.
REFERENCES
[1] S. Rahman and Rinaldy, “An efficient load model for
analyzing demand side management impacts,” IEEE

Trans. Power Syst., vol. 8, no. 3, pp. 1219–1226, Aug.
1993.
[2] S. Bu and F. R. Yu, “A Game-Theoretical Scheme in the
Smart Grid With Demand-Side Management: Towardsa
SmartCyber-Physical Power Infrastructure,”IEEETrans.
Emerg. Top. Comput., vol. 1, no. 1, pp. 22–32, 2013.
[3] E.H.Et-Tolba,M.Maarouﬁ,and M.Ouassaid,“Demandside
management algorithms and modeling in smart grids A
customer’s behavior based study,” in Proc. Int. IEEE
Renew. Sustain. Energy Conf. (IRSEC), 2013, pp. 531–
536.
[4] J. Kennedy, and R. Eberhart, “Particle Swarm
Optimization,” IEEE International ConferenceonNeural
Networks, vol.4, pp.1942-1948, 1995.
[5] Q. Bai, “Analysis of Particle Swarm Optimization
Algorithm,” Computer and Information Science
(www.ccsenet.org/cis), vol.3, no.1, 2010
[6] C.W.Gellings,Demand-SideManagement:Conceptsand
Methods. Liburn, GA: Fairmont, 1988.
[7] A. R. S. Vidal, L. a. a. Jacobs, and L. S. Batista, “An
evolutionary approach for the demand side
management optimization in smart grid,” 2014 IEEE
Symp. Comput. Intell. Appl. Smart Grid, pp. 1–7,2014.
[8] T.Logenthiran, Dipti Srinivasan,and Tan Zong Shun
“Demand Side Management in Smart Grid Using
Heuristic Optimization”IEEE Trans. On Smart
grid,vol.3,no.3,Sep 2012.

Load Shifting Technique on 24Hour Basis for a Smart-Grid to Reduce Cost and Peak Demand using Particle Swarm Optimization

More Related Content

What's hot (19)

Similar to Load Shifting Technique on 24Hour Basis for a Smart-Grid to Reduce Cost and Peak Demand using Particle Swarm Optimization (20)

More from IRJET Journal (20)

Recently uploaded (20)

Load Shifting Technique on 24Hour Basis for a Smart-Grid to Reduce Cost and Peak Demand using Particle Swarm Optimization