,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ

3DSHU,G

Demand Side Management for Household

Equipment’s

Nikhil D. Rahate Nandkishore Kinhekar

Department of Electrical Engineering Department of Electrical Engineering
Sardar Patel College of Engineering Sardar Patel College of Engineering
Andheri (West), Mumbai-400052 Andheri (West), Mumbai-400052
nikhil.rahate2491@gmail.com nandkishor.kinhekar@gmail.com

Abstract—This paper aims to minimize the peak load at utility load curve, (ii) reducing electricity bill. The first objective can
as well as reduce the electricity bill of residential consumers. A be achieved by encouraging the residential consumers and
Swarm Intelligent based Particle Swarm Optimization inform consumers about their consumption pattern by designing
methodology is used to minimize the peak load at the utility, for the more efficient DSM program. The second objective can be
that an effective load shifting Demand Side Management achieved by shifting the shift-able devices from peak hours to
technique is used. In this article, the residential consumer co- off-peak hours.
operates with the utility to manage their energy demand by
scheduling the shift-able home appliances based on consumer’s In this paper, the household appliances are categorized into
priority. The proposed article considers 14 types of shift-able two categories, (i) base load type equipment’s, (ii) shift-able
appliances which reschedule for the next day to manage the load type equipment’s. The base load type equipment’s cannot
demand. Simulation work shows that reduction in consumed be shift due to consumer’s convenience, and shift-able load type
power during peak load period and electricity bill minimization of equipment’s can only be shift depending on consumer’s
the consumer. priority. The energy management is done by scheduling the
shift-able appliances. A metaheuristics optimization technique
Keywords—Demand Side Management (DSM); Particle Swarm namely Particle Swarm Optimization is used for optimal load
Optimization (PSO); Load Scheduling.
scheduling of appliances which minimizes the different
between objective load curve and the forecasted load curve so
I. INTRODUCTION that the optimal load curve obtain, which can minimize the peak
DSM is an activity in which utilities frames the time demand of utility [3]. 14 types of shift-able devices are
schedule of domestic electric appliances to minimize the peak considered in this paper for scheduling, by considering the
on utilities and electricity bill reduction of the consumer by consumer's priority level. Each residential consumer has
performing the activities. In which consumers has to use their specific load curve, the peak of residential consumer leads to
household appliances according to time schedule given by the peak on utility. On peak period utility has to manage the
utilities. Demand Side Management help to reduces peak demand of the residential load, which can be done by either load
demand, thereby no need to operate high-cost, high-emission shedding in some areas or purchasing the power at high cost to
generating units. Demand side management (DSM) or Load fulfill the load demand. To avoid these two major problems this
Side Management optimizes energy uses and price by paper focuses on minimizing the peak demand by load
scheduling loads away from peak periods to off-peak periods, scheduling the shift-able household appliances.
in such a way that energy can be utilized efficiently, reduce load The rest of the article is organized as follows. Section II
shedding, and improves the reliability of the system. describes the system model, Section III proposed the problem
In DSM, it is important to maintain the energy balance formulation for DSM, Section IV suggest the proposed
between the utility company and the customers. It is observed methodology, Section V presents the details about simulation
that in the existing distribution system the consumers are not and Section VI concludes the paper.
aware of their daily electricity consumption. Therefore, load
factor of the residential consumer is low, to increase the load II. SYSTEM MODEL
factor the consumers should reduce their peak demand by The system model is shown in Figure 1 includes three main
avoiding the uses of many devices at the same time. parts Residential House, Distribution Transformer and Utility.
According to the demand side management program, Utility feeds the electricity to residential houses. The utility
consumers can get significant saving in electric bill and also knows the total load on each residential consumer and also
helps to the utility to reduce the new generation capacity, knows the load consumption pattern of each residential house.
reducing air pollution [1]. Household equipment scheduling
program targets the two most important objectives (i) flatten
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G
Half hourly
consumption in Shifting
Total
Shift-able kW Range
Type No of
Equipment’s in half
1st 2nd Devices
hr.
interval interval
1 Coffee Maker 1 -- 1 58
2 Toaster 0.9 -- 1 57
3 Electric Kettle 1.6 -- 1 59
4 Vacuum Cleaner 0.5 -- 3 65
5 Washing Machine 1.1 1.1 4 96
6 Clothes Dryer 1.7 -- 4 84
7 Hair Dryer 1.8 -- 2 68
8 Water Heater 1.5 -- 2 55
FIGURE 1. PROPOSED SYSTEM MODEL
9 Iron 1.15 -- 4 65
In this paper, a utility is providing electricity to 200
10 Dishwasher 1.3 -- 3 67
residential consumers through a four-distribution transformer.
Each distribution transformer is feeding a load of 50 consumers. 11 Microwave 1.2 -- 1 70
The utility collects the half hourly load data from all residential 12 Rice Cooker 0.8 -- 2 68
houses and schedules the next day load curve by managing the
13 Frying Pan 1.4 -- 2 61
shift-able load from peak hour to off-peak hour. 14 types of
shift-able devices are considered with shifting priority decided 14 Mixer 0.7 -- 2 75
by the consumer, and their power rating is given in Table 1 [1]. TOTAL 954
1st half interval shows the power rating of the shift-able device
and second half interval shows the continuation of the device. TABLE I. DATA OF SHIFT-ABLE DEVICES WITH THEIR POWER RATINGS,
Suppose the washing machine has time period of 1 hour then it PRIORITY AND TOTAL NUMBER OF DEVICES.

takes two half intervals. The shifting range of the device allows
the utility to schedule consumers load according to consumer’s B. Objective Load Data
priority. The 200 consumers having the total number of devices The Objective load curve is taken as average load of total
which is shown in the last column. 200 consumers in this paper to flatten the next day load curve
by shifting the load from peak period to off-peak period. The
III. PROBLEM FORMULATION objective load curve is found by
The paper suggested DSM methodology schedules each ∑ ( )
shift-able load such that the difference between forecasted load ( )=
curve and objective load curve minimizes based on consumer’s
priority. To minimize the difference an efficient Particle Swarm Where,
Optimization algorithm is developed in MATLAB 2014. The ( ) is objective load at time interval t in kW.
forecasted load data and objective load data is available with T is maximum number of time interval i.e. 48.
the utility given in Table 3. Load scheduling technique is
mathematically formulated as follows. C. Minimization Model
The minimization model minimizes the difference between
A. Forecasted Load data two load curves which leads to flatten load curve by reducing
The forecasted load data is the next day load curve which peaks and valleys. The minimization model given as follows.
represents the peak period and off-peak period on the system.
This paper considers total power consumed by the 200
| ( )− ( )|
consumers at time interval t as the forecasted load, which is
given by
While shifting load from the peak period to off-peak
period the consumed energy should not change so the energy
( )=
balance constraint should be satisfied which is given by

Where,
( ) is the forecasted load at time t in kW. ( )= ( )
is the total number of consumer.
Where,
( ) = Total power before load shifting in kW.
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G
( ) = Total power after shifting in kW. 10 9.30 am – 10.00 am 91.74 67.66 5.09

D. Price Minimization 11 10.00 am – 10.30 am 112.82 67.66 5.09

The price minimization model gives the actual price of
12 10.30 am – 11.00 am 105.24 67.66 5.09
particular interval.
13 11.00 am – 11.30 am 108.47 67.66 5.09
= ( )× ( )
14 11.30 am – 12.00 pm 120.72 67.66 5.09
Where,
= Final minimum cost of individual consumer 15 12.00 pm – 12.30 pm 106.94 67.66 5.09
( ) = Power consumed by the individual consumer
during t interval in kW. 16 12.30 pm – 1.00 pm 110.56 67.66 5.09
( ) = Price of electricity in rupees in t interval.
17 1.00 pm – 1.30 pm 92.17 67.66 5.09

IV. PROPOSED METHODOLOGY 18 1.30 pm – 2.00 pm 59.9 67.66 5.09

DSM program can solve by conventional methods like
19 2.00 pm – 2.30 pm 58.12 67.66 3.36
Dynamic Programming such traditional method takes a long
time to produce results and also can handle limited loads. 20 2.30 pm – 3.00 pm 60.08 67.66 3.36
Another way to find out the optimal solution using a search
based algorithm like GA, ACT, PSO such algorithm can 21 3.00 pm – 3.30 pm 52.42 67.66 3.36
manage more loads [2] such algorithm can control more loads
and also takes less time to produce optimal results. A PSO 22 3.30 pm – 4.00 pm 58.41 67.66 3.36
methodology is proposed to obtain optimal results by
scheduling the 14 types of devices [1]. The electricity price 23 4.00 pm – 4.30 pm 46.76 67.66 3.36
designed in such a way that the consumer has to pay more if
there is a peak on utility and pay less in off-peak hours. The 24 4.30 pm – 5.00 pm 46.99 67.66 3.36
developed algorithm shifts only shift-able load/ devices from
peak period to off-peak period to reduce the consumer's 25 5.00 pm – 5.30 pm 43.54 67.66 3.36
electricity bill. A set of 200 residential consumers considered in
26 5.30 pm – 6.00 pm 45.17 67.66 3.36
this article for DSM. All residential consumers have some shift-
able devices among the Table 1. Total 954 devices are available 27 6.00 pm – 6.30 pm 46.8 67.66 3.36
for shifting. A half hourly based data is considered given in
Table 2. Total 48 intervals are considered for more flexibility 28 6.30 pm – 7.00 pm 69.23 67.66 3.36
of shifting load from one interval to another interval. To get the
optimum solution for this complex problem, Particle Swarm 29 7.00 pm – 7.30 pm 81.98 67.66 3.36
Optimization methodology is chosen.
30 7.30 pm – 8.00 pm 85.98 67.66 3.36
Forecasted Objective Price in
Interval Time (Half Hour)
Load (kW) Load (kW) Rupees 31 8.00 pm – 8.30 pm 74.11 67.66 3.36

1 5.00 am – 5.30 am 34.8 67.66 3.36 32 8.30 pm – 9.00 pm 82.98 67.66 3.36

2 5.30 am – 6.00 am 38.4 67.66 3.36 33 9.00 pm – 9.30 pm 64.77 67.66 3.36

3 6.00 am – 6.30 am 51.71 67.66 3.36 34 9.30 pm – 10.00 pm 64.36 67.66 2.83

4 6.30 am – 7.00 am 61.95 67.66 5.09 35 10.00 pm – 10.30 pm 63.89 67.66 2.83

5 7.00 am – 7.30 am 78.78 67.66 5.09 36 10.30 pm – 11.00 pm 60.37 67.66 2.83

6 7.30 am – 8.00 am 101.72 67.66 5.09 37 11.00 pm – 11.30 pm 55.45 67.66 2.83

7 8.00 am – 8.30 am 98.18 67.66 5.09 38 11.30 pm – 12.00 pm 55.97 67.66 2.83

8 8.30 am – 9.00 am 105.38 67.66 5.09 39 12.00 am – 12.30 am 59.33 67.66 2.83

9 9.00 am – 9.30 am 85.18 67.66 5.09 40 12.30 am – 1.00 am 61.03 67.66 2.83
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G
41 1.00 am – 1.30 am 55.1 67.66 2.83
STEP 3. Separate the shift-able load and the base load from
the total consumer's usage data.
42 1.30 am – 2.00 am 50.19 67.66 2.83 STEP 4. Initially, shift the ith load randomly based on
consumers priority store the result as .
43 2.00 am – 2.30 am 47.12 67.66 3.36
STEP 5. Update the position and velocity of the particle
44 2.30 pm – 3.00 pm 42.47 67.66 3.36 from equation 1 and 2.
45 3.00 pm – 3.30 pm 40.32 67.66 3.36 STEP 6. In the next iteration if the result is less than
save it as .
46 3.30 pm – 4.00 pm 38.57 67.66 3.36
STEP 7. If shifting is finished then produced the results else
47 4.00 pm – 4.30 pm 36 67.66 3.36 repeat from the step 4 the procedure.

48 4.30 pm – 5.00 pm 35.26 67.66 3.36

V. PROPOSED METHODOLOGY
TABLE II. FORECASTED LOAD, OBJECTIVE LOAD AND TOD TARIFF Simulation result carried out in MATLAB 2014 (Version
8.3) shows the DSM strategy brings the forecasted load curve
PSO is a search based heuristics optimization technique close to the objective load curve. The developed algorithm
developed be Kennedy and Eberhart in 1995 [5]. In PSO the managed the all loads to their optimal interval without
solution is searched by randomly initializing the particles, disturbing the base load. One-day data is taken to apply the
which communicates with each other. Each Particle represents DSM strategy, and PSO algorithm is applied to manage the all
a solution of the function. A maximization or minimization of shift-able loads. 10 number of particles were taken and 200
the function found by storing the past best position and the iterations were carried out to get the optimal result while
swarm best position [5]. performing the simulation. The simulation result is shown in
PSO Algorithm: Figure 2. Convergence characteristics for the given objective
function is shown in Figure 3.
A PSO technique consists of the following two equations
1) Velocity Update Equation
( + 1) = × ( ) + × × − ( )

+ × × − ( ) (1)

2) Position Update Equation

( + 1) = ( )+ ( + 1) (2)
Where,
i = Particle Index.
k = Iteration Index.
w = Inertia weight.
, = Random no within range of [0 1]. FIGURE 2. PROPOSED SYSTEM MODEL
, = Acceleration co-efficient.
= Velocity of the ℎ particle.
= Position of the ℎ particle.
= Personal Position of the particle.
= Global best position of the swarm.
Following are the steps to solve the particle swarm
optimization algorithm for this problem.
STEP 1. Read the system data including all consumers half
hourly usage, forecasted load data, objective load
data, priority of loads.
STEP 2. Set the maximum iteration and chose the PSO
parameters such as acceleration coefficients taken
as 2, inertia weight is 0.9, the number of particles
FIGURE 3. CONVERGENCE CHARACTERISTICS OF OBJECTIVE FUNCTION
in the swarm.
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G
Table 3 shows the reduction in peak load during 24 hours. individual consumer is considerably reduced. Approximate
The forecasted load had the peak of 120.72 kW. After applying 20% daily electricity bill can be saved without disturbing the
DSM methodology peak demand reduces to 93.88 kW. base load and shifting load on consumer’s priority.
Peak Load Peak Load Reduction Percentage
without DSM with DSM in Peak reduction in References
(kW) (kW) load (kW) load (%)
1. 120.72 93.88 26.84 22.23
[1] T. Logenthiran, D. Srinivasan, Ei Phyu, “Particle Swarm Optimization for
TABLE III. PEAK DEMAND REDUCTION Demand Side Management in Smart Grid,” Newcastle University,
Singapore. vol. 3, no. 3, pp. 1244-1252, september 2012.
Table 4 shows the cost saving of first five residential [2] Nandkishor Kinhekar, Hari Om Gupta, “Particle Swarm Optimization
consumers after shifting loads according to consumer’s priority. Based Demand Response for Residential Consumers,” 2015.
[3] Sanjaya Nayak, et al. “Demand Side Management of Residential Loads in
Consumer No. Cost without Cost with Total Saving in a Smart Grid using 2D Particle Swarm Optimization Technique,”
DSM (Rs.) DSM (Rs.) Cost (Rs.) Anusandhan University, Bhubaneswar, India. 2015.
Consumer 1 92.62 82.05 10.57 [4] Michael Angelo A. Pedrasa , “Scheduling of Demand Side Resources
Using Binary Particle Swarm Optimization,” vol. 24, no. 3, pp. 1173-
Consumer 2 72.91 59.13 13.78 1180, August 2009.
Consumer 3 62.30 49.33 12.97 [5] Kennedy, J., Eberhart, R., Particle Swarm Optimization, Proc. IEEE Int’l.
Conf. on Neural Networks (Perth, Australia), IEEE Service Center,
Consumer 4 74.19 63.42 10.77 Piscataway, NJ, IV: 1942-1948, 1995
Consumer 5 51.44 41.56 09.88 [6] Matlab Online Help, http://www.MathWorks.com
[7] Nikhil Gudi, Lingfeng Wang, Vijay Devabhaktuni, “A demand side
TABLE IV. COST REDUCTION management based simulation platform incorporating heuristic
optimization for management of household appliances,” pp.185-193. June
2012.
DSM strategy suggested above can be applied to large no of
[8] Donya G. Dobakhshari and Vijay Gupta, “Optimal Contract Design for
the consumer so that the peak demand on utility can be reduced Incentive-Based Demand Response,” July 6-8, 2016. Boston, MA, USA.
and the individual consumer can get minimum electricity bill. pp. 3219-3224.
The above DSM methodology can be implemented in Smart [9] Wokje Abrahamse, Linda Steg, Charles Vlek, Talib Rothengatter, “A
Grid, which required two-way communication. review of intervention studies aimed at household energy conservation,”
University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen, The
Netherlands. pp. 273-291.
VI. CONCLUSION [10] N. Kinhekar, N. P. Padhy, and H. O. Gupta, “Multiobjective demand side
This paper proposed the DSM technique to manage loads of management solutions for utilities with peak demand deficit,” Int. J. Elect.
Power Energy Syst., vol. 55, pp. 612–619, Feb. 2014.
total 200 residential consumers. This load shifting strategy
[11] Nandkishor Kinhekar, Narayana Prasad Padhy, Hari Om Gupta, “Demand
applied successfully to get an optimum result and peak demand Side Management for Residential Consumers,” 2013.
reduced by 22.23% as well as the electricity bill of the