(IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO.

3, SEPTEMBER 2013)

An Optimal Power Scheduling Method for Demand

Response in Home Energy Management System
Zhuang Zhao, Won Cheol Lee, Member, IEEE, Yoan Shin, Senior Member, IEEE, and Kyung-Bin Song

Prepared By:
M. Asghar Khan
Electrical Engineering Dept.
COMSATS Institute of IT Islamabad.

OUTLINE
BACKGROUND
CONTRIBUTION
SYSTEM MODEL
PROPOSED APPROACH
SIMULATION RESULTS
CONCLUSION
2

BACKGROUND (1/6)
Smart Grid:
A modernizedelectrical grid
Uses information and communications technologyto gather and act on
information, in an automated fashion
To improve the efficiency, reliability, economics, and sustainability of the
production and distribution of electricity
HEMS (Home Energy Management System):
Any device or system in the home used to:
1. Control an energy-consuming device
2. Identify or diagnose energy saving opportunities
3. Provide information to occupants to influence how they
consume energy

BACKGROUND (2/6)
HAN (Home Area Network):
A type oflocal area network
Facilitate communication and interoperability among home appliances
Appliances capable of participating in this network are called smart appliances
AMI (Advanced Metering Infrastructure):
A system that measure, collect, and analyze energy usage, and
Communicate with smart meters either on request or on a schedule
SM (Smart Meter):
Anelectronicdevice that records consumption ofelectric energyin intervals
and
Communicates that informationat least daily back to theutilityfor monitoring
and billing purposes

BACKGROUND (3/6)
IBR (Inclining Block Rates):
IBR divide the electricity price into several steps or blocks
The first block of electricity is at the lowest price

RTP (Real Time Pricing):

RTP is generally an hourly rate which is applied to usage on an
hourly basis

Demand Response (DR):

DR refers to, actions taken to change residential electricity demand
in response to prices over time

BACKGROUND (4/6)
HG (Home Gateway):
A HGallows the connection of ahome area network(HAN) to an
internet, i.e.,wide area network(WAN)
With the development of smart grid, users have the opportunity
to schedule their power usage in home by themselves
The purpose of load scheduling is to reduce the power Peak-toAverage Ratio (PAR) and accordingly to reduce electricity
expense
6

BACKGROUND (5/6)

Peak-to-Average
Ratio (PAR) :
PAR is a measure of awaveform, such asalternating current, showing the
ratio of peak value to the average value
PAR =1 indicates no peaks, such asdirect current
PAR =
The minimum possible PAR is 1, 1:1

IHD (In-Home Display):

dedicateddisplay devicedesigned to deliver
energy-relatedinformation
e.g., energy consumption, pricing, or service
messagesfrom a utility

BACKGROUND (6/6)

CONTRIBUTION
1. A general architecture of Energy Management System (EMS)
in Home Area Network (HAN) is proposed based on Smart
Meter (SM)
2. Also, an efficient scheduling method (i.e., Genetic Algorithm
(GA) based optimization) for home power usage is presented
3. An alternative billing mechanism (i.e., Real Time Pricing
(RTP) combined with Inclining Block Rates (IBR)) is developed
9

SYSTEM MODEL (1/4)

Components of the proposed EMS
(Energy Management System) architecture are:
i.
ii.
iii.
iv.
v.
vi.

AMI (Advanced Metering Infrastructure)

SM
HG (Home Gateway)
EMC (Energy Management Controller)
Home appliances
IHD (In-Home Display)
10

SYSTEM MODEL (2/4)

Objectives of Adopting the EMS are:
i.

To minimize the expense of electricity

ii. To reduce the PAR by scheduling the pattern of electricity

usage
iii. Minimizing the Delay Time Rate (DTR) of home appliances

11

SYSTEM MODEL (3/4)

Classification of the home appliances:
1. Automatically Operated Appliances (AOAs) (e.g., washing machine, dish washer,
air conditioner)
i. Interruptible (e.g., washing machine)
ii. Non-interruptible (e.g., electric kettle)

2. Manually Operated Appliances (MOAs) (e.g., computer, television, vacuum

cleaner)
12

SYSTEM MODEL (1/4)

The home appliances that could be scheduled are only
AOAs
The AOAs mentioned in this paper are smart home
appliances
The AOAs do not interact with each other, they only
interact with the HG instead
All the operations of AOAs would have been scheduled
by the EMC at
the beginning of the day

13

PROPOSED APPROACH
A. Usage Pattern of Home Electric Appliances:
HG receives DR information and the profile of RTEP (Real Time Electricity
Price) from the utility company
. The EMC can make decision on power schedule for all AOAs in the home
. It is necessary for residents to set the time parameters for each AOA, i.e.,
a. LOT (Length of Operation Time) from start to end
b. OTI (Operation Time Interval) during which the appliance is valid to be
scheduled, and
c. Its power consumption per hour.
14

PROPOSED APPROACH
These parameters can be set on the IHD device and
then transmitted to the EMC via HG
Since each MOA is operated manually and nobody can
tell in advance when and for how long they will use an
MOA
Therefore, only the impact of AOAs on electricity cost
and PAR are considered
15

PROPOSED APPROACH
B. Final Goal of the Proposed Approach:
1 hour is divided into 5 time slots (i.e., 1 time slot=12
minutes) and so 1 day= 24 x 5=120 slots, denoted by
The shortest operation time of any appliance is set to be 12
minutes
Hence, the LOT of the air conditioner can be set to integer
multiples of the 12-minute interval
The unit of LOT is time slots
16

PROPOSED APPROACH
A denotes the set of AOAs
For each appliance
, we assume that
power consumption scheduling vector:

is the

where
denotes the power consumption value for
appliance
during the
time slot, and their unit is kWh
we assume that the power consumption values per hour
for all appliances are all fixed
17

PROPOSED APPROACH
When the power consumption value per hour of
appliance is denoted by
, during the
time slot,
the corresponding power consumption is

The objective is to optimize the power consumption

scheduling vector

18

PROPOSED APPROACH
C. RTP Combined With IBR:
.As the problem with RTP is load synchronization and
.IBR suffers due to fixed prices for most of the times
.So in this new pricing scheme the EP (Electricity Price)
will be different within the same time slot based on the
total power consumption
19

PROPOSED APPROACH
Two electricity price levels are set in IBR, and the EP
changes every hour
The EP function is,

where
denotes the total power consumption in the
home during the
hour,
is the RTEP during the
hour in a day, is the second electricity price level that
should be greater than
and
is the threshold of
power consumption at hour
of IBR.
The unit of EP is in cents/kWh

20

PROPOSED APPROACH
As one hour has been divided into five time slots, a
modification in the basic EP function is needed
Dividing
and
by 5, we obtain the total power
consumption value
and the IBR threshold
in every
12-minute, respectively
Then, the EP function can be altered as,

The number of variables becomes 120 instead of 24 due

to the time division
21

PROPOSED APPROACH
To avoid the problem of load synchronization, in this paper
the author assumed that,

According to the British Columbia Hydro, the value of is

determined to be 1.4423
It seems unrealistic to utilize this price function because it is
impossible to obtain the entire EP function ahead of the day
However, several electricity price prediction methods have
been proposed
22

PROPOSED APPROACH
D. Problem Formulation:
.For residents it is necessary to set some parameters for
each AOA
.The OTI (Operation Time Interval) will be
as the indexes of the start and the end time slots,
respectively
.Let indicates the LOT, (i.e., the number of time slots
for the operation of appliance )
.In addition,
must be greater than or equal to
.The greater the value of
is, the more possible
23
solutions there would be

PROPOSED APPROACH
The Operation Start time (OST) of appliance
is
The power consumption scheduling vector of appliance
would be determined once we have
Since
and
are all known parameters
The figure (next slide) shows the relationship of these
parameters
Four different kinds of AOA are included.

24

PROPOSED APPROACH

25

PROPOSED APPROACH
Now for each appliance
there exists a group of
parameters comprising the OTI
LOT
and
power consumption value per hour
In addition, OST
is set as a variable
The range of OST of is

The range of

is shown in the figure given in next slide

26

PROPOSED APPROACH

27

PROPOSED APPROACH
A variable vector
which is composed of the
OSTs of all AOAs
Therefore, a power consumption scheduling matrix for all
AOAs can be defined as,

Where
denotes a matrix in which each row stands for
the power schedule of a certain appliance

is the index of column

The expression
indicates that
belongs to
excluding the range

28

PROPOSED APPROACH
By summing up all the values of each column vector in
the power consumption scheduling matrix
A total power consumption scheduling vector
would be determined as

stands for the

column in the power
consumption scheduling matrix

29

PROPOSED APPROACH
E. Delay Time Rate (DTR) Minimization:
.Residents usually hope that home appliances can finish their
work as soon as possible
.Therefore, minimizing the DTR of home appliances is
considered

.Where
denotes the DTR of appliance
.The smallest and largest values of
are set to 0 and 1
.The later the appliance operates, the larger the
becomes
30

PROPOSED APPROACH
A delay parameter
is introduced and the relevant
formula can be expressed by

Since the delay parameter

geometrically
increases as
continues to increase
So, for residents, the value of this formula is expected
to be as smaller as possible
The concept is explained in the next slide
31

Explaining the

concept

32

PROPOSED APPROACH
The power consumption scheduling problem is presented as
an optimization problem, is given below:
where

and
are the weights representing the importance of the
individual objectives shown in (12) and (13)
Where
and
In (12), function
denotes the EP at the
time slot

33

PROPOSED APPROACH
After normalization, the final optimization formula is
constructed as follows:
For each appliance
since the maximum value of
is
The value of
equal to
Where

is

indicates the number

of AOAs
34

PROPOSED APPROACH
F. Genetic Algorithm (GA) Module:
The GA is adopted to optimize the OSTs of all AOAs
The GA randomly creates a solution population consisting of a certain number
of individuals
Each individual contains a solution set of all kinds of variables represented as a
chromosome
After calculating fitness values, selecting individuals, crossover, and mutation
The new solutions is obtained that include both the old and the new individuals
35

PROPOSED APPROACH

After analyzing that whether the expected generation number

has been satisfied or not, the new generation or the best
fitness solution is obtained
The GA optimization and the specific calculation process
scheme is shown in figure

36

PROPOSED APPROACH
Since the OST is the only variable in the proposed
scheme and also the constraint parameters are set in
the beginning
So the total fitness function is shown in Equation (14)
A roulette selection method in which the individual with
a better fitness value has a higher probability to be
selected for further processing is used referring to the
selection process
37

Simulation Setup
Nine kinds of AOAs in the home are selected for
simulation purpose
Some AOAs may be used more than once by users in a
day, the number of AOAs operations in one day is up to
16
Table 1 shows the information of all AOAs

38

Simulation Setup
The values of different parameters assumed are;
The power threshold
The delay parameter
For GA optimization, the population size of each
generation
is, 200
The probability of crossover between two strings
is,
90%
All strings have a probability of 2% to mutate, which is
denoted by
When the generation number reaches 1000, the
evolution process will finish

39

Simulation Setup
All known parameters used for the simulation of the
proposed approach are listed in Table 2

40

Simulation Results
A. Relationship Between Electricity Cost and DTR:
.Within the OTI, the operation times of AOAs are not fixed
due to the RTEP and the operations of other AOAs
.The average DTR can be defined for all AOAs as,

.Generally speaking, the relationship between the

electricity cost and
is a trade-off
41

Simulation Results
The figure shows the simulation result of the relationship
between

the electricity cost and the average DTR

From the figure , at the position
it implies that the major
consideration is minimizing the delay time
So in this case
and
Whereas, when the minimum electricity
cost is reached,

and
42

Simulation Results
A. Impact of Inclining Block Rates:
.The RTEP data is adopted from the Ameren Illinois Power Company,
and the date ranges from August 1st, 2012 to October 31st, 2012
(92 days)
.In this case only minimizing
the electricity cost; therefore,
and
.The average daily electricity
cost in cents for three months is
reduced about 26.06%
43

Simulation Results
The figure shows that PAR reduces from 5.22 to 3.37
(about 35.44%), to conclude that the proposed
approach is effective in reducing both the electricity
cost and PAR

44

Simulation Results
The comparison between IBR combined with RTP and

RTP scheme alone is shown in the figure

The figure shows that if
only RTP is used, the PAR
value would be
high, whereas the
Proposed RTP combined
with IBR is a better way to
reduce PAR

45

Simulation Results
The impact of IBR on peak power usage in home as
shown
The figure shows the RTEP profile in the United States
on August 7th, 2012
The EP within the 16th20th
time slots on that day is 1.646 cents/kWh, which is the
lowest during the day

46

Simulation Results
The figure shows that with the power scheduling using
RTP scheme alone
Even if the electricity cost would reduce
A large amount of power
demand would be shifted to
the 16th-20th time slots due
to the low EP at that time

47

Simulation Results
However, with the proposed power scheduling approach
based on the RTP combined with IBR pricing scheme
The power demand will be dispersive due to the
constraint of the power threshold

48

Simulation Results
C. Impact of MOAs:
Seven kinds of MOAs are involved
4-8 MOAs operations are considered
in one day that are detailed in Table 3 given below

These MOAs operate in a relatively random time slot during

a probable OTI
49

Simulation Results
Because MOAs are involved, so the electricity cost is
much more
greater than that of without
MOAs
With the proposed power
scheduling scheme, the
average electricity cost during
three months is reduced
about 15.51% in cents

50

Simulation Results
Due to the increase of the average power level, the PAR
value is relatively smaller than that of without MOAs
The PAR is reduced from
4.26 to 3.42, which is about
19.71% reduction

51

Simulation Results
D. Impact of Multiple Users:
.10 residential users are considered for the simulation
purpose
.The figure shows that even if
the number of residents
increases, the average PAR of
10 residential users aggregated
power demand is still reduced
from 4.34 to 2.84

52

Simulation Results
This figure shows that all the residents can reduce their
monthly electricity cost effectively by adopting the
proposed scheme

53

CONCLUSION
For residents, the beneficial features obtained by applying
the proposed scheme are
i.

Reduction of the electricity cost and also

ii. The delay time rate (DTR) of home appliances operations

simultaneously
.In addition, the benefit rewarded to utility companies is the
reduction of the PAR which would increase the stability of
the entire electricity system
.Surely, the proposed approach can be a reliable solution for
future EMS in HAN of smart grid
54