Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Cleaner Production 222 (2019) 231e241

Contents lists available at ScienceDirect

Journal of Cleaner Production


journal homepage: www.elsevier.com/locate/jclepro

Optimization modeling for dynamic price based demand response in


microgrids
Muhammad Arshad Shehzad Hassan a, *, Minyou Chen a, Houfei Lin b,
Mohammed Hassan Ahmed a, Muhammad Zeeshan Khan a, Gohar Rehman Chughtai c
a
State Key Laboratory of Power Transmission Equipment, and System Security and New Technology, School of Electrical Engineering, Chongqing University,
Chongqing, 400044, China
b
State Grid Pingyang Power Supply Company, Wenzhou, 325401, China
c
College of Computer Science, Chongqing University, Chongqing, 400044, China

a r t i c l e i n f o a b s t r a c t

Article history: Eco-friendly technologies regarding electricity production are urgent need of sustainable energy
Received 27 November 2018 development. The renewable energy resources (RERs) have brought green revolution in mitigation of
Received in revised form greenhouse gaseous emission resulted from traditional energy resources (TERs). Moreover, the effective
21 January 2019
utilization of these resources is influenced by pricing schemes which have limitations. Therefore, this
Accepted 7 March 2019
paper aims at optimization modeling for dynamic price-based demand response (DR) which includes
Available online 11 March 2019
flexible and inflexible loads along with the effective utilization of RERs i.e. photovoltaics (PVs) and wind
turbines (WTs) in a microgrid (MG). The optimization problem regarding profit maximization for loads
Keywords:
Dynamic pricing scheme
(flexible and inflexible) is solved via particle swarm optimization (PSO). Two cases are used to evaluate
Demand response the performance of proposed dynamic pricing scheme. The simulation results have shown that proposed
Microgrid scheme is suitable in term of profit and comfort for flexible and inflexible loads as compared to fixed
Renewable energy resources pricing scheme in both cases. In addition, the dynamic pricing scheme is exemplified as plug and play
Particle swarm optimization devices because of its easy implementation in present market structure without any modification.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction fruitful in term of energy conservation, it has also increased the life
time and efficiency of power grid by satisfying load demand and
The increasing demand of energy and its high production results ultimately gaseous emission is reduced (Palensky and Dietrich,
in scarcity of input resources and increment of environmental 2011). The DSM is sub-categorized into following programs; con-
pollution in term of gaseous emission. Therefore, energy conser- servation and energy efficiency, fuel replacement, residential or
vation/production at cost of pollution free environment is urgent commercial load management and demand response (DR).
need of time. Consequently, upcoming energy system should be DR, which is an important feature of smart grids, interacts as
economical, reliable and efficient. On the stance of these facts, bridge between utilities and system operators to allow them to
smart grid is developed to overcome these problems (Gungor et al., update the electricity infrastructure such as cost reduction by
2013; Hashmi et al., 2011). Later on, microgrid (MG) is introduced in replacing the traditional resources with intermittent one (such as
a smart grid as a key element to improve its efficiency (Lasseter, solar and wind energy), and also facilitates the users to reduce their
2011). electricity bills (Duong Tung and Le, 2014; Rahimi and Ipakchi,
To sort out energy production (along with conservation of re- 2010). The DR programs have been classified according to various
sources) problem, the researchers tried to manage the customers aspects including (a) control mechanisms/techniques, (b) offered
demand instead of supply stand alone. Hence, the demand side motivations/incentives and (c) decision variables/parameters
management (DSM) strategy is introduced. This strategy is not only (Vardakas et al., 2015). The control techniques regarding (a) are
sub-classified into centralized and decentralized modes (Alagoz
et al., 2012). In both modes, users’ interface with utility for load
* Corresponding author. consumption but differs due to direct (former) and indirect (later)
E-mail addresses: arxhad@yahoo.com, arshadshehzad@cqu.edu.cn interaction. The incentive-based and price/time-based are sub-
(M.A. Shehzad Hassan).

https://doi.org/10.1016/j.jclepro.2019.03.082
0959-6526/© 2019 Elsevier Ltd. All rights reserved.
232 M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241

classes of offered incentives DR. In both sub-classes, consumers are wholesale market with respect to integration of DR bids. The
offered with some incentives which based on their load reduction (Duong Tung and Le, 2015) has proposed a risk-aware stochastic
(Wissner, 2011). In price-based DR (Duy Thanh et al., 2011; Qdr, optimization model to ensure the profit maximization of MG
2006), price of electricity depends upon time of use pricing, real aggregator and their interaction with electricity consumers. This
time pricing and critical peak pricing. During incentive-based DR model has limitation in DR contracts i.e. DR price is fixed rather
(Iwayemi et al., 2011), electricity prices may or may not be fixed than dynamic which depends upon the real operational conditions.
which are influenced by direct load control programs, demand The optimal and practical pricing model has been proposed by
bidding programs, emergency DR programs, interruptible/curtail- (Duong Tung et al., 2016) to integrate the DR in distribution
able rates (I/C) and capacity market programs. The (c) category is network. The load serving entity (LSE) as a leader and DR aggre-
divided into energy management and task-scheduling DR schemes gator as a follower is presented in bilevel programming to create
(Dong et al., 2012). Activation of specific load in accordance with interaction between customers and their LSE. The nonlinear bilevel
pre-defined time is known as task-scheduling DR schemes mixed-integer program is transformed into single MILP through
(Koutsopoulos and Tassiulas, 2012). The energy management DR strong duality theorem and karush-kuhn-tucker (KKT). In this
schemes are achieved by adjusting the energy consumption during problem, the wholesale price for flexible load changes on hourly
peak demand hours. basis. A win-win solution is proposed by this scheme for leader
According to literature wisdom (Cui and Zhou, 2018; Ding et al., (LSE) and follower (DR aggregator). This proposed scheme is
2014; Huang et al., 2017; Xu and Lai, 2015), number of studies on DR beneficial both leader (profit) and follower (payoff). The renewable
have been subjected for industrial and residential (Chavali et al., uncertainties were not focused. In another study (Samadi et al.,
2014; Chen et al., 2012; Conejo et al., 2010; Jovanovic et al., 2016; 2012), the prices have been calculated by using mean power con-
Khalid et al., 2018; Mohsenian-Rad and Leon-Garcia, 2010; Yi sumption of customers at each time interval to enhance the profit
et al., 2013) consumers. The authors in (Xu and Lai, 2015) have of utility. Likewise, previous studies; the uncertainties are also not
practically implemented the time-based DR on industrial scale. included in (Samadi et al., 2012).
According to this implication, pool of industries is categorized via Conclusively, it is found that demand response schemes have
utility into number of groups depending upon DR potential. Specific addressed without proper description (Duong Tung and Le, 2014),
pricing scheme is designed for each individual group based on (Chen et al., 2012), (Conejo et al., 2010), (Anvari-Moghaddam et al.,
optimum load features. On the other hand, DR at residential scale is 2015), (Brahman et al., 2015) or just have fixed pricing scheme
studied through energy management. The DR price in real time (Duong Tung and Le, 2015). For example, inflexible load customers
based on stochastic and robust optimization for residential appli- have higher but constant profit throughout the energy usage than
ances is managed by (Chen et al., 2012). This scheme has considered flexible load customers in fixed pricing scheme. While, real-time
the uncertainties and applied into smart meters which automati- pricing has limitations regarding pricing signal design (either
cally informs the customers to optimize their electricity usage ambiguous design or not mentioned) and also has negative impact
accordingly within 5-min. This study has limitation regarding on small-scale flexible load customers (Chen et al., 2012). Therefore,
pricing signal design. In another study (Yi et al., 2013) have intro- current study is carried out to cope with these problems and silent
duced an optimal stopping scheme for small residential DR to features are given below:
schedule appliances in real time. According to this scheme, set of
various home equipment’s are used according to pre-defined  The dynamic pricing model is designed to increase the profit of
schedule (based on two-stage scheduling algorithm) and ulti- customers (either small or large scale flexible and inflexible
mately results in bill saving. The multi-disciplinary optimization power loads) which is based on renewable energy resources
approaches have been proposed by (Altayeva et al., 2017; Anvari- (PVs and WTs) and main grid elements of the system.
Moghaddam et al., 2015; Brahman et al., 2015) to reduce the elec-  For best validation, the particle swarm optimization (PSO) al-
tricity bills without effect of consumer’s thermal comfort. In gorithm is used to solve the DR optimization problem. Extensive
(Mahboubi-Moghaddam et al., 2018), DR has utilized through numerical results demonstrate that proposed dynamic pricing
bilevel programing to achieve marginal prices by unit commitment scheme has increased the profit of both flexible and inflexible
(upper-level), and by reducing the overall operation cost (lower- load customers as compared to fixed pricing scheme. In addi-
level). tion, the profit is flexible rather than constant. This proposed
A bilevel problem is introduced by (Asimakopoulou et al., 2013), scheme has not only eliminated the negative impact of small-
in which large number of microgrids (MGs) are managed to study scale flexible load customers but also has increased their
the interaction between energy service provider (ESP) and large profit. Another plus benefit of current scheme is its easy
central production unit. This is a step wise process which proceeds implementation in present market structure.
in following order: (a) formation and optimization of energy price
by central production unit, (b) ESP receives price signal from central The structure of this paper is organized as follows: dynamic
production unit after optimization and (c) energy generation and pricing model for DR is presented in section 2. Section 3 consists of
consumption are scheduled by ESP which further optimizes the the objective function and technical constraints. The system ar-
amount of energy from central production unit. However, this pa- chitecture and setups for grid-connected microgrid are introduced
per lacks the interaction between renewable energy sources and in section 4. The numerical results and conclusions are explained in
utility grid. In day-ahead energy markets; to maximize the profit of section 5 and section 6 respectively.
DR aggregator, a time or price based self-scheduling model is
introduced by (Parvania et al., 2013) which based on mixed-integer 2. Dynamic pricing model
linear programming (MILP). Different procurement methods of DR
including consumption of in situ (onsite) generation and energy Recently, renewable energy resources possess high investiga-
storage systems (ESSs), load shifting and load curtailment are tional values as a substituent of utility grid due to its high
considered for small to medium size consumers. In spite of this, DR economical cost and environmentally friendly. Generally, sunlight,
aggregator has some limitations regarding wholesale market wind, geo-thermal, tides, and waves are natural sources of energy
operation because of its direct interaction with wholesale market. generation. Among them, photovoltaics (PVs) and wind turbines
Hence, DR aggregator requires significant modification in (WTs) are widely used renewable energy resources because of its
M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241 233

simple installation. The general layout of grid-connected microgrid  


fix flex
having PVs and WTs along with their loads (flexible and inflexible max profit of loads ¼ Gx ðtÞ  f ðtÞ  Lp þ Lp ðtÞ (2)
loads) are presented in Fig. 1. The grid-connected microgrid is an
automated system in which excessive and shortage of power is fix
Where Gx ðtÞ is the utility function at time t. While, Lp is the fixed
automatically controlled by the host grid (Hassan et al., 2018). load and Lflex
p ðtÞ is the flexible load during time t respectively. The
In literature, numerous studies are used to increase the profits of utility function is defined as total satisfaction of consumer’s com-
utility and customers. For this purpose, various models are fort in term of power consumption at reasonable price. Therefore,
designed such as time-of-use pricing model, critical peak pricing assume consumer’s utility function will have the following features
model and real-time pricing model etc. Among them, real-time (Samadi et al., 2012):
pricing model is the most influential because price varies with
respect to real time which shows dynamic relationship between 1. The utility function should be non-decreasing because cus-
demand and supply (Yang et al., 2018). In practice, the flexible load tomers need to fulfil their maximum desires.
customers must need to lose their comfort during critical time 2. The satisfaction level/utility of consumers regarding consump-
period if they fail to do so then they will be penalty which is higher tion of first unit of electricity should exceeds the utility until the
than the retail price of inflexible load customers. Therefore, we nth unit. At this stage, utility of the users increased smoothly.
have designed the dynamic pricing model based on renewable 3. Zero utility should result from zero power utilization.
energy resources (PVs and WTs) which is not only beneficial for
inflexible load customers but also for flexible load customers in It is concluded from these assumptions that concerned function
term of profit and comfort. The dynamic pricing model is given (utility) should have quadratic relation with linear decreasing
below: marginal profit (Asr et al., 2013; Samadi et al., 2012). The utility
  function is given:
PRESs ðtÞ  k
f ðtÞ ¼ a 1  (1)
Pg ðtÞ þ PRESs ðtÞ Gx ðtÞ ¼ bx  cx2 0  x  xmax (3)

Where f ðtÞ and Pg ðtÞ are the final dynamic price and total active The energy level utilized by users in any time interval is denoted
power output from main grid at time t respectively. The total active by x. While, b and c are the parameters related to behavioral the-
power outputs from RESs (PVs and WTs) at time t is represented by ories of individuals and vary among different individuals.
PRESs ðtÞ. The basic price and constant parameter are represented by
a and k. 3.2. Technical constraints

The following equality (power balance) and inequality (gener-


ation limit) constraints are considered to solve the objective
3. Price based model for demand response function:
The power balance constraint is the basic technical constraint
3.1. Objective function which is considered in this proposed paper. According to this
constraint, the production of electricity from Pg ðtÞ and PRESs ðtÞ is
The purpose of the objective function is to maximize the profit equal to the system demand. The equation of the power balance
for load customers via dynamic pricing model as given below: constraint is given below:

fix flex
Pg ðtÞ þ PRESs ðtÞ ¼ Lp þ Lp ðtÞ (4)

The below technical constraint (generation limit) is also


Grid considered in this paper.

P min max
g ðtÞ  Pg ðtÞ  P g ðtÞ
~
(5)
WT PCC

~ 4. System architecture and setups


PV
~ load
According to Fig. 2, a radial and grid-connected microgrid is
designed in MATLAB/Simulink. Where there are m ¼ 12 distributed
generators (DGs) and loads. Here, load1 to load5 are flexible loads

~~ and load6 to load12 are inflexible loads. Moreover, in the twelve


DGs, DG2, DG3, DG5, DG7, DG10 and DG11 are PVs and DG1, DG4,
load
~
DG6, DG8, DG9 and DG12 are WTs. The PVs and WTs work in
WT
maximum power point tracking (MPPT) control mode. The value of
the constant parameter k is equal to 0.8. The parameters of the DGs
and loads are shown in Table 1 and Table 2 for case I and II.

PV
~ ~ 5. Results and discussions

load The efficiency of the proposed model is compared with fixed


AC Bus pricing scheme which have covered two aspects. According to the
Fig. 1. Grid-connected microgrid. first aspect (case I), dynamic pricing scheme is implemented in
0.4/10.5kV PCC
Grid
DG1
50kW, 0kVar ~ ~ CB1 0.03km

0-48kw
5kVar load1 cb1 0.03km
DG2
40kW, 0kVar
~ CB2

0-52kw cb2
5kVar load2

0.04km

CB3
~ DG3
60kW, 0kVar

0-46kw
0.03km cb3 load3 5kVar

0.05km

~ ~
DG4
40kW, 0kVar
0.03km CB4

0-48kw
cb4 load4
0.05km 5kVar

DG5
50kW, 0kVar
~ CB5

0-52kw cb5
5kVar load5
0.05km
~
DG6 0.03km
60kW, 0kVar
~ CB6
0.03km
15kw
load6 cb6
~
0kVar 0.04km DG7
50kW, 0kVar
CB7

cb7 20kw
load7 0kVar
0.04km
~
DG8

0.03km CB8 ~ 40kW, 0kVar

15kw
DG9
50kW, 0kVar ~ ~ CB9 0.03km
cb8 load8 0kVar

10kw
0kVar load9 cb9 0.05km
DG10
50kW, 0kVar
~ CB10

15kw cb10
0kVar load10
0.03km DG11

CB11
~ 60kW, 0kVar

10kw
0.03kmcb11 load11 0kVar

~
0.04km DG12

0.03km CB12 ~ 50kW, 0kVar

20kw
0.03km cb12 load12 0kVar

Fig. 2. Radial structure of grid-connected microgrid.


M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241 235

Table 1 50
DGs and loads setup for case I.
P1 Q1
Sources Control Capacities Load Type of Load Max. Demand 25
DG1 MPPT 50 kW, 0 kVar Load1 Inflexible 30 kW, 0 kVar
DG2 MPPT 40 kW, 0 kVar Load2 Inflexible 40 kW, 0 kVar 0
DG3 MPPT 60 kW, 0 kVar Load3 Inflexible 35 kW, 0 kVar 40
DG4 MPPT 40 kW, 0 kVar Load4 Inflexible 20 kW, 0 kVar P2 Q2

Active power and reactive power (kW, kVar)


DG5 MPPT 50 kW, 0 kVar Load5 Inflexible 20 kW, 0 kVar 20
DG6 MPPT 60 kW, 0 kVar Load6 Inflexible 15 kW, 0 kVar
DG7 MPPT 50 kW, 0 kVar Load7 Inflexible 20 kW, 0 kVar
DG8 MPPT 40 kW, 0 kVar Load8 Inflexible 15 kW, 0 kVar
0
60
DG9 MPPT 50 kW, 0 kVar Load9 Inflexible 10 kW, 0 kVar
DG10 MPPT 50 kW, 0 kVar Load10 Inflexible 15 kW, 0 kVar P3 Q3
DG11 MPPT 60 kW, 0 kVar Load11 Inflexible 10 kW, 0 kVar 30
DG12 MPPT 50 kW, 0 kVar Load12 Inflexible 20 kW, 0 kVar
0
40
P4 Q4
20
Table 2
DGs and loads setup for case II. 0
50
Sources Control Capacities Load Type of Load Max. Demand P5 Q5
DG1 MPPT 50 kW, 0 kVar Load1 Flexible 0e48 kW, 5 kVar 25
DG2 MPPT 40 kW, 0 kVar Load2 Flexible 0e52 kW, 5 kVar
DG3 MPPT 60 kW, 0 kVar Load3 Flexible 0e46 kW, 5 kVar 0
DG4 MPPT 40 kW, 0 kVar Load4 Flexible 0e48 kW, 5 kVar 60
DG5 MPPT 50 kW, 0 kVar Load5 Flexible 0e52 kW, 5 kVar
P6 Q6
DG6 MPPT 60 kW, 0 kVar Load6 Inflexible 15 kW, 0 kVar
DG7 MPPT 50 kW, 0 kVar Load7 Inflexible 20 kW, 0 kVar
30
DG8 MPPT 40 kW, 0 kVar Load8 Inflexible 15 kW, 0 kVar
DG9 MPPT 50 kW, 0 kVar Load9 Inflexible 10 kW, 0 kVar 0
DG10 MPPT 50 kW, 0 kVar Load10 Inflexible 15 kW, 0 kVar 0 0.5 1 1.5 2 2.5 3 3.5 4
DG11 MPPT 60 kW, 0 kVar Load11 Inflexible 10 kW, 0 kVar Time (sec)
DG12 MPPT 50 kW, 0 kVar Load12 Inflexible 20 kW, 0 kVar
Fig. 3. Active and reactive power outputs of DG1, DG2, DG3, DG4, DG5 and DG6.

microgrids for inflexible loads. While, in second aspect (case II) 50


flexible and inflexible loads are used in both schemes. P7 Q7
25

5.1. Case I: Comparison of dynamic and fixed pricing schemes for 0


inflexible loads 40
P8 Q8
Active power and reactive power (kW, kVar)

5.1.1. Dynamic pricing scheme 20


The active and reactive power outputs of WTs and PVs are
represented in Fig. 3 and Fig. 4, and their reactive power outputs are 0
zero. The active power outputs of WTs (DG1, DG4, DG6, DG8, DG9
50
and DG12) and PVs (DG2, DG3, DG5, DG7, DG10 and DG11) are P9 Q9
dependent on their respective natural resources i.e. wind speed and 25
intensity of the sun light. The gradual increase in active power
outputs of DG4 and DG9 is observed with the passage of time and 0
50
its peak value is analyzed at t ¼ 4s: In DG6 and DG8, the active
power outputs have gradually decreased with time (minimum at P 10 Q10
25
t ¼ 4s). Furthermore, the maximum active power outputs of DG1
and DG12 are also experienced at t ¼ 4s and their trend (active
0
power outputs) are similar to DG4 and DG9. In case of PVs, DG5 and 60
DG7 have maximum and DG2, DG3, DG10 and DG11 have minimum
active power outputs at t ¼ 1:5s: Fig. 5 demonstrates the active and
P 11 Q11
30
reactive power output of the main grid. The main grid supplies its
active power output to loads during shortfall of renewable energy 0
resources. The main grid absorbed excessive active power output 50
generated by renewable energy resources. Fig. 6 demonstrates the P 12 Q12
line voltages and frequency in the MG. The load1 has approximately 25
380 V voltages and 50 Hz frequency throughout the line. In spite of
higher disturbances in environmental conditions, the voltages and 0
0 0.5 1 1.5 2 2.5 3 3.5 4
frequency of all the loads in the MG remain unchanged throughout Time (sec)
the line. The inflexible (fixed) loads (load1 to load12) are shown in
Fig. 7 and Fig. 8. According to these figures, the load demand of Fig. 4. Active and reactive power outputs of DG7, DG8, DG9, DG10, DG11 and DG12.
236 M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241
Active power and reactive power (kW, kVar)

50
150
LP 1 LQ1
Pg Qg 25
100
0
50
50 LP 2 LQ2
25

Active and reactive loads (kW, kVar)


0 0
50
LP 3 LQ3
-50 25

0
-100 50
LP 4 LQ4
25
-150
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) 0
50
Fig. 5. Active and reactive power output of the main grid. LP 5 LQ5
25
Frequency (Hz) Line Voltage (V)

400 0
30
Voltage of Load1
LP 6 LQ6
380 15

0
360 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
50.5
Frequency of MG Fig. 7. Active and reactive power loads of inflexible load (load1 to load6).

50
30
LP 7 LQ7
49.5 15
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
0
Fig. 6. Line voltages and frequency in the MG. 30
LP 8 LQ8
15
customers is fulfilled without losing their comfort. Regarding profit,
Active and reactive loads (kW, kVar)

dynamic pricing explains that profit of inflexible load customers 0


varies with time to time due to price variations. In contrast (Duong 30
Tung et al., 2016), has used fixed price for inflexible load customers LP 9 LQ9
which might not be as profitable as dynamic pricing scheme. In 15
addition, their work is limited to small and medium scale cus-
tomers. In another study (Jiang et al., 2014), the dynamic pricing 0
30
scheme is utilized for small scale inflexible load customers. While,
present study is compatible and profitable for small and large-scale LP 10 LQ10
customers in the distribution network. Fig. 9 and Fig. 10 demon-
15
strates the dynamic price and profit of the inflexible loads. In
addition, these figures explain that profit increased with the pas- 0
30
sage of time due to low cost and maximum profit (5600cents) is
LP 11 LQ11
observed at time interval. t ¼ 2:3s: 15

5.1.2. Fixed pricing scheme 0


The fixed pricing scheme is set at 12cents and their profits can 30
be seen in Fig. 11. The profit of the customers in dynamic pricing LP 12 LQ12
scheme is higher than fixed pricing scheme in case I. Conclusively, it 15
is found that dynamic pricing scheme is beneficial for customer
loads in cases I as compared to fixed pricing scheme. The active and 0
0 0.5 1 1.5 2 2.5 3 3.5 4
reactive power output of the main grid is shown in Fig. 12 during
Time (sec)
fixed pricing scheme. In case of less outputs of the renewable en-
ergy resources, the main grid supplies the active power to the loads. Fig. 8. Active and reactive power loads of inflexible load (load7 to load12).
M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241 237

Active power and reactive power (kW, kVar)


5
Dynamic price 150
4.5 Pg Qg
Dynamic price (cents)

4 100

3.5
50
3
2.5 0
2
-50
1.5
1 -100
0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 -150
Time (sec) 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
Fig. 9. Dynamic pricing scheme.
Fig. 12. Active and reactive power output of the main grid.

5800 In contrast, main grid absorbs the active power outputs of renew-
5600 Profit able energy resources during high power outputs.

5400
5.2. Case II: Comparison of dynamic and fixed pricing schemes-
5200 based demand response
Profit (cents)

5000
In case II, flexible and inflexible loads are used in both schemes
4800 i.e. dynamic and fixed pricing scheme rather than inflexible loads
4600
4400 50
P1 Q1
4200 25
4000
0
3800 40
0 0.5 1 1.5 2 2.5 3 3.5 4 P2 Q2
Active power and reactive power (kW, kVar)

Time (sec) 20
Fig. 10. Profit of the inflexible loads.
0
60
P3 Q3
700 30
Profit
0
40
600 P4 Q4
20
Profit (cents)

0
50
500
P5 Q5
25

0
400 60
P6 Q6
30

300 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) Time (sec)
Fig. 11. Profit of the inflexible loads. Fig. 13. Active and reactive power outputs of DG1, DG2, DG3, DG4, DG5 and DG6.
238 M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241

50 100
P7 Q7 LP 1 LQ1
50
25

Active and reactive loads (kW, kVar)


0
0 100
40 LP 2 LQ2
50
P8 Q8
Active power and reactive power (kW, kVar)

20 0
100
0 LP 3 LQ3
50 50
P9 Q9 0
25 100
LP 4 LQ4
50
0
50 0
P 10 Q10
100
25 LP 5 LQ5
50
0 0
60 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
P 11 Q11
30
Fig. 16. Active and reactive power loads of flexible load (load1 to load5).

0
50
According to these figures, the active power outputs of WTs are
P 12 Q12 directly proportional to wind speed. With the passage of time, the
25
active power outputs are gradually increased in DG4 and DG9, and
decreased in DG6 and DG8. The maximum (DG4 and DG9) and
0
0 0.5 1 1.5 2 2.5 3 3.5 4 minimum (DG6 and DG8) active power outputs are observed at t ¼
Time (sec) 4s: In addition, DG1 and DG12 have similar trend with DG4 and
DG9 but maximum active power outputs are observed at t ¼
Fig. 14. Active and reactive power outputs of DG7, DG8, DG9, DG10, DG11 and DG12.

(case I). In this case, the profit of the load customers in dynamic 30
pricing scheme is compared with fixed pricing scheme. LP 6 LQ6
15

5.2.1. Dynamic pricing scheme 0


30
Respectively in Fig. 13 and Fig. 14, active and reactive power
LP 7 LQ7
outputs of uncontrollable DGs (WTs and PVs) are shown. The 15
reactive power outputs of WTs and PVs are zero.
The WTs consists of DG1, DG4, DG6, DG8, DG9 and DG12. 0
30
Active and reactive loads (kW, kVar)

LP 8 LQ8
15
Active power and reactive power (kW, kVar)

150 0
Pg Qg 30
LP 9 LQ9
100 15

0
50 30
LP 10 LQ10
15
0 0
30
LP 11 LQ11
-50 15

0
-100 30
LP 12 LQ12
15
-150 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) Time (sec)

Fig. 15. Active and reactive power output of the main grid. Fig. 17. Active and reactive power loads of inflexible load (load6 to load12).
M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241 239

4s:Regarding PVs, two different active power outputs trend are the flexible load customers were charged under regular price which
analyzed; DG5 and DG7 are similar, and DG2, DG3, DG10 and DG11 might not be beneficial for them. Although (Asgher et al., 2018;
are identical. In addition, the maximum (DG5 and DG7) and min- Chen et al., 2012; Jiang et al., 2014), have also used dynamic pric-
imum (DG2, DG3, DG10 and DG11) active power outputs can be ing scheme for small residential loads but their findings were not
seen at. t ¼ 1:5s: suitable for large flexible loads. However, the proposed dynamic
The active and reactive power output of the main grid is shown pricing scheme is appropriate for both small- and large-scale
in Fig. 15 which have direct relation with load demand and inverse flexible load customers.
relation with outputs of renewable energy resources. It can be Fig. 19 demonstrates the profit of the flexible loads. The
explaining as inadequate active power outputs of PVs and WTs are maximum load profit (1325cents) is observed at t ¼ 2:2s due to the
resulted in high load demand (flexible and inflexible loads due to low price (3.8cents). The total active flexible loads can be seen from
dynamic price) which are fulfilled through main grid. Hence, the Fig. 20. According to this figure, the total active flexible load varies
increased active power output is observed at initial time period among different time zones based on dynamic price. If the price
(Fig. 15). In addition, 25 kW total reactive power of the flexible loads signal is high, the flexible load will gradually decrease and vice
are supplied by the main grid. versa. In case of more flexible loads, the profit of the flexible loads
The active and reactive power loads of DG1 to DG12 are pre- will have increased with the passage of time.
sented in Fig. 16 and Fig. 17. The load1 to load5 are flexible and
load6 to load12 are inflexible loads. In flexible loads, price and load 5.2.2. Fixed pricing scheme
demand have inverse relation. It can also be seen (Fig. 16) that load The fixed pricing scheme is set at 12cents. According to fixed
demand varies with different time durations; maximum load de- pricing scheme, the load demand is strictly influenced by price i.e.
mand is observed at t ¼ 2s and minimum load demand at the higher price resulted in lower demand. In other words, the load
t ¼ 0  0:5s due to price variations (Fig. 18). Besides flexible load demand is highly reduced by flexible load customers due to higher
customers, it also prejudiced the inflexible load customers without prices during fixed pricing scheme as compared to dynamic pricing
losing their comfort. In previous study (Duong Tung and Le, 2015), scheme. Consequently, the load profit also reduced (Fig. 21) due to

6.5
Dynamic price
6
Dynamic price (cents)

5.5

4.5

3.5

3
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
Fig. 18. Dynamic pricing scheme.
240 M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241

Active power and reactive power (kW, kVar)


1500
150
Profit
Pg Qg
1400
100
1300
Profit (cents)

50
1200

1100 0

1000 -50

900
-100
800
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) -150
0 0.5 1 1.5 2 2.5 3 3.5 4
Fig. 19. Profit of the flexible loads.
Time (sec)
Fig. 22. Active and reactive power output of the main grid.

250
Total active flexible loads low load consumption by flexible customers in fixed pricing
Total active flexible loads (kW)

scheme. In addition, both flexible and inflexible load customers can


240 lose their comfort. Finally concluded, this scheme is not beneficial
as dynamic pricing scheme for both flexible and inflexible load
customers.
230 The active and reactive power outputs of WTs and PVs are same
in both pricing schemes (dynamic and fixed pricing schemes). The
behavior of active and reactive power output of the main grid
220 varies; the injection (Fig. 15) in case of dynamic pricing scheme is
due to continuous load demand and absorption (Fig. 22) in fixed
pricing scheme is because of reduction in load demand.
210
6. Conclusions

200 This paper aims at profit maximization for both flexible and
0 0.5 1 1.5 2 2.5 3 3.5 4 inflexible load customers. For this purpose, the dynamic pricing
Time (sec) scheme for demand response in microgrids is designed which
Fig. 20. Total active flexible loads.
utilizes the renewable energy resources and main grid in efficient
way. The demand response optimization problem in this scheme
has been solved through PSO algorithm. The PSO algorithm is used
because of its higher efficiency and accuracy as compared to others
80 algorithm.
Profit Two main cases are utilized in order to check the performance of
the proposed dynamic pricing scheme through simulation results.
70 In case I, the dynamic and fixed pricing schemes are utilized for
inflexible loads and their results are compared. The simulation
results of case I elaborated that the profit of load customers through
dynamic pricing scheme is higher than fixed pricing scheme. In
Profit (cents)

60
case II, flexible and inflexible loads are used in dynamic and fixed
pricing schemes. In this case, the profit of flexible load customers is
also higher in dynamic pricing scheme as compared to fixed pricing
50
scheme. Comparatively, the dynamic pricing scheme is cost saving
for inflexible load customers as compared to fixed pricing scheme.
In all, it is concluded that dynamic pricing scheme is relatively more
40 beneficial for flexible and inflexible load customers in both cases.
Furthermore, this scheme can easily be implemented in present
market infrastructure without any additional substitution.
30
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) Conflicts of interest

Fig. 21. Profit of the flexible loads. There are no conflicts of interest to declare.
M.A. Shehzad Hassan et al. / Journal of Cleaner Production 222 (2019) 231e241 241

Acknowledgement Duy Thanh, N., Negnevitsky, M., de Groot, M., 2011. Pool-based demand response
exchange-concept and modeling. IEEE Trans. Power Syst. 26 (3), 1677e1685.
Gungor, V.C., Sahin, D., Kocak, T., Ergut, S., Buccella, C., Cecati, C., Hancke, G.P., 2013.
This work was supported in part by the Science and Technology A survey on smart grid potential applications and communication re-
Project of State Grid Zhejiang Electric Power Company (under Grant quirements. Ieee Trans. Ind. Inf. 9 (1), 28e42.
No. 5211W4180001) and in part by the National “111” Project of Hashmi, M., Hanninen, S., Maki, K., 2011. Survey of Smart Grid Concepts, Archi-
tectures, and Technological Demonstrations Worldwide.
China (under Grant No. B08036). Hassan, M.A.S., Chen, M., Li, Q., Mehmood, M.A., Cheng, T., Li, B., 2018. Microgrid
control and protection state of the art: a comprehensive overview. J. Electr. Syst.
References 14 (2), 148e164.
Huang, X., Hong, S.H., Li, Y., 2017. Hour-ahead price based energy management
scheme for industrial facilities. Ieee Trans. Ind. Inf. 13 (6), 2886e2898.
Alagoz, B.B., Kaygusuz, A., Karabiber, A., 2012. A user-mode distributed energy Iwayemi, A., Yi, P., Dong, X., Zhou, C., 2011. Knowing when to act: an optimal
management architecture for smart grid applications. Energy 44 (1), 167e177.
stopping method for smart grid demand response. Ieee Network 25 (5), 44e49.
Altayeva, A., Omarov, B., Im Cho, Y., 2017. Multi-objective optimization for smart Jiang, T., Cao, Y., Yu, L., Wang, Z., 2014. Load shaping strategy based on energy
building energy and comfort management as a case study of smart city plat- storage and dynamic pricing in smart grid. Ieee Trans. Smart Grid 5 (6),
form, high performance computing and communications. In: IEEE 15th Inter- 2868e2876.
national Conference on Smart City; IEEE 3rd International Conference on Data Jovanovic, R., Bousselham, A., Bayram, I.S., 2016. Residential demand response
Science and Systems (HPCC/SmartCity/DSS). 2017 IEEE 19th International
scheduling with consideration of consumer preferences. Applied Sciences-Basel
Conference on. IEEE, pp. 627e628. 6 (1).
Anvari-Moghaddam, A., Monsef, H., Rahimi-Kian, A., 2015. Optimal smart home
Khalid, A., Javaid, N., Guizani, M., Alhussein, M., Aurangzeb, K., Ilahi, M., 2018. To-
energy management considering energy saving and a comfortable lifestyle. Ieee wards dynamic coordination among home appliances using multi-objective
Transactions on Smart Grid 6 (1), 324e332. energy optimization for demand side management in smart buildings. Ieee
Asgher, U., Rasheed, M., Al-Sumaiti, A., Rahman, A., Ali, I., Alzaidi, A., Alamri, A.J.E.,
Access 6, 19509e19529.
2018. Smart energy optimization using heuristic. Algorithm Smart Grid Integr. Koutsopoulos, I., Tassiulas, L., 2012. Optimal control policies for power demand
Sol. Energy Sources 11 (12), 3494.
scheduling in the smart grid. IEEE J. Sel. Area. Commun. 30 (6), 1049e1060.
Asimakopoulou, G.E., Dimeas, A.L., Hatziargyriou, N.D., 2013. Leader-follower stra- Lasseter, R.H., 2011. Smart distribution: coupled microgrids. Proc. IEEE 99 (6),
tegies for energy management of multi-microgrids. Ieee Transactions on Smart 1074e1082.
Grid 4 (4), 1909e1916. Mahboubi-Moghaddam, E., Nayeripour, M., Aghaei, J., Khodaei, A.,
Asr, N.R., Zhang, Z., Chow, M.-Y., 2013. Consensus-Based Distributed Energy Man- Waffenschmidt, E., 2018. Interactive robust model for energy service providers
agement with Real-Time Pricing, 2013. Ieee. Ieee Power and Energy Society
integrating demand response programs in wholesale markets. Ieee Trans. Smart
General Meeting. Grid 9 (4), 2681e2690.
Brahman, F., Honarmand, M., Jadid, S., 2015. Optimal electrical and thermal energy Mohsenian-Rad, A.-H., Leon-Garcia, A., 2010. Optimal residential load control with
management of a residential energy hub, integrating demand response and price prediction in real-time electricity pricing environments. Ieee Trans. Smart
energy storage system. Energy Build. 90, 65e75. Grid 1 (2), 120e133.
Chavali, P., Yang, P., Nehorai, A., 2014. A distributed algorithm of appliance sched-
Palensky, P., Dietrich, D., 2011. Demand side management: demand response,
uling for home energy management system. Ieee Trans. Smart Grid 5 (1), intelligent energy systems, and smart loads. Ieee Trans.Ind. Inf. 7 (3), 381e388.
282e290.
Parvania, M., Fotuhi-Firuzabad, M., Shahidehpour, M., 2013. Optimal demand
Chen, Z., Wu, L., Fu, Y., 2012. Real-time price-based demand response management response aggregation in wholesale electricity markets. Ieee Trans. Smart Grid 4
for residential appliances via stochastic optimization and robust optimization. (4), 1957e1965.
Ieee Trans. Smart Grid 3 (4), 1822e1831. Qdr, Q.J.U.D.E., 2006. Benefits of Demand Response in Electricity Markets and
Conejo, A.J., Morales, J.M., Baringo, L., 2010. Real-time demand response model. Ieee Recommendations for Achieving Them. Washington, DC, USA, Tech. Rep.
Trans. Smart Grid 1 (3), 236e242.
Rahimi, F., Ipakchi, A., 2010. Demand response as a market resource under the
Cui, H., Zhou, K., 2018. Industrial power load scheduling considering demand smart grid paradigm. Ieee Trans. Smart Grid 1 (1), 82e88.
response. J. Clean. Prod. 204, 447e460. Samadi, P., Mohsenian-Rad, H., Schober, R., Wong, V.W.S., 2012. Advanced demand
Ding, Y.M., Hong, S.H., Li, X.H., 2014. A demand response energy management side management for the future smart grid using mechanism design. Ieee Trans
scheme for industrial facilities in smart grid. Ieee Trans. Ind. Inf. 10 (4), Smart Grid 3 (3), 1170e1180.
2257e2269.
Vardakas, J.S., Zorba, N., Verikoukis, C.V., 2015. A survey on demand response
Dong, Q., Yu, L., Song, W.-Z., Tong, L., Tang, S., 2012. Distributed demand and programs in smart grids: pricing methods and optimization algorithms. Ieee
response algorithm for optimizing social-welfare in smart grid. Ieee. In: 2012
Commun. Surv. Tutorials 17 (1), 152e178.
Ieee 26th International Parallel and Distributed Processing Symposium, Wissner, M., 2011. The Smart Grid - a saucerful of secrets? Appl. Energy 88 (7),
pp. 1228e1239. 2509e2518.
Duong Tung, N., Hieu Trung, N., Le, L.B., 2016. Dynamic pricing design for demand Xu, F.Y., Lai, L.L., 2015. Novel active time-based demand response for industrial
response integration in power distribution networks. IEEE Trans. Power Syst. 31 consumers in smart grid. Ieee Trans. Ind. Inf. 11 (6), 1564e1573.
(5), 3457e3472.
Yang, C., Meng, C., Zhou, K., 2018. Residential electricity pricing in China: the
Duong Tung, N., Le, L.B., 2014. Optimal bidding strategy for microgrids considering context of price-based demand response. Renew. Sustain. Energy Rev. 81,
renewable energy and building thermal dynamics. Ieee Trans. Smart Grid 5 (4), 2870e2878.
1608e1620. Yi, P., Dong, X., Iwayemi, A., Zhou, C., Li, S., 2013. Real-time opportunistic scheduling
Duong Tung, N., Le, L.B., 2015. Risk-constrained profit maximization for microgrid for residential demand response. Ieee Trans. Smart Grid 4 (1), 227e234.
aggregators with demand response. Ieee Trans. Smart Grid 6 (1), 135e146.

You might also like