See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357128362

A Genetic Algorithm-Based Demand Side Management Program for

Implementation of Virtual Power Plant Integrating Distributed Energy
Resources

Chapter · January 2022

DOI: 10.1007/978-981-16-7076-3_41

CITATIONS READS

6 191

4 authors:

Sarthak Mohanty Subhasis Panda

Siksha O Anusandhan University The Gandhi Institute for Education and Technology
9 PUBLICATIONS 126 CITATIONS 15 PUBLICATIONS 155 CITATIONS

SEE PROFILE SEE PROFILE

Binod Kumar Sahu Pravat Kumar Rout

Siksha O Anusandhan University 259 PUBLICATIONS 3,286 CITATIONS
152 PUBLICATIONS 2,244 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Sarthak Mohanty on 22 December 2021.

The user has requested enhancement of the downloaded file.

A Genetic Algorithm-Based Demand
Side Management Program
for Implementation of Virtual Power
Plant Integrating Distributed Energy
Resources
Sarthak Mohanty, Subhasis Panda, Binod Kumar Sahu,
and Pravat Kumar Rout

Abstract The concept of Virtual Power Plant (VPP) provides flexibility in the opera-
tion of current Smart Grid (SG) based electric power systems. Demand Side Manage-
ment (DSM) programs present a lot of scope in the optimization of the load consump-
tion and curtailing of loads during peak usage periods. VPP and DSM together
enable communication-based dispatching of loads depending upon the consump-
tion pattern of consumers and generation capacity of Distributed Energy Resources
(DER). Various optimization techniques have been proposed for efficient and cost-
effective dispatch schedules within the intended power networks. In this paper, a
Genetic Algorithm (GA) based optimization approach is presented to allow for reduc-
tion of the overall cost of the market price of electricity and reduction of burden during
peak periods of use after integration of DERs and Demand Response (DR) strategy.

Keywords Genetic algorithm · Virtual power plant · Demand side management ·

Demand response

1 Introduction

The present power system scenario has seen a transition from centralized generation
and distribution to the setting up of a smaller-scale interconnected microgrid (MG)
system [1, 2]. Individual MGs allow for localized setup of power generation and
supply systems based upon the load requirement and usage patterns. MGs also allow
for a more flexible, local, and independent operation without compromising on secu-
rity, quality, resiliency, and reliability [3]. To facilitate their operation in handling

S. Mohanty (B) · S. Panda · B. K. Sahu

Department of Electrical Engineering, Siksha ‘O’ Anusandhan University, Bhubaneswar 751030,
India
P. K. Rout
Department of Electrical and Electronics Engineering, Siksha ‘O’ Anusandhan University,
Bhubaneswar 751030, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 469
M. Mishra et al. (eds.), Innovation in Electrical Power Engineering, Communication,
and Computing Technology, Lecture Notes in Electrical Engineering 814,
https://doi.org/10.1007/978-981-16-7076-3_41
470 S. Mohanty et al.

various power demands, Distributed Energy Resources (DERs) are often integrated
with MGs.
These DERs may either be based on conventional energy sources or renewables.
Renewable source-based DERs have increasingly sparked the interest in setting up
demand for greener sources of energy, mainly due to the current energy crisis, their
abundance of generating sources, and for being highly reproducible. DER operation
allows for the utility to compensate for various shortfalls in the meeting of load
demands during peak periods of operation. This brings down the burden on elec-
trical utility and allows for a less stressed network to function, thereby increasing
the reliability of the Distribution Network (DN). These small-scale power generation
units are eventually replacing conventional generation sources due to lower carbon
footprints and government incentivized installation schemes. With the current trend
of the shift from conventional grid systems to smart grid systems, greater emphasis
has been put on the development of communication and dispatch capabilities of these
DER units within MG systems. DER units coupled with Information and Communi-
cation Technologies (ICTs) have enabled faster and more efficient dispatch of power
when and wherever it is required.
The interconnection of these MGs and DERs have allowed for the development of
the Virtual Power Plant (VPP) concept in recent years [4], with optimized integration
of DERs and focus on improvement of ICT and dealing with the intermittent nature of
most of the Renewable Energy Sources (RES). This concept allows for smaller DERs
to participate in coordination with the controllable loads and Energy Storage Systems
(ESSs). It integrates the ICT infrastructure either through a cloud or a stand-alone set
up to connect, monitor, and coordinate the operation of various controllable loads and
operating schedule of loads using Demand Side Management (DSM) techniques [5,
6] through the Internet of Things (IoT) connected loads and devices at the consumer
premises.
Proper scheduling of these DER sources allows for better and lower energy
production costs and curtailment of emission of greenhouse gases, thus maximizing
profits and making the whole supply and distribution network sustainable. VPP with
its ICT features allows for a more market-centric generation of power with the gener-
ating units playing an important part in the regulation of electricity prices in the
market.
Excess power generation can thus be sold in electricity markets, thereby making
the whole system more profitable in operation during times of peak generation and
low demand for power. It also allows for various features to be implemented in the
current conventional MG set up segregating it from the former as in structural and
operational differences as well as the services offered by it. Some of these features
are discussed below:
• Along with the integration of various DERs, ESSs, VPP allows for control over
controllable loads and dispatching of RES to account for unpredictability in the
operation of the MG system [7].
• It gives greater scope for the participation and conglomeration of DERs, ESSs,
and IoT-connected devices in the SG setup.
A Genetic Algorithm-Based Demand Side Management … 471

• They can be coupled with other nearby grid setups using open communication
protocols to allow for the various constituent units to participate in electricity
wholesale markets.
• They can facilitate intercommunication over a larger geographical area and diverse
MG setups by using newer communication protocols and technologies [7].
• They can not only manage the basic operations undertaken by MGs but also make
the overall system offer a higher degree of control and operation by offering
ancillary services [8] like frequency regulation, contingency protocols, Demand
Response (DR) programs [9, 10] like peak load shaving, and load shifting as well
as price regulation of electricity market prices.
In this paper, a VPP setup is implemented for the DSM-based DR program using
Genetic Algorithm (GA) to allow for Load Shifting and Peak Load Shaving along
with the integration of DERs.

2 Genetic Algorithm

The Genetic Algorithm is a search-based algorithm incorporating natural selection

and gene-based heredity. It forms a subdivision of a large area of a computational
process called Evolutionary Computation. It is a probability-based evaluation process
in nature, but on the contrary, they are effective in local random search to identify
the best possible solutions, as historic data makes it more complex and intricate.
The solutions obtained after initialization of the population go through a process
of recombination as well as mutation of genes. This results in the offspring having
different characteristic values as that of the parents and this process continue for
multiple generations. Each chromosome is evaluated using a distinct fitness value
which is calculated using a predefined objective function. The individuals who are
more fit have a greater probability of mating to reproduce more fit individuals. This
entire process results in the creation of fitter individuals having better solutions
in each succeeding generation and the process is continued till the final stopping
criterion is fulfilled.

2.1 Operations Involved in GA

Generally, GA comprises three main operations for obtaining fitness values:

• Selection
• Crossover
• Mutation
472 S. Mohanty et al.

Fig. 1 Genetic algorithm flow chart

The whole process is illustrated in Fig. 1, which starts with the population initial-
ization with chromosomes and genes being encoded. After that, the fitness value of
all the chromosomal genes is evaluated by applying the fitness function.
After the initialization process, the selection of chromosomes having larger fitness
values are selected and then processed through crossover and mutation operations
based on crossover and mutation factor and process. When the iterative process is
completed, better fitness value individuals replace the parent chromosomes and the
whole cycle is repeated until the stopping criterion is fulfilled. The chromosomes
referring to solutions have been processed through the entire process in GA, and the
best individuals having optimal chromosomal values are returned. The whole process
is continued till the solution has been satisfactorily obtained satisfying the allowable
tolerance value.

3 Problem Setup

The problem setup has been based on the scheduling of loads during 24 h’s time
period. The load has been studied before Photovoltaic (PV) Arrays and Wind Turbines
(WT) have been integrated and optimized using single-objective GA to present lower
electricity consumption costs for the consumer and also to reduce peak loads during
DER generation periods.
The developed model presents a VPP comprising three-generation sources: PV.
Arrays, Wind Turbine Power Plants (WTPP), and Conventional Power Plant (CPP)
represented by the electricity grid supply. The power output of the PV and WT-based
sources is simulated using models in SIMULINK and the power output is logged
accordingly. The total electrical power consumption load logged by the PV inverter
system at the premises over 24 h is taken into consideration after averaging out daily
load consumption over a week.
The VPP is responsible for the timely dispatch of DER generation to deliver
contracted energy to customers in hourly blocks and to minimize the power consumed
from the electric grid supply.
A Genetic Algorithm-Based Demand Side Management … 473

4 Problem Formulation

The VPP optimization problem is based on the minimization of the operation cost of
VPP by a reduction in consumption of power from the CPP and ensuring maximum
dispatch possible from DERs. A modified version of the objective function used in
is incorporated as the objective function in this paper.
Objective function


N 
T 
N 
T
 
Profit = (Cu .Pub ) − Cu .Pu + Cdr .Pdr + C pv .Ppv + Cw .Pw (1)
i=1 t=1 i=1 t=1

where profit is in Rs/KWh, C u is the cost of each kWh of utility generation in Rs/kWh,
Pub is the number of units of energy consumed from the utility grid before application
of DSM in kWh, Pu is the number of units consumed from the utility grid after DSM
and GA application in kWh, C dr is the cost of each kWh of DR-based generation
in Rs/kWh, Pdr is the number of units consumed from the utility grid through DR
reduction in kWh, C pv is the cost of each kWh of PV-based generation in Rs/kWh,
Ppv is the number of units consumed from PV plant generation in kWh, C w is the
cost of each kWh of WT-based generation in Rs/kWh and Pw is the number of units
consumed from WT plant generation in kWh.
Subject to:
Constraints
The utility, PV, and WT power consumption must remain between their minimum
and maximum limits.

Pumin ≤ Pu ≤ Pumax ; ∀t ∈ T (2)

min
Ppv ≤ Ppv ≤ Ppv
max
; ∀t ∈ T (3)

Pwmin ≤ Pw ≤ Pwmax ; ∀t ∈ T (4)

Pdr ≤ Pu ; ∀t ∈ T (5)

Equalities

PF G = Pu + Pdr − Ppv − Pw (6)

where, Pumin and Pumax are the minimum and maximum values of power consumption
min max
from the utility in kWh, Ppv and Ppv are the minimum and maximum values of PV
474 S. Mohanty et al.

power generation in kWh, Pwmin and Pwmax are the minimum and maximum values of
WPP power generation in kWh, and PF G is the final net generation from the utility
grid and through reduction of peak load using DR, PV, and WPP generation sources
in kWh.

5 Case Study

The VPP considered in this case study consists of WPP with an installed capacity of
61.8 kW, and PV with an installed capacity of 176.6 kW. Here the PV case study is
done on a grid-tied PV plant installed at OUAT (Odisha University of Agriculture
and Technology), Bhubaneswar, Odisha.
The installed PV system selected for the case study is installed on the building
rooftop of the OUAT premises, Bhubaneswar, Odisha, India. The arrangement
comprises 570 PV modules over a total area of 1760 m2 . The total power gener-
ation capacity is 176.6 kWp based on the campus load requirements. The WAREE
WS-310 multi-crystalline Silicon modules of 310 Wp capacity and conversion effi-
ciency of 15.97% under standard test conditions (STC) are used. The PV modules
are installed with the base at an angle of 10° tilt angle. Table 1 illustrates the technical
specifications of the PV plant and the PV module specifications carried out in this
case study. Figure 2 shows the SIMULINK PV array model implemented in this
paper.
For the WT plant source, a SIMULINK/MATLAB model is implemented using
as reference documentation comprising three individual wind turbine generators of
20 kW each. The maximum aggregated power output from the WPP is 60 kW. For
the choice of simulation, RyseEnergy E-20 HAWT is chosen as a test case model.
The technical specifications of the chosen WT system are given in Table 2. The
SIMULINK model of E-20 HAWT is shown in Fig. 3.
The VPP is on a bilateral contract with customers on the electricity market and
is contracted to deliver energy at a preset market price after the integration of DER
energy production. The power delivery is on a contractual basis of an hourly period.
The RES sources produce maximum power as they can deliver at any given time and
the remaining is covered by the CPP if required. Figure 4 shows the hourly electricity
consumption of the case study logged by the PV inverter system before integration
of DSM principles, PV, and WPP sources.
The hourly market prices for the utility are obtained using the Indian Energy
Exchange’s (IEX) Day-Ahead Market Data Tool available on IEX website with the
average data chosen according to the time frame taken for the load consumption data.
The hourly market prices at Market Clearing Price (MCP) is applied in the utility
contract power delivery is illustrated in Fig. 5.
A Genetic Algorithm-Based Demand Side Management … 475

Table 1 Technical
Parameters Specifications
specifications of 176.6 kW
PV plant and PV module No. of connected PV 570
specifications modules
PV panel technology Multi-crystalline Si
Manufacturer of PV module WAREE
No. of series strings 19
No. of parallel strings 30
Rated power of PV Plant 176.7 kWp
Tilt angle 10°
Total area span 1760 m2
Tracking system Perturb and Observe
Installed number of inverter 3
units
Manufacturer of inverter SUNGROW
VOC 45.20 V
ISC 9.14 A
VMP 36.70 V
IMP 8.45 A
Power thermal coefficient (γ) –0.3845
Efficiency % of the module 15.97%
Fill factor 0.75
Dimension of each PV 1960 mm × 990 mm × 40 mm
module
Solar cells per module 72 (12 × 6)
Nominal maximum module 310 Wp
Power
Operating temperature –40 to 85 °C

Fig. 2 PV system model in SIMULINK

476 S. Mohanty et al.

Table 2 Technical
Parameters Specifications
specifications of 20 kW WT
system Generator type Permanent magnet
Configuration Horizontal axis
Manufacturer of WT RyseEnergy
No of blades 3
Blade length 4.5 m
Rotor diameter 9.8 m
Swept area 75.4m2
Nominal rotor speed 120 rpm
Cut-In speed 2 m/s
Rated wind speed 9 m/s
Cut-Out speed 30 m/s
Survival speed 70 m/s
Weight of rotor 1000 kg
Pitch Downward active pitch
Yaw Assisted yaw
Turbine design class IEC 61400-2 Class I
Operating temperature –20 to 50 °C
Lattice 15–36 m
Monopole 18–27 m
Tilt-Up 18–27 m
Rated power 18 kW
Maximum power 20 kW

Fig. 3 WT system model in SIMULINK

A Genetic Algorithm-Based Demand Side Management … 477

Fig. 4 Total power consumption before integration of PV, WPP, and DSM techniques

Fig. 5 Hourly utility market prices

6 Results of Simulation and Discussion

Simulation of PV and WPP is carried out for an entire day time frame based on the
meteorological data supplied by SolCast Database and National Renewable Energy
Laboratory (NREL) datasets. The same is simulated using MATLAB/SIMULINK
and the PV and WPP output are obtained. The power generated from PV and WPP
sources are illustrated in Fig. 6.
Next, GA constraints are defined and the parameters used in the GA are
represented by a function as follows [11]:
 
GA = pi , I, λ, length, FF, NG , e, c, m, S, eqc (7)

where, pi = Initial population size; I = Encoding of chromosomes; λ = Size of

population; length = Chromosome length; FF = Fitness function; NG = Generation
number; e = Elitism function; c = Crossover function; m = Mutation function;
S = Stopping criterion; and eqc = Equality constraints.
478 S. Mohanty et al.

Fig. 6 PV and WPP

generated output power

In the optimization method, the following parameters are assigned to solve the
GA problem: (1) Number of Generations: 400, (2) Population Size: 50, (3) Crossover
Rate: 0.7, (4) Mutation Rate: 0.03, and (5) Elitism Rate: 0.25.
The execution of GA on forecasted hourly values of load resulted in peak shifting
of loads from peak usage periods to a more levelized usage pattern before the inte-
gration of RES. This eventually results in lower utility power purchase costs. The
resultant optimized load pattern through GA is illustrated in Fig. 7.
After completion of the GA optimization, the optimized load values are integrated
with the power generated from PV and WPP, and the units of consumption of utility
are calculated. The final power consumed from the utility is then used to calculate
the final cost of the utility consumption. The reduction in peak load shaving and
peak load shifting operations is calculated using Load Demand Response (LDR).
The comparative graph between final utility consumption and the consumption after
LDR techniques is illustrated in Fig. 8.
The final objective of the optimization is to maximize VPP profits [12]. This has
been achieved by the reduction in the cost of power consumption from CPP and
by supplementation of different pricing mechanisms of RES and DR-based sources.

Fig. 7 Load after

application of DSM
principles using GA
algorithm
A Genetic Algorithm-Based Demand Side Management … 479

Fig. 8 Final utility power

consumption and generation
using LDR

The final difference in the cost of hourly power consumption for the system as a
whole has been illustrated in Figs. 9 and 10 indicate the before-after cost reduction

Fig. 9 VPP cost before and

after GA algorithm and DER
integration

Fig. 10 VPP profit over

unoptimized scheduling
480 S. Mohanty et al.

policies of the system as a whole with integrated DER components and the VPP
profit maximization percentage chart respectively.

7 Conclusion

The objective of this research paper is to provide a profit-maximizing framework

for VPP optimal operation. The main importance to achieve profit maximization is
to influence the shift from CPP-based generation towards non-conventional genera-
tion sources which provide cheaper energy consumption tariff rates and cleaner less
carbon-emitting sources [13]. Based on the provided results, it is observed that.
optimization techniques like GA present convenient algorithms in DSM and DR-
based programs. Other Artificial Intelligence (AI) and Nature-inspired techniques can
induce similar optimization results due to their suitability in dealing with forecasted
market data without distortion of data falling into either local maxima or minima.
Future work will be oriented towards the improvement of the presented optimiza-
tion model by the inclusion of dynamic pricing of CPP and DR reduction price of
utility based on PV and WPP production. Further work can be focused upon the
inclusion of ESS and uncertainty modeling. Universal interaction scheme modeling
of VPP with other market entities can provide greater scope for participation in the
market framework to improve profit maximization/cost reduction policies.

References

1. Pudjianto D, Ramsay C, Strbac G (2005) Microgrids and virtual power plants: concepts to
support the integration of distributed energy resources. Proc Inst Mech Eng Part A: J Power
Energy 222(7):731–741
2. Dulau LI, Abrudean M, Bica D (2014) Applications of virtual power plants approaches. In:
2014 international conference and exposition on electrical and power engineering (EPE). IEEE
3. Pasetti M, Rinaldi S, Manerba D (2018) A virtual power plant architecture for the demand-side
management of smart prosumers. Appl Sci 8(3):432
4. Nikonowicz, ŁB, Milewski J (2012) Virtual power plants-general review: structure, application
and optimization. J Power Technol 92(3):135–149
5. Nosratabadi SM, Hooshmand RA, Gholipour E (2017) A comprehensive review on microgrid
and virtual power plant concepts employed for distributed energy resources scheduling in power
systems. Renew Sustain Energy Rev 67:341–363
6. Esther BP, Sathish Kumar K (2016) A survey on residential demand side management
architecture, approaches, optimization models and methods. Renew Sustain Energy Rev
59:342–351
7. Behrangrad M (2015) A review of demand side management business models in the electricity
market. Renew Sustain Energy Rev 47:270–283
8. Macedo MNQ et al (2015) Demand side management using artificial neural networks in a smart
grid environment. Renew Sustain Energy Rev 41:128–133
9. Mesarić P, Krajcar S (2015) Home demand-side management integrated with electric vehicles
and renewable energy sources. Energy Build 108:1–9
 A Genetic Algorithm-Based Demand Side Management … 481

10. Aghajani GR, Shayanfar HA, Shayeghi H (2017) Demand side management in a smart micro-
grid in the presence of renewable generation and demand response. Energy 126:622–637
11. Shukla A, Pandey HM, Mehrotra D (2015) Comparative review of selection techniques in
genetic algorithm. In: 2015 international conference on futuristic trends on computational
analysis and knowledge management (ABLAZE). IEEE
12. Narkhede MS, Chatterji S, Ghosh S (2013) Multi objective optimal dispatch in a virtual power
plant using genetic algorithm. In: 2013 international conference on renewable energy and
sustainable energy (ICRESE). IEEE
13. Kuzle I, Zdrilić M, Pandžić H (2011) Virtual power plant dispatch optimization using
linear programming. In: 2011 10th international conference on environment and electrical
engineering. IEEE

View publication stats