EJEEE

https://doi.org/10.62909/ejeee.2023.003 Edison Journal for Electrical and Electronics Engineering

Article

Tilos Island's ideal microgrid size for wind, solar, and batteries

Yousif Ali Latif Al-Zuhairy 1, * and Faraqid Q. Mohammed 2

1 Dep. of computer and communication engineering, Islamic University of Lebanon, Lebanon;

yl76354@net.iul.edu.lb

2 National School of Electronics and Telecoms, University of Sfax, Tunisia; faraqid@enis.tn

* Correspondence: Tel.: +961-3866468

Abstract: This article describes a power plant that is versatile regarding its modeling and associ-

ated with a Multiple Objectives Particle Swarm optimization in order to determine the optimal size

of each component of the power plant. The simulation is appropriate for a variety of power sources,

storage devices and loads. The method is utilized on a Wind Turbine/ Photovoltaic Device/ Battery

System setup located in Tilos, Greece. The optimization is intended to reduce the expense of the

system and the energy derived from alternative sources that are not renewable. The results produce

a Pareto front that represents the expense of the equipment and the degree of autonomy of the mi-

cro-grid. The most effective solution to a specific expense associated with energy importation is

demonstrated as an example.

Keywords: Particle swarm optimization; hybrid power plants; techno-economic research

1. Introduction

In certain instances, the error value rose to 59.5 percent (depending on the weather

and renovation scenarios combination considered) [1]. The average increase in slope co-

efficient over the course of a decade was between 3.8 and 8 percent, which is consistent

with a drop in the number of heating hours throughout the heating season from 22 to 139

hours (depending on the combination of weather and renovation scenarios considered).

Conversely, function intercept rose by 7.8–12.7% every ten years (depending on the cou-

pled scenarios). The proposed values could be used to adjust the function parameters for

the scenarios taken into account and raise the heat demand estimator's accuracy.

Such a power plant is costly and may not turn a profit if it is not scaled correctly [2].

Numerous methods, including the Genetic Algorithm[3, 4], and the Particle Swarm Algo-

rithm [5, 6], have been used to study this topic in the literature [7]. All of these references,

nevertheless, are concentrated on certain power plant configurations. This paper's meth-

odology employs a 12-variable modeling that may be applied to a variety of micro-grid

layouts [8]. The Multi Objective Particle Swarm Optimization (MOPSO) technique is uti-

lized to reduce the dependence on external energy sources and system costs [9]. Following

optimization, this external energy cost is utilized to determine the optimal plant configu-

ration for a specific location and consumption profile. On the Greek island of Tilos, the

algorithm is used to size a wind and solar power plant connected to a battery bank. The

flexible plant modeling, its configuration for the case under study, its power sources, the

energy conversion components, the energy management plan, and the economic assump-

tions are all covered in the following section. The optimization issue and the MOPSO al-

gorithm are briefly presented in Section 3. Lastly, Section 4 presents the optimization out-

comes.

2. Materials and Methods

2.1 Flexible plant modeling

Citation: Al-Zuhairy, Y.A.L. and

F.Q. Mohammed, Tilos Island's ideal

microgrid size for wind, solar, and

batteries. Edison Journal for electri-

cal and electronics engineering, 2023.

1: p. 11-16.

Academic Editor: Prof. Dr. Sergey

Shabunin

Received: 10/4/2023

Revised: 15/5/2023

Accepted: 25/5/2023

Published: 1/6/2023

Submitted for possible open access

publication under the terms and

conditions of the Creative Commons

Attribution (CC BY) license

(https://creativecommons.org/license

s/by/4.0/).

EJEEE 2023, Vol.1 12 of 16

This work uses a modular approach to modeling. One or more Renewable Energy

Systems (RES) and one External Storage System can be combined to create a broad variety

of plant designs that it can emulate (ESS). The simulated plants must always supply a load

or have a tolerance for a loss of power supply. If the RES power is insufficient, they can

be connected to the main grid or a controlled source, such as a diesel generator, and they

may or may not export the excess energy generated. Nine Generic Conversion Systems

(GCS) provide this flexibility; they can be turned on or off based on the configuration that

is depicted. The algorithm configuration for the plant used as an example in this paper is

shown in Figure 1. It consists of a bank of sodium nickel chloride batteries, a photovoltaic

array, and a wind turbine. The facility has to provide electricity to about 800 people and

is situated at Tilos, Greece [10]. If the energy from the RES is insufficient to supply the

demand, electricity can be imported from the nearby island of Kos via an underwater ca-

ble or generated using a diesel generator.

Figure 1. Procedure arrangement for the plant design

2.2 Renewable Energy Systems

The power output of the PV panels and wind turbine is calculated using in-situ me-

teorological weather data. Prior to optimization, calculations are performed using a uni-

tary installed power, and the result is multiplied by the installed power in reality. The

plant can also import power using a diesel generator or an underwater cable in addition

to these RES [11].

2.3 Power management

To identify the optimal solution, the optimization algorithm requires values that rep-

resent the plant performances. A specific plant's behavior is simulated using weather and

consumption data over an extended period of time in order to assess its performance. In

order to prevent seasonal phenomena and ensure that the timeframe is reflective of the

location, it should be at least one year. It is imperative that the power management method

be sufficiently simple to execute rapidly, given that the optimizer will simulate many con-

figurations. In our instance, WT control converts the power generated by the wind turbine

to the voltage and frequency of the grid. PV inverters are used in a similar manner to

transform the power generated by PV panels [12]. The load is supplied by this electricity.

The remaining power is sent to Charge so that it can be converted to DC and kept in the

battery bank if the RES power is higher than the consumption. If it is feasible, energy is

exported to the main grid when the batteries are full. The batteries are depleted and con-

verted to AC through discharge if the RES are insufficient to power the load. Should that

prove insufficient, the residual energy can be obtained by importing it from either the

diesel generator or the grid. Ultimately, there is a Loss of Power Supply, and the plant is

penalized by the optimizer if the import power limitation prevents it from meeting the

load [13].

EJEEE 2023, Vol.1 13 of 16

To prevent the algorithm from returning a costly solution with almost no energy in-

put, an economic requirement must also be lowered [14].

2.4 Economics

A power plant's cost estimation is a challenging task because there are many eco-

nomic criteria involved, and they can vary greatly. For instance, the price of an installed

PV panel dropped by 83% in just seven years [15]. They are also susceptible to sudden

changes and rely on the location, labor costs, and supplier. The PV panels and inverters

values are taken from [16] and [17], the wind turbine values are from [18], and the battery

values are from [19] and [20]. The prices utilized in this paper are merely illustrative, ac-

cording to the authors, who also emphasize that the paper concentrates on modeling and

optimization techniques. The installation cost of each component varies based on its size.

The equipment lifespan (Year) and the study duration (Year), which is set at 25 years, are

used to determine how many replacements (Year) are needed. After then, the installation

cost multiplied by the actualization rate (R), which reflects the yearly cost volatility, equals

the purchase cost. An annualized cost is calculated by dividing the purchase price by the

length of the study.

( )

BA kS



−

=

(1)

It is estimated that the annual maintenance cost will be a small percentage of the

installation cost. Finally, the yearly cost of a certain piece of equipment is:

1A BA M

C C C



(2)

The yearly cost of each piece of equipment is then added up to determine the Annu-

alized Cost of System (ACS). The second optimization target to be minimized is the ACS.

The optimization outcomes will give rise to a Pareto front since it clashes with the im-

ported energy.

2.5 Particle Swarm Optimization

The goal of the optimization issue is to reduce the imported energy and the Annual-

ized Cost of System (ACS) while ensuring that the Probability of Loss of Power Supply

(Positive). It can be expressed like this:

 



arg(min) ( )

( ) 0

x and X x ACS x

LPSP x

Find such as =

(3)

Where V is the vector that defines the study domain, 𝑋𝑋 the wind turbine nominal

power, PV array peak power, PV inverter rated power, and battery bank capacity. Multi-

Objective Particle Swarm Optimization (MOPSO) is used to address this four-parameter

optimization issue [21]. Large-scale issues can be resolved using this stochastic approach

that lacks gradients. It functions by shifting the particles in the research domain, which

stand in for different plant arrangements. The particle velocity is determined by the plant

performances, namely 𝐴𝐴𝐴𝐴𝐴 and 𝐸𝐸𝐼𝐼𝐴𝐴, in order for them to approach the best possible

solution, if any. A list of nondominated plants in the form of a Pareto front is produced

by the method.

3. Results

In Figure 2, the dark blue line represents the algorithm's solutions. With its 𝐴𝐴𝐴𝐴𝐴𝐴

and the percentage of imported energy, each point in the graph represents an ideal plant:

𝐸𝐃𝐼𝐼𝐼𝐼 𝐸𝐸𝐿𝐿𝐎𝐶 ⁄. The installed arrangement is indicated by the blue circle, while the bat-

tery bank capacity, PV array peak power (yellow), and wind turbine nominal power

(green) are represented by the thinner lines (light blue). These findings suggest that wind

power should be the main energy source. The plant depends entirely on wind and imports

for energy over twenty-five percent; neither solar power nor energy storage is used. It

becomes profitable to increase the size of the PV array below this point. Nevertheless, this

EJEEE 2023, Vol.1 14 of 16

electricity should be connected to a larger storage unit because it is unavailable at night.

This results in a sharp rise in price and pricey autonomous gains.

Figure 2. Procedure optimization outcome for a Tilos

The cost of importing energy and the cost of producing energy with the diesel gen-

erator now determine the best course of action. These expenses ought to be handled inde-

pendently and might change every hour. They are assumed to be constant and equal for

the sake of example so that the data can be presented in a three-dimensional graph. Let C

be the cost of purchasing energy (i.e. the importation and the diesel cost). The ratio of

annual expenditures to energy consumption for each plant is known as the production

cost 𝐶𝐶𝑝𝑝.

1BA

ACS C E

(4)

The plant in the Pareto front that minimizes 𝐶𝐶𝑝𝑝 is plotted in red in Figure 3a for a

given 𝐶𝐶𝐵𝐵. Ultimately, assuming 𝐶𝐶𝐵𝐵 specifies the ACS, the imported energy, and the

production cost and helps determine the ideal plant component size. These findings are

summarized on the same figure in Figure 3b. With instance, the ideal plant for VL = 220

€/MW.h consists of a 1 MW wind turbine, a 200 kWp PV array with connected inverters

that provide 350kW of nominal electricity, and a 400 kW.h power bank. This plant has a

production cost of 87 €/MW.h. and an energy autonomy of 80%. In Figure 2, this setup is

indicated by a red circle. For somewhat better performances than its real equivalent, it

needs less storage and more wind power.

Figure 3a production cost depending on 3bAnnualized Cost of System

the optimal solution Figure

4. Conclusions

EJEEE 2023, Vol.1 15 of 16

This work presents an algorithm that can optimize the component sizes of a hybrid

renewable power plant that is connected to a storage system. The plant model is flexible

enough to accommodate different setups and accounts for the equipment's nonlinear cost

as well as power-dependent efficiency. Utilizing a Multi Objective Particle Swarm tech-

nique, the optimization issue is resolved.

On the Greek island of Tilos, the algorithm has been deployed to a wind turbine,

photovoltaic array, and battery bank power plant. The goals are to decrease the annual-

ized cost of the system and the imported energy in order to avoid assuming an energy

importation cost prior to the optimization. The best option for various importation costs

can be calculated after the Pareto front is found. The ACS, the imported energy, the cost

of producing energy, and the size of each plant component (WT size, PV installed power,

inverter nominal power, and storage capacity) make up the solution.

The optimization problem has been effectively resolved by this algorithm. It can be

enhanced by putting into practice a more effective energy management plan, having the

option to deploy numerous storage units, having more accurate power source models, or

having storage eventually age. To provide more practical answers, the economic factors

for the component purchasing and maintenance expenses also need to be improved.

Conflicts of Interest: Declare conflicts of interest or state “The authors declare no conflict of

interest.”

EJEEE 2023, Vol.1 16 of 16

References

1. Bahramian, M. and K. Yetilmezsoy, Life cycle assessment of the building industry: An overview of two decades of

research (1995–2018). Energy and Buildings, 2020. 219: p. 109917.

2. Sijm, J., K. Neuhoff, and Y. Chen, CO2 cost pass-through and windfall profits in the power sector. Climate policy,

2006. 6(1): p. 49-72.

3. Sefrioui, M. and J. Périaux. A hierarchical genetic algorithm using multiple models for optimization. in International

Conference on Parallel Problem Solving From Nature. 2000. Springer.

4. Sivanandam, S., et al., Genetic algorithm optimization problems. Introduction to genetic algorithms, 2008: p. 165-

209.

5. Wang, D., D. Tan, and L. Liu, Particle swarm optimization algorithm: an overview. Soft computing, 2018. 22: p.

387-408.

6. Gad, A.G., Particle swarm optimization algorithm and its applications: a systematic review. Archives of

computational methods in engineering, 2022. 29(5): p. 2531-2561.

7. Schutte, J.F., et al., Parallel global optimization with the particle swarm algorithm. International journal for

numerical methods in engineering, 2004. 61(13): p. 2296-2315.

8. Abazari, A., et al., Wind turbine participation in micro‐grid frequency control through self‐tuning, adaptive fuzzy

droop in de‐loaded area. IET Smart Grid, 2019. 2(2): p. 301-308.

9. Bazmi, A.A. and G. Zahedi, Sustainable energy systems: Role of optimization modeling techniques in power

generation and supply—A review. Renewable and sustainable energy reviews, 2011. 15(8): p. 3480-3500.

10. Kaldellis, J.K., Supporting the clean electrification for remote islands: The case of the greek tilos island. Energies, 2021.

14(5): p. 1336.

11. Duchaud, J.-L., et al., Multi-Objective Particle Swarm optimal sizing of a renewable hybrid power plant with storage.

Renewable Energy, 2019. 131: p. 1156-1167.

12. Yaramasu, V., et al., High-power wind energy conversion systems: State-of-the-art and emerging technologies.

Proceedings of the IEEE, 2015. 103(5): p. 740-788.

13. Geidl, M. and G. Andersson, Optimal power flow of multiple energy carriers. IEEE Transactions on power

systems, 2007. 22(1): p. 145-155.

14. Chaudhari, K., et al., Hybrid optimization for economic deployment of ESS in PV-integrated EV charging stations.

IEEE Transactions on Industrial Informatics, 2017. 14(1): p. 106-116.

15. Van der Zwaan, B. and A. Rabl, The learning potential of photovoltaics: implications for energy policy. Energy

policy, 2004. 32(13): p. 1545-1554.

16. Moser, D., et al., Identification of technical risks in the photovoltaic value chain and quantification of the economic

impact. Progress in Photovoltaics: Research and Applications, 2017. 25(7): p. 592-604.

17. Hacke, P., et al., A status review of photovoltaic power conversion equipment reliability, safety, and quality assurance

protocols. Renewable and Sustainable Energy Reviews, 2018. 82: p. 1097-1112.

18. Hernández-Callejo, L., S. Gallardo-Saavedra, and V. Alonso-Gómez, A review of photovoltaic systems: Design,

operation and maintenance. Solar Energy, 2019. 188: p. 426-440.

19. Jossen, A., Fundamentals of battery dynamics. Journal of power sources, 2006. 154(2): p. 530-538.

20. Campagna, N., et al., Battery models for battery powered applications: A comparative study. Energies, 2020. 13(16):

p. 4085.

21. Deb, K. and N. Padhye, Enhancing performance of particle swarm optimization through an algorithmic link with

genetic algorithms. Computational Optimization and Applications, 2014. 57: p. 761-794.

EJEEE 2023, Vol.1 17 of 16