ANALYZING THE IMPACT OF PHOTOVOLTAIC AND BATTERIE SYSTEMS ON THE LIFE OF A DISTRIBUTION TRANSFORMER

(1)

ANALYZING THE IMPACT OF

PHOTOVOLTAIC AND BATTERIE

SYSTEMS ON THE LIFE OF A

DISTRIBUTION TRANSFORMER

MOHAMED MOHAMED ALI

School of Business, Society and Engineering Course: Degree Project

Course code: ERA401 Credits: 30 ECTS

Program: MSc. Sustainable Energy Systems

Supervisor: Dr. Amir Vadiee Examiner: Dr. Allan Hawas

Costumer: Eskilstuna Energi Och Miljö, Dr. Pietro Elia Campana, FREE Project Date: 2021-06-13

(2)

ABSTRACT

This degree project presents a study case in Eskilstuna-Sweden, regarding the effect of the photovoltaic (PV) systems with battery energy storage system (BESS) on a power distribution transformer, and how they could change the transformer lifespan. For that, an extensive literature review has been conducted, and two MATLAB models were used to simulate the system. One model simulates the PV generation profile, with the option of including battery in the system, and the other one simulates the transformer loss of life (LOL) based on the thermal characteristics. Simulations were using hourly time steps over a year with provided load profile based on utility data and typical meteorological year weather data from SMHI and STRÅNG. In this study, three different scenarios have been put into consideration to study the change of LOL. The first scenario applies various levels of PV penetrations without energy storage, while, the other scenarios include energy storage under different operating strategies, self-consumption, and peak shaving. Similarly, different battery capacities have been applied for the purpose of studying the LOL change. Thus, under different PV

penetrations and battery capacities, results included the variation of LOL, grid power, battery energy status, and battery power. Moreover, results concluded that the PV system has the maximum impact on LOL variation, as it could decrease it by 33.4 %, and this percentage could increase by applying different battery capacities to the system. Finally, LOL

corresponding to the battery under peak shaving strategy varies according to the battery discharge target. As different peak shaving targets were used to control the battery discharge, and hence, study the impact on the transformer and estimate its LOL.

Keywords: Distribution Transformer, PV Penetration, BESS, Self-Consumption, Peak Shaving, HSP, ONAN, Thermal Model, RPF, Paper Insulation, Oil Immersed Transformers.

(3)

PREFACE

This paper has been written as a part of a project with Eskilstuna Energi Och Miljö and presented as a master’s degree project at Mälardalen University in Sweden. For that, I’d like to thank my supervisor Dr. Amir Vadiee for his continuous follow up and guidance, Dr. Bengt Stridh for his valuable suggestions, and special gratitude to Dr. Pietro Elia Campana for his technical assistance and contribution. Finally, I am very grateful for the support I have received from my family and friends.

Västerås-Sweden in May 2021

(4)

CONTENT

1 INTRODUCTION ... 1 1.1 Background ... 2 1.2 Purpose/Aim ... 3 1.3 Research Questions ... 3 1.4 Delimitation ... 3 2 METHOD ... 3 2.1 Data Collection ... 4 2.2 Load Profile ... 4 2.3 Simulation ... 5 2.4 System Description ... 5

2.4.1 Scenario 1 - PV System without BESS ... 6

2.4.2 Scenario 2 - PV System with BESS (Self Consumption) ... 6

2.4.3 Scenario 3 - PV System with BESS (Peak Shaving) ... 6

3 LITERATURE STUDY ... 7

3.1 Types of Transformers ... 7

3.2 Transformer Thermal Behaviour ... 8

3.3 Lifetime of Transformer ... 8

3.4 PV Generation ... 9

3.4.1 Tilt and Azimuth Angels ... 9

3.4.2 Reverse Power Flow (RPF) ... 10

3.5 Battery Energy Storage System (BESS) ... 10

4 CASE STUDY ... 10

4.1 OptiCE Model ... 11

4.1.1 PV System Model ... 11

4.1.2 Battery Model ... 11

4.1.2.1. OPERATIONAL STRATEGY: ... 12

(5)

4.2.1 Assumptions and Input Data ... 13

4.2.2 Initial Conditions ... 13

4.2.3 Paper Aging Calculation ... 13

4.2.4 Relative Aging Rate ... 14

4.2.5 Loss of Life (LOL) ... 14

4.3 The Effect of The Load Profile ... 15

4.4 Power and Angle Control ... 16

4.4.1 PV System without BESS ... 16

4.5 Peak Shaving Strategy ... 19

5 RESULTS ... 20

5.1 Scenario 1 - PV System without BESS ... 20

5.2 Scenario 2 - PV System with BESS (Self Consumption) ... 21

5.2.1 Simulating 500 battery ... 22

5.2.2 Simulating 3000 battery ... 24

5.3 Scenario 3 - PV System with BESS (Peak Shaving) ... 25

5.3.1 Peak shaving at 10 MW ... 25 5.3.1.1. SIMULATING 500 BATTERY ... 26 5.3.1.2. SIMULATING 3000 BATTERY ... 27 5.3.2 Peak shaving at 12 MW ... 29 5.3.2.1. SIMULATING 500 BATTERY ... 30 5.3.2.2. SIMULATING 3000 BATTERY ... 31 5.3.3 Peak shaving at 14 MW ... 32 5.3.3.1. SIMULATING 500 BATTERY ... 33 5.3.3.2. SIMULATING 3000 BATTERY ... 34 6 DISCUSSION ... 36 6.1 Scenario 1 ... 36 6.2 Scenario 2 ... 37 6.3 Scenario 3 ... 38 6.3.1 Peak shaving at 10 MW ... 38 6.3.2 Peak shaving at 12 MW ... 38 6.3.3 Peak shaving at 14 MW ... 39

(6)

7 CONCLUSIONS ... 40 8 SUGGESTIONS FOR FURTHER WORK ... 41 9 REFERENCES ... 42

(7)

LIST OF FIGURES

Figure 1: One year, hourly electricity load profile of the commercial buildings ... 4

Figure 2: Schematic structure of the system under study ... 5

Figure 3: Algorithm for Operation Strategy and Battery management (Campana, Zhang, & Yan, Opti-CE, 2016) ... 12

Figure 4: Annual load imported from the grid without PV and BESS ... 16

Figure 5: Variation of LOL under different PV rated powers (BESS not included) ... 17

Figure 6: Variation of LOL under different azimuth angels (BESS not included) ... 18

Figure 7: The variation of the hourly commertial load when applying PV penetration ... 19

Figure 8: Peak shaving targets applied on the load profile ... 19

Figure 9: The total PV profile generated when simulating the optimum parameters ... 20

Figure 10: The energy taken and given from the grid for a system without a battery ... 21

Figure 11: Variation of the LOL under different battery capacities (self-consumption) ... 22

Figure 12: Energy exchanged with the grid at 500 battery ... 22

Figure 13: Available battery energy for each hour of the year at 500 battery ... 23

Figure 14: Battery charge and discharge on an hourly basis of the year (power output) ... 23

Figure 15: Energy exchanged with the grid at 3000 battery ...24

Figure 16: The amount of energy in the battery for each hour of the year ...24

Figure 18: Variation of the LOL under different battery capacities ...26

Figure 19: Energy exchanged with the grid at 500 battery ...26

Figure 20: The amount of energy in the battery for each hour of the year ... 27

Figure 24: Battery charge and discharge on an hourly basis of the year (power output) ...29

Figure 25: Variation of the LOL under different battery capacities (peak shaving at 12 MW) 29 Figure 26: Energy exchanged with the grid at 500 battery ... 30

Figure 32: Variation of the LOL under different battery capacities (peak shaving at 14 MW). ... 33

(8)

LIST OF TABLES

Table 1: Transformers classifications according to the power range ... 7

Table 2: Thermal Characteristics ... 13

Table 3: Variables used in OptiCE model ... 15

Table 4: Simulation results of LOL for a system without BESS ... 17

Table 5: The optimum variables for simulation ... 18

Table 6: LOL variation under different battery capacities (self-consumption) ... 21

Table 7: LOL variation under different battery capacities (peak shaving at 10 MW) ... 25

Table 8: LOL variation under different battery capacities (peak shaving at 12 MW) ...29

Table 9: LOL variation under different battery capacities (peak shaving at 14 MW) ... 32

NOMENCLATURE

Symbol Description Unit

𝐴𝐴𝑃𝑃𝑃𝑃 PV array area [𝑚𝑚2]

𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Battery capacity [𝑊𝑊ℎ]

𝐺𝐺𝑔𝑔,𝑏𝑏 Global tilted irradiance [𝑊𝑊/𝑚𝑚2]

𝐾𝐾11 Thermal model constant [−]

𝑃𝑃𝑃𝑃𝑃𝑃 PV system hourly power output [𝑊𝑊]

𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Battery Power [𝑊𝑊]

𝑃𝑃𝑙𝑙𝑙𝑙𝑏𝑏𝑙𝑙 Power consumption [𝑊𝑊]

𝑇𝑇𝑆𝑆𝑆𝑆𝑆𝑆 The temperature under standard test conditions [°𝐶𝐶]

𝑇𝑇𝑏𝑏 Ambient temperature [°𝐶𝐶]

𝑉𝑉𝑛𝑛 The relative aging rate at time step (n) [−]

𝑡𝑡0 Average oil time constant [𝑚𝑚𝑚𝑚𝑚𝑚]

𝑡𝑡𝑛𝑛 Time at the interval (n) [𝑚𝑚𝑚𝑚𝑚𝑚]

𝑡𝑡𝑤𝑤 Winding time constant [𝑚𝑚𝑚𝑚𝑚𝑚]

𝛳𝛳ℎ(𝑛𝑛) Hot spot temperature at time step (n) [°𝐶𝐶]

𝛳𝛳𝑏𝑏 Ambient temperature [°𝐶𝐶]

𝛳𝛳𝑙𝑙 Top oil temperature at the load considered [°𝐶𝐶]

(9)

Symbol Description Unit 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Battery efficiency [%]

𝜂𝜂𝑖𝑖𝑛𝑛𝑖𝑖 Inverter efficiency [%]

𝜎𝜎𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Battery self-discharge [%/𝐻𝐻]

𝜎𝜎𝑠𝑠𝑙𝑙 Hourly self-discharge rate [−]

µ Temperature coefficient of output power [%/°𝐶𝐶]

𝐿𝐿𝐿𝐿𝐿𝐿 Transformer loss of life [𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌]

𝑁𝑁 The total number of intervals during the period considered [−]

𝑁𝑁𝐿𝐿𝐶𝐶𝑇𝑇 Nominal operating cell temperature [°𝐶𝐶]

𝑅𝑅 The ratio of load losses at rated current to no-load losses [−] 𝑆𝑆 PV system effective absorbed global solar irradiance [𝑊𝑊/𝑚𝑚2_]

𝑚𝑚 Number of intervals [−]

𝑡𝑡 Time step [𝐻𝐻]

𝑣𝑣 Wind speed [𝑚𝑚/𝑠𝑠]

𝑥𝑥 Oil exponent [−]

𝑦𝑦 Winding exponent [−]

𝛥𝛥𝛳𝛳ℎ(𝑛𝑛) Hot spot to top oil temperature raises at the _{load considered at time step (n)} [𝐾𝐾]

𝛥𝛥𝛳𝛳ℎ𝑟𝑟 Hot spot to top oil temperature raises at rated current [𝐾𝐾]

𝛥𝛥𝛳𝛳𝑙𝑙𝑟𝑟 Top oil temperature rises at rated losses [𝐾𝐾]

𝜂𝜂 Battery bank efficiency [%]

ABBREVIATIONS

Abbreviation Description

[BESS] Battery Energy Storage System

[HST] Hot-spot Temperature

[LOL] Loss of Life

[ONAN] Oil Natural Air Natural

[PV] Photovoltaic

[RPF] Reverse Power Flow

(10)

1 INTRODUCTION

Nowadays technology is playing an important role in our lives, and due to the rapid

development of technology, there is an increase in energy consumption. So far, most of the energy sources being used are fossil fuels, which causes the atmosphere to get affected by the emission of greenhouse gases (GHG). This phenomenon leads to an increase in the average temperature of the earth’s surface. Hence, renewable energy sources are essential as they produce a low level of GHG and provide an important solution when it comes to energy efficiency improvement on the demand side (Queiroz, Amaral Lopes, & Martins, 2020). According to (Awasthi, et al., 2020), it is important to promote the production and

implementation of green technology to limit and reduce environmental degradation. Solar, wind, and biomass are some of the alternative energy sources that must be implemented and used to replace all traditional energy sources. Hence, as solar is one of the most plentiful sources of energy, PV solar panels are used for converting solar energy into electrical energy. Consequently, PV systems usually produce load profile relying on solar energy which can vary according to the daytime or natural events, therefore many of the building with PV systems are connected to the power grid to insure energy stability, as they charge from the installed system and sell surpluses to the grid in case of over-generation registered, or import from the grid when energy generation is not covering the required load (Queiroz, Amaral Lopes, & Martins, 2020).

The same study (Queiroz, Amaral Lopes, & Martins, 2020) reflected another perspective, as a building with an installed PV system connected to the grid could provide an advantage and prolong the distribution transformer’s life, as it can prevent the distribution grid from overloading and effect the distribution transformer which considered one of the most important components in the electricity grid. On the other hand, when the PV system overproduces energy, surpluses got exported to the grid and create a reverse power flow (RPF) which could affect the transformer negatively.

Usually, in electrical power grids, transformers are important to monitor because of their high cost and reliability (Uçar, Bağrıyanık, & Kömürgöz, 2017), as the distribution systems rely on the perception of load impacts on distribution transformer insulation systems. Moreover, (IEEE Guide for Loading Mineral-Oil-Immersed Transformers, 1995) states that the overall life of the transformer depends on the insulation life, as overloading leads to reach limits temperature values which can weaken the dielectric insulating properties of the

insulation material to reduce the life of the transformer. The same source highlighted the calculation of the temperature of the winding hottest spot as it’s considered an important parameter to control temperature overloads. According to (Pezeshki, Wolfs, & Ledwich, 2014), an oil-paper insulation system can be affected by the mechanical, electrical, and thermal stresses, and oil-immersed transformer life depends on the load applied, moisture and oxygen content of the oil, and ambient temperature. Moreover, unbalanced loading leads

(11)

to increase losses, which affect the insulation and consequently reduce the lifetime of the transformer.

Despite different factors like oxidation and moisture could affect the paper insulation degradation, and since these factors could be controlled with the transformer’s oil preservation system, this research is only considering transformer aging caused by heat, moreover, it deals with transformer aging as the mechanical degradation of its paper insulation through thermal stress generated from the current flowing through the windings and leads. As (McBee, 2017) states, the material capability to resist short circuit stress decrease in a direct proportion by reducing the tensile and dielectric strength of the insulation, which could lead to a possible failure during a downstream fault event.

1.1 Background

The impact of different PV penetration levels on a distribution system has been studied under various perspectives and analysed for different applications, moreover, many simulation techniques are utilized to model the behaviour of the system to assess the effect on

distribution transformers. In one hand, (Soleimani, Affonso, & Kezunovic, Transformer Loss of Life Mitigation in the Presence of Energy Storage and PV Generation, 2019) studied the impact of the BESS with PV installed for an electrical vehicles (EV) station, and how the entire system is affecting the transform’s life, considering the battery nominal power and capacity. For that, electricity consumption for different PV and EV penetration cases were simulated for a residential building in Texas, considering the effect of the battery and how it could reduce the aging progress. Result concluded that applying PV and BESS leads to reduce the EV penetration on the distribution grid. While, high penetration of EV without PV and BESS system support, decrease the transformer lifetime dramatically. Moreover, (Affonso & Kezunovic, 2019) carried a study regarding a similar system, coordinating between EVs and BESS to avid transformer overload in a system powered by PV. The research included economic analysis to reduce energy consumption costs and schedule the BESS charge and discharge for the customer using price optimization. Results included in the paper verified the importance of the PV-BESS system, as PV energy supplied reduced the energy purchased from the grid, moreover, BESS shaved peak load demand as it comes to play during peak hours, hence, the system is considered profitable as long as minimum charging fees could cover the cost.

On the other hand, (Bangash, Farrag, & Osman, 2018) investigates the impact of renewable energy high penetration on a low voltage system in the UK, and as a result of that RPF occurs and causes malfunctioning to the grid, and battery system is installed to confine RPF. The same solution concluded by both papers (Jackson, Walker, & Mithulananthan, 2014) & (Unahalekhaka & Sripakarach, 2020) as the last one presented an analysis for the most efficient BESS size to install to reduce RPF in transmission systems and smoother the load curve for the distribution system.

(12)

1.2 Purpose/Aim

The objective of this degree project is to study the impact of commercial buildings integrated with PV and BESS and analyse their effect on the operating lifetime of grid power

distribution transformers.

1.3 Research Questions

• How can different PV and battery configurations affect the lifetime of a power distribution transformer?

• What is the minimum LOL possible for the transformer in different PV and battery configurations?

• Which component has the maximum effect on the system?

1.4 Delimitation

The entire study is based on a specific load profile, and according to one of the research questions, it is required to specify the point which defines the minimum LOL for the

transformer, which, in this case, results are only valid for the acquired load profile. Moreover, the simulation model did not take into consideration the effect of snow on the PV modules production, which could affect the results concluded in this research.

As the entire simulation has been done with the same weather data for one year, the

effectiveness of the proposed method might vary from one year to another, depending on the variation of the climate change. Moreover, the thermal characteristics parameters used in the transformer thermal model were retrieved assuming an oil-immersed transformer cooled with the natural circulation of oil and air. Hence, results are only valid for this specific type of transformers, as other types would have different thermal characteristics. Finally, the aging of the paper insulation of the transformer can be accelerated by high temperature, high water content, oxidization, and acidity. The study conducted in this research considers only the thermal aging factor, hence, the use of the thermal model took place to estimate the loss of life, as other models like the chemical kinetic model, could yield different results.

2 METHOD

To fulfil the objective of this project, an extensive literature review regarding the effect of PV and batteries on the life of transformers was made from different sources. Several scientific databases were used, such as ScienceDirect, IEEE, and Diva, with keywords such as high PV penetration, transformer life, and reverse power flow for the purpose of the research. This

(13)

degree project evaluates how PV systems with BESS affect distribution transformer lifetime by applying different PV penetration levels considering multiple battery sizes, to create different thermal stresses on the distribution transformer, consequently, to give an

estimation of the maximum level of PV penetration on the system. For that, MATLAB, the programming software for mathematical solutions and simulation ability has been used.

2.1 Data Collection

To study the lifetime of the transformer, a thermal model has been developed and simulated by using a commercial load profile supplied from Eskilstuna Energi och Miljö as input data. Moreover, to ensure the simulation of PV with BESS is accurate, the input data for the OptiCE model have been updated. Therefore, SMHI and STRÅNG websites are used to retrieve the weather data for Eskilstuna for the same period as the provided load profile, including climatological data like ambient temperature, radiation, and wind speed, and locational data including latitude, and longitude.

2.2 Load Profile

The load profile is defined as the regular load characteristics of electric demand as a function of time. In this research, commercial buildings load profile is provided by Eskilstuna Energi och Miljö for 2 years and four months, in one-hour resolution. In the simulation, a load profile of one year has been considered to analyse the transformer annual LOL. As Figure.1 shows, the minimum electricity consumption is during the summer, while the consumption increases during wintertime.

Figure 1: One year, hourly electricity load profile of the commercial buildings

0 2 4 6 8 10 12 14 16 18 20 1 259 517 775 1033 1291 1549 1807 2065 2323 2581 2839 3097 3355 3613 3871 4129 4387 4645 4903 5161 5419 5677 5935 6193 6451 6709 6967 7225 7483 7741 7999 8257 8515 Lo ad [M W ] Time [H]

(14)

2.3 Simulation

For the simulation, one year of commercial electric power load data was used to examine the effect of insulation aging. Considering most distribution systems are designed for a particular load profile based on usage patterns. The pattern of electric power demand will change once the PV system is added. For that, two models have been used as follows:

•_{To simulate the PV system with the effect of BESS, a model has been retrieved from the} open-source code OptiCE to generate different PV penetrations and apply different battery

capacities. As the load profile is the main factor affecting the lifetime of the transformer, the integration of PV and BESS will change the loading pattern (Behi, Arefi, Pezeshki, & Shahnia, 2017). Hence, different PV and BESS penetrations would yield the result of different load profiles. Each load profile is used as an input for the transformer thermal model.

•_{To estimate the winding hot-spot temperature (HST), a thermal model has been developed} and provided by Dr. Pietro Elia Campana, as it has been simulated considering thermal characteristics, ambient temperature data, and loading. Different grid load profiles created from the OptiCE model have been used to study the effect on the transformer and estimate its operational life.

2.4 System Description

A simplified system for the commercial facility with PV and battery system for charge and discharge has been created as Figure.2 illustrates. The system is designed to charge the building using energy gathered from the sun and sell surpluses after charging the battery to the grid. Moreover, in case of no electricity is produced from the PV system and the battery is fully discharged, the building charges directly from the grid.

Figure 2: Schematic structure of the system under study

To analyse the behaviour of the transformer, and hence, the effect of the system under the given load profile on the LOL of the transformer, the following scenarios have been taken into consideration to simulate:

Main Grid _Transformer

Commercial Building Energy Source

BESS PV System

(15)

2.4.1 Scenario 1 - PV System without BESS

In this scenario, the battery is not included in the system as it can be shown in Figure. 2 the highlighted part with the dotted line, and basically, this scenario has been conducted to define the optimum tilt angle, azimuth angle, and PV rated power of the system, which consequently, apply the minimum effect on the life of the transformer while covering the given load profile. The following concepts have been followed:

•_{PV charges the building during the daytime according to the load required from the system.} • Under over-production from the PV system, the excess electricity is exported to the grid. • When no solar energy available, the building charges from the main grid.

2.4.2 Scenario 2 - PV System with BESS (Self Consumption)

Here, and after defining the optimum parameters in the first scenario, different battery capacities are simulated to study the effect of the batteries on the system, and whether it is going to reduce or prolong the lifetime of the transformer. The study has considered the following concepts:

• PV charges the building first, then the battery in case an energy surplus occurs.

•_{Under over-production and a charged battery, the excess electricity is exported to the grid.} • When no solar energy available, the building charges from the battery.

• In case of no solar energy and the battery is fully discharged, the building charges from the main grid.

2.4.3 Scenario 3 - PV System with BESS (Peak Shaving)

Similarly, this scenario follows the same concept as the previous one, with a different battery strategy, as a peak shaving value has been defined to control the charge and discharge of the battery to make the battery always charge below the specific value and only discharge above it, to apply peak shaving. The following concepts have been taken into consideration:

• PV charges the building first, then the battery in case an energy surplus occurs.

• Under over-production and a charged battery, the excess electricity is exported to the grid. •_{According to the load, a certain load target is taken to control the battery charge and}

discharge.

•_{When no solar energy available or the PV system is not fully covering the load, the building} charges from the battery, as it performs peak shaving and discharges only when the load is more than the chosen target.

•_{In case of no solar energy and the battery is fully discharged, the building charges from the} main grid.

(16)

3 LITERATURE STUDY

(Soleimani & Kezunovic, 2020) Studied the overuse of EV as it could result in higher

electricity consumption, which could create issues such as power quality degradation, under-voltage, and increased demand on distribution transformers. Typically, distribution

transformers do not have online control, and if they are constantly operated in an overload state, they will age faster, and the risk of failure will increase. In addition to an unplanned outage, their increased frequency of failure will result in higher repair or replacement costs. For that, to provide synthetic data of EVs load in a residential complex and model their stochastic action, Monte Carlo simulation is used. Data for load, temperature, energy price and PV generation are collected, and a one-year case study is created. The results show that using the proposed solution for optimum use of PV generation and BESS would minimize the transformer loss of life caused by EVs. Similarly, in a commercial building parking garage operated by PV, (Affonso & Kezunovic, 2019) offers a smart charging strategy that allows EVs and BESS to work together efficiently, to avoid transformer overloading. The study also shows that the transformer is subjected to overload due to high ambient temperatures, as the EV uncoordinated charging method reaches its thermal limitations during the summer season, hence, rapid deterioration occurs. Results concluded that Transformer overloading and failure can be avoided with the PV-BESS technology, as the anticipated smart charging method of PV generation can help reduce grid energy consumption, while BESS can assist during peak periods.

3.1 Types of Transformers

(Godina, Rodrigues, Matias, & Catalão, 2015) classified transformers according to the power range as it is shown in Table.1. This study is considering a medium power transformer with a power range of 40 MW, assuming a power factor of 0.9.

Table 1: Transformers classifications according to the power range

Type Power Range Phase

Small Distribution

Transformers 50 to 200 kVA Single-phase

Distribution

Transformers 200 to 2000 kVA Three-phase

Cast Resin

Transformers 250 to 4000 kVA Three-phase

Large Distribution

Transformers 2000 to 20000 kVA Three-phase Medium Power

Transformers 30 to 250 MVA Three-phase

Large Power

Transformers 250 MVA- 1000 MVA Three-phase or Single-phase

The same authors classified the cooling method for the oil-immersed transformer to oil natural air natural (ONAN), oil natural air forced (ONAF), oil forced air forced (OFAF),

(17)

oil forced water forced (OFWF), oil-directed water forced (ODAF), and oil-directed water forced (ODWF). This study is considering the ONAN cooling method.

Oil natural air natural (ONAN) is considered the most popular transformer cooling

mechanism, as it uses natural convection to cool the oil. Hot oil flows to the upper part of the transformer tank in this process, while cold oil fills the empty space. When the transformer is filled, the hot oil that flowed to the upper side will dissipate heat in the atmosphere and cool down and sink, causing the transformer oil in the tank to circulate constantly (Godina, Rodrigues, Matias, & Catalão, 2015).

3.2 Transformer Thermal Behaviour

Usually, transformers malfunction occurs when voltage stress exceeds the insulation's dielectric power. As transformer's insulation is made of a special type of paper that could degrade chemically over time, resulting in a loss of dielectric strength and eventual failure. Typically, the rate of deterioration is mainly determined by the operating temperature. Thus, transformers that appear to run at higher loads, with correspondingly higher temperatures, will be predicted to fail faster than those that run at lower loads, considering all other factors are equivalent. Moreover, events as short circuits on the transmission grid can cause brief thermal spikes that are particularly damaging to the insulation. (Hong, Meeker, & McCalley, 2009). Furthermore, as (Queiroz, Amaral Lopes, & Martins, 2020) state, oil and winding temperatures are not evenly distributed, thus, the loading guide for oil-immersed power transformers, (IEC 60076-7, 2005) determines two critical temperatures that influence the transformer's operating conditions. First one, the top-oil temperature (TOT), as it is the hottest point of the oil in the transformer's tank. Second one, the hotspot temperature (HST), as it is the hottest point of the insulating oil within the transformer's windings, which is by necessity higher than the top-oil temperature to a considered load due to the current flowing on the windings.

3.3 Lifetime of Transformer

The lifetime of the distribution transformer depends on the transformer insulation life, as it primarily determined by the transformer loading, which is influenced by the magnitude and efficiency of the transformer load, ambient temperature, moisture content and oxidation level of the oil as (Behi, Arefi, Pezeshki, & Shahnia, 2017) states. The same source mentions that, proper transformer utilization considering loading, ambient temperature, and thermal characteristics is necessary to achieve better output for transformer investment. As a reason, to be able to predict the winding hot-spot temperature and top-oil temperature respectively, a thermal model is required. to achieve this goal.

According to (Freitas, Santos, & Brito, 2018), the normal sizing procedure contemplates the installation of oversized equipment to ensure a lower load level on the transformer in case of additional capacity is needed to allow for potential demand development. Usually, a power

(18)

distribution transformer (PT) is estimated to have an operating lifetime of 30 years.

Moreover, (Behi, Arefi, Pezeshki, & Shahnia, 2017) states that, distribution transformers are one of the most important elements of electric distribution systems, with a life expectancy of 35 years on average. Many distribution transformers have already outlived their planned service life by more than half. (Freitas, Santos, & Brito, 2018) adds, as service time increases, transformer insulation loses its dielectric and mechanical power. This raises the risk of failure and reduces the remaining life. A power transformer's average predicted working life, according to industry guidelines, is about 40 years. It is generally agreed that after this time, the risk of catastrophic failure is extremely high.

3.4 PV Generation

PV electricity calculations are usually done with commercial software like PVsyst or Polysun, which rely on climatic data from Meteonorm in most cases. Meteonorm® 7.2 is based on data from over 8,000 weather stations around the world, five geostationary satellites, and a globally calibrated aerosol climatology using data from 1991 to 2010 for solar irradiance and 2000 to 2009 for temperature. However, with a land area of around 450,000 𝑘𝑘𝑚𝑚2_{, Sweden is}

only defined by irradiance measurements from 12 stations and data from a single geostationary satellite that covers around a third of the world. Most of the commercially available software only allows for single-location simulations, with only a few allowing for batch simulations over a larger area. Therefore, the spatial and temporal climatic data provided by the Swedish Meteorological and Hydrological Institute (SMHI), STRÅNG, and MESAN models are used as inputs in the model (Campana, et al., 2020).

3.4.1 Tilt and Azimuth Angels

(Hafeza, Soliman, El-Metwally, & Ismail, 2017) studied the variation of tilt angles for different solar energy technologies including analysis for the best electricity output for the solar system. The study concluded that the optimum tilt angle is low during summer and spring and high during winter and autumn as it can vary from 0° to more than 30° according to the location and the season. According to (Killinger, et al., 2018), in most cases, the tilt angle range can be up to 50° with the ability to conduct the study with step size 10°. While, (Jacobson & Jadhav, 2018) estimates the optimum tilt angles for every country around the world. The study included Sweden, Stockholm area with a latitude of 59.65° and longitude of 17.95° the optimum tilt angle is approximately 41°. Hence, for this research, tilt angles from 35° to 55° with a step size of 10° would be considered for the study case. As the most

optimum angle is the one that yields a PV production profile that generates a loading pattern with the minimum effect on the transformer lifetime. Similarly, and since the provided data did not include the azimuth angle, a step size of 45° has been taken considering -90° to be east, 90° to be west, and 0° is south (Killinger, et al., 2018).

(19)

3.4.2 Reverse Power Flow (RPF)

(Unahalekhaka & Sripakarach, 2020) defines the reverse power flow according to the increase of penetration of PV systems, which provide more power than load, as the RPF occurs from the delivery into the transmission system has an effect on the overall power system. The same authors studied the optimum size and installation position of BESS to reduce RPF in transmission systems. Results of the study concluded that a BESS would lower the RPF while also improving the smoothing of the delivery load curve. It will also help to reduce energy waste and peak power demand. (Hatta, Asari, & Kobayashi, 2009) proposed batteries energy management to minimize RPF, including load control to reduce the storage capacity for economical purposes as the cost of the storage battery is high. The study

concluded that load control reduces RPF by 19%. Moreover, (Rahman, Aburub,

Moghaddami, & Sarwat) studied the same phenomena and suggested to use a reverse power relay (RPR), as the use of RPR as a defensive scheme not only prevents the power system from damage caused by reverse power flow but also ensures that the load receives continuous power. In this research, RPF will be minimized by controlling the load applied and the

battery capacity.

3.5 Battery Energy Storage System (BESS)

(Soleimani & Kezunovic, 2020) the study is covering the impact of electrical vehicles (EVs) on transformers, considering BESS on the system, as the cost of a battery energy storage device (BESS) has decreased, paving the way for it to be considered a viable alternative for mitigating the negative effects of EVs on different utility assets. BESS may be used for a variety of tasks, including energy arbitrage, frequency and voltage assistance, peak shaving, and traffic management. This research is taking energy storage to help to slow down the accelerated aging of transformers. Therefore, a Powerwall Tesla battery with a capacity of 13500 𝑊𝑊ℎ is taken for simulation (Tesla, 2021) assuming minimum SOC of 0.2, maximum SOC of 1.0, the efficiency of 0.7, and self-discharge of 0.0002.

4 CASE STUDY

The project is regarding a system in Eskilstuna, Sweden, it consists of a transformer under a load of commercial buildings integrated with PV and BESS. The request has been proposed to Dr. Pietro Elia Campana from the grid operator Eskilstuna Energi och Miljö as a task under FREE project required to investigate the effects on a distribution transformer with a high penetration of photovoltaics, and including energy storage, as buildings are supplied by the transformer and connected to the grid. For that two MATLAB models are used.

(20)

4.1 OptiCE Model

According to (Campana, Zhang, & Yan, Opti-CE, 2016), the OptiCE model is an open-source code written in MATLAB language, for simulation and optimization. Also, it could be used for either off-grid or on-grid applications, to design hybrid power systems. The model is

validated by (Campana, et al., 2020), as the development of the model is based on two transposition and PV models, three decomposition models, and five irradiance databases. In this study, the model is used to simulate and analyse the performance of the PV system with the integration of energy storage, and the PV module simulated in this study is assumed as SunPower SPR-E20-327. According to (Campana, et al., 2020), the following models have been used for both PV system and Battery:

4.1.1 PV System Model

The simulation uses input data including the annual load and the climatical data, and calculates the photovoltaic production profiles as an hourly power output from the PV system, by using the following equation, considering the temperature under standard test conditions (𝑇𝑇𝑆𝑆𝑆𝑆𝑆𝑆) equals to 25°𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃= 𝜂𝜂𝑃𝑃𝑃𝑃,𝑆𝑆𝑆𝑆𝑆𝑆�1 +_𝜂𝜂 µ 𝑃𝑃𝑃𝑃,𝑆𝑆𝑆𝑆𝑆𝑆(𝑇𝑇𝑏𝑏− 𝑇𝑇𝑆𝑆𝑆𝑆𝑆𝑆) +_𝜂𝜂 µ 𝑃𝑃𝑃𝑃,𝑆𝑆𝑆𝑆𝑆𝑆 9.5 5.7 + 3.8𝑣𝑣 (𝑁𝑁𝐿𝐿𝐶𝐶𝑇𝑇 − 20) 800 �1 − 𝜂𝜂𝑃𝑃𝑃𝑃,𝑆𝑆𝑆𝑆𝑆𝑆�𝐺𝐺𝑔𝑔,𝑏𝑏� 𝐴𝐴𝑃𝑃𝑃𝑃𝑆𝑆, 𝐸𝐸𝐸𝐸. 1 4.1.2 Battery Model

The battery model depends on another function called (Operational Strategy), as the PV power profile and the battery capacity are fed into the function to calculate the battery SOC profile and the grid power profile depending on the battery model strategy.

The battery SOC for charging and discharging is calculated according to the following equations from (Kanase-Patil, Saini, & Sharma, 2011):

𝑆𝑆𝐿𝐿𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡) = 𝑆𝑆𝐿𝐿𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡 − 1)�1 − 𝜎𝜎𝑠𝑠𝑙𝑙(𝑡𝑡)� + �𝑃𝑃𝑃𝑃𝑃𝑃(𝑡𝑡) −𝑃𝑃𝑙𝑙𝑙𝑙𝑏𝑏𝑙𝑙_𝜂𝜂 (𝑡𝑡)

𝑖𝑖𝑛𝑛𝑖𝑖 � 𝜂𝜂(𝑐𝑐ℎ𝑌𝑌𝑌𝑌𝑎𝑎𝑚𝑚𝑚𝑚𝑎𝑎) 𝐸𝐸𝐸𝐸. 2

𝑆𝑆𝐿𝐿𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡) = 𝑆𝑆𝐿𝐿𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡 − 1)�1 − 𝜎𝜎𝑠𝑠𝑙𝑙(𝑡𝑡)� + �𝑃𝑃𝑙𝑙𝑙𝑙𝑏𝑏𝑙𝑙_𝜂𝜂 (𝑡𝑡)

𝑖𝑖𝑛𝑛𝑖𝑖 − 𝑃𝑃𝑃𝑃𝑃𝑃(𝑡𝑡)� 𝜂𝜂(𝑑𝑑𝑚𝑚𝑠𝑠𝑐𝑐ℎ𝑌𝑌𝑌𝑌𝑎𝑎𝑚𝑚𝑚𝑚𝑎𝑎) 𝐸𝐸𝐸𝐸. 3

While, the battery power for charge and discharge relies on the battery capacity and SOC whether its maximum or minimum, as it’s expressed with the following equations:

𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡) = − �𝑆𝑆𝐿𝐿𝐶𝐶_{100 − �}𝑚𝑚𝑏𝑏𝑚𝑚 𝑆𝑆𝐿𝐿𝐶𝐶(𝑏𝑏−1)₁₀₀∗ (1 − 𝜎𝜎𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏)�� ∗𝐶𝐶_𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

(21)

𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏(𝑡𝑡) = ��𝑆𝑆𝐿𝐿𝐶𝐶(𝑏𝑏−1)₁₀₀∗ (1 − 𝜎𝜎𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏)� −𝑆𝑆𝐿𝐿𝐶𝐶₁₀₀𝑚𝑚𝑖𝑖𝑛𝑛� ∗ 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏∗ 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝐸𝐸𝐸𝐸. 5

4.1.2.1. Operational Strategy:

Figure 3: Algorithm for Operation Strategy and Battery management (Campana, Zhang, & Yan, Opti-CE, 2016) Figure.3 shows the operation strategy at a given time (t), as the PV power production is compared with the load. The battery is charged only if there is an overproduction of PV power. If the battery is fully charged and there is an excess of production, it gets exported to the main grid (hence the minus sign in Eq.4). On the other hand, If the electricity produced from PV is less than the load, the battery is discharged, and If the PV power production is less than the load and the battery is fully discharged, the required load is met using the electric grid.

4.2 Transformer Thermal Model

A MATLAB model has been developed to simulate the lifetime of the transformer, as the insulation deterioration can be considered as a time function of temperature, oxygen concentration, water content, and acidity level. The thermal model considered only the temperature of the insulation as the controlling parameter since the highest deterioration at the highest temperature operating on a specific part. As a result, the winding hot-spot temperature is used to describe the rate of aging. The following part shows the thermal modelling equations according to (IEC 60076-7, 2005), as the hot spot temperature profile was determined using the load and atmospheric weather measurements as inputs into the IEC thermal model (Martin, Goodwin, Krause, & Saha, 2014).

(22)

4.2.1 Assumptions and Input Data

A load profile starting from January 2016 to April 2018 at one-hour interval, has been provided by the grid operator in Eskilstuna. To calculate the hotspot temperature, the transformer type is assumed to be as an oil-immersed, cooled with natural circulation of oil and air (ONAN), taking into consideration a medium to large size power transformer. One year load profile, at one-hour resolution has been taken as an input data for the model, and according to (IEC 60076-7, 2005), the following parameters are taken to estimate the temperature of both top oil and hot spot:

Table 2: Thermal Characteristics Parameter Value 𝜟𝜟𝜭𝜭𝒐𝒐𝒐𝒐 35 𝜟𝜟𝜭𝜭𝒉𝒉𝒐𝒐 45 𝒕𝒕𝟎𝟎 210 𝒕𝒕𝒘𝒘 10 𝒙𝒙 0.8 𝒚𝒚 1.3 𝑹𝑹 6 𝑲𝑲𝟏𝟏𝟏𝟏 0.5 𝑲𝑲𝟐𝟐𝟏𝟏 2 𝑲𝑲𝟐𝟐𝟐𝟐 2 4.2.2 Initial Conditions

The initial conditions can be determined using the following equations based on the previous thermal characteristics and the IEC standards:

𝛳𝛳𝑙𝑙(0) = �1 + 𝐾𝐾 2_𝑅𝑅 1 + 𝑅𝑅 � 𝑚𝑚 ∗ 𝛥𝛥𝛳𝛳𝑙𝑙𝑟𝑟+ 𝛳𝛳𝑏𝑏 𝐸𝐸𝐸𝐸. 6 𝛥𝛥𝛳𝛳ℎ1(0)= 𝑘𝑘21∗ 𝐾𝐾𝑦𝑦∗ 𝛥𝛥𝛳𝛳ℎ𝑟𝑟 𝐸𝐸𝐸𝐸. 7 𝛥𝛥𝛳𝛳ℎ2(0) = (𝑘𝑘21− 1) ∗ 𝐾𝐾𝑦𝑦∗ 𝛥𝛥𝛳𝛳ℎ𝑟𝑟 𝐸𝐸𝐸𝐸. 8

states that, the winding to oil raise of temperature depends on the load factor, and it can be defined by the following formula:

𝐾𝐾 =_{𝑅𝑅𝑌𝑌𝑡𝑡𝑌𝑌𝑑𝑑 𝐶𝐶𝐶𝐶𝑌𝑌𝑌𝑌𝑌𝑌𝑚𝑚𝑡𝑡}𝐿𝐿𝐿𝐿𝑌𝑌𝑑𝑑 𝐶𝐶𝐶𝐶𝑌𝑌𝑌𝑌𝑌𝑌𝑚𝑚𝑡𝑡

4.2.3 Paper Aging Calculation

According to the loading guide for oil-immersed power transformers (IEC 60076-7, 2005), the change in top oil temperature on a specific time step can be calculated by:

(23)

𝑑𝑑𝛳𝛳𝑙𝑙(0)=_𝐾𝐾 𝑑𝑑𝑡𝑡 11∗ 𝑡𝑡𝑙𝑙�� 1 + 𝐾𝐾2_𝑅𝑅 1 + 𝑅𝑅 � 𝑚𝑚 ∗ 𝛥𝛥𝛳𝛳𝑙𝑙𝑟𝑟− [𝛳𝛳𝑙𝑙− 𝛳𝛳𝑏𝑏]� 𝐸𝐸𝐸𝐸. 9

Hence, the temperature at a given time (𝑚𝑚) can be calculated based on (𝑚𝑚 − 1) value as follows:

𝛳𝛳𝑙𝑙(𝑛𝑛) = 𝛳𝛳𝑙𝑙(𝑛𝑛−1)+ 𝑑𝑑𝛳𝛳𝑙𝑙(𝑛𝑛) 𝐸𝐸𝐸𝐸. 10

Hot spot temperature can be calculated by the following: 𝑑𝑑𝛥𝛥𝛳𝛳ℎ1 =_𝐾𝐾 𝑑𝑑𝑡𝑡 22∗ 𝑡𝑡𝑤𝑤[𝑘𝑘21∗ 𝛥𝛥𝛳𝛳ℎ𝑟𝑟𝐾𝐾 𝑦𝑦_{− 𝛥𝛥𝛳𝛳} ℎ1] 𝑑𝑑𝛥𝛥𝛳𝛳ℎ2=_(1/𝐾𝐾𝑑𝑑𝑡𝑡 22) ∗ 𝑡𝑡𝑙𝑙[(𝑘𝑘21− 1) ∗ 𝛥𝛥𝛳𝛳ℎ𝑟𝑟𝐾𝐾 𝑦𝑦_{− 𝛥𝛥𝛳𝛳} ℎ2]

Similarly, as 𝐸𝐸𝐸𝐸. 8, the 𝑚𝑚𝑏𝑏ℎ value for each 𝑑𝑑𝛥𝛥𝛳𝛳ℎ1 and 𝑑𝑑𝛥𝛥𝛳𝛳ℎ2 are calculated by:

𝛥𝛥𝛳𝛳ℎ1(𝑛𝑛)= 𝛥𝛥𝛳𝛳ℎ1(𝑛𝑛−1)+ 𝑑𝑑𝛥𝛥𝛳𝛳ℎ1(𝑛𝑛) 𝐸𝐸𝐸𝐸. 11

𝛥𝛥𝛳𝛳ℎ2(𝑛𝑛)= 𝛥𝛥𝛳𝛳ℎ2(𝑛𝑛−1)+ 𝑑𝑑𝛥𝛥𝛳𝛳ℎ2(𝑛𝑛) 𝐸𝐸𝐸𝐸. 12

Total hot spot temperature rises at time step (n) can be determined by:

𝛥𝛥𝛳𝛳ℎ(𝑛𝑛)= 𝛥𝛥𝛳𝛳ℎ1(𝑛𝑛)− 𝛥𝛥𝛳𝛳ℎ2(𝑛𝑛) 𝐸𝐸𝐸𝐸. 13

From 𝐸𝐸𝐸𝐸. 8 and 𝐸𝐸𝐸𝐸. 11, the hot spot temperature at time step (𝑚𝑚) is given by:

𝛳𝛳ℎ(𝑛𝑛)= 𝛳𝛳𝑙𝑙(𝑛𝑛)+ 𝛥𝛥𝛳𝛳ℎ(𝑛𝑛) 𝐸𝐸𝐸𝐸14

4.2.4 Relative Aging Rate

For a thermally upgraded paper, the relative ageing rate (𝑉𝑉) at time step (𝑚𝑚) can be calculated according to the following equation:

𝑉𝑉𝑛𝑛= 𝑌𝑌� 15000110+273− 15000

𝛳𝛳ℎ(𝑛𝑛)+273� 𝐸𝐸𝐸𝐸. 15

4.2.5 Loss of Life (LOL)

Finally, transformer loss of life for a certain period is calculated with the following equation: 𝐿𝐿𝐿𝐿𝐿𝐿 = � 𝑉𝑉𝑛𝑛∗ 𝑡𝑡𝑛𝑛

𝑁𝑁 𝑛𝑛=1

(24)

4.3 The Effect of The Load Profile

In this part, the transformer thermal model created above has been used to test the actual effect of the load profile, as the given annual load profile shown in Figure.1 has been simulated by the OptiCE model taking into consideration no PV system and no BESS by setting the following operational variables to zero:

Table 3: Variables used in OptiCE model Variable Value Tilt angle 0° Azimuth angle 0° PV rated power 0 𝑀𝑀𝑊𝑊 Battery capacity 0 𝑊𝑊ℎ

Hence, the following Figure.4 shows the amount of the energy imported from the grid, as the minus sign indicates, when applying the variables in Table.3. By importing the generated grid load profile in the transformer thermal model, the result yields LOL of 1.3391 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌, or in another word, the given annual load profile consumes the transformer and decrease its operational life by 1.3391 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌 every year. That means a transformer with an operational life of 30 years is going to live only for 22.4 years. This analysis coincides with (Affonso &

Kezunovic, 2019), as it states that, if the value of the relative aging rate is larger than one, the transformer age is speeding up. Therefore, for the current system to be efficient, the LOL value must be less than or equal to one, as it’s a directly proportional function of the relative aging rate. Moreover, for an assumed transformer life of 30 years, the value of LOL obtained should increase the assumed lifetime, hence the following formula is derived from (Affonso & Kezunovic, 2019):

𝐴𝐴𝑠𝑠𝑠𝑠𝐶𝐶𝑚𝑚𝑌𝑌𝑑𝑑 𝐿𝐿𝑂𝑂𝑌𝑌𝑌𝑌𝑌𝑌𝑡𝑡𝐿𝐿𝑚𝑚𝑌𝑌𝑂𝑂 𝐿𝐿𝑚𝑚𝐿𝐿𝑌𝑌

𝐿𝐿𝐿𝐿𝐿𝐿 ≥ 30 𝑦𝑦𝑌𝑌𝑌𝑌𝑌𝑌𝑠𝑠 𝐸𝐸𝐸𝐸. 17

As 𝐸𝐸𝐸𝐸. 17 indicates, the PV penetration with BESS must impact the system to yield LOL less than 1, and with controlling the variables in Table.3, less values of LOL produce better results.

(25)

Figure 4: Annual load imported from the grid without PV and BESS

4.4 Power and Angle Control

In this part, the first scenario has been used as a base ground to define the optimum PV angles and rated power for the system, for that, the following analysis has been done:

4.4.1 PV System without BESS

To define the angle which applies the most effect on the PV production, and consequently the life of the transformer. According to 3.4.1, a sensitivity analysis is performed considering tilt angle varies between 35° and 55° with a step size of 10° and azimuth varies between -90° and 90° with a step size of 45°, while in each case the rated PV power has been taken within the range of 15-30 MWp with a step size of 5 MWp. Simulation results are presented in the following table for the LOL:

(26)

Table 4: Simulation results of LOL for a system without BESS Azimuth (°) Tilt (°) PV Rated Power (MWp)

15 20 25 30 0 35 45 0.9503 0.9008 0.9561 0.9073 0.8782 0.8936 0.8849 0.901 55 0.9649 0.9161 0.8918 0.9032 45 35 45 0.9989 0.9451 1.009 0.9556 0.9085 0.8910 0.9182 0.8987 55 1.0207 0.9673 0.928576 0.9060 -45 35 45 0.9248 0.8869 0.9275 0.8941 0.8921 0.9740 0.9090 1.0203 55 0.9326 0.90149 0.9211 1.0478 90 35 45 1.0380 0.9821 1.0508 0.9952 0.9387 0.9064 0.9509 0.9168 55 1.0600 1.0043 0.9592 0.9234 -90 35 45 0.9394 0.9109 0.9491 0.9330 0.9391 1.0896 1.0014 1.2973 55 0.9604 0.9566 1.0738 1.5787

As Table.4 shows, results represent the annual LOL of the transformer, as the lifetime has been consumed the least when applying 25 MWp rated PV power at tilt 35° and azimuth 0° to yield the result of 0.8782 Year. Moreover, an increment happens in the transformer lifetime utilization at 30 MWp which reflects RPF at this stage.

To define the most optimum PV-rated power, smaller step sizes have been taken from 15 to 25 MWp at tilt angle of 35° and azimuth of 0°. The simulation yielded 26 MWp as the optimum PV-rated power that causes the minimum LOL for the transformer when applying PV system without BESS with a value of 0.8775 Year. Results can be shown in the following figure:

Figure 5: Variation of LOL under different PV rated powers (BESS not included)

0.865 0.87 0.875 0.88 0.885 0.89 0.895 0.9 0.905 20 21 22 23 24 25 26 27 28 29 30 LO L [ Yea r] PV Rated Power [MWp]

(27)

Next step, to simulate 26 MWp at different azimuth angels from -90° to 90° with a smaller step size of 30° for more accuracy. Results ensure that the minimum LOL of 0.8775 Year is obtained with an azimuth angle of 0° Figure.6.

Figure 6: Variation of LOL under different azimuth angels (BESS not included)

Hence, the following values would be considered to study the effect of different battery capacities to achieve the minimum LOL for the current system:

Table 5: The optimum variables for simulation

Parameter Value

PV Rated Power 26 MWp Tilt Angle 35 Degree Azimuth Angle 0 Degree

Figure.7 shows the difference in the load after applying the values obtained in the previous step, as the negative values refer to RPF. The variables in Table.5 would be simulated under different battery capacities for the other scenarios to study the effect of the battery storage on the system. Moreover, applying energy storage on this load profile is expected to decrease the effect of RPF, and hence, provide even less LOL for the transformer.

0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 -90 -60 -30 0 30 60 90 LO L [ Yea r] Azimuth Angle [°]

(28)

Figure 7: The variation of the hourly commertial load when applying PV penetration

4.5 Peak Shaving Strategy

According to the load profile, three different values are taken to study the effect of the battery on the system, when operating peak shaving strategy. The results of each one is going to be analysed and compared with the other ones. For that, targets of 10 MW, 12 MW, and 14 MW have been set as three cases to apply additional constrain on the battery. Figure.8 shows the target applied on the load with PV penetration. Hence, the battery starts discharging when the PV system is not covering the required load and the load is more than the target.

Figure 8: Peak shaving targets applied on the load profile

-15 -10 -5 0 5 10 15 20 1 267 533 799 1065 1331 1597 1863 2129 2395 2661 2927 3193 3459 3725 3991 4257 4523 4789 5055 5321 5587 5853 6119 6385 6651 6917 7183 7449 7715 7981 8247 8513 Lo ad [M W ] Time [H]

Load Without PV Penetration Load With PV Penetration

-15 -10 -5 0 5 10 15 20 1 267 533 799 1065 1331 1597 1863 2129 2395 2661 2927 3193 3459 3725 3991 4257 4523 4789 5055 5321 5587 5853 6119 6385 6651 6917 7183 7449 7715 7981 8247 8513 Lo ad [M W ] Time [H]

(29)

The green graph shows the change that happens on the load profile after applying the variable in Table.5. The yellow line on the bottom represents the first target, as below this line, the battery always charges and starts discharging only above 10 MW. Similarly, the red line in the middle represents the second target, and in this case, the battery charges more and comes into action less than the previous one. Because the target is higher, and relatively less load is covered. Finally, the black line at the top shows the third target of 14 MW, and in this case, the battery comes into action the least among all scenarios, as it discharges only when the load is at 14 MW or above.

5 RESULTS

This section shows the total PV power generated for all scenarios when simulating the optimum variables in Table.5. Moreover, it also presents the result regarding the grid power for each scenario, while, the battery energy status and the battery power result are presented for the second and the third scenarios, taking into consideration two

different observation points for each one.

5.1 Scenario 1 - PV System without BESS

Figure 9: The total PV profile generated when simulating the optimum parameters

Figure.9 shows the total PV power generated when simulating the variables in Table.5, as the PV production profile is the same for all scenarios. While, Figure.10 shows the energy

imported and exported to the grid, considering positive values for the energy exported to the grid and the negative values for energy imported from the grid.

(30)

-Figure 10: The energy taken and given from the grid for a system without a battery

5.2 Scenario 2 - PV System with BESS (Self Consumption)

Applying different battery sizes in the previous scenario above can change the transformer annual LOL. The simulation has been done considering a battery size of 13500 𝑊𝑊ℎ, and changing the capacity of the battery by multiplying it, as can be presented in the next table:

Table 6: LOL variation under different battery capacities (self-consumption) Multiply 𝟓𝟓𝟎𝟎𝟎𝟎 𝟏𝟏𝟎𝟎𝟎𝟎𝟎𝟎 𝟏𝟏𝟓𝟓𝟎𝟎𝟎𝟎 𝟐𝟐𝟎𝟎𝟎𝟎𝟎𝟎 𝟐𝟐𝟓𝟓𝟎𝟎𝟎𝟎 𝟑𝟑𝟎𝟎𝟎𝟎𝟎𝟎 LOL [Year] 0.8701 0.8629 0.8572 0.8523 0.8488 0.846

Multiply 3500 4000 4500 5000 5500 6000 6500

LOL [Year] 0.8449 0.8437 0.8432 0.8428 0.8426 0.8425 0.8425

Figure.11 shows the variation of LOL under different battery capacities, for that two different points have been taken to observe the graph generated from the simulation and to study the effect of changing the battery capacity on the system. The points taken are 500 and 3000 multiply.

(31)

Figure 11: Variation of the LOL under different battery capacities (self-consumption)

5.2.1 Simulating 500 battery

Figure.12 shows the result of the grid power profile generated when simulating the optimum values with a battery capacity multiplied by 500, operating a self-consumption strategy. While Figure.13 and Figure.14 show the battery energy status and the battery power respectively for the same simulation.

Figure 12: Energy exchanged with the grid at 500 battery

0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86 0.865 0.87 0.875 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 LO L [ Yea r] Number of Batteries

(32)

Figure 13: Available battery energy for each hour of the year at 500 battery

Figure 14: Battery charge and discharge on an hourly basis of the year (power output)

The battery energy status indicates the amount of electricity left in the battery on an hourly basis, considering the minimum and the maximum SOC set for the simulation, while the battery power shows the amount of electricity the battery charge it (when SOC<maximum), or neglect it (when SOC ≥ maximum) and get exported to the main grid. The positive value is electricity charged by the battery, while the negative one is electricity dumped by the battery when it’s fully charged.

(33)

5.2.2 Simulating 3000 battery

Figure.15 shows the result of the grid power profile generated when simulating the optimum values under battery capacity multiplied by 3000, operating a self-consumption strategy. While Figure.16 and Figure.17 show the battery energy status and the battery power respectively for the same simulation.

(34)

5.3 Scenario 3 - PV System with BESS (Peak Shaving)

Similarly, as the second scenario, different battery sizes have been applied while the change of the transformer annual LOL is observed. The simulation has been done considering a battery size of 13500 𝑊𝑊ℎ, and changing the capacity of the battery by multiplying it,

considering peak shaving strategy with three different targets of 10 MW, 12 MW, and 14 MW. For each target, two different points of 500, 3000 battery multiply have been taken to

observe the graph generated from the simulation and to study the effect of changing the battery capacity on the system.

5.3.1 Peak shaving at 10 MW

Different battery sizes have been applied, to observe the change of the transformer annual LOL. The simulation has been done considering a battery size of 13500 𝑊𝑊ℎ, and changing the capacity of the battery by multiplying it, as can be presented in the next table:

Table 7: LOL variation under different battery capacities (peak shaving at 10 MW) Multiply 𝟓𝟓𝟎𝟎𝟎𝟎 𝟏𝟏𝟎𝟎𝟎𝟎𝟎𝟎 𝟏𝟏𝟓𝟓𝟎𝟎𝟎𝟎 𝟐𝟐𝟎𝟎𝟎𝟎𝟎𝟎 𝟐𝟐𝟓𝟓𝟎𝟎𝟎𝟎 𝟑𝟑𝟎𝟎𝟎𝟎𝟎𝟎

LOL [Year] 0.8566 0.8524 0.8487 0.8463 0.8435 0.8414

Multiply 3500 4000 4500 5000 5500 6000 6500

(35)

Figure 18: Variation of the LOL under different battery capacities

5.3.1.1. Simulating 500 battery

Figure.19 shows the result of the grid power profile generated when simulating the optimum values under battery capacity multiplied by 500, operating a peak shaving strategy under the target of 10 MW. While Figure.20 and Figure.21 show the battery energy status and the battery power respectively for the same simulation.

0.82 0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 LO L [ Yea r] Number of Batteries

(36)

Figure 20: The amount of energy in the battery for each hour of the year

5.3.1.2. Simulating 3000 battery

(37)

(38)

LOL [Year] 1.2514 1.2495 1.2481 1.2474 1.2467 1.246

Multiply 3500 4000 4500 5000 5500 6000 6500

LOL [Year] 1.2453 1.2445 1.244 1.2436 1.2426 1.242 1.2413

Figure 25: Variation of the LOL under different battery capacities (peak shaving at 12 MW)

1.236 1.238 1.24 1.242 1.244 1.246 1.248 1.25 1.252 1.254 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 LO L [ % ] Number of Batteries

(39)

5.3.2.1. Simulating 500 battery

(40)

5.3.2.2. Simulating 3000 battery

(41)

LOL [Year] 1.8768 1.8755 1.8743 1.8734 1.8726 1.8719

Multiply 3500 4000 4500 5000 5500 6000 6500

(42)

Figure 32: Variation of the LOL under different battery capacities (peak shaving at 14 MW).

5.3.3.1. Simulating 500 battery

1.862 1.864 1.866 1.868 1.87 1.872 1.874 1.876 1.878 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 LO L [ Yea r] Number of Batteries

(43)

5.3.3.2. Simulating 3000 battery

(44)

(45)

6 DISCUSSION

6.1 Scenario 1

To decrease the LOL of 1.3391 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌 achieved by applying the variables in Table.3, and consequently apply Eq.17, a PV system is considered in this scenario. Table.4 shows the LOL variation when simulating the PV system under different azimuth angles, tilt angles, and PV rated powers. As can be seen, in all simulations, the value of LOL starts at a high value and decreases gradually until a certain point depending on the angles and the rated power, before it starts to rise again. This phenomenon is connected to the amount of energy exchanged with the grid, as by applying a PV rated power of 15 MWp, the LOL result is high because 15 MWp is not enough to cover the load profile, and there is a considerable amount of electricity imported from the main grid to meet the load. Moreover, increasing the PV-rated power gradually leads to a decrease in the amount of electricity needed from the grid and

consequently LOL, until it reaches its minimum value. Further PV penetration leads to an increment of the LOL result because the energy produced is more than the load profile, hence, excess energy gets exported to the main grid creating RPF. This phenomenon happens in all simulations shown in Table.4, except for simulations at azimuth angles 45° and 90°, which indicates a lack of energy production at these angles. The analysis yielded a minimum LOL of 0.8775 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌 achieved at azimuth 0°, tilt 35°, and PV rated power 26 MWp as

presented in Table.5. LOL achieved is almost around 33.4% less than 1.3391 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑠𝑠. Finally, as a result, the simulation generates a grid energy profile presented in Figure.10, as the