“The Netherlands under construction,

(1)

“The Netherlands under construction,

an optimal division of resources within a renewable hybrid system”

Technology and Operations Management / Supply Chain Management MSc Thesis

University of Groningen, Faculty of Economics and Business Newcastle University Business School

Supervisor: prof. dr. R.H. (Ruud) Teunter Second supervisor: dr. R. (Rebecca) Casey Student: Stan Landewé Student number: S2959429/B9060814 Date: 07.12.2020

Word Count: 9668

(2)

Abstract

Within this simulation study, an optimal balance between renewable energy sources will be targeted. A so-called renewable hybrid energy system is constructed, that collectively covers the electricity demand of the Netherlands for 25%, 50%, 75% or 100%. The energy sources that are used, are wind and solar. Excessive energy production is stored via the energy carrier hydrogen. The ultimate goal is to cover the complete demand within the Netherlands, which is tackled stepwise through the renewable energy targets. Moreover, the prices related to the different sources are integrated to see to what extent these renewable energy sources are able to compete with conventional sources. This research finds that offshore wind energy will play a central role in the energy transition, covering large parts of the sustainable electricity generation. The role of onshore wind locations and solar energy becomes more crucial towards the higher sustainability percentages, where the matching of supply and demand is more decisive. Hydrogen turns out to be too expensive with the current efficiency and price ranges during the energy transition, but still plays a significant role in the 100% situation.

(3)

Introduction

With a population that keeps on growing together with an increase of personal income, the demand for energy is growing and is expected to keep on growing in the upcoming years (Zhang et al., 2016). As Treffers et al. (2005) states in case no changes will be made to the production and energy system, a strong increase in greenhouse gas (GHG) emission will occur on a global basis due to growth in energy demand. The economy is expected to keep on growing in the years up to 2050, which consequently influences the demand for energy and materials. In case this energy will still be generated based on conventional resources, the GHG emission will increase significantly over the years. Since purposely slowing down economic growth is not a realistic option, the current population is set for an important challenge: keep the emissions to an acceptable level to avoid harming the environment of future generations. Ram et al. (2020) thereby states that the electricity generation is accountable for 25% of the GHG emission, which makes this sector essential in the energy transition. The European Commission (EC) acknowledges the current trend and has developed the Green Deal. Within this set of policies, the EC targets to be climate neutral in 2050 and decouples the economic growth from the resource use. For energy producers, this means that major changes have to be implemented in the upcoming decades to be able to reach this target.

In order to cope with this transition, countries are at the top of their development and are investigating ways to increase their sustainable energy production. Sustainable resources that are currently operating and running are wind and solar energy. Based on data from Enerdata (2020), wind and solar covered approximately 9% of the global electricity consumption in 2019.

This puts the situation in perspective and stresses the importance of investments in this area.

When solely focusing on Europe, the EWEA expects that 14% of the European electricity demand will be generated by wind power in 2030 (Association, 2011). To do so, European countries are scaling up their sustainable energy production of which a large part will be generated in the offshore sector (Rudion et al., 2010). An example of this is the North Sea Wind Power Hub, which is a large-scale offshore wind project that enables the supply of cost-efficient and ramp-up energy (Hub, 2019).

A remark that should be made, is that resources relying on the weather also have a downside.

Solar radiation and wind both have a variable and unpredictable nature, and should therefore be carefully assessed (Monforti et al., 2014). In the ideal situation, wind and solar energy would complement each other in such a way that storing the energy will be kept to a minimum. In this

(5)

5 way, it enables the cost of energy to be significant lower. However, the unpredictability of those sources is forcing energy producers to store the energy in case fluctuation will take place on either the supply or the demand side.

In order to capture wind and sun energy in a sustainable way, hydrogen storage offers a potential solution. As Ozarslan (2012) states, hydrogen is an energy carrier that can be produced from multiple resources via various energy conversion processes. Hydrogen is currently the only storage system that can offer long-term seasonal storage (Ozarslan, 2012). It is thereby seen as a source which has great potential to provide energy to multiple sectors and enables a low carbon energy system (Zhang et al., 2016). By using such a storage function, abrupt variations can be absorbed and grid stability problems can be prevented (Dufo-López and Bernal-Agustín, 2013).

In order to see how feasible the current developments are, this paper aims to find an optimal balance of solar, wind and hydrogen, which jointly cover the electricity demand of the Netherlands for a chosen percentage. Since the potential of wind and solar energy is already proven, will those resources be taken for granted. An optimal balance with integrated storage will be calculated, and the economic implications will be included as well.

(6)

6

Theoretical Background

Within this section, the renewable energy resources (RES) described in the introduction will be elaborated on. By means of the latest developments on sustainable resources and their storage, a base is formed for the latter part of this research work.

Sustainable resources

With the current decreasing demand for oil and gas, due to the current COVID-19 crisis, multiple parties are seeking opportunities to speed up the energy transition. An increase of the use of renewable energy resources will be essential to achieve a reduction in greenhouse gas emissions (Walker et al., 2016). The current sustainable resources that have a promising future which are included in this research are wind and solar.

Wind and Solar energy

Wind and solar energy are commonly referred to as variable renewable energy resources (VRES). For these resources, both on- and offshore positioning is being used. Currently, companies are mainly looking at offshore opportunities due to the scarcity of available land and an increasing population (Solanki et al., 2017). Especially wind farms are targeting the offshore sector since it brings multiple advantages; wind speed is significantly higher compared to onshore, social acceptance factors such as visibility of the wind farms and noise are less important, easier to find suitable locations and the possibilities to enlarge the wind turbines (Sannino et al., 2006).

Solar panels, on the other hand, are currently mainly located on shore, however there’s also done some investigation to operate offshore. Solanki et al. (2017) states that implementing offshore plants will enhance the solar panel efficiency, improve the cleaning of PV cells and reduce the evaporation losses. Hollow cubes are attached to the PV cells and as such a floating system is created. However, this will need further investigation to see how feasible this idea is.

Solar energy is able to produce electricity directly via the use of Photovoltaic (PV) cells or indirectly by using concentrated solar power (CSP) technology (Hayat et al., 2019). Especially CSP has a promising future due to its high capacity and ability to store energy. Solar energy can thereby be used in multiple industries and with multiple purposes since it is able to provide both heating and energy. The solar industry excels in terms of availability, cost effectiveness, accessibility and capacity in comparison to other RES (Kannan and Vakeesan, 2016). A combination of both sources seems to be the ideal due to seasonal availabilities. Whereas the generation of wind power is stronger in the winter compared to the summer, the generation of solar power works the other way around (Heide et al., 2010).

(7)

7 Although, the use of these VRES is highly promising, their weather dependence is also a serious drawback. Because of this dependence, VRES has an electricity generation that is variable and uncertain over time. In case there will be a period with low amounts of sun and wind, less energy is generated. Moreover, the supply of VRES is uncertain beforehand since it can only be known after realization. Another form of uncertainty is that the supply is location-specific, since the potential of VRES is not necessarily correlated to load centers (Kondziella and Bruckner, 2016). The variability is a main topic of interest and brings various ideas that could possibly cope with this. A challenging part of VRES is to integrate this electricity into traditional power systems, which could be problematic in a way as it leads to suboptimal preparation of the system (Huber et al., 2014). Singh et al. (2010) also highlight the problems of network stability, voltage regulation and power-quality issues due to the high penetration level of aggregated energy systems. These issues are not inescapable but will come at a cost in case the capacity of these sources will be upscaled. It will thereby be essential to integrate multiple energy sources into a grid network to deliver consistent energy (Kannan and Vakeesan, 2016). This could be done with the creation of a renewable hybrid system, which is an energy system consisting of two or more energy sources, power conditioning equipment, a controller and an possible energy storage system (Nema et al., 2009).

Flexibility of power grids

Besides creating a stable base load of energy supply, flexibility of power grids is of importance when coping with this seasonality. This flexibility will be needed in order to achieve a high penetration of both wind and solar energy. When this flexibility is lacking, a large part of the produced energy might be curtailed (Denholm and Hand, 2011). Power grids can cope with variation and uncertainty of VRES generation to a percentage of ten percent, without creating technical problems or having a significant amount of extra cost (Doherty and O'malley, 2005).

However, in case the integrated capacity of VRES will be upscaled, it will be harder to match the supply and demand side (Gallo et al., 2016). Therefore, the flexibility of power grids needs to improve. A power system is seen as flexible when it is able to deal with uncertainty and variability in generation and demand at reasonable additional cost (Ma et al., 2013). Huber et al. (2014) states that flexibility of power grids can be enhanced in the following ways: creation of flexible power plants, storage of energy, integrated demand-side management (DSM), geographic dispersion of generators and extension of interconnections. Within this paper hydrogen will be added as a storage system, which will capture the excessive production of solar and wind and thereby enhance the flexibility.

(8)

8

Hydrogen

To cover the mismatch between supply and demand, energy storage (ES) is a highly potential system that decouples supply-demand and enables time shifting power delivery (Gallo et al., 2016). Discussion is still ongoing on which energy carrier is most suitable, but that is beyond the scope of this research, which will focus on hydrogen storage. Multiple studies have already proven the functionality and efficiency of hydrogen as an energy carrier, and it is thereby expected to grow significantly in the upcoming decades (Zhang et al., 2016). The key philosophy behind this concept is a system that utilizes hydrogen in order to deliver energy.

The surplus or intermittent power will be used to produce hydrogen through water electrolysis.

This is seen as effective method of storing energy (Walker et al., 2016). In this way curtailment can be reduced and the utilization of VRES can be increased. With the electrolytic production of hydrogen via Power-to-gas, long term storage will be possible. With a targeted penetration of 80 % by VRES, the Power-to-gas storage could reduce the wind and solar capacity by 23%

and the curtailment of this energy production by 87%. A major advantage of hydrogen is that it able to store large amounts of energy over a long period, without significant losses during storage. Besides, integration of wind and solar energy is made possible with the generation of hydrogen by power to gas facilitation (Braun, 2008).

Cost of renewable resources

The described developments within the energy transition, will however all come at a cost. It will therefore be challenging to compete with conventional sources. At the end of the day, the price of energy must be attractive enough for end consumers. To achieve this, subsidies are given to promote the use of renewable energy sources. Over the period between 2000 and 2010, an increase of 1 % (1c€) in subsidies has led to an increase of renewable generation of 0,4 to 1

% (18-26%) among the 5 largest countries within Europe (Nicolini and Tavoni, 2017). This proves the effectiveness of subsidies within the renewable energy sector. These subsidies will foremost help to overcome the learning cost that arise with these new renewable energy sources.

In the past years, the installation cost for both wind- and solar project have decreased significantly states the International Renewable Energy Agency (IRENA, 2018).

(9)

9 Moreover, the cost associated with renewable energy is not limited to the generation of energy.

When incorporating the flexibility of a system, this will affect the cost relating to this system as well. The integration of VRES into energy systems will lead to cost accounted for profile, balancing and grid-related cost (Kondziella and Bruckner, 2016). These costs could be seen as a reduction in revenue for suppliers of energy, or it could be seen as additional cost that are accounted to specific market participants. The storage cost of energy via hydrogen will also affect the feasibility of a renewable hybrid system. However, research has already shown how these costs could be reduced. When great amounts of hydrogen will be produced in a single plant, the monetary capital investments per unit of product will reduce drastically (Mazloomi and Gomes, 2012). Alternatives to regular storage, are underground storage of hydrogen gasses and compressed air. Optionally this could be stored in existing gas or oil reservoirs, which would therefore not lead to significant higher cost (Ozarslan, 2012). The underground storing and piping of gas is proven to be capable of transporting hydrogen. A great positive impact of this is that no additional systems have to be created which is also beneficial in an economical way.With price and cost developments in mind, at a certain time the subsidies will be reduced significantly or even removed, the renewable sources have to figure a way to remain competitive.

(10)

10

Model description

This study will construct a model, in which real-life data from the Netherlands will be used to create an optimal mix of renewables that jointly covers a certain percentage of the total electricity demand. This model will operate as an autonomous system, meaning that there will be no import or export of electricity through other countries. The data that is collected consist of weather patterns, windmill efficiency and output, solar energy efficiency and output, hydrogen storage and lastly the electricity consumption in the Netherlands.

Renewable energy sources that are predicted to dictate the stage are solar and wind power.

Therefore, a model is created that relies on the production of energy through these sources. A conversion of weather patterns (i.e. wind speed and solar irradiance) into power output is applied to get a realistic overview of the energy generation via these renewable sources. The electricity generation of both sources is taken together which is shown as a daily supply throughout the year. In case there is a supply overage, meaning that the aggregate electricity supply of a certain day is higher than the electricity demand, will the electricity be converted into hydrogen through electrolysis and stored. This hydrogen can be transformed back into electricity at a later stage when there is a shortage of electricity supply. In this way, curtailment of electricity generation is averted.

The sustainability rate is calculated by summing all the sustainable generated electricity (solar, wind and hydrogen storage resp.) and dividing this with the total electricity demand of that year.

The total of the obtained sustainability rate throughout the year has to be above the chosen targeted percentage (25%, 50%, 75% and 100% resp). The part which is not covered by renewable energy, either production or supply through storage, will be supplemented by conventional resources to still meet the demand.

(11)

11

Model content

The aim of the simulation is to find an optimal balance between wind, solar and hydrogen given certain restrictions which minimizes the cost. A route towards 100% renewable electricity will be generated by means of the sustainability targets. This is done by varying with the variables of solar panels and both onshore and offshore wind turbines. The storage function will change as a result of the number of solar panels and wind turbines. The objective is to minimize the costs which includes the costs per solar panel, onshore wind turbine costs, offshore wind turbine costs, conventional energy costs, electrolysis costs, underground hydrogen storage costs and the costs of converting hydrogen back to electricity by a fuel cell. By multiplying these costs with the number of solar panels, wind turbines, amount of conventional energy and hydrogen used, the total yearly costs are calculated which are opted to be minimized.

For each sustainability target, the simulation will run at least 10 years in order to get a reliable output regarding the optimal balance which minimizes the costs. Since the model contains daily data summing up to a year, each run can be seen as a year. A more elaborate explanation on the number of replications is given in the outputs section.

Conceptual model

To construct a conceptual model that can be used for modelling, the structure proposed by Robinson (2014) is used. The conceptual model is thereby used to represent the real world in a visual and comprehensive way, which is shown in Figure 1.

Modelling objective

The modeling objective is the objective that is aimed to be achieved. Robinson (2014) subdivides the objective into three different parts:

Achievement: Minimize the total costs while reaching a percentage of (25, 50, 75 or 100%) renewable electricity generation in the Netherlands.

Performance: The following performance measurements are included:

- Total electricity generation by the different sources of energy.

- Amount of energy captured via hydrogen.

- Prices related to the different sources. This includes investment cost (CAPEX), operation cost (OPEX) and storage cost.

(12)

12 Constraints: The following constraints are included:

- Average wind speed and solar irradiance within the Netherlands. This will limit the generation of energy via wind and solar power for a certain capacity. If more energy is aimed to be generated, the capacity should be increased.

- Targeted renewable energy production. This percentage will determine the minimum generation of energy through renewables. The conventional supply will be limited to a certain extent, whereby the other part has to be covered by renewables. An optimal combination of the capacity of renewables have to be calculated to reach this percentage, and by setting a certain percentage for the conventional supply the capacity will be limited as well.

- Electricity Demand. The electricity demand will indicate the approximate target that has to be reached. The model could overproduce at certain days, in order to deliver enough in the following period(s).

Inputs Simulation

variables

Outputs

• Electricity Demand

• Wind speed onshore

• Wind speed offshore location 1

• Wind speed offshore location 2

• Solar irradiance

• Produced energy per source

• (Mis) matching of Supply and Demand

• Cost per energy source

• Hydrogen storage

• Nr of solar panels

• Nr of onshore windturbines

• Nr of offshore windturbines 1

• Nr of offshore windturbines 2

• Scenario (25, 50, 75, 100%)

•

Figure 1: Overview of conceptual model

(13)

13 Simulation runs

In Figure 1, all the components used in the model are shown. Each run will start with the generation of the input variables, which will be generated according to their distribution. This is included to reflect the fluctuating weather patterns and demand, which also changes from year to year in real life. Based on these input variables, the model will calculate the optimal number of solar panels and wind turbines which will collectively cover the electricity consumption for a certain percentage (25, 50, 75 or 100% resp.). For each scenario, the simulation will run at least 10 years to generate a reliable and representative outcome. Since the electricity demand and weather patterns will change each run, the amount of solar panels and wind turbines need to change as well. The outputs show the average over all runs of the produced energy, (mis) matching of supply and demand, costs per energy source and the amount of hydrogen that will be needed for each chosen sustainability target. The inputs, simulations variables and outputs shown will be described individually in the section below.

Inputs

Electricity demand

The energy consumption in the Netherlands is gathered from the Nederlandse Energie Data Uitwisseling (NEDU, 2020) which shows the fraction of electricity usage throughout the year for each quarter hour. The dataset also subdivides this into public and private usage, which is taken together since the electricity demand of the country is researched. To get a reliable dataset, the electricity demand of all the available years is used (2016 to 2019). The fraction of usage is multiplied with the electricity usage of that year (Enerdata, 2020) to get the usage in kWh per year. Thereafter a normal distribution, which shows the average daily demand, for each week is calculated. This is based on the average and standard deviation of electricity demand, of that week over the year. The model generates new daily demand for each week following this normal distribution. When looking at the average daily electricity consumption throughout the year (Figure 2), the consumption is significantly higher throughout the winter months. This is an important observation since the system should be able to operate at both higher and lower demand situations. The seasonality of wind and solar should therefore be carefully studied, which must be able to cope with this fluctuation in demand. The storage via hydrogen could thereby operate as a back-up function.

(14)

14

Figure 2 Average electricity consumption

Note: This shows the average daily consumption based on a weekly normal distribution, the shown amount is not the total of one week.

Wind Speed

In order to get a representative overview of the Netherlands, it is chosen to gather data from three different locations in the Netherlands. Due to time and complexity limitations it is chosen to limit the number of locations. The locations which are chosen to gather data from are Ijmuiden, Terschelling and De Bilt. Ijmuiden and Terschelling will operate as reference points for offshore wind parks, whereas De Bilt will represent the onshore locations of the Netherlands due to its central location. The data which is used is collected from the Koninklijk Nederlands Meteorologisch Insitituut (KNMI, 2020). The available dataset shows the daily windspeed for a particular location. The average windspeed for each group of 7 consecutive days over the years 2010-2019 is taken, which makes the distribution reliable. In this way, 70 data points are collected (7 days * 10 years) which makes sure that the weekly distribution is not based on outliers. By calculating the standard deviation over the same period, it is possible to create a normal distribution. The normal distribution will generate the daily wind speed for each week based on the corresponding distribution. Below the average windspeed is shown (Figure 3), which highlights the higher windspeed in the winter months and shows that the offshore locations logically have higher windspeeds.

0 50000000 100000000 150000000 200000000 250000000 300000000 350000000 400000000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53

Electricty consumption (kWh)

Week number

Electricity consumption troughout the year

(15)

15

Figure 3: Average windspeed

Below the normal distribution of the wind speed in Ijmuiden in week 1 and 26 is displayed in Figure 4. This shows the chance that a certain wind speed will be generated in the model. For each week a new distribution is created, which generates the daily average wind speed. As can be seen, the mean of the first week is significantly higher compared to the mean in week 26 (9 versus 6,294 resp.). This is in line with the forecasts, which normally show a higher wind speed during the winter and a lower during the summer months. When looking at the standard deviation, the value of the first week is fairly larger than week 26. This can also be seen in the figure, where the dots are more spread along the distribution. For week 26, the dots are more concentrated around the mean value. This indicates that the wind speed during the summer months fluctuates less and is therefore more predictable.

0,00 20,00 40,00 60,00 80,00 100,00 120,00

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53

Wind speed (m/s)

Week number

Average windspeed (2010-2019)

De bilt Terschelling Ijmuiden

(16)

16

µ = 9

σ = 3,447

µ = 6,294

σ = 2,331

Figure 4: Normal distribution of wind in Ijmuiden in weeks 1 and 26

Solar irradiance

To know the amount of kWh that can be produced, the solar irradiance in the Netherlands have to be determined first. This is done by picking one location (De Bilt resp.) and collecting the irradiance data for this location. De Bilt is chosen due to its central location in the Netherlands, which is assumed to be representative for the Netherlands. The used irradiance data is found on the website of the KNMI (2020). Similar to the wind data, a normal distribution is created by collecting the average and standard deviation for each week of the year of a certain timeframe.

The data is gathered for the timeframe between 2010 and 2019 and calculated for each week.

In this way, the distribution is based on 70 (7 days * 10 years) observation which makes it more reliable. The distribution is showing the average daily solar irradiance on a day within the

0 0,002 0,004 0,006 0,008 0,01 0,012 0,014

0 5 10 15 20

Probability

Wind speed (m/s)

1 Jan - 7 Jan (Week 1)

0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0,018

0 2 4 6 8 10 12 14

Probability

Wind speed (m/s)

25 June - 1 July (Week 26)

(17)

17 generated week. To make this clearer, one week will be highlighted. The average daily irradiance in week 27 is 5,9 kwh/m2 (see Figure 5). Within week 27, each day the generated irradiance will fluctuate around this value following the distribution. In another week, the generated daily data will follow the distribution of that particular week. Where the data from the KNMI is shown in joules per cm2, the unit used in the model is kWh per m2. The data is converted by multiplying this with the factor of 0,00277778. Since the solar panels are also covering a certain surface in m2, will this scale make it easier for the calculation. As can be seen in the figure below, the solar irradiance is significantly higher during the summer months, which makes sense since there are more sun hours. Whereas the wind speed is generally higher during the winter months, it is the other way around for solar irradiance.

Figure 5: Average solar irradiance throughout the year

Balancing

As mentioned in the theoretical background, the patterns that the wind and solar follow counterbalance each other during the seasons. In Figure 6 this is displayed by showing the average wind speed of Terschelling throughout the year, compared to the average solar irradiance within the Netherlands. Where the average wind speed flattens towards the summer months, the solar irradiance behaves the opposite way and increases during the summer. During the winter months, this is the other way around (i.e. higher windspeeds and lower solar irradiance). Another observation is that the wind speed can be seen as more stable compared to solar throughout the year. Although, the wind speed is lower during the summer months, it is still generating enough speed to be useful for wind power. The solar irradiance on the other hand, shows a serious diminution towards the winter months.

5,903695199

0 1 2 3 4 5 6 7

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53

Solar irradiance (kWh/m2)

Week nr

Average solar irradiance throughout the year

(18)

18

Figure 6: Balancing attribute Solar vs Wind

Simulation Variables Wind energy generation

In order to calculate the power output of wind energy at certain wind speeds, the type and size of wind turbines have to be selected. As stated in the report of WindEurope (2018) is the average size of newly installed offshore wind turbines in 2018 calculated on 6,8 MW. The assumption is made that the futuristic wind turbines will have a similar size and potentially be slightly larger. Therefore, in the simulation model the Aerodyn SCD 8.0/168 is used which has a capacity of 8 MW. The wind speed to power output conversion from Bauer and Matysik (2020) of this turbine is used to calculate the total kWh produced. As can be seen in Figure 7 below, the turbine has a cut-in and cut-out windspeed between which it delivers power. For this turbine, the cut-in windspeed is 3,5 m/s and the cut-out windspeed is 25 m/s. As can be seen in Figure 7 and 8, the power output increases exponentially till the wind speed reaches around 12 m/s.

From this point on there is a stable output which is the maximum output as well. Above 25 m/s the offshore turbine will stop working to prevent from getting damaged.

0 1 2 3 4 5 6 7

0,00 1,00 2,00 3,00 4,00 5,00 6,00 7,00 8,00 9,00

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53

Solar irradiance (kWh m2)

Wind speed (m/s)

Week nr

Solar/Wind speed

Wind speed Terschelling Solar irradiance

(19)

19

Figure 7: Power output offshore wind turbine

According to WindEurope (2019), was the weighted average onshore turbine size 2,7 MW in 2018. To work with similar data, the Enercon E-115 EP3 3.000 is used for the simulation which has a capacity of 3 MW (Bauer and Matysik, 2020). The cut-in speed is stated on 2,5 m/s, whereas the cut-out speed is listed on 34 m/s. Above the 34 m/s, the onshore turbine will stop working. What is noticeable, is that the maximum power output is being delivered over a larger range. This has to do with the relatively large cut-out wind speed.

Figure 8: Power output onshore wind turbine

The costs of offshore wind turbines are assumed to be following the costs calculated by a research of Planbureau voor de Leefomgeving (PBL) (Pisca, 2019). In order to calculate the total yearly costs per turbine, the total cost per kWh are multiplied with the power of the turbine which is based on a lifetime of 25 years. For the calculations of Ijmuiden, the data of the wind farm of Ijmuiden is used, for Terschelling the data of windfarm Boven de Wadden Eilanden is used. The costs for a turbine in Ijmuiden are calculated on 15,6 million euros. Whereas the turbines around Terschelling are less expensive with an amount of 14,2 million euros. These

0 2000 4000 6000 8000 10000

0 5 10 15 20 25 30 35

Power output (kWh)

Wind speed (m/s)

Power output offshore wind turbine

0 500 1000 1500 2000 2500 3000 3500

0 10 20 30 40

Power output (kWh)

Wind speed (m/s)

Power output onshore wind turbine

Model Aerodyn SCD 8.0/168 Power

output 8000

K/w Cut in

windspeed 3,5 m/s Rated wind

speed 11,5

m/s Cut out

windspeed 25 m/s

Model Enercon E-115 EP3 3.000 Power

output 3000 K/w

Cut in

windspeed 2,5 m/s Rated

wind speed

12,8 m/s

Cut out

windspeed 34 m/s

Table 1: Attributes offshore wind turbine

Table 2: Attributes onshore wind turbine

(20)

20 costs include both investment costs as well as operation and maintenance cost. Since the model is operating on a yearly basis, the total costs have to be divided with the lifetime. Therefore, the yearly costs will account for €624.000 for Ijmuiden and €568.000 for Terschelling.

The onshore wind turbines are assumed to be around 2.60 million euros based on an indication by Renewablesfirst (2020). This amount covers for 69% the investment costs, where the remaining 31% is covering operation and other project related costs. The lifetime of onshore wind turbines is assumed to be similar to the offshore turbines, meaning 25 years. Dividing the total onshore wind costs by the lifetime gives us the costs of €104.000.

Solar energy generation

Similar to the conversion of wind to power, the same holds for solar irradiance to power.

Depending on the type of solar panel a certain output of energy will be generated. One aspect that is of crucial importance at the conversion of solar irradiance, is the efficiency loss. Due to dust, high temperatures, shadow, snow and other factors is the conversion of irradiance into power limited and will not be the optimal 100%. The efficiency of this conversion is currently one of the highest with the Crystalline silicon solar cells, which can still operate at a competitive price (Snaith, 2013). The efficiency range of this solar cells is between 20 and 25 % (Jacoby, 2016). The solar irradiance will be multiplied with this efficiency rate, which is following a uniform distribution. The average size of an solar panel according to (Fattenval, 2020a) is 168 x 100 cm, which is therefore covering a surface of 1,68m2. In order to calculate the output of these solar panels, the irradiance will be multiplied with the efficiency and solar panel size. The average price of these solar panels according to Fattenval (2020b) is listed on €429,40. This includes installation and inspection costs. It is thereby chosen to only include solar parks to cover the demand of the Netherlands. The size of the solar parks that can be constructed is fixed at 10.000 solar panels. By choosing a solar park around this size, the simulation is still able to operate at a reasonable pace. When reducing the size of the park, the run time of the simulation will increase significantly due to a larger range of solar panels which can be chosen. Moreover, is the peak power produced by 1 solar park at a similar range as the wind turbines that are selected in this research.

(21)

21 Hydrogen storage

The hydrogen system within this model consist of 3 parts: electrolysis, storage and the fuel cell.

The electrolyzer splits water into hydrogen and oxygen using electricity. When performing this activity, a fraction of energy will be lost due to this conversion. In the literature, different efficiency percentages and different capital costs are used. This efficiency varies between 65%

and 83% (Dufo-López and Bernal-Agustín, 2013; Zakeri and Syri, 2015; Proost, 2019). Within this research, an optimistic and futuristic view is taken, with an efficiency of 83%. The costs of these fuel cells is ranging between 700 and 1700 euros (Dufo-López and Bernal-Agustín, 2013;

Schmidt et al., 2017; Proost, 2019). Since some of these systems are not completely developed yet, the pricing is assumed to end up in the middle. The cost associated with using this electrolysis in this system are assumed to be €1200, - for a 1 MW system. Besides the capital costs, an additional % for the operation cost (OPEX) will be included as well. Chardonnet et al.

(2017) lists these OPEX costs as 2-4% of the capital cost. These costs include maintenance, spare parts and replacement of auxiliary components but excludes electricity, water consumptions and stacks replacement. Within the model, an additional 3% is added to the CAPEX costs to cover the additional operation cost. In order to calculate the number of needed electrolyzers, the following equation will be used:

Total electrolyzer cost: 𝑚𝑎𝑥 𝑓𝑙𝑜𝑤 𝑒𝑙𝑒𝑐𝑡𝑟𝑖𝑐𝑖𝑡𝑦 𝑡𝑜 𝑠𝑡𝑜𝑟𝑎𝑔𝑒 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟 ∗ 𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑧𝑒𝑟 𝑐𝑜𝑠𝑡

As soon as the electricity is captured in the hydrogen, the hydrogen needs to be stored. The storage, will lead to costs which are assumed to be €0,21 per kWh for underground storage based on research by Zakeri and Syri (2015). For each day that an underground storage location will be used, this price per kWh has to be paid. Consequently, will it be more expensive to store for a larger amount of time. A research performed by Caglayan et al. (2020), has shown the potential storage possibilities in the Netherlands, which is not expected to play a limiting factor.

Therefore, no storage cap will be included. The total storage costs are calculated with the following equation:

Total Storage cost: 𝑇𝑜𝑡𝑎𝑙 𝑠𝑡𝑜𝑟𝑎𝑔𝑒 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟 ∗ 𝑢𝑛𝑑𝑒𝑟𝑔𝑟𝑜𝑢𝑛𝑑 𝑠𝑡𝑜𝑟𝑎𝑔𝑒 𝑐𝑜𝑠𝑡

As soon as the demand exceeds the supply of electricity via the solar and wind sources, electricity has to be extracted to meet a certain percentage of sustainability. When transforming the hydrogen back into electricity, a fuel cell is used. This can be seen as the reverse operation of the electrolyzer. The hydrogen will be combined with oxygen in order to generate electricity (Dufo-López and Bernal-Agustín, 2013). Similar as for the electrolyzer, a certain efficiency

(22)

22 rate and cost is associated with this operation. The efficiency of the fuel cell is significantly lower than for the electrolyzer and ranges between 40 and 55% (Dufo-López and Bernal- Agustín, 2013; Chardonnet et al., 2017). The costs of a fuel cell varies between 540 and 3000 euros, depending on the size of the fuel cell (Mekhilef et al., 2012; Dufo-López and Bernal- Agustín, 2013; Zakeri and Syri, 2015; Chardonnet et al., 2017). Within this model a fuel cell of 1 MW is taken, which has an efficiency of 50% and is priced at €2000, -. The roundtrip efficiency, which stands for the efficiency of the whole process, is therefore 41,5%. The same percentage of 3% as the electrolyzer is included to cope with the OPEX, which makes the price a total of €2060, -. The total fuel costs are calculated by means of the following equation:

Total fuel cell cost: 𝑚𝑎𝑥 𝑓𝑙𝑜𝑤 𝑠𝑡𝑜𝑟𝑎𝑔𝑒 𝑡𝑜 𝑔𝑟𝑖𝑑 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟 ∗ 𝑓𝑢𝑒𝑙 𝑐𝑒𝑙𝑙 𝑐𝑜𝑠𝑡

Below (Table 3), an overview is given of the different costs and efficiencies associated with transforming electricity into hydrogen and the other way around.

Electrolyzer

Efficiency 83,00%

CAPEX € 1200,00

OPEX € 36,00

Total cost € 1236,00

Costs per kWh (1

MW system) € 1,236

Storage

Under ground € 0,21

Fuel Cell

Efficiency 50%

CAPEX € 2000,00

OPEX € 60,00

Total cost € 2060,00

Cost per kWh (1

MW system) € 2,060

Roundtrip efficiency 41,5%

Table 3: Overview of costs and efficiency hydrogen system

(23)

23 In order to calculate the total costs per kWh for hydrogen, the total costs of hydrogen are divided by the supplied kWh hydrogen. In this way, the costs of converting into hydrogen, storing and converting back into electricity are taken together in the hydrogen price. Also, the storage which will not be supplied is thereby integrated.

Cost per kWh hydrogen: 𝑇𝑜𝑡𝑎𝑙 𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑧𝑒𝑟 𝑐𝑜𝑠𝑡 + 𝑇𝑜𝑡𝑎𝑙 𝑠𝑡𝑜𝑟𝑎𝑔𝑒 𝑐𝑜𝑠𝑡 + 𝑇𝑜𝑡𝑎𝑙 𝑓𝑢𝑒𝑙 𝑐𝑒𝑙𝑙 𝑐𝑜𝑠𝑡 𝑇𝑜𝑡𝑎𝑙 𝑘𝑊ℎ ℎ𝑦𝑑𝑟𝑜𝑔𝑒𝑛 𝑠𝑢𝑝𝑝𝑙𝑖𝑒𝑑

To understand the hydrogen function, it is important to know how the in- and outflow behaves for each period. The following equation can be used to clarify this:

𝐇_𝐧= 𝐻_𝑛−1+ 𝑆𝑂_𝑛∗ 𝐸𝐸 −^𝑆𝑆^𝑛

𝐸𝐹

Table 4: Parameters hydrogen storage system

Sustainability targets

Within this simulation study, 4 sustainability targets will be used (25%, 50%, 75%, 100%).

Each sustainability target will represent a minimum percentage of electricity consumption that should be covered collectively by the renewable energy sources.

Outputs

The outputs that will forth come out of the model are listed below (Table 5). Since the inputs of the generated model are stochastic, the outputs of the model have to be stochastic as well (Robinson 2013). The aim is to estimate the parameters of the output distributions, which are displayed by a mean (µ) and standard deviation (σ).

Parameters

𝐇_𝐧: Hydrogen storage in month n, n=1 … 12 𝑆𝑂_𝑛: Oversupply renewables in month n, n=1 … 12 EE: Efficiency Electrolyzer

𝑆𝑆_𝑛: Supply shortage in month n, n=1 … 12 EF: Efficiency Fuel cell

(24)

24 Produced energy per source

-Amount of kWh produced by Solar -Amount of kWh produced by Wind -Amount of storage needed by hydrogen

-Amount of kWh covered by conventional resources (Mis) matching of Supply and Demand

-% of demand covered by solar and wind -% of demand covered by hydrogen

-% of demand covered by conventional resources -% oversupply by sources

Cost per energy source

- Cost per kWh of wind - Cost per kWh of solar - Cost per kWh of hydrogen

Table 5: Overview outputs model

Since the model starts in a condition in which there is no storage, a warm-up period of 1 year is performed to create a realistic starting condition. The number of runs is based on the graphical method (Robinson, 2014). The minimum number of replications following this method is 10.

This method plots the cumulative mean of the output of the generated outputs. With more outputs obtained, the displayed cumulative mean graph should become flatter until there is only a marginal change. At this point, a larger number of runs will not have a significant effect on the reliability of the output. Therefore, the optimal number of replications is reached which is called the steady state. The output is varying around a constant mean according to the steady state distribution. Moreover, a confidence interval is applied, which level is listed on 95%.

Below, the cumulative mean together with the confidence interval for the solar and wind output is shown in the 100% situation (Figure 9 and 10). As can be seen, are both outputs quite stable and are not changing when performing more runs. To be sure that the output is stabilized, 15 runs are performed which confirms this stabilization. By performing multiple consecutive runs, the reliability of the model will increase.

(25)

25

Figure 9: Graphical method displaying the coverage of electricity demand including oversupply by solar

Figure 10: Graphical method displaying the coverage of electricity demand including oversupply by all wind locations 0,00

0,10 0,20 0,30 0,40 0,50 0,60

1 2 3 4 5 6 7 8 9 10 11 12 13 14

% of electricity demand

Run (years)

% covered by solar

kWh solar Upper bound Lower bound

0,60 0,70 0,80 0,90 1,00 1,10 1,20 1,30 1,40 1,50

1 2 3 4 5 6 7 8 9 10 11 12 13 14

% of electricity demand

Run (years)

% Total wind

kWh wind total Upper bound Lower bound

(26)

26

Results

For the sustainability targets of 100% and 75%, the results will be discussed in detail. The targets of 25 and 50 percent behave in a similar way as the 75%, which will therefore be discussed in a comparative way at the end.

100% situation

In Figure 11, the matching of supply and demand is shown when all the demand is covered by renewables. The renewable supply in this figure, contains the electricity supplied to the grid by both solar and wind energy. The oversupply is deducted from the total electricity generated, to only show the part which is actually delivered to the grid. To match the demand, the remaining part should be covered by the hydrogen storage since all the demand has to be covered by renewable energy. Collectively, this sums up to the total demand as can be seen in the figure.

This figure is therefore reflecting the part which is supplied directly from the renewable sources and the part which is supplied indirectly via the hydrogen storage.

Figure 11: Supply and demand matching

Figure 12: Total wind/solar generation 0

2E+09 4E+09 6E+09 8E+09 1E+10 1,2E+10

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Electricity (kWh)

Months

Supply and demand matching

Hydrogen supplied (after conversion) Renewable supply

0 5E+09 1E+10 1,5E+10 2E+10

Electricity (kWh)

Month

Total wind/solar generated

Wind Solar Demand

(27)

27 In Figure 12, the total electricity generated per month together with the demand is shown. The difference with the prior figure, is that the oversupply is shown as well. Within the hybrid system, this oversupply will be captured through hydrogen storage. Moreover, the division of the renewable sources is made visible in this figure as well. As can be seen, wind energy is chosen as the major source. The dominance is particularly visible in the winter months, in which almost all the electricity demand is covered. Towards the summer months does this fall significantly, which is backed up by the solar that increases in this period. Moreover, it can be seen in Figure 11 that the complementary supply through hydrogen storage increases in the winter. This can also be explained by the solar energy which falls away in the winter months, together with the higher electricity demand. Since the electricity is now mostly dependent on the wind, which can fall away sometimes, the importance of storage increases. Therefore, the role of storage during the winter months is essential to deliver a stable supply of electricity.

Table 6 displays the coverage of renewable sources as a fraction of the electricity demand.

Besides the importance of the wind energy, hydrogen storage stands out as well with a coverage of 13%.

Table 6: Coverage of demand per source Table 7: kWh costs per source 100% situation

Wind energy

As can be seen in Table 6, the offshore location of Ijmuiden is responsible for almost half of the total demand. Mainly because of the relatively high wind speed, combined with the low electricity price, this offshore location is the main producer of electricity. Although the prices of the wind turbines at Terschelling and onshore location are larger (Table 7), it is still responsible for a significant part of the wind generation. This can be related to the fluctuating wind speeds, meaning that it can be windier at one location compared to another. To illustrate this, the power output and wind speed of the different locations in the first week of January is shown below in Figure 13 and Table 8. It should be noted that the inputs are randomly generated for each year, therefore this figure can purely be used to illustrate this fluctuating behavior.

Solar 15%

Wind on 10%

Wind off IJ 47%

Wind off T 15%

Hydrogen 13%

Total 100%

Cost per energy source

Cost per kWh wind onshore € 0,05 Cost per kWh wind offshore IJ € 0,03 Cost per kWh wind offshore T € 0,04 Cost per kWh solar € 0,08 Cost per kWh hydrogen € 9,61

(28)

28 As can be seen in Figure 13, Ijmuiden is responsible for most of the kWh hours produced in this week. However, at 3 and 7 January the wind is not strong enough to produce energy at this location. As displayed in Table 8, the wind conditions more favorable at the other locations which therefore take care of a part of the electricity supply. Due to the lower cut-in speed, the onshore location is also useful in case the wind speed is relatively low. This illustrates the

selection of multiple locations, despite some being more expensive.

Hydrogen storage

As can be seen in Figure 14, the storage level is decreasing in the first 4 months, thereafter it starts to increase during the summer months. This trend can be explained by the higher production of solar energy in the summer months together with the relatively stable wind generation. Moreover, the electricity consumption during the summer is lower as well resulting in more energy injected in the hydrogen storage. Together will this lead to an increase of storage in the summer, while decreasing in the winter months.

De bilt Terschelling Ijmuiden

1-jan 4,50 6,02 11,04

2-jan 4,26 8,60 12,00

3-jan 4,06 8,57 4,16

4-jan 2,74 3,08 14,44

5-jan 3,64 4,55 12,65

6-jan 6,29 4,42 5,99

7-jan 6,21 9,30 5,45

0 50000 100000 150000 200000 250000

1-jan 2-jan 3-jan 4-jan 5-jan 6-jan 7-jan

kWh produced

Day

Power output wind

De bilt Terschelling Ijmuiden Figure 13: Power output wind week 1

Table 8: Wind speed (m/s) week 1

(29)

29

Figure 14: Hydrogen storage

Figure 15: Injected storage vs supplied storage

Figure 15 shows the injected and supplied storage, which eventually leads to an increase or decrease of the overall storage. Injected storage is the oversupply of electricity which is injected into the storage, meaning after the conversion, whereas the supplied storage is the electricity subtracted from the storage, prior to conversion, and delivered to the grid. In general, they follow the same trend, showing that the injected amount is being subtracted in the same month as well. In the months where the supplied storage is higher than the injected oversupply, the overall storage decreases and vice versa.

Noticeable in the hydro storage function (Figure 14) is the storage at the end of the year, which is significantly lower than the storage in January. To explain this difference, Figure 16 comes in handy. In this figure, the efficiency of the storage can be extracted by looking at the amount

0 2E+10 4E+10 6E+10 8E+10 1E+11 1,2E+11 1,4E+11

Electricity (kWh)

Months

Hydrogen storage

0 1E+09 2E+09 3E+09 4E+09 5E+09 6E+09

Electricity (kWh)

Months

Injected vs supplied storage

Oversupply injected Storage supplied

(30)

30 that is injected compared to the supplied electricity. The injected storage here is prior to the conversion, whereas the supplied electricity is shown after the conversion. In this way, it can be seen how much of the injected electricity is eventually supplied to the grid. As can be seen, the electricity that goes into the storage is significantly higher than the storage that is extracted.

The efficiency of the electrolyzer and fuel cell can be accounted for this. With a roundtrip efficiency of approximately 41,5%, the injected electricity needs to be about 2,5 times as a high as the extracted electricity.

Figure 16: Flow of hydrogen storage -4E+09

-2E+09 0 2E+09 4E+09 6E+09 8E+09

Electricity (kWh)

Months

Storage balance

Oversupply injected Hydrogen supplied (after conversion)

(31)

31

75% situation

Below the supply and demand matching for the 75% situation is shown (Figure 17). A similar figure as in the previous situation is shown, in which the supplied renewable electricity is shown together with the supplied hydrogen. As can be seen, the 75% coverage through renewables is not equally distributed throughout the year. Again, in the summer months, the coverage by renewables and hydrogen is the largest compared to the demand. When moving further in the year, this coverage is lower, which therefore will be covered by conventional supply.

Figure 17: Supply and demand matching

Below the total electricity generation trough renewables is shown. As in in the 100% situation, most of the renewable production is coming from wind power. However, the amount of solar production as a fraction of the total renewables is greater. As a result of the increase of conventional electricity supply, the total wind electricity generation decreased and the amount of solar mostly remained the same. Moreover, it can be seen that most of the electricity is generated in the period from December till June.

Figure 18: Total electricity generated through wind and solar 0

2E+09 4E+09 6E+09 8E+09 1E+10 1,2E+10

Electricity (kWh)

Months

Renewable supply hydrogen supplied (after conversion) Conventional Demand

0 2E+09 4E+09 6E+09 8E+09 1E+10 1,2E+10

Electricity (kWh)

Months

Solar Wind Demand

(32)

32 The division of the electricity generation is, as mentioned, dominated by the wind production.

However, where the wind generation decreased, the solar generation as a percentage of the electricity demand almost remained the same. The relative occupancy by solar is therefore increasing with the sustainability target. This is also displayed in Table 9, which shows the generation per source as a fraction of the total electricity demand. Where the solar generation was 15% in the 100% situation, it now increased to 19%.

Table 9:Division as % of total electricity generated

The use of conventional energy also has its effect on the hydrogen storage. This now accounts for just 3%, compared to the 13% in the 100% situation. Since less renewable electricity and especially less wind turbines are used, the oversupply of electricity is lower. This has to do with the flexible and cheap backup by conventional electricity which only supplies electricity in case necessary. Contradictory, do the renewable sources generate electricity regardless what the demand is. This results in its way to an increase in storage in case there is oversupply of electricity. The hydro storage and conventional supply thereby have a contradicting relationship, whenever the hydro storage increases, the conventional supply decreases. The relative increase of solar indicates that using solar is more beneficial than generating wind electricity in the winter, storing it and thereafter use it in the summer.

Wind energy

Where in the 100% situation Ijmuiden was accounting for 47%, this is now much smaller as can be seen in Table 9. The other wind locations even grew its relative coverage, making the division of wind sources more balanced. This can be explained by the fact, that the power output of the offshore location at Ijmuiden is relatively high. When keeping this high fraction of offshore generation at Ijmuiden, this could lead to oversupply more frequently. By dividing this among the different locations, it is able to better match the demand and therefore create less oversupply.

Solar 19%

Wind on 14%

Wind off IJ 24%

Wind off T 16%

Hydrogen 3%

Conventional 24%

Total 100%

“The Netherlands under construction,