• No results found

The potential of wind power on the Swedish ancillary service markets

N/A
N/A
Protected

Academic year: 2021

Share "The potential of wind power on the Swedish ancillary service markets"

Copied!
109
0
0

Loading.... (view fulltext now)

Full text

(1)

The potential of wind power on the

Swedish ancillary service markets

HANNES WIKLUND

KTH ROYAL INSTITUTE OF TECHNOLOGY

(2)

on the Swedish ancillary

service markets

HANNES WIKLUND

Master in Electric Power Engineering Date: February 9, 2021

Supervisor: Daniel Kulin & Evelin Blom Examiner: Lennart Söder

School of Electrical Engineering and Computer Science Host company: Svensk Vindenergi

(3)
(4)

Abstract

An increasing proportion of variable renewable energy in the Swedish power system is leading to greater needs of system flexibility. A key aspect of han-dling this is frequency flexibility where actors can either increase or decrease their production or consumption when required. This type of service is called an ancillary service and are procured by the transmission system operator in specifically designed markets. These markets are predicted to grow signifi-cantly in the coming years and there is a demand for new actors. Wind power has the capability to provide these services via pitch control of the turbines, but this has not been common practice in Sweden previously.

This thesis explores the potential of wind power to enter the Swedish an-cillary service markets. The overarching goal is to provide a solid introduction to the workings of the markets and how wind power can join, a description of hindrances, and what the potential financial impact on a wind farm is. First, a detailed review of the technical requirements of the markets and the price de-velopment is provided in addition to a review of similar international markets. Then an interview study is performed where actors in the wind power industry give their view on the topic. Lastly, an economic modelling of an already ex-isting generic wind farm that is participating on the ancillary service markets FCR-N, FCR-D, aFRR Down, and FFR is presented. 2018, 2019, and 2020 as well as all four pricing areas are evaluated. The model utilizes historical wind and price data to generate a production estimation and a Monte-Carlo simulation of the estimation error is included.

(5)

Sammanfattning

En större andel intermittent förnybar energi i det svenska elsystemet leder till större behov av systemflexibilitet. En nyckel till att hantera det är frekvens-flexibilitet där en aktör kan öka eller sänka sin produktion eller konsumtion när det behövs. Denna typ av tjänst kallas för stödtjänst och sådana reserver upphandlas av Svenska kraftnät i specifika marknader. Dessa marknader pro-gnosticeras att påtagligt öka i omsättning de närmsta åren och det finns ett behov av nya aktörer. Vindkraft har möjligheten att leverera stödtjänster men det är inte vanligt i dagsläget.

Den här uppsatsen utreder potentialen för vindkraft att delta på de svens-ka stödtjänstmarknaderna. Det övergripande målet är att leverera en solid in-troduktion till hur marknaderna fungerar och hur vindkraft kan delta, en be-skrivning av potentiella hinder och vad den potentiella ekonomiska effekten på en vindfarm kan vara. Först ges en detaljerad beskrivning av de tekniska kraven på marknaderna och hur prisutvecklingen sett ut. Sedan presenteras en intervjustudie där aktörer i vindkraftsindustrin intervjuats kring deras syn på ämnet. Sist har en ekonomisk modellering tagits fram där en generisk vind-farm som deltar på FCR-N, FCR-D, aFRR Ned och FFR simuleras. Resultat för 2018, 2019 och 2020 samt alla fyra prisområden tas fram. Modellen an-vänder historisk vind- och prisdata för att ta fram produktionsprognoser och Monte-Carlo-simulerar sedan ett prognosfel.

(6)

1 Introduction 2

1.1 Introduction . . . 2

1.2 Purpose and goal . . . 4

1.3 Relevance . . . 4

1.4 Demarcation . . . 5

2 Background 6 2.1 The Swedish electric power system . . . 6

2.1.1 System topology . . . 6

2.1.2 Power balance . . . 7

2.2 The electricity market . . . 7

2.2.1 Electricity market players . . . 8

2.2.2 Electricity trading . . . 9

2.3 Ancillary services . . . 11

2.3.1 Frequency related services . . . 12

2.3.2 Voltage support and system restoration . . . 15

2.4 Swedish ancillary service markets . . . 15

2.4.1 Frequency containment reserve . . . 16

2.4.2 Frequency restoration reserve . . . 21

2.4.3 Fast frequency reserve . . . 25

2.4.4 Planned ancillary service market changes by Svenska kraftnät . . . 32

2.4.5 Market turnover projections . . . 32

2.5 Wind turbines . . . 34

2.5.1 Wind turbine types . . . 34

2.5.2 Wind turbine capabilities . . . 35

2.6 Previous studies . . . 38

2.6.1 Bibliometric analysis . . . 38

2.6.2 Examples of previous studies . . . 39

(7)

3 Methods 40

3.1 Methodology . . . 40

3.2 Simulation model . . . 42

3.2.1 Turbine specifications . . . 43

3.2.2 Wind farm conditions . . . 44

3.2.3 Production estimation error . . . 47

3.2.4 Adjustments of model between different markets . . . 49

3.2.5 Bidding strategies . . . 56

4 Results 59 4.1 Interview study . . . 59

4.1.1 Wind turbine manufacturers . . . 60

4.1.2 Wind power industry actors . . . 64

4.2 Simulation results . . . 67

4.2.1 Normal operation of wind farm . . . 67

4.2.2 Bidding on FCR-N . . . 68

4.2.3 Bidding on FCR-D . . . 74

4.2.4 Bidding on aFRR Down . . . 79

4.2.5 Bidding on FFR . . . 84

5 Discussion 86 6 Conclusions and future studies 91 6.1 Conclusions . . . 91

6.2 Future studies . . . 92

(8)

aFRR Automatic Frequency Restoration Reserve CE Central European

CIGRE International Council on Large Electric Systems DFIG Doubly-Fed Induction Generator

ENTSO-E European Network of Transmission System Operators for Electricity FCR-D Frequency Containment Reserve Disturbance

FCR-N Frequency Containment Reserve Normal FFR Fast Frequency Reserve

GB Great Britain

HVDC High Voltage Direct Current IE/NI Ireland and Northern Ireland IEA International Energy Agency

IEC International Electrotechnical Commission IEEE Institute of Electrical and Electronics Engineers mFRR Manual Frequency Restoration Reserve

PPA Power Purchase Agreement

SMHI Swedish Meteorological and Hydrological Institute TSO Transmission System Operator

(9)

Introduction

This chapter gives an introduction to the project which concluded in this thesis. The purpose and goal are specified, the relevance of the project is discussed, and the demarcations are explained.

1.1

Introduction

The Swedish power system, as many power systems across the world, is in the coming years expected to undergo substantial changes. These changes are driven by an increasing availability of variable renewable energy (VRE) at low cost and further deployment of distributed energy resources (DER). Progress in digitalisation and electrification of industrial processes and transportation will both increase and diversify the load side of the power system. This new dynamic is posing challenges on the Swedish grid which traditionally has been comprised of stable power sources such as nuclear and hydro power.

The current changes in power systems have been defined by IEA as 6 differ-ent phases of transformation characterized by the level of VRE integration. In Sweden, this transformation has reached phase 3 which means that VRE

gen-eration determines the opgen-eration pattern of the system. A key concern in

han-dling the inherent variability that this transformation entails is power system

flexibility. Traditionally, flexibility has been provided by conventional power

sources such as hydro-power with its controllable power output. VRE are now emerging as an alternative flexibility resource. Proper policy is crucial in fa-cilitating this potential resource. Regulatory framework has been updated in several countries, including Australia, Ireland, Spain, Canada, and to a certain extent Sweden, as well as having introduced market reform to activate

(10)

bility from VRE resources [1].

In Sweden, the proportion of electricity produced from wind power has in-creased sharply from almost zero in the 1990s to above 10 % in 2018 [2]. This increase is expected to continue, and wind power is the main driver of the Swedish power system transformation. Being sparsely populated with long coastlines the technical potential of land based and off-shore wind power in Sweden is high. Additionally to increasing the production variability, wind power also reduces the rotational inertia in the power system by not being syn-chronously connected. The rotational inertia in a power system is crucial to handling power imbalances as it allows kinetic energy to be stored or released by the rotating turbines when a mismatch between production and consump-tion in the system has occurred. A reducconsump-tion in rotaconsump-tional inertia makes the system lighter which in turn decreases its resilience to balance contingency situations.

While wind power affects the system negatively by increasing the produc-tion variability and reducing the amount of rotaproduc-tional inertia, there is a techni-cal potential to limit this effect. Modern wind turbines are generally control-lable with pitch control and are connected to the grid via power converters. Pitch control can keep the power output at a set value, either in a curtailed (decreased) state or at an optimal level, and then ramped up or down when desired. Furthermore, the power converters can mimic the behavior of a syn-chronously connected generator via control algorithms. This capability mean that wind turbines can participate in the primary and secondary control of the power system and strengthen the inertial response [3]. To incentivise wind power actors to provide these ancillary services to the system, fair compen-sation must be provided via ancillary service markets. Some such markets already exists in Sweden and are turning over a significant amount of money while even more markets are under way.

(11)

1.2

Purpose and goal

The purpose of this project is to explore the potential of wind power to con-tribute to the safe and reliable operation of the Swedish electrical power sys-tem by participating in the markets for ancillary services. To investigate this, the attitude of actors in the wind power industry towards participating in these markets today and what the plans for the future are will be explored. A simula-tion modelling of the economic potential of a wind farm will also be designed. The goal is to provide Svensk Vindenergi with a synthesis of the current technical readiness, business maturity, and an assessment of the possibility and feasibility to participate on the markets today and in the future. To reach the goal of the study the following research questions are relevant to answer. Market:

• How are the current markets for ancillary services in Sweden constructed and what are the future trends?

Technical ability:

• Which ancillary services are wind power adapted to provide today? Financing:

• What impact could providing ancillary services have on the financial structure and profitability of a wind farm?

1.3

Relevance

(12)

specifically were found.

The increasing need of system flexibility is leading to both an expansion of the current Swedish ancillary service markets [4] as well as the construction of new markets such as the fast frequency reserve. These circumstances have led to a demand for new participants to enter the markets. While this demand could be satisfied by resources such as batteries, electric boilers, or load flexibility it is important to explore all possibilities. Given the technical capabilities of wind turbines, a study into which other practical obstacles can occur for actors and showing that an economic potential exists could provide an incentive to participate.

1.4

Demarcation

This project is delimited to focus on the geographical region of Sweden. From an electric power system point of view this naturally includes some aspects of the Nordic system to which Sweden is synchronously connected but the focus is on the Swedish ancillary service markets. Furthermore, some comparisons are made to the situation in other power systems and markets but this is only done with the purpose of further illustrating aspects of the situation in Swe-den. The reason for this is that technology developed for another market might be applicable on the Swedish market which makes it relevant to compare. For instance, the technical requirements on markets in other countries and power systems can be of interest.

(13)

Background

This chapter gives a background to the project in which the relevant systems, technologies, and markets are explained. This includes a brief description of the Swedish electricity system and market, a general overview of ancillary ser-vices, an account of the Swedish ancillary service markets, and the technical capabilities of wind turbines to provide such ancillary services. Lastly, a study of related literature is presented.

2.1

The Swedish electric power system

In this section workings of the Swedish electric power system is detailed, how the trading is done, and who the players are.

2.1.1

System topology

The electric power system is central requisite for the workings of any advanced society. Traditionally, the system is divided into three levels: the national grid level, the regional grid level, and the local grid level. The national grid is de-signed to transmit large amounts of electricity over long distances with as low losses as possible and the voltage level is kept at 220-400 kV, the highest in the system. It can be described as the highways of the electricity system con-necting the relatively large share of hydro power in northern Sweden with the main consumption which is in the central and southern parts. It also connects to its counter parts in the Nordic system and via HVDC-links to some other countries, enabling export and import [5].

Svenska kraftnät is the Swedish transmission system operator which owns

(14)

and operates the national grid [6]. The regional grid has lower voltage levels at 40-130 kV with the objective of connecting the national grid with the main consumption areasi.e., the local grids. The local grids are operated by private actors with concession and the voltage is transformed to first 10-20 kV and then 400 kV where it connects to the normal consumer [5].

2.1.2

Power balance

In an electric power system, there must always be an instantaneous balance between production and consumption. This is due the physical property of electricity of it being transmitted at the speed of light [7]. Furthermore, since alternating current is used, most electricity production is connected to the grid via synchronous generators which are rotating at the same frequency. This global frequency is referred to as the system frequency and is an important metric for evaluation of the stability of the system. Exceptions to generation being synchronously connected are solar power which generates DC current that is transformed to AC current, wind power which is disconnected via elec-tronic converters and power imported via HVDC-links [8].

The nominal system frequency in the Nordic system is 50 Hz but it is al-lowed a variation of ±0.1 Hz [9]. These variations originate from power im-balances in the system where stored kinetic energy in the rotating generators is used to make up the difference. To large variations in frequency may damage electric equipment, leading greater faults, it is therefore of critical interest for the TSO to maintain the frequency close to its nominal value [10]. To achieve this, a division of responsibility, further described in Section 2.2.1, is used and certain ancillary services have been defined, further described in Section 2.3.

2.2

The electricity market

(15)

The financial trading is mainly conducted on the marketplace Nasdaq Com-modities [11]. Wholesale electricity markets are characterized by having volatile prices. The price can in some cases vary by multiples of hundreds or even thousands across one day. Due to many market participants being risk ad-verse, it is common practice to reduce their exposure by trading on various forward markets. This reduces the risk by letting producers and consumers secure a fixed price shielded from the price volatility [12].

2.2.1

Electricity market players

The players of the Swedish electricity market relevant to this study are intro-duced here.

Transmission system operator

The transmission system operator in Sweden is Svenska kraftnät on behalf of the Swedish government [6]. Among the responsibilities of the TSO lies the management of the national grid and ensuring the safe and reliable operation of it. This involves ensuring that the power balance is kepti.e., keeping the frequency close to its nominal value. Svenska kraftnät is also responsible for the preparedness of crises [10].

Electricity producers

An electricity producer is a player that produces and supplies the grid with electricity. This can for example be large scale hydro or nuclear power nected to the transmission grid or smaller wind power plants that can be con-nected to the regional grid. A common definition is that it is the owner of the production facility, but it can also be viewed as the operator of the facility [13].

Electricity grid companies

(16)

that it would be economically unfeasible to allow competition. This means that the revenue framework of the grid companies is regulated [14].

Electricity trading companies

Electricity trading companies buys electricity, usually on the Nord Pool spot market, and sells it on to their customers. The trading of electricity is a liber-alized market with competition, in contrast to what is true for the grid compa-nies. Electricity trading companies are thus free to determine the pricing. To reduce the risk associated with price volatility, many electricity trading com-panies trade with financial contracts on the Nasdaq Commodities market [15].

Balance responsible parties

Determined by law, there has to be a balance responsible party for every outlet point and every inlet point of the grid. This responsibility usually falls to the electricity trading company or power producer, but the service can also be procured by a third party. In order to have this responsibility the party needs to have a contract with Svenska kraftnät. The balance responsible party balances the consumption that it is responsible for with production and trading at Nord Pool. The TSO then calculates any imbalances caused by the balance responsible and adjust the costs accordingly [13].

2.2.2

Electricity trading

In this section, a basic description of the electricity trading in the Nordic sys-tem that is relevant to this thesis is provided.

Day-ahead market

(17)

The daily process of the bidding on the day-ahead market begins at 10:00 CET each day with the publication of available capacities on interconnectors and in the grid. Following this, the buyers and sellers have until 12:00 CET to submit their bids to the auction on Nord Pool for physical delivery hours in the next day. Submitted offers are matched through an advanced algorithm where a single price is set for each hour for each price area where the supply and demand curves meet. This algorithm also takes network constraints in the grid into account. After the clearing prices are announced around an hour after the bid submissions, the results of each individual buyer and seller are reported. This comes with a physical obligation to deliver on the submitted bid, for both a consumer and producer. Any imbalances due to failure of these obligations are dealt with in the balancing market [16].

Intraday market

The intraday market works in conjunction with the day-ahead market to help ensure the balance between production and consumption as the trading within the intraday markets is closer to the physical delivery hour. Keeping the bal-ance is of interest for both market participants and the power system as it re-duces the need for reserves and other associated costs. Furthermore, the in-traday market allows participants to adapt to sudden or unexpected changes in consumption or production capacity. To allow this the market is open 24 hours a day, 365 days a year with 15-, 30-, hourly, and block products to provide the flexibility needed for market participants [17].

Balancing market

(18)

Power purchase agreements

Power purchase agreements (PPA) are a type of future contracts signed bilater-ally between a power producer and a large consumer that is increasingly driv-ing the growth of renewable energy production in the Nordic system, primarily wind power. These agreements allow the producer and consumer to secure the price for a certain volume over a long period in order to limit the risks taken on the highly volatile electricity market. A PPA can be purely financial but is most often physical which means that that the agreement pertains a certain amount of electricity during a period to a pre- determined price. Such an agreement is often demanded from a wind power producer up to a certain share of the projected capacity in order to secure financing for a project. For the consumer party of the PPA, the business logic is firstly risk minimizing in that the actor secures itself from price volatility on the spot market. Furthermore, signing a PPA can serve as a juridical evidence for a consumer that it buys its electricity from renewable sources. This leads to a companies being able to reach their sustainability targets which adds further value to the agreement [19].

2.3

Ancillary services

Ancillary services, defined as such in [20], are a set of functions that support the safe and reliable operation of a power grid. These services are either per-formed by or procured by the transmission system operator (TSO) from other stakeholders and can be used to maintain the desired power flow, help the sys-tem recover after a significant event, handle active power imbalances, and other similar applications. Hertem, Renner, and Rimez [21] state that ancillary ser-vices as per definition are wide-ranging in number and types and by studying the literature a standardized definition of exactly which services are included in the term does not appear to exist but is instead made on a case-by-case basis. The procurement of ancillary services is performed through various mar-kets designed for a specific service and are usually offered from the transmis-sion system point of view. At the moment, distribution systems mainly take part in the control of voltage, but this may change in the future. The design of these markets can have different shapes depending on the ancillary service it aims at trading and differ between systems/countries [22].

(19)

[23]. In this categorization there is an overlap between stability and frequency maintenance such that the stability category seems superfluous. Another at-tempt at classifying ancillary services was made in a CIGRE report on in-ternational practices on ancillary services [24]. In it, there are categories for frequency control, network control (including parts that are not ancillary ser-vices), voltage control and black start. There is also further categorization into services that are instantaneous, continuous, and event-driven. Based on the mentioned sources as well as Holttinen et al. [25] the classification used for this project is proposed in this section with the aim of being clear and sim-ple. The three main categories are as follows. As described in Section 2.4 all Swedish ancillary service markets are within the frequency related services category. For this reason, voltage support and system restoration fall outside the scope of this project.

1. Frequency related services 2. Voltage support

3. System restoration

2.3.1

Frequency related services

The frequency in an AC power system is a global parameter in that it is the same at every measurement point across the entire synchronous system. It is essential that the frequency remains close to its nominal value, 50 Hz in the ENTSO-E area, to maintain a safe and reliable operation of the system. However, there is always a natural fluctuation in frequency due to a continuous mismatch in active power production and consumption. Due to this natural fluctuation a set of parameters have been defined in [26] for evaluation of the frequency in the ENTSO-E area. The definition of these parameters are cited from [26] accordingly:

• Frequency recovery range "means the system frequency range to which the system frequency is expected to return in the GB and IE/NI syn-chronous areas, after the occurrence of an imbalance equal to or smaller than the reference incident, within the time to recover frequency" [26]. • Time to recover frequency "means for the synchronous areas GB and

(20)

• Frequency restoration range "means the system frequency range to which the system frequency is expected to return in the GB, IE/NI and Nordic synchronous areas, after the occurrence of an imbalance equal to or smaller than the reference incident within the time to restore fre-quency" [26].

• Standard frequency range "means a defined symmetrical interval around the nominal frequency within which the system frequency of a syn-chronous area is supposed to be operated" [26].

• Standard frequency deviation "means the absolute value of the fre-quency deviation that limits the standard frefre-quency range" [26].

• Steady state frequency deviation "means the absolute value of quency deviation after occurrence of an imbalance, once the system fre-quency has been stabilised" [26].

The parameter values of the ranges defined above (Frequency recovery, Fre-quency restoration, Standard freFre-quency, FreFre-quency deviation) vary between different systems based on the characteristics of the specific system. Smaller islandic systems such as the GB system for instance, allow greater frequency variation as compared to larger systems such as the interconnected CE system. The reasoning for this stems from the direct relation between the amount of synchronous inertia of a system and the resulting frequency deviation. The same active power imbalance in the GB system and the CE system, will result in a larger frequency deviation in GB, due to the amount of rotating inertia being lower [27]. The parameter values of the frequency ranges in CE, GB, IE/NI and the Nordic system [26] are set according to Table 2.1.

Table 2.1: Frequency parameters [26]

Parameter CE Nordic System GB IE/NI

Standard frequency range [mHz] ±50 ±100 ±200 ±200 Maximum steady-state deviation [mHz] 800 1000 800 1000 Maximum instantaneous deviation [mHz] 200 500 500 500

(21)

replacement reserve or tertiary control which is activated within 10-15 min-utes [28]. Inertial response is a physical property of the grid whereas the three control steps are power reserves with respective markets designed by the TSO. However, markets with reserves aiming at mimicking the inertial response are increasingly common and are usually called Fast Frequency Reserves.

A Inertial response: Inertial response refers to the innate property of large synchronous generator by which additional active power can be extracted from the rotating mass i.e., kinetic energy. This process is automatic and acts to overcome an immediate imbalance between sup-ply and demand of active power in an electrical grid [29]. The trigger can be a sudden load increase, causing a mismatch between electricity production and consumption. The synchronous generators in the grid will respond by releasing some "stored" kinetic energy from the rotat-ing mass of the turbine, compensatrotat-ing for the active power imbalance. By releasing some kinetic energy the rpm of the generators will now decrease and the frequency in the grid is lowered as a result [30]. Any given synchronous systems total inertia is an important parameter for the stability of the system as it has a direct impact on the rate of change of frequency during a power mismatch-event. Systems with a high pen-etration of wind power tend to have a lower system inertia than systems with a larger proportion of traditional power sources (e.g. nuclear, hy-dro, coal, etc.). This is due to the fact that wind power is connected to the synchronous system via power electronics, further explained in Section 2.5.1 [31].

B Primary Control: The first step of the process to restore the frequency to its nominal value is the primary control or frequency containment reserve. Active power reserves are kept being activated within seconds of a dip in frequency. These reserves are activated for a limited period of time [26]. Power companies are compensated through various market structures for providing this service by the responsible TSO [32]. C Secondary Control: The secondary control is made up of another set of

(22)

D Tertiary control: The last step of the frequency control is known as ter-tiary control or the replacement reserve (RR). This active power reserve is made available to support the restoration of the FRR to be prepared for future imbalances in the system. RR is activated manually by the system operator as a result of an optimisation decision [26, 21].

2.3.2

Voltage support and system restoration

Voltage support and system restoration fall within the definition of ancillary services used in this project. However, there are no open markets related to these services in Sweden that wind turbines could participate in. Thus, a brief explanation is given for context, but the information will not be needed for the rest of the document.

Voltage support can broadly be divided into two separate targets. Voltage profile management in steady state and maintaining dynamic voltage stability. Steady state management means keeping the voltage profiles within acceptable limits in the longer term. A time frame of hours is commonly used. Dynamic stability refers to voltage management in the time frame of seconds to minutes. Keeping the voltage within acceptable limits is crucial to the safe and reliable operation of a power system [34].

System restoration relates to services that support an electrical power sys-tem’s return to normal operation after a blackout. From a generator point of view this mainly means black start capability but may in the future also include islanding operation. Islanding operation is the ability of a portion of the power system to maintain operation while disconnected from the rest of the system [35].

2.4

Swedish ancillary service markets

(23)

which are further described in this section. Price and volume development of the markets during 2018, 2019, and 2020 are explored as these are the years that the model explore in Section 3.2.

2.4.1

Frequency containment reserve

The primary control of the Swedish power system is achieved through a fre-quency containment reserve. This has the main objective of stabilizing the frequency after the occurrence of an imbalance. It consists of active power reserves such as generators, storage, and demand response. It is procured for every hour of the year a day in advance and is automatically activated when an imbalance occurs [37]. In the Nordic system FCR is divided into two sep-arate markets with the yearly volume target being decided mutually between the TSOs. FCR-N (Frequency Containment Reserve Normal) stabilises the frequency after differences in production and consumption during normal op-eration and FCR-D (Frequency Containment Reserve Disturbed) is activated during larger disturbances in the system [38].

The yearly volume target of FCR-N in the Nordic system has the last few years been 600 MW. How this volume is divided between the countries is de-cided with an allocation key that has seen Svenska kraftnät procure 225-237 MW over the last five years [39]. FCR-N is a symmetrical product meaning that all bids need to be available for both up and down regulating with the same capacity. It is thus not allowed to bid only downwards regulating for example. If a bid is activated, then the provider is compensated in two steps. One ca-pacity payment according to pay-as-bid where the bid price is regulated based on the cost of the provider, with a margin for profit and risk [40]. An activated bid is subsequently also compensated for the provided energy with is done ac-cording to either the upward regulating- or downward regulating price which set at Nord Pool for the given hour [41].

(24)

amount of energy provided by an FCR-D bid is usually quite small [41]. There are also plans for introducing a new market called FCR-D Down. As the name indicates this is a similar market to the current FCR-D but accept-ing bids in the opposite regulation direction. This market is meant to handle over-frequencies in the grid caused by for instance a sudden trip in an export-ing HVDC line. Accordexport-ing to current plans, this market will start the pre-qualification process at the start of 2021. The technical requirements will be reflected to the already existing FCR-D market but handle bids in the down-ward direction instead. That is, the technical requirements such as activation time etc. will be the same [43]. The technical requirements of both current FCR markets are presented in Table 2.2.

Table 2.2: Technical parameters of the Swedish FCR markets [41]

Product FCR-N FCR-D Minimum bid size 0.1 MW 0.1 MW Activated

Automatically in the frequency range 49.9-50.1 Hz

Automatically when the frequency drops below 49.9 Hz

Activation time 63 % within 60 s and

100 % within 3 min

50 % within 5 s and 100 % within 30 s

Volume target Around 230 MW

for Sweden 573 MW for Sweden

Up/Down regulation Symmetric product Only up Compensation Capacity: Pay-as-bid Energy: Up/down regulating price on Nord Pool Capacity: Pay-as-bid

Time procured One and

two days before

One and

two days before

(25)

the daily average prices across the year is displayed in Figure 2.1. The daily average is used instead of the hourly average price in order to enhance the read-ability of the graph as it reduces volatility.

Figure 2.1: Daily average prices of FCR-N and FCR-D during 2018, 2019, and 2020

(26)

the individual prices of the different price areas.

Figure 2.2: 2018 FCR daily average prices compared to the Nordic system price

Figure 2.3: 2019 FCR daily average prices compared to the Nordic system price

(27)

Figure 2.4: 2020 FCR daily average prices compared to the Nordic system price

FCR price exceeds the Nordic system price. This is a more concrete expression of the relation between the two different prices. It allows an observer to quickly get an idea of how competitive the prices on FCR markets are compared to the spot market. Given a hypothetical market participant that can offer the same capacity on FCR or energy on the spot market but can only decide on one, this represents the proportion of days that the ancillary service markets would be profitable. In the case of a wind farm operator interested in participating in either FCR market, this indicates how often bidding would occur. However, a reader making the comparison should still consider the pay-as-bid structure which means that the compensation likely is slightly different to the average price presented in the graphs.

Table 2.3: Proportion of hours that FCR prices exceed system price

(28)

2.4.2

Frequency restoration reserve

The secondary control in Sweden is performed by the frequency restoration reserve. This consists of two separate markets in the Nordic system. It is the automatic aFRR market and the manual mFRR market. The main purpose of these reserves is to restore the frequency to its nominal 50 Hz after the fre-quency containment reserve has initially halted a deviation in frefre-quency at a constant but outside the nominal level. In order to provide a power reserve to either market a participant need to go through a pre-qualification process in the same manner as that of the FCR markets. In this process Svenska kraftnät checks that the prospective resource can meet all technical requirements of the market [38].

The automatic frequency restoration reserve aFRR is currently procured by Svenska kraftnät a week before the hour of delivery. In order to participate the resource needs to be able to reach 100 % activation within 120 seconds. Bids on aFRR can be either upward or downward regulating separately. This differs from FCR in that pure downward regulation bids are now possible. The mar-ket has both a capacity compensation according to the pay-as-bid structure and an energy compensation according to the Nord Pool hourly upward or down-ward regulating price. The smallest bid size is 5 MW and can afterdown-wards be increased in multiples of 5 MW up to 30 MW which is the maximum [45]. A more comprehensive summary of the technical requirements are presented in Table 2.4.

(29)

Table 2.4: Technical parameters of the Swedish FRR markets [45]

Product aFRR mFRR

Minimum bid size 5 MW 10 MW Activated

Automatically when the frequency deviates from 50 Hz

Manually on the request of Svenska kraftnät

Activation time 100 % within 120 s Within 15 minutes Volume target Around 150 MW for

Sweden

-Up/down regulation Both up and down Both up and down Compensation Capacity: Pay-as-bid Energy: Up/down regulation price on Nord Pool Energy: Up/down regulation price on Nord Pool

Time procured The week before

Bidding up until 45 minutes before delivery hour

(30)

Figure 2.6: 2019 aFRR Down hourly average prices compared to the Nordic system price

Figure 2.7: 2020 aFRR Down hourly average prices compared to the Nordic system price

(31)

months. The aFRR price also exceeds the system price during a larger number of hours.

That not all hours are procured is an important aspect to consider when the economic potential of the market and the plausibility of entering is evaluated. The number of procured hours in the three years that are considered in this the-sis are therefore presented in Table 2.5. Additionally, another important aspect that arises from this fact is which hours during the day the reserve is procured. If this is during high load hours the potential increase in profitability of enter-ing the market decreases as the spot prices would be higher. A distribution of the procured hours have been compiled and is presented in Figure 2.8.

Table 2.5: Total number of procured hours on aFRR Down. Data for 2020 is the first 7295 hours of the year.

Number of hours 2018 1878 2019 2839 2020* 3170

Figure 2.8: Distribution of procured aFRR Down hours across the day

(32)

to 2019 and then continuing to increase in 2020. It should be noted that the data for aFRR Down provided by Svenska kraftnät for 2020 only included the first 7295 hours of the year. This means that the actual outcome will be even more procured hours further consolidating the trend. The time of day these hours have been procured coincides with the peak load during mornings and evenings as can be observed in Figure 2.8. This may be a disadvantage to a prospective participant as the electricity prices will be higher during these hours.

A compilation of the proportion of hours that the average aFRR price ex-ceeds the Nordic system price similar to that made for the FCR markets is of interest. In the same manner, this metric gives a basic understanding of how competitive the market prices are compared to the spot market. It should be noted that a significant proportion of all hours in a year are not procured on aFRR. These non-procured hours obviously have an aFRR price of 0. In or-der to not let this influence the metric, those hours have been excluded in the comparison. The results are presented in Table 2.6.

Table 2.6: Proportion of hours that aFRR Down prices exceed system price. The second column shows proportion for all hours in the data set and the third column excludes non-procured hours. Data for 2020 includes the first 7295 hours of the year.

aFRR Down All hours Excluding 0-price hours 2018 1.8 % 8.5 %

2019 2.5 % 7.6 % 2020* 34.6 % 61.2 %

2.4.3

Fast frequency reserve

(33)

to Svenska kraftnät introducing Fast Frequency Reserve (FFR) in 2020 which is a market for a synthetic inertia reserve meant to strengthen the inertial re-sponse of the system [50].

Currently, FFR is only procured during the summer as this is the period with the most pressing need for synthetic inertia. The underlying reason for FFR is to handle loss of production and this means that the market is only open for upward regulating bids. The technical requirements open up for some vari-ation in the specifics of the bid. The parameters that determine the bids are activation, deactivation, recovery time, and repeatability. There are three op-tions for activation level where it varies slightly. This is further illustrated in Table 2.7 [51].

A FFR providing entity must be able to repeat the cycle 15 minutes af-ter an activation instant. However, the entity can stay active if the frequency stays below 49.8 Hz and start a deactivation sequence once the frequency again exceeds that. During the deactivation sequence there are two options for the support duration requirement. One long support duration option of 30 seconds and one short option of 5.0 seconds. For the long support option there is no limitation on the rate of deactivation, during section (C) in Figure 2.9, which can be step wise if desired. In the case of the short duration option the rate of deactivation is limited to 20 % of the pre-qualified FFR capacity per second with no single step change exceeding this [51].

(34)

Figure 2.9: Illustration of FFR recovery requirements. Modified from [51]

.

Table 2.7: Technical parameters of the Swedish FFR market [51]

Product FFR

Pricing Marginal price

Activation frequency [Hz] 49.5 (A), 49.6 (B), 49.7 (C) Activation time [s] 0.7 (A), 1.0 (B), 1.3 (C) Duration of active

power contribution [s] 5 or 30 seconds

Repeatability Available for reactivation

within 15 minutes

Overshoot Maximum 35 % Volume target 70 MW

(35)

volume target of 70 MW.

Figure 2.10: Weekly accumulated procured volume of FFR during 2020

Figure 2.11: Hourly procured volume of FFR during 2020

(36)

during each hour. This is similar to the workings of the spot market but di-verges from the aFRR, FCR-N, and FCR-D markets. In Figure 2.12 the hourly price has been plotted with the Nordic system price as comparison. While FFR is not activated for an entire hour (the requirements demand an activation time of either 5 or 30 seconds) it is still procured on an hourly basis. This makes the comparison between the FFR capacity price [SEK/MW] and the system price [SEK/MWh] valid in the case where an actor decides to use its capacity to bid on one of the markets.

Figure 2.12: Hourly FFR price during 2020

(37)

International FFR markets

The synthetic inertia capabilities of wind turbines described further in Sec-tion 2.5.2 have been developed as a result of demand from markets in other power systems. The specifics of these capabilities are thus tuned for the tech-nical parameters of those markets which is of importance when assessing the feasibility of wind turbines participating on the newly designed Swedish FFR market. In order to provide a foundation for such an analysis a review of the technical requirements in other markets with FFR-like markets is presented be-low. These markets include the one in the Hydro-Québec power system which in 2010 was the first grid operator to connect wind turbines with FFR capabil-ities. Wind farms larger than 10 MW are required to provide FFR since then and the technical requirements are presented in Table 2.8 [52].

Table 2.8: Technical parameters of the compulsory FFR requirements of wind farms larger than 10 MW connected to the Hydro-Québec system [52].

System Hydro-Québec Activation frequency [Hz] Adjustable in region

-0.1 to -1.0 Hz

Activation time [s] ≤1.5 second Duration of active

power contribution [s] ≥ 9 seconds Recovery time 2 minutes Maximum reduction of

generation during recovery ≤ 20% Operating level over which

availability is required ≥ 25%

EirGrid/SONI is the collaborative grid operator of the combined transmis-sion system operators of Ireland and Northern Ireland. In 2011 a comprehen-sive framework of new system ancillary services was designed under the name DS3. Included in this was a fast frequency response service with it own techni-cal requirements [53]. The technitechni-cal requirements of this system are presented in Table 2.9.

Differences in technical requirements

(38)

Table 2.9: Technical parameters of the FFR requirements in the EirGrid/SONI system as part of the DS3 system services [53]

.

System EirGrid/SONI Activation During frequency

event

Activation time [s] 2 seconds Duration of active

power contribution [s] 8 seconds Recovery

The extra energy provided during activation must exceed the energy reduction

during recovery

is directed at wind power. Since this has led to wind turbines already actively providing this service in the system, the requirements can be viewed as well aligned to the capabilities of wind power. One of those requirements that differ from both the Nordic FFR market and the Irish/Northern Irish market is the size of the power boost (an increase in power output during a short space of time). The power boost is defined to be greater than 6 % of the rated capacity in Hydro-Québec. In contrast, the Swedish market accepts bids of a size de-fined by the market participant itself. Svenska kraftnät has dede-fined FFR as a resource neutral market, similar to the EirGrid/SONI market, which explains the reasoning for this. Furthermore, the activation time in the Hydor-Québec market is higher than that of the Swedish requirements. A longer activation time means that an actor has more time to increase its production when the bid is activated.

(39)

2.4.4

Planned ancillary service market changes by

Sven-ska kraftnät

Svenska kraftnät are planning to implement some changes to the markets that will affect the possibilities of wind power to participate. The first change, or rather introduction, is the formation of the new market FCR-D Down. This market is briefly described in Section 2.4.1, and will in essence be the same as the already existing FCR market but instead handle downwards regulating bids. It will be similar in the sense that all technical requirements will mimic those of the already existing market. This market represents a good opportu-nity for wind power actors to join as downwards regulation is better adapted to the conditions of wind power. Furthermore, the bids on the market will only be activated during major disturbances. A prospective wind farm would pro-duce at the optimal level only to curtail the production when a bid is activated. Thus, being able to generate spot market income whenever the bids are not activated [43].

Additionally, Svenska kraftnät will also conduct a review of the technical requirements of FCR. The purpose of this is to investigate the possibilities of simplifying the testing procedure. Furthermore, there is a goal to determine a common Nordic structure of FCR for both the technical requirements and pre-qualification process [54]. In conclusion, this may lead to a situation where new market participants experience lower entry barriers.

A common Nordic market for aFRR will be also introduced in 2021 as op-posed to the Swedish market currently in operation. This will include bigger volumes and more operating hours than the current market. Contrary to the current situation where aFRR is procured a week before the operating hour, this new market will be procured the day ahead similarly to how FCR is pro-cured. The accepted bid size will also change. Where it currently only accepts bids in blocks of 5 MW, the change will lead to bids of at least 1 MW being accepted. The pricing will also be according to marginal price and not pay-as-bid as is currently the case [54].

2.4.5

Market turnover projections

(40)

this represents the total costs of ancillary services to the system that is being paid by Svenska kraftnät to participants. This serves as a basis for any future analysis of potential market share. Included in the total cost are the markets described previously in this Section: FCR-N, FCR-N, aFRR, and mFRR. The costs up until 2019 are based on the actual outcome whereas 2020 and onwards are projections. In these projections, FFR and FCR Down are included [55]. The published results by Svenska kraftnät are adapted into Figure 2.13.

Figure 2.13: Total cost of all Swedish ancillary service markets [55]

(41)

Figure 2.14: The total cost of the ancillary service markets explored in this thesis

2.5

Wind turbines

In this section the categorization of different types of wind turbines used in this study is introduced and the technical capabilities of those to provide ancillary services are discussed.

2.5.1

Wind turbine types

This section introduces the technical characteristics that define different types of wind turbines and compares them. Traditionally, this has been divided into four separate types according to the standard IEC 61400-27. Type 1 and 2 are characterized by being synchronously connected to the grid. These types are becoming increasingly rare and most newly installed models are type 3 or 4. Due to being synchronously connected to the grid, type 1 and 2 do not have the capabilities discussed in Section 2.5.2 that relates to providing the ancillary services previously discussed [56]. Therefore, type 3 and 4 are more relevant to this project and are described below.

Type 3 wind turbine

(42)

voltage source converter connecting the machine rotor windings and the power grid. This enables an independent control of the production and consumption of reactive power to a certain extent. Furthermore, the power converter largely decouples the mechanical drive train from the electrical grid providing less fluctuation in power and less stress on the mechanical system [57].

Having a power converter connected between the rotor windings and the power grid also allows for some speed control. The operating speed range of type 3 wind turbines depend on the back-to-back converter ratings with respect to the ratings of the generator. Conventionally, it is common practice to size the power electronic converter to around 25-35 % of the generator rating due to economic reasons. This allows for speed control to an operating speed of ±25-35% of the rated speed [56].

Type 4 wind turbine

The type 4 wind turbine concept has been designed using squirrel cage in-duction generators, wound field synchronous machines, and permanent mag-net synchronous machines, with geared and gearless (direct drive) concepts. What separates type 4 from type 3 is that the back-to-back voltage source con-verter has the same rating as the generator, and that the generator has no direct connection to the power grid. Type 4 turbines are thus completely decoupled from the grid unlike type 3 where there is still a connection between the stator windings and the grid. This allows the generator to operate at any speed from zero to maximum rated speed as well as greater reactive power control than in type 3 [56].

Due to the converters fully matching the rating of the generators, type 4 turbines are typically more expensive than type 3. However, the setup means that active and reactive power is fully controlled in the converter. This control is also faster than that in type 3 turbines. In summary type 4 is more efficient, more reliable (does not need a gearbox), and have relatively higher reactive power capabilities than type 3 [56].

2.5.2

Wind turbine capabilities

(43)

Active power control

Active power control is an important functionality of wind power turbines with respect to providing ancillary services. Ramping up and curtailment of the ac-tive power output is possible with most available type 3 and 4 wind turbines [56]. This can be achieved through pitch regulation i.e., where the blades of the turbine are turned to change the pitch angle. This will create less lift and more drag in order to decrease or increase the rotational speed of the blades. Pitch regulated control is usually used during winds above rated speed in order to limit the rotational speed under a certain threshold. The reason for this is to limit the physical forces on the components of the turbine. However, this capability can also be used to participate in the primary control of the power system through the ancillary service markets. By pitching the blades and thus controlling the active power output, it is possible to provide both an upward and downward regulating service[58].

In the case of downward regulation, the wind turbine produces at the opti-mal level, only to be curtailed when required by the power system. If upward regulating is preferred, the power production is kept at a curtailed level. The margin between the curtailed power output level and the optimal production level is then used for the upward regulating capacity bid. When activated the power output is ramped up to the level required by the power system [58].

Inertial response

Regarding inertial response wind turbines can be divided into two groups. Fixed/semi fixed-speed turbines i.e., type 1 and 2, and variable-speed turbines i.e., type 3 and 4. Both type 1 and type 2 innately provide some rotational in-ertia as they are synchronously connected to the system. Type 3 and 4 do not have this innate ability due to being semi- or fully decoupled from the system via the power converters. However, with additional control systems and aux-iliary equipment it is possible to provide synthetic inertia. That is, controlling the turbines output to mimic the behaviour of real inertia [56].

There are two different situations in which wind turbines can provide syn-thetic inertia and mainly two different subsequent strategies. During periods with low wind speeds below rated speed, Vr, (but higher than the cut-in wind

(44)

Figure 2.15: A generic power curve of a wind turbine. Vinis the cut-in wind speed of the turbine, Vris the rated speed of the wind turbine, and Vout is the cut-out wind speed of the turbine.

extracted and released as output to the grid. To store enough energy, the rotor is usually rotating slightly above the optimal rpm before the event. After re-leasing the excess energy, the turbine needs some time to regenerate the kinetic energy and get up to speed again. Over time this strategy is essentially energy neutral in that it does not reduce the total energy output significantly [56].

(45)

2.6

Previous studies

The state of research in the area of this study is analyzed in this section both through a bibliometric analysis and a short literature study.

2.6.1

Bibliometric analysis

To provide an overview of the current and previous state of research on the subject a bibliometric analysis was carried out on 23 September 2020. The well-established scientific database Web Of Science (WOS) was used for this purpose. A search query according to TS=("wind" AND ("power" OR "tur-bine" OR "turbines") AND (("frequency" AND "ancillary" AND "service") OR ("synthetic" AND "inertia"))) was explored.

Figure 2.16 shows the number of academic publications as indexed by WOS across time. In total, 205 publications were found when employing the search query. By limiting the search based on the year published, the trend of research on the topic can be observed accordingly. The first studies found with the search query were just before 2010. Afterwards, a clear increase in the years 2015-2018 up until today is observed.

(46)

2.6.2

Examples of previous studies

Several studies as well as pilot projects have been made exploring the poten-tial of using wind power for various ancillary services. In December 2019, the Danish TSO Energinet published "Ancillary services from new technolo-gies - Energinet". This report presents an overview of how different emerg-ing energy technologies can be used for ancillary services, among which is wind power. The chapter on wind power describes the technical capabilities to provide different types of frequency support, inertia support, and voltage & reactive power support [59].

In [60] the authors propose a fast frequency support scheme in which wind turbine systems are used. In the proposed scheme, the rotor speeds of the wind turbine systems are proposed to be fully recovered to the optimal operating points during the secondary frequency control instead of during the primary frequency control. This enables the frequency nadir (the lowest point after a drop) to be raised significantly and eliminate a second frequency dip. Results are simulated on two different IEEE test systems to verify the effectiveness of the scheme.

A German study [61] aim to show the possibilities of pitch-controlled wind power plants to provide FCR and FRR with respect to the dynamic system be-haviour. By simulating a system with an integrated wind park the study shows that pitch-controlled wind power plants have the possibility to support other conventional power producing units in providing ancillary services but not as a stand-alone solution. That is, wind power is capable of providing primary frequency response but participation from other sources is needed to guaran-tee system stability.

(47)

Methods

The methods used in this study consists of three semi-overlapping phases. First, a reading phase in which the focus was to gain an understanding of the subject area and which aspects had already been studied. Based on this, the purpose and goal of the project as well as a time plan was specified in com-munication with the supervisors and examiner. After this initial phase of the project, an interview study and subsequent modeling of a wind farm operating on the Swedish ancillary service markets was conducted. In this chapter the methodology leading to the choice of the method is discussed and the specifics are detailed.

3.1

Methodology

Given that part of the purpose of the study was to investigate the attitude of ac-tors in the wind power industry towards participating in the ancillary service markets, an interview study appeared a natural choice of method. In con-junction with a review of relevant literature, an interview study of industry actors is therefore used to explore the research questions stated in Section 1.2. In order to have some leeway to probe respondents for specific knowledge a semi-structured approach was used. This still allows for the views of many respondents to be explored systematically and comprehensively with the use of an interview guide [67]. Lastly, a model is built which simulates a wind farm’s electricity production. This model is then used to investigate different bidding strategies on the markets for ancillary services using historical data of the wind, electricity price, and price on the ancillary service markets. In essence, the model is an illustrative case study with the purpose of showing what the impact of bidding on these markets could be for the profit of a wind

(48)

farm.

Sampling is of critical importance to any interview study, both for the quantitative and qualitative aspects of it. In order to reach the goal of this study the sampling followed a four-step approach proposed in [68]. In the first step a sample universe is specified by choosing which inclusion and exclusion criteria apply for participation. The second step is deciding a sample size in which a balancing of epistemological and practical concerns is made. The third and fourth steps are choosing a sampling strategy and sample sourcing respectively.

In the first step the sample universe is specified. Sample universe refers to the total population that the subsequent sample can be drawn from and is spanned by the inclusion criteria. Inversely, the exclusion criteria define the area of the total population from which the sample cannot be drawn. For this study the set of inclusion criteria were two-fold due to two samples being drawn. The first set was defined as "a representative of a wind turbine manu-facturer and with knowledge of the technical capabilities of the wind turbines to provide ancillary services". For the second set, the criteria were "repre-sentative of a wind power industry player with knowledge of the company’s organization and operation". The exclusion criteria were anyone not fitting the description and with the addition of them not being a representative of a wind turbine manufacturer in the second set (as this was included in the first set).

The sample size is as previously mentioned a balancing act between epis-temological and practical concerns. That is, a larger size, while not the only factor that influence generalisability, is preferred as it generally leads to a better representation of the population. However, it is desirable to set a provisional number of respondents during the design stage in order to limit the duration of the project. This does not necessarily mean inflexibility to change since an appropriate range of respondents can be determined instead of a fixed num-ber. There are according to Silverman [69] strong grounds to alter sample size within agreed parameters as a project progresses when collecting qualitative data as collecting qualitative data may lead to challenges that are difficult to predict. The size of the first sample set of wind turbine manufacturers was decided to be 5 respondents. For the second set a range of between 10 to 20 respondents was chosen, depending on the quality of responses.

(49)

to find the respondents to be included in the sample. A sample choice strategy can broadly be divided into two categories; (1) random/convenience sampling strategies and (2) purposive sampling strategies [68]. For this study, a conve-nience sampling strategy was employed in which the contacts of Svensk Vin-denergi were approached with a first-come-first served approach by selecting willing participants until the sample size quotient was filled. The reasoning behind the validity of this strategy choice was that the sample universe had been restricted to players in the wind power industry which already would make potential generalisation of the results narrowed and justifiable.

Having established an understanding of the possibilities and specific con-ditions setting the framework of potential bidding strategies through the inter-view study, a case study modelling of the economic potential of a wind farm operating on the ancillary service market was used. The purpose of this was to answer the research question on how providing ancillary services can impact the financial structure and profitability of a wind farm. Any case study is re-stricted in its generalisability by the conditions of the case. The generalisabil-ity depends on the topic of investigation, the data used, and the conclusions drawn. Furthermore, any model will also be restricted by the assumptions made in it. This is important to consider when analysing the results.

However, the value in producing a model for economic evaluation of a specific case is that historical market and weather data can be used. This means that the correlation between electricity prices and wind conditions is naturally included. Furthermore, the structure of the model can easily be adapted to any wind farm and serve as inspiration for other case evaluations. Since the wind farm evaluated in the model is generic in its layout and technology choices, the results can be viewed as a baseline for the potential of any wind farm. Assumptions were made in a conservative fashion when possible in order to limit overestimation in the results.

3.2

Simulation model

(50)

section below.

3.2.1

Turbine specifications

The wind turbine that was used in the model is a generic 4 MW wind turbine. The reasoning for using this was that the necessary data of the power curve and thrust coefficient as well as data such cut-in wind speed, rotor diameter, hub height, etc. was readily available. The relevant specifications of the turbine is presented in Table 3.1. Additionally, the power curve of the selected turbine is presented in Figure 3.1.

Table 3.1: Specification for the wind turbine used in the model [70]

Specification Turbine

Capacity [kW] 4000 Cut-in wind speed [m/s] 3 Cut-out wind speed [m/s] 24.5 Number of blades 3 Hub height [m] 155 Rotor diameter [m] 150

(51)

3.2.2

Wind farm conditions

The site selected for the fictional wind farm is just outside of Hällum in south/-central Sweden, price area SE3. Any issues relating to grid connection, land leasing, etc. that could have impacted a real project has been disregarded as it is irrelevant for the purpose of this study. Instead, it was assumed that the wind farm is already built and operational on the site, allowing for different bidding strategies to be explored. The reason for choosing the specific spot is that it is close to one of SMHI’s wind metering stations from which wind data is openly available and used in the simulation. A summary of the prevailing wind directions is presented as a wind rose in Figure 3.2.

A specific layout of the wind farm was determined in order to include the wake effect in the model. First a decision on the total capacity of the farm was taken at 96 MW. This size allows the wind farm sufficient margins to partici-pate on the ancillary service markets that are explored. A total capacity of 96 MW results in 24 individual wind turbines that are placed in three lines fac-ing the prevailfac-ing wind direction. The distance between adjacent turbines was selected as 4 rotor diameters and the distance between the rows was selected as 7 rotor diameters as this would limit the wake effect while still covering a reasonably small area of land. Based on this layout the wake effect in the wind farm could be calculated.

The wind data measurements are taken at 10 meter above sea level. In order to simulate wind conditions at hub height the wind profile power law is utilized to scale up the wind speeds. The wind profile power law relates wind speeds at one height to another according to equation 3.1 where α is an empirically derived coefficient that varies with different parameters such as elevation, time of day, season, nature of terrain, wind speed, temperature, etc. However, these variations impact the simplicity and applicability of the wind profile power law and multiple empirical methods for calculating α have been developed. The one used in this thesis was developed by Justus et al. and is expressed according to Equation 3.2 [71].

(52)

Figure 3.2: Wind rose of the wind data used for the site

Where,

u = Wind speed at hub height uref = Wind speed measurement

z = Hub height

zref = Height of wind speed measurement

(53)

wind conditions at Hällum with conditions on other sites. In the model, the actual data points are used to simulate the production outcome.

k = σ U −1.086 (3.3) c = U Γ(1 + 1k) (3.4) Where,

σ = The standard deviation U = The mean wind speed

Γ = The gamma function

Table 3.2: Weibull parameters of the raw wind speed data and the up-scaled data set

Data set Raw data Up-scaled data Mean wind speed [m/s] 3.14 6.23

Standard deviation 2.36 3.69 Shape factor 1.36 1.77 Scale factor 3.43 6.99

(54)

Figure 3.3: Wind farm layout and wake effects

3.2.3

Production estimation error

To include the inherent uncertainty of wind power production estimations in the model, a Monte-Carlo approach was used for the production estimation er-ror. This means that a production estimation is determined by the model using the historical wind data. This estimation is used for day-ahead bidding on the spot market or one of the ancillary service markets. Afterwards, the produc-tion outcome is determined by multiplying the outcome of the Monte-Carlo simulation of the production estimation error as a factor.

In broad terms, the production estimation and production outcome are cal-culated in two sequential steps. Firstly, the historical wind data combined with the wind turbine data and wake effects as described previously in this chapter are used to calculate a production estimation for each hour. Secondly, the in-verse transform method is used to determine the production estimation error in order to generate the power production on the hour of delivery. This pro-duction outcome is then the basis for the income calculations.

(55)

Figure 3.4: Probability density function fXof the production estimation error on a 12h fore-cast horizon [72]

fX of the production estimation is needed. This estimation may differ

depend-ing on a case-by-case basis dependdepend-ing on the accuracy of the prognosis tool used. However, for the purpose of this thesis a broad estimate is considered satisfactory. Herre et al. explored this production uncertainty of wind power in "Exploring Wind Power Prognosis Data on Nord Pool: The Case of Sweden and Denmark" and produced a distribution for a 12h forecast horizon accord-ing to Figure 3.4 [72]. This same forecast error distribution is assumed for each of the forecasted delivery hours in the model. In reality, the uncertainty would vary across a forecasted day, but this effect has been disregarded in the model.

Utilizing the inverse transform method is done by first generating a pseudo-random number U from a uniform distribution Unif(0,1). This number is then plugged in to the inverse cumulative distribution function to generate a fac-tor X according to X = FX−1. The cumlative distribution function FX of the

References

Related documents

Stöden omfattar statliga lån och kreditgarantier; anstånd med skatter och avgifter; tillfälligt sänkta arbetsgivaravgifter under pandemins första fas; ökat statligt ansvar

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

Generally, a transition from primary raw materials to recycled materials, along with a change to renewable energy, are the most important actions to reduce greenhouse gas emissions

För att uppskatta den totala effekten av reformerna måste dock hänsyn tas till såväl samt- liga priseffekter som sammansättningseffekter, till följd av ökad försäljningsandel

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Syftet eller förväntan med denna rapport är inte heller att kunna ”mäta” effekter kvantita- tivt, utan att med huvudsakligt fokus på output och resultat i eller från

Industrial Emissions Directive, supplemented by horizontal legislation (e.g., Framework Directives on Waste and Water, Emissions Trading System, etc) and guidance on operating

The predicted impact of an increased number of wind power plants would be positive, since an additional wind power plant increase the potential for wind