Gotland as a microgrid - Energy storage systems frequency response in grids with high level of renewable energy penetration

(1)

Juni 2018

Gotland as a microgird - Energy storage systems frequency response in grids with high level of renewable energy penetration

Firas Qasem Daraiseh

Masterprogram i förnybar elgenerering

(2)

Teknisk- naturvetenskaplig fakultet UTH-enheten

Besöksadress:

Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0

Postadress:

Box 536 751 21 Uppsala

Telefon:

018 – 471 30 03

Telefax:

018 – 471 30 00

Hemsida:

http://www.teknat.uu.se/student

Gotland as a microgird - Energy storage systems frequency response in grids with high level of renewable energy penetration

Firas Qasem Daraiseh

The Swedish island of Gotland , situated about 100 km from mainland Sweden in the Baltic Sea, represents a power system with a high wind power penetration. The island is connected to the mainland Sweden exclusively via two HVDC cables that provide the only source of active power and frequency control. The two cables can operate in different configurations, i.e. import or export power from or to mainland. However, in order to ensure the N-1 criterion, one of the cables currently always must import power from the mainland. This means that the available power exporting capacity is limited to the rated power of one of the cables.

Therefore, in the case of having a fault on the exporting HVDC cable during low load demand and high wind power production, the power system will suffer from high active power transients that will increase the frequency above the acceptable threshold.

Consequently, the protection system will trip the over-frequency relays, triggering cascading outages on the island that might eventually lead to blackout if the problem is not addressed correctly. Thus, increasing the renewable energy production on Gotland is currently considered as a risk that will increase the probability of instable over-frequency contingencies. This has led the local grid operator to cap the installed wind power capacity to its current level. Therefore, the ability to preserve the stability of the power system during islanded operations until the HVDC cables fault is cleared or the emergency reserves are online is essential for the growth of installed wind power capacity.

The main objective of the thesis is to examine the capability of a centralized energy storage along with or without wind curtailment.

The ESS is tested for maintaining the frequency stability during the unintentional islanding through dynamic studies using the software PSS/E. The results show that an ESS prevents frequency instabilities and provide frequency response during HVDC cables fault albeit of the absence of any form of rotating inertia. The results show that for today’s 185 MW of installed wind power capacity, an energy storage of 50 MW power capacity will reduce over-frequency instabilities in the case of HVDC cables fault from 13% to 1%. The analysis finds that the power capacity of the

energy storage depends on the exported power from the HVDC cables at the instant of fault, which eventually relates to the installed

wind power capacity. finally, the study shows that using wind power curtailment will significantly decrease the energy capacity of the energy storage.

TVE-MFE 18 002 Examinator: Irina Temiz Ämnesgranskare: Urban Lundin

Handledare: Vincent Gliniewicz & Erica Lidström

(3)

Dissertation in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE IN RENEWABLE ELECTRICITY PRODUCTION

Uppsala University

Department of Engineering Sciences Division of Electricity

[Firas Daraiseh]

[dd MON yyyy]

(4)

Dissertation in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE IN RENEWABLE ELECTRICITY PRODUCTION

Uppsala University

Department of Engineering Sciences Division of Electricity

Approved by

University Supervisor, Prof. Urban Lundin

Vattenfall Supervisor, Vincent Gliniewicz Erica Lidström

Examiner, Prof. Juan de Santiago

[2018]

(5)

Abstract

The Swedish island of Gotland , situated about 100 km from mainland Sweden in the Baltic Sea, represents a power system with a high wind power penetration. The island is connected to the mainland Sweden exclusively via two HVDC cables that provide the only source of active power and frequency control. The two cables can operate in different configurations, i.e. import or export power from or to mainland. However, in order to ensure the N-1 criterion, one of the cables currently always must import power from the mainland. This means that the available power exporting capacity is limited to the rated power of one of the cables. Therefore, in the case of having a fault on the exporting HVDC cable during low load demand and high wind power production, the power system will suffer from high active power transients that will increase the frequency above the acceptable threshold. Consequently, the protection system will trip the over-frequency relays, triggering cascading outages on the island that might eventually lead to blackout if the problem is not addressed correctly. Thus, increasing the renewable energy production on Gotland is currently considered as a risk that will increase the probability of instable over-frequency contingencies. This has led the local grid operator to cap the installed wind power capacity to its current level. Therefore, the ability to preserve the stability of the power system during islanded operations until the HVDC cables fault is cleared or the emergency reserves are online is essential for the growth of installed wind power capacity.

The main objective of the thesis is to examine the capability of a centralized energy storage along with or without wind curtailment. The energy storage system (ESS) is tested for maintaining the frequency stability during the unintentional islanding through dynamic studies using the software PSS/E. The results show that an ESS prevents frequency instabilities and provide frequency response during high-voltage direct current (HVDC) cables fault albeit of the absence of any form of rotating inertia. The results show that for today’s 185 MW of installed wind power capacity, an energy storage of 50 MW power capacity will reduce over-frequency instabilities in the case of HVDC cables fault from 13% to 1%. The analysis finds that the power capacity of the energy storage depends on the exported power from the HVDC cables at the instant of fault, which eventually relates to the installed wind power capacity. finally, the study shows that using wind power curtailment will significantly decrease the energy capacity of the energy storage.

(6)

Acknowledgements

Along the route I trode there were peaks and valleys, it has been a long journey since I started my master’s degree at Uppsala University. But time flies leaving its shadow behind and what felt like yesterday is coming for an end.

Firstly, I would like to thank Vattenfall Research and Development AB for giving me the opportunity to do my master’s thesis with them. And I would like express my sincere gratitude for both of my supervisors at Vattenfall, Vincent Gliniewicz and Erica Lidström. I would like to cordially thank you for the great times we had, the laughs in our meetings, and the continuous support.

Secondly, I would like to thank Uppsala University for the two great years that I have spent there. I would like to thank my master’s thesis University supervisor, Urban Lundin, for the close follow up and the offered time. And I would like to show my appreciation for my program study counselor, Juan de Santiago, for the endless help.

Finally, I am happily expressing my cordial gratitude to my family. Mom and dad, I do believe that it is quite simple to write my thanks, but words will never express how much I appreciate that you are always there for me and always been by my side.

(7)

List of Tables

Table 4.1: Energy storage energy capacity - present case study . . . . 22 Table 4.2: Energy storage energy capacity - future case study 1 . . . . 35

(10)

List of Figures

Figure 2.1: Energy storage technology characteristics . . . . 6

Figure 2.2: ESS basic topology . . . . 7

Figure 2.3: Gotland’s illustrative microgrid . . . . 8

Figure 2.4: HVDC cables different configurations . . . . 9

Figure 3.1: Gotland Blurred Microgrid . . . . 11

Figure 3.2: ESS PSS/E general model . . . . 12

Figure 3.3: REGC_A generator/converter generic model . . . . 12

Figure 3.4: REEC_C electrical controller generic model . . . . 13

Figure 3.5: ESS operating P-Q diagram . . . . 13

Figure 3.6: REPC_A plant controller generic model . . . . 14

Figure 3.7: REEC_A electrical controller generic model . . . . 15

Figure 3.8: Droop characteristics sample . . . . 16

Figure 4.1: Gotland’s simplified single line diagram . . . . 21

Figure 4.2: Frequency response of present study case . . . . 22

Figure 4.3: Present case study hourly data . . . . 23

Figure 4.4: Average utilization factor of diurnal present case study data . . . . . 24

Figure 4.5: Diurnal utilization factor of present case study box-plots . . . . 24

Figure 4.6: HVDC cables average diurnal utilization factor of present study case data . . . . 25

Figure 4.7: HVDC cables diurnal utilization factor of present study case box-plots 25 Figure 4.8: Averagy monthly utilization factor of present study case data . . . . 25

Figure 4.9: CDF of HVDC cables power flow - present case study . . . . 26

Figure 4.10: Frequency response with centralized ESS - present case study . . . . 27

Figure 4.11: Active power of the ESS - present study case . . . . 28

Figure 4.12: Frequency response with centralized ESS and different wind curtailment options - present case study . . . . 29

Figure 4.13: Active power of the ESS with different wind curtailment options - present case study . . . . 29

Figure 4.14: Energy storage’s energy capacity for different durations - present case study . . . . 30

Figure 4.15: Active power at the terminal of the WTGs - present case study . . . 30

Figure 4.16: ESS power capacity sensitivity analysis . . . . 31

Figure 4.17: p(P_HVDC<=0)sensitivity analysis . . . . 32

(11)

Figure 4.18: ESS power capacity sensitivity analysis for Installed wind power capacity . . . . 33 Figure 4.19: p(P_HVDC<=0)sensitivity analysis for Installed wind power capacity 33 Figure 4.20: ESS power capacity sensitivity analysis for Installed solar power

capacity . . . . 34 Figure 4.21: p(P_HVDC<=0)sensitivity analysis for Installed solar power capacity 35 Figure 4.22: Future case study hourly estimated data . . . . 35 Figure 4.23: CDF of HVDC cables power flow - future case study . . . . 36 Figure 4.24: Frequency response with centralized ESS- future case study . . . . . 37 Figure 4.25: Frequency response with centralized ESS and different wind curtail-

ment options - future case study . . . . 38 Figure 4.26: Active power of the ESS during an HVDC cables fault with different

wind curtailment options - future case study . . . . 38 Figure 4.27: Energy storage’s energy capacity for different durations - future case

study . . . . 39 Figure 4.28: Active power at the terminal of the WTGs plant - future case study . 39

(12)

List of Acronyms

AC alternating current

CIGRE International Council for Large Electric Systems CDF cumulative distribution function

DC direct current

DER distributed energy resource DNO distribution network operator

DOE Department of Energy

ESS energy storage system

EPRI The Electric Power Research Institute

GEAB Gotlands Energi AB

HVDC high-voltage direct current

IEC International Electrotechnical Commission IEEE Institute of Electrical and Electronics Engineers RES renewable energy resource

SCES super capacitor energy storage

SOC state of charge

TSO transmission system operator

PSS/E Power System Simulation for Engineering WECC Western Electricity Coordinating Council

WTG wind turbine-generator

(13)

1 Introduction

The world present energy politics are transforming toward the commitment for reducing greenhouse emissions, eliminating air pollution, securing energy, controlling climate changes, and facing the global warming phenomena in hopes of meeting the Paris climate agreement that was adopted on the 12th of December 2015 [1]. Therefore, the world must march towards meeting the agreement goals by increasing the hosting capacity for the renewable energy resources (RESs) in the next three decades [2]. In this chapter the brief introduction of the thesis problem, motivation, objectives, and the the thesis outline will be demonstrated.

1.1 Motivation and Problematization

Gotland is one of Sweden’s most resourceful territories in terms of wind and solar energy, which is connected to mainland Sweden only via the two aforementioned HVDC cables.

Nonetheless, due to the HVDC cables configurations restriction and the concern of losing connection to the synchronous grid, the installation of new RESs has been halted until further notice. Moreover, as of 2017, Svenska Kraftnät The Swedish transmission system operator (TSO) canceled any plans of installation of any new transmission capacity between Gotland and mainland Sweden [3]. Thus, the necessity for innovative solutions must be investigated, which allows Gotland to exploit its RESs and increase the penetration level without reducing the reliability.

Currently, Gotland power system is representing an HVDC connected microgrid that is only run in grid-connected mode. Moreover, the two HVDC cables that are connecting Gotland microgrid can operate in different configurations,i.e. importing or exporting power from or to mainland Sweden electrical grid. Furthermore, export configuration is only used to export power to mainland Sweden when the generated renewable power is higher than the load demand. In order to ensure the N-1 criterion, one of the cables always has to import power from the mainland, which means that the available exporting power capacity is limited to the rated power of one of the cables. Consequently, losing one or both of the HVDC cables will deprive the island from exporting excess generated power to Sweden mainland. Thus, in the case of having a fault on the exporting HVDC cable or both HVDC cables during low load demand and high renewable power production, the power system will suffer from high active power transients and over-frequency. The active power transients will increase the frequency of electrical grid beyond acceptable levels, which will trip the over-frequency relays of the local

(14)

wind plants triggering cascading outages on the island that might eventually lead to blackout. The blackout will lead to the shutdown of every RES until the emergency back-up generators ramp-up or the HVDC cables get reconnected unless the problem of active power transient and over-frequency is addressed correctly.

Since increasing the renewable energy production on Gotland is considered as a risk that will increase the odds of instable over-frequency and active power transients contingencies, the first approach to Gotland’s problem has led the local grid operator, Gotlands Energi AB (GEAB), to cap the installed RESs capacity to its current level. Thus, facilitat- ing the power system stable transition and operations as a microgrid in islanded mode is crucial to the growth of the renewable energy production capacity on Gotland. The master thesis will focus on studying energy storage systems as the technical solution that will address the frequency stability issues and enable the stable transition Gotland microgrid until the fault is cleared or emergency reserves become online.

1.2 Objectives

The broad objective of the thesis is to investigate the solutions for increasing the renewable energy penetration in electrical grids in general and on Gotland specifically and tackling the challenges that are mentioned in section 2.1. However, the thesis main objective will focus on studying ESSs as a solution that will provide frequency stability for over-frequency events during low load and high renewable power generation periods.

The case study that will be analyzed is Gotland mircrogrid and it will be examined for ESS frequency response during the HVDC cables fault under different levels of exported power, which eventually translates into different levels of RESs penetration.

The thesis will aim to accomplish the following tasks respectively to achieve the main objective for the study:

• The descriptive analysis of the HVDC cables power flow patterns, renewable energy generation profiles, and load demand profiles for the previous three years on Gotland microgrid case study.

• The examination of Gotland present microgrid transient response to the unintentional islanding due to the loss of the HVDC cables while exporting high level of excess renewable power.

• The development of a centralized ESS model that can be utilized in Gotland microgrid as a solution to preserve the system frequency stability.

• The investigation of the integration of the centralized ESS effect on the transient response after the unintentional islanding.

• The development of wind plant models that can automatically curtail power as frequency down regulation in the case of over-frequency conditions, which can be integrated in Gotland microgrid model.

(15)

• The investigation of the integration of the controllable wind plants with the centralized ESS in Gotland microgrid. In addition to, analyzing the effect on the transient response after the unintentional islanding.

• The analysis of the required ESS along with and without wind curtailment power and energy capacities that secures the seamless unintentional islanding for different scenarios of high levels of renewable energy penetration.

1.3 Thesis Outline

The thesis start with a brief introduction in chapter 1 about the challenges, current situation, the objectives, and the focus of the thesis. The literature review and recent research background is examined in chapter 2. The methodology, the power system and its components modeling, and the theoretical background are discussed in chapter 3.

The results of our simulations are presented in chapter 4. The discussion and the analysis of our results is presented in chapter 5. Finally, chapter 6 concludes the thesis findings and discuss future work insights.

(16)

2 Literature Review

The growing penetration of RESs lead to new challenges that reduced the overall stability of electrical grids. Gotland microgrid stability has also suffered from the high level of installed wind power capacity. In this chapter a brief review of the challenges introduced by RESs, Gotland specific case, and ESSs technologies will be examined.

2.1 Challenges Introduced by High Renewable Energy Penetration on The Electrical Grid

Recently, the renewable energy penetration has been increasing steadily to reduce the greenhouse gases emissions, meet the global and local environmental targets, reduce the electrical operating costs, and guarantee cost-effective energy resources. However, this growth has its tolls on the electrical power system transient and steady-state operations, which forces TSOs, distribution network operators (DNOs), and electric utility service providers to investigate the challenges that affect the grid reliability and seek solutions to operate at the highest possible level of reliability, stability, and security [4]. Consequently, the obstacles have to be identified in order to facilitate the renewable energy generation growth .

Classical power systems in a typical electric grid relies on dispatchable generators such as hydroelectric power plants and natural gas power plants to balance the electric power system load and peak demand. Furthermore, it relies on supplying the base load by nuclear or coal power plants. However, the introduction of RESs is transforming the classical power system. Nowadays, RESs are considered, usually, non-dispatchable resources that produces energy when the resources are available in a variable manner.

The variability introduced from the renewable energy generation, load demand, and the mismatch between renewable energy generation and load demand will create an augmented need for high ramp-rate dispatchable generators and accurate forecasts to accommodate for those issues.

The power system stability is defined by Institute of Electrical and Electronics Engi- neers (IEEE) and International Council for Large Electric Systems (CIGRE) as the ability of the interconnected electric power system as a whole to recover into an operating steady-state after being subjected to a disruption. The power system stability can be classified into rotor angle stability, voltage stability, and frequency stability. The different categories of power system stability are coupled with each other, even if weakly coupled.

Power system instabilities does not occur purely independent of each other. Cascading

(17)

outages and blackouts befall eventually as a consequence of more than one form of instability [5].

High penetration of non-dispatchable RESs decreases the power system ability to recover during disturbances. Thus, reducing the power system stability. Keeping the stability of power systems that are highly penetrated with RESs generation within the standard defined limits will necessitate the introduction of novel solutions. Such solutions could be synthetic inertia resources, high ramp rate dispatchable energy storages, and active, reactive, voltage and frequency controls for the RESs that allow them to be dispatchable in some manner [6].

Typically, power systems are designed with protection schemes for unidirectional flow of fault currents. Nonetheless, due to the bidirectional power flow introduced from the distributed generation of RESs, the classical unidirectional protection coordination schemes are rendered ineffective [7]. Moreover, directly connected synchronous generators contribute with high short circuit currents during fault while RESs such as solar and wind have relatively lower contribution. Hence, new protection coordination schemes with higher sensitivity or different fault detection methods for radial power systems are mandatory to provide reliable system protection in power systems highly penetrated with renewable energy [8].

In addition to, other challenges are introduced due to distributed generation such as islanding detection, anti-islanding, unintentional islanding, or operating in island mode;

thus, smart advanced controls are required to ensure the continuous reliable, secure, and stable electrical grid operations. Furthermore, the ability to restore the system from a blackout is compromised under high renewable energy penetration and new system restoration methods needs to be found [9]. Some of the challenges introduced by high renewable energy penetration can be summarized as the following:

• Variability and uncertainty in generation and demand.

• Power system stability.

• Protection coordination.

• Seamless unintentional islanding and islanded mode control.

• Black-start capability.

2.2 Energy Storage Systems (ESSs) for Grid Applications

ESSs have the potential to offer various services to the electrical grid and enable the growing penetration of RESs. ESSs have potential applications in energy management, operating and ramping services, frequency response, voltage support and power quality.

They also have the upper hand over conventional generation methods by offering almost instantaneous start-up, high ramp-up rates, and the ability to operate in charging mode.

This thesis will focus on studying energy storage application for frequency response to maintain the frequency stability.

(18)

The ESSs have many different technology types such as chemical storage in lithium- ion or lead-acid batteries, electrical storage in super capacitor energy storage (SCES), mechanical storage in flywheels, and thermal storage in molten salt or hot water. The different technologies vary between each other in terms of life span, cost, power density, energy density, self-discharge rate, and many other characteristics. Therefore, the choice of the type of technology is dependent on the application and on-site conditions [10].

Figure 2.1 shows the different types of technologies and their respective power and energy capacity.

The basic topology of all inverter-based types of ESSs is the same and consist of three main parts as shown in figure 2.2. The energy source is the first part, which charges and discharges under the constriction of its state of charge (SOC) limits. The power conversion system is the second part, which represents a power electronics circuit that is build as an AC/DC rectifier that charges the energy source from the grid, and a DC/AC inverter that feeds power to the grid. The third part is the control and monitoring system, which provides the control and protection of the power conversion system such

Figure 2.1: Energy storage technology characteristics [11].

(19)

as voltage regulation, frequency response, and short circuit protection.

Figure 2.2: ESS basic topology.

2.3 Gotland’s Present Case Study

Gotland electrical grid can be considered as a microgrid that is always running in grid- connected mode. Furthermore, the power system of Gotland can be representative of the future power systems with high penetration of renewable energy penetration where multiple microgrids are interconnected with each other via HVDC cables and the reliability and stability of each is threatened by islanded operations. Other microgrid islands such as Bornholm-Denmark, Kythnos-Greece, and Dangao-China have been studied in the literature for different reliability and stability issues as in [12], [13], [14], and [15]. Figure 2.3 shows a representation of Gotland microgrid.

The United States Department of Energy (DOE) has defined microgrids as an assembly of distributed energy resources (DERs) and loads within clear electrical borders that can be represented as a single controllable point with respect to the main grid. A microgrid can be operated in both connected or islanded mode [17]. The International Electrotech- nical Commission (IEC) has also defined microgrids as alternating current (AC) or direct current (DC) electrical systems with flexible loads and DERs operating at medium or low voltage. There are two classifications of microgrids: isolated microgrids and non- isolated microgrids. Isolated microgrids are not connected to any form of other main electrical power system. Non-isolated microgrids act as controllable entity and run in two different modes: island and grid-connected mode [18].

(20)

Figure 2.3: Illustrative microgrid figure of Gotland [16].

Gotland has both high wind and solar resources that converted the island into a renewable energy generation hub. Moreover, the present maximum load demand is around 190 MW, the total installed wind power capacity is 185 MW, and the total installed solar power capacity is 3 MWp [19] and the maximum allowed installed capacity of RES that was imposed by the local grid operator of Gotland is 195 MW. In 2016, the yearly wind energy generation corresponded to 0.43 TWh. However, despite of the fact that the installed wind power generation capacity exceeds the maximum load demand of Gotland, only 45-50% of the total yearly energy demand of Gotland [20] was met from RESs as a consequence of the intermittent nature of the wind resources.

The electrical grid of Gotland is connected to mainland Sweden exclusively via two 100 km long HVDC cables, Gotland 2 and 3 that were installed 1983 and 1986 respectively.

The two HVDC cables are rated with 130 MW transmission power capacity per cable (pole) and 150 kV transmission voltage [21]. Furthermore, as of 2017, Svenska Kraftnät The Swedish TSO canceled any plans of installation of any new transmission capacity between Gotland and mainland Sweden.

As discussed in chapter 1.1 the two HVDC cables can be operated in different configuration by changing the polarity of the cables. Figure 2.4 shows the different configuration of the HVDC cables. The first configuration of both cables importing is the most active configuration where the load demand is higher than the RESs generation. The second configuration is not allowed on Gotland microgrid due to the N-1 criterion. The third

(21)

configuration is active when the RESs generation is higher than the load demand and it is called in "rundkörning" in Swedish. During the third configuration the second cable will be run on minimum power setting of 3 MW while the first cable is exporting power.

The local grid operator, GEAB, tries to minimize the duration that the cables are run in the third configuration as much as possible due to reliability issues. It can be explained that reversing the polarity of the cable reduces the lifespan of the cable and jeopardizes the stability of the microgrid since only one of the cables will be working during the short period of reversing. Furthermore, while exporting the loss of the exporting cable will lead to high active power transients. Moreover, Gotland HVDC cables provides frequency response and are the only source of dispatchable active power on the island.

Therefore, it is only reasonable to search for solutions that will increase the reliability of the HVDC cables different operations.

Figure 2.4: HVDC cables different configurations.

(22)

3 Methodology

The case study of Gotland microgrid is provided by Vattenfall. The case study includes the power system model on Power System Simulation for Engineering (PSS/E) platform.

Furthermore, it includes hourly load demand, renewable energy generation, and HVDC cables power flow for three previous years. This chapter will describe the modeling platform, explain the modeling method and design of the ESS, WTG, and the plant controller models. Moreover, it will explain the descriptive analysis of the provided data, and describe the basic extrapolation method for future estimates of wind, solar, and load demand.

3.1 Power System Modeling in PSS/E

PSS/E is a software tool that is used by power system planning and operation engineers, universities, and researchers to simulate and model electrical grids and its components for wide range of analysis.

3.1.1 Gotland Microgrid Model

The electrical network model for Gotland microgird was provided by Vattenfall Research and Development AB for the purposes of this master thesis. Basically, the model given consisted all the electrical components models including the HVDC link, the wind generators, the transformers, the synchronous compensators, the transmission cables, and the loads. The model was developed originally by Vattenfall Research and Development AB as a part of a master thesis project in 2008 [22]. Afterwards, the model had been upgraded, updated, and validated on multiple instances to comply with the ongoing changes on Gotland microgrid as in [23], [24], and [25]. The current model dates back to the last update in 2012 [26], which has 170 MW of installed wind plants capacity and 190 MW of load demand modeled as compared to today’s 185 MW of installed wind plants capacity and the 3 MWp of installed solar power capacity. The single-line of the microgid is shown censored in figure 3.1 due to confidentiality concerns. The microgrid consists of 139 bus and different system voltages such as 130 kV, 70 kV, and 13 kV. Moreover, the microgrid consists around 75 WTGs of different types, multiple synchronous condensers, and an HVDC light cable connecting the 70 kV grid.

(23)

Figure 3.1: Gotland microgrid blurred single line diagram [26].

3.1.2 The ESS Model

The ESS model that is used is selected from PSS/E generic models library in order to keep the results general and represent no vendor-specific equipment. The model is originally built by Western Electricity Coordinating Council (WECC) and validated by The Electric Power Research Institute (EPRI) for different power system simulation software including PSS/E [27]. The WECC ESS model assumes that the ESS represent a single plant that is connected to the grid in steady-state operation. The dynamic model consists of different modules, that represent the topology of an ESS, connected together for the sake of transient operations as discussed further in [28], and [29]. The block diagram of the generic model is shown in figure 3.2, where the default module parameters for REGC_A, REEC_C, and REPC_A can be found and has been selected from [27].

REGC_A, as shown in figure 3.3 represents the generator/converter module. It acts like an inverter interface with the grid that processes active (I_pcmd) and reactive (I_qcmd) current commands input from the electrical controller (REEC_C) and sends an output of active (I_p) and reactive (I_q) current into the network [27]. REGC_A generator/converter module can be used for modeling different RESs, such as wind or solar plant. REEC_C, as shown in figure 3.4 represents the electrical controller module. It receives active (P_{re f}) and reactive (Q_{re f}) power references inputs from the plant controller (REPC_A) with a feedback from the generator output and the voltage terminal, and sends output as

(24)

active (I_pcmd) and reactive (I_qcmd) current commands to the generator/converter module (REGC_A). REEC_C electrical controller module is unique for ESSs since it can operate in the four quadrants of the P-Q diagram if the SOC of the ESS allow it, as shown in 3.5. Finally, REPC_A , as shown in figure 3.6 represents the plant controller. It processes the active power and the frequency to model the active power control or the frequency response, and processes the reactive power and the voltage to model the reactive power control or the voltage regulation, and then sends output active (P_{re f}) and reactive (Q_{re f}) power references to the electrical controller (REEC_C). The active power part of the plant controller will be discussed thoroughly in subsequent subsections because it emulates the inertial and primary frequency response of the ESS. REPC_A can also be used as a plant controller in other RESs models. Appendix A shows an example of the modeled ESS dynamic parameters in PSS/E.

Figure 3.2: ESS model using WECC modules as suggested by [28], [30].

Figure 3.3: REGC_A generator/converter generic model provided by WECC [27].

(25)

Figure 3.4: REEC_C electrical controller generic model provided by WECC [30].

Figure 3.5: ESS operating P-Q diagram [30].

(26)

Figure 3.6: REPC_A plant controller generic model provided by WECC [27].

3.1.3 The WTGs Model

The wind turbine-generators (WTGs) technologies are mainly divided into four groups, but the main focus here is on doubly fed induction generators (Type 3) and synchronous generators with full converter (Type 4) because nearly all newly manufactured wind turbine generators falls into either one of those categories. The steady-state and dynamic models of the four types of wind turbines were provided by Vattenfall in the original PSSE model in [26]. The provided original model wind turbines did not have any form of frequency response to disturbances and they acted like non-dispatchable generators.

Thus, updated models were proposed and implemented for both type 3 and type 4 WTGs that had frequency response incorporated in their dynamic modules.

The type 4 and type 3 WTGs updated model were built from multiple modules like the ESS model. Moreover, the modules is presented and updated by WECC and tested by EPRI for different simulation platforms including PSS/E as shown in details in [31], and [32]. The basic building modules for type 4 WTG consisted of REGC_A, REEC_A, and REPC_A [33]. Furthermore, The basic building modules for type 3 WTG consisted of REGC_A, REEC_A, WTGT_A, WTGAR_A, WTGPT_A, WTGTRQ_A, and REPC_A [33]. The default module parameters for aforementioned modules for type 3 and type 4 WTGs have been selected as suggested in [27].

REGC_A, as shown in figure 3.3, is also used in the ESS model as the the generator/converter module. REGC_A serves the same purpose as an inverter interface with the grid as discussed in details in subsection 3.1.2. REEC_A, as shown in figure 3.7 represents the electrical controller module for both type 3 and type 4 WTGs. REEC_A is very similar to REEC_C in terms of operation and building blocks. The difference is that REEC_A oper- ates only in two quadrants of the P-Q diagram, where it only generates active power and it does not consume it. Moreover, REEC_C has an extra block as compared to REEC_A

(27)

to emulate the SOC of an ESS. WTGT_A represents the drive-train module, WTGAR_A represents the aero-dynamic module, WTGPT_A represents the pitch controller module, and WTGTRQ_A represents the torque module. The detailed description and figures of the aforementioned modules is found in [33], [27], [31], and [32]. Finally, REPC_A , as shown in figure 3.6 represents the plant controller. REPC_A serves the same purpose in WTG models as in ESS models, where it control both active and reactive power within the P-Q quadrants of operations that are imposed by the electrical controller. Appendix B shows an example of the modeled type 3 WTG dynamic parameters in PSS/E.

3.1.4 The Active Power Controller & Frequency Response

The lower part of figure 3.6 of the plant controller (REPC_A) represents the active power controller that emulates the frequency response. REPC_A provides inertial response and primary frequency response through the PI controller that receives the frequency signal from the grid and sends the reference active power signal to the electrical controller.

First, the frequency is compared with the nominal reference frequency Freq_{re f}. If the difference is higher than the frequency dead-band threshold f_dbd, which is the range where changes in frequency will not activate the power controller, then the frequency change will be multiplied by the droop gain coefficient D_dn or Dup. The droop gain

Figure 3.7: REEC_A electrical controller generic model provided by WECC [33].

(28)

coefficient D_dnor D_uprepresents the active power change to the change in frequency in pu/Hz. Both the the frequency dead-band threshold f_dbdand the droop gain coefficient D define the droop characteristics of the energy storage. Figure 3.8 shows an example of droop characteristics graph with f_dbd= ±0.1 Hz and D=2.5 pu/Hz.

Then, the signal is added to the current plant active power Plant_{re f} before being sent to the PI controller to represent the increase or the decrease that should be added to the power output. The PI controller has proportional Kp and integral K_i gain coefficients.

The proportional gain Kpcoefficient and the integral gain K_i coefficient represent the inertial and primary frequency response since the input signal is power deviation from the plant power reference. Finally, the signal goes through the plant controller lag T_lag before being sent to the electrical controller as the power reference signal P_{re f}. The exact values for the droop characteristics and PI controller proportional and integral gain for the plant controllers will be shown in chapter 4 and discussed in chapter 5.

Figure 3.8: Droop characteristics sample.

(29)

3.2 Data Description and Design Criteria

The sample of the hourly data was provided by Vattenfall for Gotland for the duration of three years, from the 1st of July 2014 to the 1st of July 2017. The sample included the total power flow over the HVDC link, which represents the power imports from and exports to the mainland Sweden. Moreover, the sample included the total wind and solar power production on Gotland for the entire period, which if added to the HVDC links power flow can represent the total load consumption on Gotland at that point of time. As discussed previously in chapter 2.3 the present maximum load is 190 MW, the total wind plant capacity is 185 MW, and the total solar power capacity is 3 MWp which in coordination with the hourly data provided will decide the design parameters and sensitivity analysis criteria as explained in the subsequent section.

3.2.1 Design Criteria

One of the objectives of the thesis, regardless of the RESs penetration level and frequency response solution adapted, is to examine the transient response during the unintentional islanding due to the loss of the HVDC link when the power generated is higher than the load consumption. Consequently, designing the solution for the worst case scenario would mean designing for maximum generation capacity and minimum load consumption, and that is in the present case an excess of 185 MW of wind power with 40 MW load consumption [34]. It is worth mentioning the worst case scenario, but it is completely unrealistic and costly to be choosing the solution parameters for that case because the probability of having the worst case and having a fault on the HVDC link at the same time is nearly zero. Therefore, the design criteria for the chosen solution and the sensitivity analysis should follow a realistic probability for the fault event to happen, and it is defined as the following:

p(instable over− f requency) =p(P_wind+P_solar≥P_load ∩ HVDC cable f ault) (3.1)

P_load=P_HVDC+P_wind+P_solar (3.2)

p(P_wind+P_solar≥P_load) =p(P_HVDC≤0) (3.3) Where,

P: The active power (MW).

P_HVDC: The active power flow on the HVDC cables (MW), where the positive sign represents importing power and the negative sign represents exporting power.

P_wind: The total wind produced active power (MW).

P_solar: The total solar produced active power (MW).

P_load: The active power load demand (MW).

And,

p: The probability of an event.

(30)

p(instable over− f requency): The probability of having an event that causes instability in frequency of the microgrid.

p(P_wind+P_solar ≥ P_load): The probability of generating more power that the load demand.

p(HVDC cable f ault): The probability of having a fault on the HVDC cables that leads to losing one or both of them.

p(P_HVDC ≤0): The probability of exporting power from Gotland to mainland on the HVDC cables.

Thus,

p(instable over− f requency) =p(P_HVDC≤₀ ∩ HVDC cable f ault) _(3.4) For the sake of simplification, it is assumed that the event of the HVDC cables fault is independent of the event exporting the excess energy, then,

p(instable over− f requency) =p(P_HVDC≤0) ∗p(HVDC cable f ault) (3.5) Therefore, reducing the probability of either of the events will minimize the probability of having an instable over-frequency and unstable frequency stability in general. Reducing the probability of having faults on the HVDC links is out of scope of this thesis. Moreover, increasing the installed power capacity will increase the probability of exporting power on the HVDC cables and eventually increase the probability of having an instable over- frequency. Therefore, solutions such as an energy storage or wind spilling (wind power curtailment) that can minimize the exported power during fault conditions can maintain the stability and allow the increase of installed RESs. Hence, at the moment of the fault for a stable frequency response:

PHVDC≤P_solution (3.6)

Where,

P_solution: The power capacity of the solution that can be instantaneously charged of discharged (MW).

The optimal value of P_solution can be found from the cumulative distribution function for the HVDC cables power flow data and descriptive statistics that were given. But the main idea of the design is to find P_solutionthat reduces the value p(P_HVDC<=0)at the moment of the fault to a chosen value that will be defined in chapter 4.

3.2.2 Present Data and Future Estimation

Finding the correlation between available wind and solar resources, and the generated power by the installed wind capacity and solar capacity is out of the scope of this thesis.

Moreover, forecasting the future wind or solar resources on hourly basis for future scenarios with high precision is out of the scope of this thesis too. Thus, the increase

(31)

in the generated wind and solar power due to the growth of the installed capacity will be estimated following a basic linear extrapolation approach for the sake of simplicity.

Therefore, the future estimates will be a replica of the present data but with linear extrapolation to represent the growth in the capacity.

The wind hourly power production values for the three years is provided for a constant 185 MW wind plant capacity, and due to the nature of the variability of wind resources and geographical separation between the wind turbines, the wind plants do not produce power equally neither at 100% of their capacity. As a matter of fact the minimum generated power can be as low as 0 MW, and the maximum generated power was observed to reach up to 150 MW. The hourly future estimation of the wind hourly power production will rely on the average of three years hourly production and the constant total wind capacity as in the following equation:

Pf uture wind capacity(t) = Future wind capacity∗Average o f wind power generation(t) Total present wind plants capacity

(3.7) Pf uture wind(t) =Pf uture wind capacity(t) +Average o f wind power generation(t) (3.8) The solar hourly power production values for the three years is provided for different total capacities that were increasing gradually over the span of the three years, but it will be assumed that the increase in the capacity was annual for the simplicity of calculations of the future estimates. The current total capacity is 3 MWp and the growth was assumed around 1 MWp/year as suggested by GEAB. The future estimation of the solar hourly power production will rely on the weighted average of three years hourly production assuming an increase of 1 MWp/year as in the following equation:

Pf uture solar capacity(t) = Future solar capacity∗Weighted average o f solar power generation(t) Average present solar power capacity

(3.9) Pf uture solar(t) =Pf uture solar capacity(t) +Weighted average o f solar power generation(t)

(3.10) The hourly load consumption is provided for nearly constant maximum load demand over the three years, but the future hourly load consumption will be considered to increase annually by a constant value as in the following equation:

Pf uture load(t) =Average o f present load consumption(t) ∗ (1+i)ⁿ (3.11) Where,

i: Load growth constant.

n: Number of years.

(32)

Therefore, substituting in equation 3.2, the future estimation of the HVDC links hourly power flow data can be calculated by the following equation:

Pf uture HVDC(t) =Pf uture load(t) − (Pf uture wind(t) +Pf uture solar(t)) (3.12) The installation of new wind and solar energy plants capacity on Gotland does not follow any mathematical pattern, but it follows the policies, investment availability, and future plans. Moreover, the load is assumed increase very lowly since the number of inhabitants on Gotland is not assumed to grow. It is worth to mention also that plans of electrification of the industrial sector of Gotland will increase the maximum load demand by around 200 MW, which will require an installation of a new transmission capacity and is out of the scope of the thesis.

(33)

4 Results

Gotland’s microgrid was described in details in chapter 3. The main objective is to examine the frequency stability after the addition of a centralized ESS along with or without wind curtailment. Figure 4.1 shows a simplified diagram of the investigated power system model, where the ESS is connected to main bus connecting Gotland to mainland Sweden. In this chapter the descriptive and sensitivity analysis of the data, for different installed capacity of RESs and different exporting probability scenarios, are presented. The exporting probability p(P_HVDC <=0)along with the probability of having a fault on the HVDC cables defines the probability of having an unstable frequency that might lead to a blackout. Moreover, the frequency response of the ESS along with or without wind curtailment during the HVDC cables fault will be represented.

Figure 4.1: The studied power system simplified diagram.

4.1 The HVDC Cables Fault

The frequency stability of the microgrid of Gotland is dependent on the HVDC cables, since currently they solely provide the frequency response for the microgrid. It is guaranteed that the loss of the HVDC cables will cause the microgrid to lose its frequency stability without any preemptive methods. Hence, The microgrid model dynamic frequency response is tested during the HVDC cables disconnection and figure 4.2 shows the simulation of the HVDC cables fault while having 40 MW of exported power on one of the HVDC cables. It is clear from the figure that the frequency kept on increasing (solid line) until over-frequency relays started tripping the wind turbines (dashed line).

(34)

0 2 4 6 8 10 12 14 16 18 20 Time (s)

50 50.2 50.4 50.6 50.8 51 51.2 51.4 51.6 51.8

Frequency (Hz)

Frequency Response

Disconnection of both of the HVDC Cables at t=5 s while having an excess power production of 40 MW.

Figure 4.2: Frequency response of present study case during HVDC cables fault.

4.2 Present Study Case - 2018

The present study case of Gotland represent the power system as shown in figure 3.1 and the simplified diagram shown in figure 4.1. The microgrid includes 185 MW of installed wind power capacity, 3 MWp of installed solar power capacity, and 190 MW maximum load demand. The data for the hourly generated wind and solar power, the hourly power flows over the HVDC cables, and the hourly load consumption was provided for the previous three years as explained in chapter 3.2. Figure 4.3 shows the aforementioned data sets from July-2014 until July 2017. The ESS is chosen to be of 50 MW power capacity to be able to charge an active power transient of 40 MW with fast ramp-up rate. The different energy capacities are dependent on the wind curtailment option as shown below in table 4.1. The findings are explained in details in the subsequent subsections.

Table 4.1: Energy storage energy capacity for different durations of present case study.

Wind Curtailment Option/Du-

ration One Minute Five Minutes One Hour

No Wind Curtailment 0.60 MWh 3.20 MWh 38.95 MWh

Primary Wind Curtailment 0.42 MWh 2.10 MWh 25.00 MWh Primary Wind Curtailment with

Secondary Auxiliary Signal 0.35 MWh 0.45 MWh 0.45 MWh

Gotland as a microgrid - Energy storage systems frequency response in grids with high level of renewable energy penetration

Juni 2018

Gotland as a microgird - Energy storage systems frequency response in grids with high level of renewable energy penetration

Firas Qasem Daraiseh

Masterprogram i förnybar elgenerering

Gotland as a microgird - Energy storage systems frequency response in grids with high level of renewable energy penetration

Abstract

Acknowledgements

Table of Contents

List of Tables

List of Figures

List of Acronyms

1 Introduction

1.1 Motivation and Problematization

1.2 Objectives

1.3 Thesis Outline

2 Literature Review

2.1 Challenges Introduced by High Renewable Energy Penetration on The Electrical Grid

2.2 Energy Storage Systems (ESSs) for Grid Applications

2.3 Gotland’s Present Case Study

3 Methodology

3.1 Power System Modeling in PSS/E

3.2 Data Description and Design Criteria

4 Results

4.1 The HVDC Cables Fault

4.2 Present Study Case - 2018