• Non ci sono risultati.

Adaptive protection algorithms for smart distribution systems tested in a hardware-in-the-loop setup

N/A
N/A
Protected

Academic year: 2021

Condividi "Adaptive protection algorithms for smart distribution systems tested in a hardware-in-the-loop setup"

Copied!
90
0
0

Testo completo

(1)

POLITECNICO DI MILANO

School of Industrial and Information Engineering Master’s Degree

in Electrical Engineering

ADAPTIVE PROTECTION ALGORITHMS FOR SMART

DISTRIBUTION SYSTEMS TESTED IN A

“HARDWARE-IN-THE-LOOP” SETUP

Supervisor: Prof. Samuele Grillo

Co-Supervisor: Prof. Enrico Ragaini

Internship-Supervisor: Alexandre Oudalov

Master of Science Thesis:

Mario Bertolo

Matr. 841954

(2)
(3)

“Science is none other than a prevision if it hasn’t as ultimate aim the human life improvement.” - Nikola Tesla

(4)
(5)

Acknowledgements

First of all, I want to dedicate this work to my family, my father Fabio, my mother Silvia, my sister Ilaria and my grandmother Romana. You have always been present and next to me during my university experience, full of satisfaction but also with a lot of difficult challenges, and your helps, your advices and your presence have been precious and indispensable. I love you. Thanks also to all my relatives, who always give me moments of happiness, and to Jarno and Bianchina.

Then, I would like to express my gratitude to the people that were essential for this project: the first “thanks!” goes to my Professor Samuele Grillo, my thesis Supervisor. Thanks to him I became aware about the possibility to undertake this work experience in ABB and, once the internship was finished, with his help I was able to summarise all the results and the information acquired during the work experience and complete this final thesis work. Thanks for all the time made available for me and for all the useful tips and corrections. The second “thanks!” goes to my Professor Enrico Ragaini, my thesis co-Supervisor. He introduced me in the ABB world, allowing me to join this project. His availability and his helps were essential, especially when he allowed me to spent part of the time of the internship in ABB SACE. The time I spent there was fundamental for the success of the project since I became familiar with the electrical devices used to develop the setup. I want to say thanks also to all the person that worked with me during the weeks spent in Bergamo. In particular, to Massimo Belometti and Paolo Bettinelli, my Professor’s colleagues in ABB SACE, who gave me advice from a technical point of view; to Sergio Moreno and Andrea Cesana, both Politecnico students and interns in ABB SACE, who helped me figure out several difficult features related to LabVIEW and to electrical devices; to Nicola Dadda, a very young and great guy with excellent programming and technical abilities, who also provided to me an essential support.

The third “Thanks!” goes to my internship Supervisors, Alexandre Oudalov and Marija Zima, who managed my work in Baden. In particular, I want to express my gratitude to Alexandre because, with his devotion and professionalism, he was able not only to explain

(6)

IV

me very difficult topics in a very simple and clear way, making me enthusiastic about the topic, but also because he made me realize the importance of working hard to achieve predefined goals, and this teaching will be precious for my entire carrier.

I would like also to express my gratitude to all the Professors of my Master course, especially to Maurizio Delfanti, a great and nice teacher who always showed his willingness to help his students.

Another “thanks” goes to all my friends: to Andrea Cusumano, medical doctor under training and faithful gym companion, who always gave to me good advices especially when I had to choose to undertake this university experience at Politecnico; to Alessio Caramiello, lifelong friend and high school and middle school classmate, with whom I spent a lot of fun moments, especially in this last year; to Andrea Sperati and Stefano Brignoli, my lifelong friends too, fellow countrymen and elementary classmates, always there willing to help me; to Gigio, Pold, Cami and Biaf, high school friends with which I had a lot of funny experiences and holidays; to Carolina, Pold’s girlfriend, who introduced me to Margherita, my lovely girlfriend; to my “Superamici” Roberto and Simone; to all my bachelor classmate: Mala, Claudio, Teo, la Fede, Crema, Pacchia, Calvi, Angelo, il Bobo, Albi, Bruni, il Cangino, Cesare, Trave, Maru; to my friends of Milano Bovisa: Bona, Sara, Rache, Ivan, Guen, Gerva, Teo, Chiara, Ettore, Richi; thanks a lot to every one!!

Thanks also to my Master classmates: to Stefano Bonomi, with whom I spent a lot of time studying for exams and who gave me a lot of explanation about the matter, making the difficulties easier to overcome; to Luca Quistini, with whom I spent more time having fun rather than studying; to Gianluca Godi, Luca Uboldi, Alessandro Daverio and Michele Lussana, which always helped my providing explanation and useful notes; to “il grande e irreprensibile” Francesco Quacci, with whom I studied for the last exam; to Alessio Mazzola, Hani el Khoury, Alberto Cosmai, Luca Bertoletti, Mattia Bettinelli, Alex Castro, Okan Erim, Farhan Rana, Alaudin, Adeen, Giulia, Jogi…..thank you very much!

A great thanks to all the people I knew during my work experience in Baden: to Giorgio, Fabio and Francesco, Italian friends who helped me to familiarize with the new environment since the beginning of my internship; to my office colleagues, Jonathan, Burak, Qiao, Vidak, Nemanja and Chrysella. I want to especially thank her very much since she helped me to understand some LabVIEW codes; to Arno, Vincent, Huiting, Nikolas, Vlad; to all the “table soccer” guys: Thomas, Manuel, Daniele, Alessandro, Andrea,

(7)

V Aurelien, Raphael; to Antti, Cyrill, Matilde, Matze, Sarha, Marin, Eva…thank you guys! It was one of my best experience in my life!

The last thanks, maybe one of the most important, is for Margherita, my beloved, beautiful and nice girlfriend, always by my side, who has been able to give me the strength to move on, even in the more difficult moments, to give me advice and to lovely bear with me, especially in these last challenging year. I will be always grateful with you and with your family!

(8)
(9)
(10)

VIII

Abstract

(EN)

This thesis is based on the research work carried out from the 1st June 2016 to the 30th November 2016, in the Power System team of the Automation department at ABB Switzerland Ltd Corporate Research Center, Baden-Dattwil and in the laboratories at ABB SACE Division in Bergamo.

The goal of the project was to develop and test new protection methods and algorithms for the distribution grids in presence of distributed generation units.

The first part of the internship was dedicated to the study of recent publications, research articles and papers on the subject, in order to understand the main problems related to protection devices in the new smart grid environment.

The second part, was spent at ABB SACE Division in Bergamo. In this company, information on how to use and set up the protection relays EMAX 2 and on how to develop programs with LabVIEW was provided, in order to develop a simulation environment and to interface it with the protection devices.

In the third part, again in Switzerland, using two EMAX 2 relays, the distribution system real-time simulation with loads and distributed generation has been developed. When this simulation was achieved, the previously conceived protection algorithms were implemented in LabVIEW, interfacing the software with two protection devices and creating in this way a “hardware-in-the-loop” setup.

The thesis consists of seven chapters: the first and the second ones introduce the distribution system’s evolution from a totally passive electrical system component to an active smart grid and the issues related to the protection system (protection blinding, sympathetic tripping, etc.,); in chapters three and four some of the main protection methods described in the literature and, more specifically, those that have been developed and simulated during this project are described; in the fifth chapter the “hardware-in-the-loop” setup is described in detail and in the last two chapters the simulation results along with some suggestion about future works and applications are presented.

(11)

IX

Abstract

(IT)

Questo lavoro di tesi si basa sull’esperienza di ricerca maturata dal 1 giugno 2016 al 30 novembre 2016, all’interno del team Power System del dipartimento di Automation, presso il centro ricerca ABB di Baden-Dattwil, Svizzera e presso i laboratori di ABB SACE Division a Bergamo.

L’obiettivo del progetto è stato quello di sviluppare e testare nuovi metodi e algoritmi di protezione per le reti di distribuzione elettrica in presenza di unità di generazione distribuita.

La prima parte dell’internship è stata dedicata allo studio dell’argomento mediante l’analisi di recenti pubblicazioni, al fine di comprendere le principali criticità relative ai dispositivi di protezione nel nuovo contesto della smart grid.

La seconda parte, che ha previsto due settimane di collaborazione presso ABB SACE Division (Bergamo), è stata dedicata all’apprendimento del funzionamento dei relè EMAX 2 e della programmazione con il sistema LabVIEW, al fine di sviluppare un ambiente di simulazione e di interfacciarlo con i dispositivi di protezione.

Nuovamente in Svizzera, adoperando i dispositivi EMAX 2 precedentemente analizzati, è stata programmata una simulazione real-time di un sistema di distribuzione, con carichi e unita di generazione distribuita. Terminato lo sviluppo di questa simulazione, gli algoritmi di protezione precedentemente ideati sono stati implementati in LabVIEW, interfacciando il software con i due relè e creando in questo modo un setup “hardware-in-the-loop”.

La tesi è strutturata in sette parti: la prima e la seconda introducono l’evoluzione del sistema di distribuzione elettrica e le problematiche relative al sistema di protezione in questo scenario innovativo; nel terzo e nel quarto capitolo sono descritti alcuni metodi di protezione presenti nelle pubblicazioni e negli articoli scientifici e i metodi che sono stati ideati e sviluppati durante l’internship; nella quinta parte è fornita una descrizione dettagliata del setup e nella sezione finale sono presentati i risultati delle simulazioni degli algoritmi implementati e alcune riflessioni riguardo lo sviluppo di futuri progetti applicabili al setup e a casistiche reali.

(12)
(13)

Table of contents

ADAPTIVE PROTECTION ALGORITHMS FOR SMART DISTRIBUTION SYSTEMS TESTED IN A “HARDWARE-IN-THE-LOOP” SETUP ... I ACKNOWLEDGEMENTS ... III ABSTRACT (EN) ... VIII ABSTRACT (IT) ... IX TABLE OF CONTENTS ... XI LIST OF FIGURES ... XV LIST OF TABLES ... XIX NOMENCLATURE ... XXI

CHAPTER 1. EVOLUTION OF ELECTRIC POWER SYSTEM ... 1

1.1SMART GRID PARADIGM ... 1

1.2THE ROLE OF THE DISTRIBUTION SYSTEM ... 3

1.3MICROGRID CONCEPT ... 4

CHAPTER 2. TECHNICAL ISSUES OF THE PROTECTION SYSTEM... 7

2.1INTRODUCTION ... 7

2.2SHORT CIRCUIT LEVEL VARIATION IN ISLAND MODE ... 8

2.3PROTECTION BLINDING ... 10

2.4SYMPATHETIC TRIPPING ... 11

CHAPTER 3. PROTECTION SCHEMES FOR MICROGRID ... 13

3.1INTRODUCTION ... 13

3.2PROTECTION SCHEME BASED ON FAULT CURRENT DIRECTION ... 13

3.3MATRIX STRATEGY BASED ON FAULT CURRENT DIRECTION ... 15

(14)

XII

3.5PROTECTION SCHEME BASED ON ZONAL DIVISION ... 19

CHAPTER 4. DESCRIPTION OF THE SIMULATED PROTECTION ALGORITHMS ... 21

4.1INTRODUCTION ... 21

4.2PROTECTION METHOD BASED ON VARIATION OF IMPEDANCE MATRIX ... 21

4.3PROTECTION METHOD BASED ON AREA CONCEPT ...22

4.3.1 Local logic area controller ... 23

4.3.2 Short circuit level ... 23

4.3.3 The adaptive algorithm ... 25

CHAPTER 5. HARDWARE-IN-THE-LOOP SETUP ...27

5.1INTRODUCTION ... 27

5.2REAL SCENARIO ... 27

5.3SOFTWARE COMPONENT ...29

5.3.1 Synchronous generator modelling and implementation ... 30

5.3.2 Power flow implementation ... 33

5.3.3 Short Circuit implementation ... 35

5.3.4 Modbus TCP/IP ...36

5.4HARDWARE COMPONENT ... 37

5.4.1 ABB SACE EMAX 2 ... 37

5.4.2 Ekip Connect ... 40

5.4.3 Function generator Keithley 3390 ... 41

5.5SETUP LAYOUT ...42

CHAPTER 6. THE PROTECTION ALGORITHMS SIMULATED IN THE HIL SETUP ... 45

6.1INTRODUCTION ... 45

6.2IMPLEMENTATION OF THE PROTECTION ALGORITHM ... 45

6.2.1 Impedance matrix variation algorithm ... 46

6.2.2 Area concept algorithm ... 47

6.3LOGICAL INTERACTION BETWEEN IED USING EKIP LINK ... 49

6.4SIMULATION RESULTS ... 50

6.4.1 Simulation results with Ekip Link ... 50

6.4.2 Simulation results with Impedance matrix variation algorithm ... 53

6.4.3 Simulation results with Area concept algorithm ... 56

(15)

XIII

(16)
(17)

List of figures

Figure 1.1: Smart grid ... 2

Figure 1.2: The structure of the electric power system ... 3

Figure 1.3: Hierarchical structure of EPS ... 4

Figure 1.4: Microgrid ... 5

Figure 2.1: Simulation of fault current in grid-connected and island mode ... 8

Figure 2.2: Fault current in branch 3-4 (grid-connected mode and island mode) ...9

Figure 2.3: Fault current in branch 3-4 (grid-connected mode and island mode) ... 10

Figure 2.4: LV power system used to study protection blinding ... 10

Figure 2.5: Fault current affected by blinding phenomena ... 11

Figure 2.6: Distribution system used to study sympathetic tripping ... 12

Figure 3.1: Tested microgrid ... 14

Figure 3.2: Simulation of fault current in branch 1.3 ... 14

Figure 3.3: Simulation of fault current in branch 2.3 ... 15

Figure 3.4: Low voltage PV microgrid ... 15

Figure 3.5: Protection scheme ... 18

Figure 3.6: DT generated for fault detection ... 18

Figure 3.7: Division of DN into main zone and DG zone ... 19

Figure 4.1: Flowchart of the adaptive method ...22

Figure 4.2: Exemplary communication between protection areas ... 23

Figure 4.3: MVA method explanation ...24

Figure 4.4: MVA method explanation ... 25

Figure 4.5: Flowchart of the adaptive method (offline phase) ...26

Figure 4.6: Flowchart of the adaptive method (online phase) ...26

Figure 5.1: Protection devices located in a real environment ... 28

(18)

XVI

Figure 5.3: Protection devices located in a real environment ...29

Figure 5.4: Simulated power system during normal condition ...29

Figure 5.5: Simulated power system during fault condition ... 30

Figure 5.6: Time parameters of the Control and Simulation loop (LabVIEW library) ... 30

Figure 5.7: Angle stability during fault event ... 31

Figure 5.8: Control panel of synchronous generator ... 32

Figure 5.9: Overall script of the mathematical model ... 32

Figure 5.10: Model of the simulated power system ... 33

Figure 5.11: LabVIEW script of the power flow subVI (input values) ... 33

Figure 5.12: LabVIEW script of the power flow subVI (Gauss-Seidel iteration) ...34

Figure 5.13: LabVIEW script of the short circuit subVI (input values) ... 35

Figure 5.14: LabVIEW script of the short circuit subVI ... 35

Figure 5.15: Modbus TCP/IP script ...36

Figure 5.16: ABB SACE EMAX 2 ... 38

Figure 5.17: Touch interface ... 38

Figure 5.18: Ekip Link module (left) and Modbus TCP/IP module (right) ...39

Figure 5.19: Device used for the setup ...39

Figure 5.20: Ekip Connect layout ... 40

Figure 5.21: Function generator Keithley 3390 ... 41

Figure 5.22: LabVIEW script of the function generator ... 41

Figure 5.23: Final setup ...42

Figure 5.24: Power system layout ...42

Figure 6.1: Flowchart of the adaptive method ... 46

Figure 6.2: LabVIEW script for the protection algorithm ... 47

Figure 6.3: Implementation of the Area concept in LabVIEW ... 48

Figure 6.4: LabVIEW script for the protection algorithm ... 48

Figure 6.5: Logical interaction through Ekip Link ... 49

Figure 6.6: Power system during fault condition (Ekip Link) ... 51

Figure 6.7: Fault clearing time 0.2 seconds (Ekip Link) ... 51

Figure 6.8: Generator stability during the fault (Ekip Link) ... 51

Figure 6.9: Power system during fault condition (island mode) (Ekip Link)... 52

Figure 6.10: Time duration of the adaptive operation (0.8 seconds) ... 52

Figure 6.11: Fault clearing time 0.4 seconds (island mode) (Ekip Link) ... 52

Figure 6.12: Generator stability during the fault (island mode) (Ekip Link) ... 53

Figure 6.13: Power system during fault condition (Impedance matrix) ... 54

(19)

XVII

Figure 6.15: Generator stability during the fault (Impedance matrix) ... 54

Figure 6.16: Power system during fault condition (island mode) (Impedance matrix) ... 55

Figure 6.17: Time duration of the adaptive operation (0.7 seconds) ... 55

Figure 6.18: Fault clearing time 0.4 seconds (island mode) (Impedance matrix) ... 56

Figure 6.19: Generator stability during the fault (island mode) (Impedance matrix) ... 56

Figure 6.20: Power system during fault condition (Area concept) ... 57

Figure 6.21: Fault clearing time 0.3 seconds (Area concept) ... 57

Figure 6.22: Generator stability during the fault (Area concept) ... 57

Figure 6.23: Power system during fault condition (island mode) (Area concept) ...58

Figure 6.24: Time duration of the adaptive operation (0.7 seconds) ...58

Figure 6.25: Fault clearing time 0.3 seconds (island mode) (Area concept) ... 59

(20)
(21)

List of tables

Table 3.1: Protection Algorithm ... 16

Table 3.2: Differential features for DT ... 17

Table 4.1: Formulas to calculate the MVA value ...24

Table 5.1: Setup operations ...43

(22)
(23)

Nomenclature

EPS Electric Power System DG Distributed Generation RES Renewable Energy Source EESS Electrical Energy Storage System DN Distribution Network

MG Microgrid

LV Low Voltage

MV Medium Voltage

HIL Hardware In The Loop

OC Overcurrent

MGCC Microgrid Central Controller PV Photovoltaic

PS Power System

PCC Point of Common Coupling

SCADA Supervisory Control and Data Acquisition DFT Discrete Fourier Transform

DT Decision Tree

CT Current Transformer VT Potential Transformer IED Intelligent Electronic Devices LLC Local Logic Controller

(24)

XXII

CB Circuit Breaker

RC Rogowski Coil

(25)

1

CHAPTER 1.

EVOLUTION OF ELECTRIC POWER

SYSTEM

1.1 Smart grid paradigm

Modern society depends critically on a secure supply of energy. Growing concerns for primary energy availability and aging infrastructure of current electrical transmission and distribution networks are increasingly challenging security, reliability and quality of power supply. Very significant amounts of investment will be required to develop and renew these infrastructures, while the most efficient way to meet social demands is to incorporate innovative solutions, technologies and grid architectures [1].

Over the last few decades, there has been a paradigm change in the way electricity is generated, transmitted, and consumed. However, fossil fuels continue to form a dominant initial source of energy in the industrialized countries [2]. The steady economic growth of some of those industrialized nations gradually exposed the inadequate nature of the energy policy that is highly dependent on foreign fossil fuels. On the other hand, a hidebound power grid that faces new challenges posed by higher demands and increasing digital and nonlinear loads has placed new reliability [2].

As a result, power generation, transmission and consumption have been subject of studies in order to see what solutions will address the above challenges, transforming the power grid into a more efficient, reliable, and communication-based system [2]. According to the European Technology Platform of smart grids [1], a smart grid is an electricity network that can intelligently integrate the actions of all users connected to it – generators, consumers and those that assume both roles – in order to efficiently deliver sustainable,

(26)

Chapter 1. EVOLUTION OF ELECTRIC POWER SYSTEM

2

economic and secure electricity supplies. A smart grid employs innovative products and services (high levels of utilization for renewable energy sources) together with intelligent monitoring, control, communication and self-healing technologies [1].

In summary, a smart grid (Figure 1.1) makes use of innovative products and services together with intelligent monitoring, control, communication, and self-healing technologies to:

- better facilitate the connection and operation of generators of all sizes and technologies [3];

- allow consumers to play a part in optimizing the operation of the system [2]; - accommodate intermittent generation and storage devices [3];

- enable new products, services, and markets [2];

- integrate electric vehicles into the distribution network [3];

- provide consumers with greater information and options for choice of supply [2] [3];

- significantly reduce the environmental impact of the whole electricity supply system [3];

- maintain or even improve the existing high levels of system reliability, quality and security of supply [3];

- maintain and improve the existing services efficiently [3];

- foster market integration towards a European integrated market [2] [3].

(27)

3

1.2 The role of the distribution system

The electric power system (EPS) is a product of rapid urbanization and infrastructure developments in various parts of the world in the past century (Figure 1.2). Though it is present in many different parts of the world, the utility companies have generally adopted similar technologies and structures [2]. The growth of EPSs, however, has been influenced by economic, political, and geographic factors that are unique to each utility company. Despite such differences, the basic topology of the existing EPS has remained unchanged [2].

Figure 1.2: The structure of the electric power system

Since its inception, the power industry has operated with clear demarcations according to a complex infrastructure that may be subdivided into the following major subsystems: generation, transmission system, distribution system and loads and each of these subgroups has shaped different levels of automation, evolution, and transformation [2]. As Figure 1.3 demonstrates, the traditional electricity grid is a strictly hierarchical system in which power plants at the top of the chain ensure power delivery to customer’s loads at the bottom of the chain. The system is essentially a one-way pipeline where the source has no real-time information about the service parameters of the end points [2].

Despite that this way of exploit electrical energy has lasted for a long time, nowadays distribution grids are being transformed from passive to active networks [1]. This type of network eases the integration of distributed generation (DG), renewable energy sources (RES), demand side integration and electrical energy storage systems (EESS), and creates opportunities for novel types of equipment and services, all of which would need to conform to common protocols and standards [1]. The main function of an active distribution network (DN) is to efficiently link power generation with consumer demands,

(28)

Chapter 1. EVOLUTION OF ELECTRIC POWER SYSTEM

4

Figure 1.3: Hierarchical structure of EPS

allowing both to decide how best to operate in real-time. Power flow assessment, voltage control and protection require cost-competitive technologies and new communication systems with information and communication technology playing a key role [1].

The realization of active DN requires the implementation of radically new system concepts and microgrids (MG), also characterized as the “building blocks of smart grids”, are perhaps the most promising and novel network structure for this scope [1].

1.3 Microgrid concept

A microgrid (Figure 1.4) is a Low Voltage (LV) or Medium Voltage (MV) network with DG sources, together with storage devices and controllable loads (e.g. water heaters, air conditioning, etc.,) with a total installed capacity in the range of few kilowatt to a couple of megawatt [4]. The feature of MGs is that although they operate most of the time connected to the upper level voltage DN, they can be automatically transferred to islanded mode, in case of faults in the upstream network. Some authors emphasize the peer-to-peer and plug-and-play concepts of generating units, which can be installed in any point of the electrical system, with the possibility to place generators near the electrical and heat loads [4].

From the customer’s point of view, MGs provide both thermal and electricity needs, and, in addition, enhance local reliability, reduce emissions, improve power quality by

(29)

5 supporting voltage and reducing voltage dips, and potentially lower costs of energy supply [1].

From the grid’s point of view, the matching of DG with nearby loads implies that the flows which characterize the transmission and distribution circuits are reduced and thus reducing power losses, while supplying the same load. Moreover, MGs take additional advantages to the local utility, by providing dispatchable power which can be used during peak power conditions and alleviating or postponing distribution system upgrades [4]. Unfortunately, MGs have not only benefits: connection of DG unit changes properties and profiles of the distribution system (DS), leading to the emergence of many potential problems [4]. For example, there could be problems of stability, voltage and frequency regulation, bi-directional power flow, load shedding, etc., and this kind of issues are responsible of the failure of many automatisms of the electrical devices [4]. In this work, the attention was focused on malfunctions of the protection devices during fault event caused by the installation along the DN of some generating device.

In the next chapters, are described primarily the types of malfunctions that affect the protection system and then some methods that could be implemented to avoid them.

(30)

Chapter 1. EVOLUTION OF ELECTRIC POWER SYSTEM

(31)

7

CHAPTER 2.

TECHNICAL ISSUES OF THE

PROTECTION SYSTEM

2.1 Introduction

As previously mentioned, DNs are in transition towards future smart grids and the amount of DGs connected to networks is expected to increase steadily. The addition of active distributed energy resources, which include conventional fuel powered DGs and RES powered DGs, brings up new requirements for protection that needs to adapt to changes in the network topology and configuration [4]. The direct consequence of the changes in the direction of power flow and short circuit levels due to the increasing number of DG units is that the requirements for protection devices are getting more complicated [4] and the traditional, non-directional overcurrent based, protection system is not anymore adequate [5]. Typical issues, which compromise the good operation of protection devices, are the following:

- Variation of the short circuit level in the island mode: connection and disconnection of local generators and lines and the passage form grid-connected to islanded mode bring to have different topological status of the electrical network and, consequently, different available short circuit levels. This kind of issues make protection devices ineffective and not suitable during a fault event [4];

- Protection blinding: reduction of fault sensitivity and tripping speed due to the presence of DG units along the line [4];

(32)

Chapter 2. TECHNICAL ISSUES OF THE PROTECTION SYSTEM

8

- Sympathetic tripping: unwanted tripping of healthy feeder for a fault on another feeder due to DG fault contribution [3].

Now this kind of issues are analysed more in details, with examples and results of simulations, in order to better understand them and to focus on which kind of solutions could be adopted.

2.2 Short circuit level variation in island mode

An MG is designed to operate connected to the utility grid and, in some cases (quantity local generation matches loads, fault at upper level, etc. [4],), disconnected from it, operating in islanded mode. Consequently, when a fault occurs, the value of fault currents depends on the configuration of the MG [4]. The short circuit level may also change due the status and the type of local generators which contribute to the short circuit. In other words, the short circuit level depends on the system topology and on the type of the generators interfaced with the electrical system and the currents seen by the protection device when a fault occurs are generally lower than the fault currents seen when the MG is connected to the utility [4]. Moreover, the power output of RES DGs is often unpredictable and the behaviour of a microgrid when a fault occurs changes constantly [3].

Consider the power system represented in Figure 2.1:

Figure 2.1: Simulation of fault current in grid-connected and island mode

As previously mentioned, MG could be disconnected form the utility grid and works independently. When generators at bus 2 and at bus 3 are providing enough power, R1

(33)

2.2 Short circuit level variation in island mode

9 opens and the power system switches to the islanded mode. Considering this change from the point of view of network protections, in case of fault at bus 4, R2 must deal with a short circuit current significantly smaller in the islanded mode since there is no more the contribution of the main grid, and without an appropriate adjustment of protection settings the device will not work properly (non-tripping or excessive delay).

In Figure 2.2, are shown the results of a computation of a three-phase-fault at bus 4, first in grid-connected mode and then in islanded mode and it is evident the large difference between the two cases. In this case we took in consideration the availability of the generator at bus 2 and of the generator at bus 3.

Figure 2.3 instead shows the results of a scenario analogue to the previous one but taking in consideration the availability of only the generator at bus 2 and with a slight different in load profile.

Figure 2.2: Fault current in branch 3-4 (grid-connected mode and island mode)

After these considerations, it is evident how much important and necessary is the development of powerful adaptive protection methods in this new microgrid scenario. In the final chapters, it will be shown a hardware in the loop (HIL) simulation of an EPS implementing powerful protection methods that could allow to avoid malfunctions of this type.

(34)

Chapter 2. TECHNICAL ISSUES OF THE PROTECTION SYSTEM

10

Figure 2.3: Fault current in branch 3-4 (grid-connected mode and island mode)

2.3 Protection blinding

Blinding of overcurrent (OC) protection may occur when a DG unit is located between the main grid and the fault location, since the fault current seen by the protective device at the beginning of the feeder decreases, because of the DG short-circuit current contribution [6]. In order to describe the phenomena, we consider the distribution system presented in Figure 2.4 [4].

The study scheme consists in a radial feeder, 300 m long, protected by a single CB, indicated with CB_F and installed at the beginning. Every 100 m, three phase loads, 300 kVA each, are connected at the feeder and, at the point indicated with 2, a synchronous generator (SG) can be switched on or off [4].

(35)

2.4 Sympathetic tripping

11 If a fault occurs in the position indicated with 3 in Figure 2.4, the short circuit current seen by the protection device can take low values, due to the short circuit contribution of SG installed in other points of the considered network [4]. This causes the non-tripping or excessive delay of CB_F, despite existence of a fault in the related feeder. The phenomenon becomes more and more relevant as the size of the SG increases, and as its internal impedance decreases [4]. As a consequence, the current passing through the faulty zone is higher than the case without SG, but the current seen of the CB_F is smaller and, if the settings of the protection relay are not modified, the device might trip with an unacceptable time delay and the whole system could be damaged by thermal effects [4]. The diagram in Figure 2.5 shows the fault current, seen by CB_F, in case of connection/disconnection of a 1 MVA SG. After the fault instant (0.9 s), CB_F is interested by a current which is about 600 A less than the current when the SG is disconnected. In order to ensure a correct and fast intervention in the both cases, with and without SG, a modification of the settings is fundamental [4].

Figure 2.5: Fault current affected by blinding phenomena

2.4 Sympathetic tripping

One potential impact of interconnecting distributed generation units is the so-called sympathetic tripping of OC protection devices, where healthy feeders trip unnecessarily for a fault on another feeder [6].

Sympathetic tripping comes from DG units with high short circuit current contribution (typically rotating machines such as diesel or gas turbine units) and can be observed in radial feeders that are fed from a common source [3].

(36)

Chapter 2. TECHNICAL ISSUES OF THE PROTECTION SYSTEM

12

Figure 2.6: Distribution system used to study sympathetic tripping

Figure 2.6 shows two radial feeders with a common connection to the main grid. With no DG units installed, if a fault occurs at bus 3, the short circuit will be completely fed from the utility source with no contribution from the other feeder. As a result, R2 will not respond to the short circuit and the fault will be cleared by R3 as intended. In contrast, if a DG is installed on the healthy feeder, the short circuit contribution from this feeder will be relevant and if R2 has a faster characteristic than R3 it could respond to the fault inappropriately and interrupt the loads of the healthy feeder before R3. To summarize, the sympathetic tripping or bi-directionality issue could potentially occur when the following conditions are met:

- Two or more radial feeders are fed from a common source;

- DG units with a relatively high short-circuit contribution are installed in a feeder; - OC relays at the feeder where DG unit is installed has faster characteristic than the

(37)

3.2 Protection scheme based on fault current direction

13

CHAPTER 3.

PROTECTION SCHEMES FOR

MICROGRID

3.1 Introduction

Until now we focused on the new paradigm of the distribution system, passing from a totally passive role to a smart grid, integrating DGs and reaching level of automation completely innovative compared to the past.

Having analysed some typical issues that can arise in this new environment, we are going to study in detail some of the many protection techniques that are object of accurate research in these last years.

3.2 Protection scheme based on fault current direction

In [7] were developed protection schemes for different kinds faults in autonomous MG, aiming at finding innovative protection schemes that involve minimum implementation costs for the distribution network operator and try, as far as possible, to make use of traditional appliance. The tested MG [8] is shown in Figure 3.1 and it is assumed that an effective microgrid central controller (MGCC) with communication, data acquisition and automation systems is available to perform several control functions (such as optimization of distributed energy resources use, minimization of power and energy losses, optimization of voltage profiles, optimal generators dispatching, power flows management, etc.). Further, the MGCC functions should be coordinated with the protection scheme to reduce interruptions frequency and duration by allowing for

(38)

Chapter 3. PROTECTION SCHEMES FOR MICROGRID

14

network reconfiguration and assisting the local DG regulators to maintain stability, if possible, in the MG or in its subsystems separated after a fault [7].

In the protection scheme tested, two directional varmetric earth-fault relays with local control logic are installed in each branch. Each relay “looks” into the branch direction (forward tripping direction), as shown in Figure 3.2 (yellow arrows).

Figure 3.1: Tested microgrid

The two relays of a given branch will trip if they both sense the fault in the positive direction. Assume that fault occurs in branch 1.3 and all relays that sense the fault in their tripping direction, i.e. P11, P21, P31, P42, at the same time will ask information about the fault current direction to the coupled relay on the same branch. If a questioned device replies by telling it has sensed a positive direction fault, then the two coupled relays trip and clear the fault [7]. In the considered example, this happens to devices P31 and P32. Branch 1.4 is isolated, but both loads could keep being supplied if a generator were connected to one of the two nodes.

(39)

3.3 Matrix strategy based on fault current direction

15 Figure. 3.3 shows the scheme in operation in case of fault on branch 3.2. All feeder relays that sense the fault current with the same direction as the tripping one, indicated by yellow arrows (i.e. P31a, P32a, P32b and P33b) are activated and send their status to the coupled relay on the same branch (P31b, P32b, P32a and P33a, respectively). In the considered example P32a and P32b will trip and clear the fault.

Figure 3.3: Simulation of fault current in branch 2.3

This preliminary work has the main purpose to propose a simple solution from a computational point of view, since the only variable used here is the direction of the current seen by each relay. On the other hand, it needs very sophisticated devices (like SCADA and automation systems) able to continuously control and monitor the state of the protection devices and, last but not the least, it needs a very reliable, efficient and fast communication network able to work in interval of time of the order of milliseconds.

3.3 Matrix strategy based on fault current direction

In [9] a new strategy to protect MGs using the fault direction information to achieve fault location and isolation is proposed.

In Figure 3.4 is presented a LV photovoltaic (PV) microgrid with several PV units connected to the grid (PCC).

(40)

Chapter 3. PROTECTION SCHEMES FOR MICROGRID

16

The whole system is composed by an MGCC linked to protection devices on site. These devices detect the current direction information and communicate these informations to the control center which, running a protection algorithm, sends trips order to the selected devices and isolates the fault region [9].

The algorithm implemented by the MGCC is summarized in Table 3.1: Table 3.1: Protection Algorithm

Step1: Computing switch-branch matrix A 𝑎𝑖,𝑗= 1, 𝑠𝑤𝑖𝑡𝑐ℎ 𝑖 𝑑𝑖𝑟𝑒𝑐𝑡𝑙𝑦 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝑏𝑟𝑎𝑛𝑐ℎ 𝑗 (𝑏𝑟𝑎𝑛𝑐ℎ 𝑑𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚 𝑡ℎ𝑒 𝑠𝑤𝑖𝑡𝑐ℎ); 𝑎𝑖,𝑗= −1, 𝑠𝑤𝑖𝑡𝑐ℎ 𝑖 𝑑𝑖𝑟𝑒𝑐𝑡𝑙𝑦 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝑏𝑟𝑎𝑛𝑐ℎ 𝑗 (𝑏𝑟𝑎𝑛𝑐ℎ 𝑢𝑝𝑠𝑡𝑟𝑒𝑎𝑚 𝑡ℎ𝑒 𝑠𝑤𝑖𝑡𝑐ℎ); 𝑎𝑖,𝑗= 0, 𝑠𝑤𝑖𝑡𝑐ℎ 𝑖 𝑛𝑜𝑡 𝑑𝑖𝑟𝑒𝑐𝑡𝑙𝑦 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝑏𝑟𝑎𝑛𝑐ℎ 𝑗; Step2: Judging if there is a fault on the bus 𝑑𝑖𝑟(𝑃𝐶𝐶) = 1, 𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑎𝑢𝑙𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑛 𝑐𝑜𝑢𝑙𝑝𝑙𝑖𝑛𝑔; 𝑑𝑖𝑟 (𝑆𝑗) = 0, 𝑗 = 1,2 … 𝑛 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑎𝑢𝑙𝑡 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑖𝑛 𝑒𝑎𝑐ℎ 𝑜𝑓 𝑡ℎ𝑒 𝑛 𝑑𝑒𝑣𝑖𝑐𝑒𝑠 𝑖𝑠 𝑛𝑜𝑡 𝑓𝑜𝑟𝑤𝑎𝑟𝑑;

In this step, the algorithm can establish that the fault is in the PCC and not in the branches

Step3: Judging if there is a fault on the feeder 𝑑𝑖𝑟(𝑃𝐶𝐶) = 1, 𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑎𝑢𝑙𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑛 𝑐𝑜𝑢𝑙𝑝𝑙𝑖𝑛𝑔; 𝑑𝑖𝑟 (𝑆𝑗) = 0, 𝑗 = 1,2 … 𝑘 − 1, 𝑘 + 1, 𝑛; 𝑑𝑖𝑟 (𝑆𝑘) = 1, 𝑘 𝑖𝑠 𝑡ℎ𝑒 𝑓𝑎𝑢𝑙𝑡𝑒𝑑 𝑓𝑒𝑒𝑑𝑒𝑟;

In this step, the algorithm can establish that the fault is in the k feeder

Step4: Location of faulted branches: computing fault direction matrix B 𝑏𝑖= 𝑑𝑖𝑟(𝑖) = 1, 𝑓𝑎𝑢𝑙𝑡 𝑎𝑡 𝑡ℎ𝑒 𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑤. 𝑟. 𝑡 𝑑𝑒𝑣𝑖𝑐𝑒 𝑖; 𝑏𝑖= 𝑑𝑖𝑟(𝑖) = 0, 𝑓𝑎𝑢𝑙𝑡 𝑡ℎ𝑒 𝑟𝑒𝑣𝑒𝑟𝑠𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑤. 𝑟. 𝑡 𝑑𝑒𝑣𝑖𝑐𝑒 𝑖 𝑜𝑟 𝑛𝑜 𝑓𝑎𝑢𝑙𝑡; Step5: Location of faulted branches: computing branch fault information matrix B 𝐺 = 𝐵𝑥𝐴; 𝑔𝑖= 0, 𝑛𝑜 𝑓𝑎𝑢𝑙𝑡 𝑜𝑛 𝑏𝑟𝑎𝑛𝑐ℎ 𝑖; 𝑔𝑖= 1, 𝑓𝑎𝑢𝑙𝑡 𝑜𝑛 𝑏𝑟𝑎𝑛𝑐ℎ 𝑖;

(41)

3.4 Data-mining based protection scheme

17 Once the G matrix is computed, a trip order to each protection device related to the faulted branch is sent, the fault is cleared and the system can continue to provide the service [9]. This time, even if the method relies on a limited number of measurement (the direction of fault current), it could be easily implemented for a DN with a reduced number of busses but not so easily for a more extended one. Related to the dimension of the grid, there could be also problem linked to the control systems (SCADA) which has to increase their computational power, their internal memory and tools as the dimension of the system increases. Finally, all this operation could be implemented only with a reliable and fast communication network. Otherwise, the exchange of information would not allow to interrupt fault current in a suitable time.

3.4 Data-mining based protection scheme

In [11] a data-mining based protection scheme for MG using decision tree is presented. Current and voltage signals are pre-processed at both ends of the faulted feeder using discrete Fourier transform (DFT) and results are used to compute a set of differential features, which are used to build decision trees (DTs) (for primary and backup protection) for the final relaying decision [11]. These features are presented in Table 3.2:

Table 3.2: Differential features for DT

FEATURES FOR BACKUP PROTECTION

𝑋1𝑃= ∆ (

𝑑𝑉

dt) , the differential rate of voltage [ 𝑝. 𝑢

𝑠 ] ; 𝑋2𝑃= ∆ (

𝑑𝑓

dt) , the differential rate of frequency [ 𝐻𝑧

𝑠] ; 𝑋3𝑃= ∆ (

𝑑𝑃ℎ𝑖

dt ) , the differential rate of power angle difference; 𝑋4𝑃= ∆ (

𝑑𝑉𝑛𝑒𝑔

dt ) , the differential rate of change neg seq voltage; 𝑋5𝑃= ∆ (

𝑑𝐼𝑛𝑒𝑔

dt ) , the differential rate of change of neg seq current;

FEATURES FOR BACKUP PROTECTION

𝑋1𝐵= ∆ (

𝑑𝑉

dt) , the differential rate of voltage [ 𝑝. 𝑢

𝑠 ] ; 𝑋2𝐵= ∆ (

𝑑𝑓

dt) , the differential rate of frequency [ 𝐻𝑧

𝑠] ; 𝑋3𝐵= ∆ (

𝑑𝑃ℎ𝑖

dt ) , the differential rate of power angle difference; 𝑋4𝐵= ∆ (

𝑑𝑉𝑛𝑒𝑔

dt ) , the differential rate of change neg seq voltage; 𝑋5𝐵= ∆ (

𝑑𝐼𝑛𝑒𝑔

(42)

Chapter 3. PROTECTION SCHEMES FOR MICROGRID

18

The proposed intelligent protection scheme is shown in Figure 3.5 [11]. The post fault instantaneous current and voltage signals are retrieved using current transformer (CT) and voltage transformer (VT) respectively at the ends of the faulted feeder and then pre-processed through DFT pre-processor. The DFT pre-processor estimates amplitude, phase and frequency and subsequent related differential features [11].

In the end, with this algorithm carried out by the developed DT (Figure 3.6) protection devices give a high level of protection to the PS under study. Of course, compared to the other methods, this is very complex from a computational point of view and needs very high-performance processors and computing tools to accomplish the task even in the case of not very extended DNs. Communication system is essential and, continuing to analyse different protection methods and techniques, we can notice how much this element is becoming a fundamental element of the whole electrical system [15].

Figure 3.5: Protection scheme

(43)

3.5 Protection scheme based on zonal division

19

3.5 Protection scheme based on zonal division

In [13] an adaptive zonal protective scheme for DSs integrated with DG units is proposed. The DN (Figure 3.7) is divided into zones with the help of boundary circuit breakers, which basically are the breakers in the point of connection between the area of the network containing the DG (DG zone) and the area with no DG (main zone) and the operation of these relays is completely controlled by main control center.

The following assumptions are made in the scheme:

• all the relays in the system are numerical relays having proper communication facility with the control center. This is very well feasible in the smart grid environment;

• all the DGs are equipped with frequency and voltage control mechanisms. If not, DGs have to be interrupted during fault period in the system.

Figure 3.7: Division of DN into main zone and DG zone

The functioning of proposed scheme involves the following steps:

- as soon as the fault is sensed by any one of the relays in the system, it should be communicated to the control center before the operating time of the relay located at the farthest end of the system;

- the control center should give the trip signal to the boundary circuit breaker to break the circuit;

(44)

Chapter 3. PROTECTION SCHEMES FOR MICROGRID

20

- simultaneously trip signals should be given to the DGs to interrupt the power fed into the grid if they are not having frequency and voltage control capability; - after fault clearance by coordinated operation of relays, the control center should

give the command to boundary circuit breakers to connect the circuit.

This chapter analysed some of the numerous protection methods that in these years have been studied and proposed. Obviously, the target is to develop protection algorithms which could be adaptable to most electrical systems, minimizing costs and maximizing all that potentiality needed for a good operation of the protection system (speed, sensitivity, selectivity) [14].

After the analysis of the presented methods, we can conclude that there are some elements (such as communication systems, SCADA systems and other computational devices) which in the past were not fully exploited but now are becoming more and more fundamental elements [15].

Now we are going to study in detail some proposed protection methods that could be easily implemented and adapted to many cases and we will see HIL simulation of a power system during normal and fault condition, working with these methods.

(45)

21

CHAPTER 4.

DESCRIPTION OF THE SIMULATED

PROTECTION ALGORITHMS

4.1 Introduction

In this chapter, we want to present two adaptive protection algorithms that could be both practical and powerful methods applicable in the new smart grid environment. These methods are the results of a research work developed in the ABB Corporate Research Center in Baden-Dättwil.

The first one is based on use of the Impedance matrix, and the variation of this mathematical element every time a new power sources is connected to the power system. The second one is based on the division of the network in protection areas and each area communicates with the others its own short circuit level.

In the following part of the chapter is provide a detailed analysis of the algorithms.

4.2 Protection method based on variation of Impedance matrix

This method is similar to the others previously described ([7], [9], [11], [13]) from a structural point of view: in fact, it is assumed that an effective MGCC with communication, data acquisition and automation systems is available to perform several control functions and it is in direct communication with intelligent electronic devices (IED) on site (i.e. digital relays).

If in the previous methods the MGCC has to coordinate protection devices during fault event, and so with a very limited time action and needing a very reliable, fast and faultless

(46)

Chapter 4. DESCRIPTION OF THE SIMULATED PROTECTION ALGORITHMS

22

communication system, this time the control center coordinates IEDs based on the Impedance matrix variation. In fact, the Impedance matrix is computed in the MGCC based on the utility grid, line impedances and on the sub-transient reactance of each DG units connected to the PS [27]. When a DG unit is connected to the grid, the switching device connecting the unit transmits to the control center the value of the unit sub-transient reactance, and this information is used to update the new value of the matrix. In this way, every time a DG unit is connected or disconnected from the grid, the matrix changes accordingly and also the value of the short circuit current in case of fault [16]. Once this change is detected, the MGCC sends an adaptive order to all protection devices and makes them eligible for the new situation.

Relying on this concept, a flowchart can be developed for the adaptive protection method like in Figure 4.1:

Figure 4.1: Flowchart of the adaptive method

4.3 Protection method based on area concept

The method divides the DN in multiple protection areas, each comprising a busbar, a protection device and an area controller [17]. Upon a change in connect status of a distribution feeder line or any kind of DG to one of the busbar, the logic controller

(47)

re-4.3 Protection method based on area concept

23 calculates the short circuit level of its own area and, based on the recalculated short circuit level, the protection settings of devices may be adapted [17]. Each controller is in direct communication with nearby controllers, which receive this new value through a direct communication channel and so they can also adapt their protection devices following the new situation. The objective of this method is to provide a decentralized protection of active DN with a dynamic setting adaptation in different parts of the network without the need to know a complete system topology, reducing the burden on communication infrastructure which links different parts of the network and enabling an automatic system adjustment after integration of new distributed generators or feeder extension [17].

4.3.1 Local logic area controller

The protection area can be seen as a black box with connection ports, which can exchange power in normal and fault conditions with the direct neighboring areas via external ports and with its local generators and loads with internal ports [17]. Each port is connected to a local logic controller (LLC) via bi-directional communication lines such as bus, hardwired, wireless, etc. [17]. The LLC can receive and send information from/to each port and also talk to the direct neighboring areas, in order to coordinate protection and emergency control actions (Figure 4.2).

Figure 4.2: Exemplary communication between protection areas

4.3.2 Short circuit level

In order to evaluate the short circuit level in each protection area, the so called “MVA method” for fault analysis [3] [17] [33] is exploited. In the MVA method each network

(48)

Chapter 4. DESCRIPTION OF THE SIMULATED PROTECTION ALGORITHMS

24

component is replaced by a block representing the component’s contribution to the short circuit level [33]. When the MVA diagram is traced, we can calculate the short circuit level at a specific point: starting from this point, different elements are combined as series and parallel components.

This method establishes the short circuit power in a system in a simple and fast way [3]. The method works with the equivalent short circuit power of each component, and these values are managed like admittances [3]. Admittances are calculated considering each component connected to an infinite-power bus (Figure 4.3).

Figure 4.3: MVA method explanation

It isn't necessary know from the beginning the MVA value, it can be obtained with the formulas shown in Table 4.1:

Table 4.1: Formulas to calculate the MVA value

Utility grid 𝑉𝑙𝑖𝑛𝑒−𝑡𝑜−𝑙𝑖𝑛𝑒2 𝑋𝑔𝑟𝑖𝑑 Transformer 𝑀𝑉𝐴𝑡𝑟𝑎𝑛−𝑛𝑜𝑚𝑖𝑛𝑎𝑙 𝑋𝑡𝑟𝑎𝑛−𝑝.𝑢. Generator 𝑀𝑉𝐴𝑔𝑒𝑛−𝑛𝑜𝑚𝑖𝑛𝑎𝑙 𝑋𝑔𝑒𝑛−𝑝.𝑢.′′ Motor 𝑀𝑉𝐴𝑚𝑜𝑡−𝑛𝑜𝑚𝑖𝑛𝑎𝑙 𝑋𝑚𝑜𝑡−𝑝.𝑢.

(49)

4.3 Protection method based on area concept

25 Figure 4.4 shows a simple example. At the fault point there will be an amount of short circuit power depending on the contribution of the elements. Since they behave like admittances, the formulas to get a result are the following:

series components → 𝑀𝑉𝐴1,2=

𝑀𝑉𝐴1∗ 𝑀𝑉𝐴2 𝑀𝑉𝐴1+ 𝑀𝑉𝐴2

;

parallel components → 𝑀𝑉𝐴1,2= 𝑀𝑉𝐴1+ 𝑀𝑉𝐴2 ;

Figure 4.4: MVA method explanation

4.3.3 The adaptive algorithm

The algorithm is divided in two phases: the offline phase and the online phase. The off-line phase is the phase before the PS gets into operation. In this phase, as we can see from the flowchart presented in Figure 4.5, the system analyses the topology of the DN, defining each area and its LLC and the short circuit power level is computed. After these

(50)

Chapter 4. DESCRIPTION OF THE SIMULATED PROTECTION ALGORITHMS

26

computations, protection devices settings are adapted accordingly and the PS starts to supply the service operating in a secure way.

During the online phase (Figure 4.6) each LLC continuously checks if there are connections or disconnections of any kind of power sources in the area under its domain and the consequent variations of the short circuit power level, communicating this value to every neighbour. In parallel to this action, it receives the short circuit level of nearby areas and, with this two information, protection settings are updated accordingly.

Figure 4.5: Flowchart of the adaptive method (offline phase)

(51)

27

CHAPTER 5.

HARDWARE-IN-THE-LOOP SETUP

5.1 Introduction

The developed setup has the purpose to test the adaptive protection methods described in the previous chapter, running different simulation of an LV electric power system (with loads and a DG unit) in a LabVIEW environment and, at the same time, performing a digital-to-analog conversion of the simulations values and use them as input for real protection devices (ABB SACE EMAX 2). These IEDs are in direct communication with LabVIEW through a Modbus TCP/IP communication channel and the network topology simulated by the software can be updated based on the CBs status (OPEN/CLOSED). When the configuration changes, the analog values used to feed protection devices change accordingly, creating in this way an “hardware in the loop setup” composed by and hardware component (protection devices) and a software component (LabVIEW).

5.2 Real scenario

Before analysing in detail each component of the setup and explaining how they are coordinated and how they interact among each other, a short description of the measuring process that IEDs effectuate in a real environment to compute the value of the current flowing in a feeder is given.

Let consider the schema shown in Figure 5.1 (taken from EMAX 2 technical catalogue [18]).

(52)

Chapter 5. HARDWARE-IN-THE-LOOP SETUP

28

In this picture, the three phases (indicated with L1, L2 and L3) and the neutral line (indicated with N) are shown. In each phase, a Rogowski Coil (RC) (red squares) is installed.

RC is a current transformer which operates on the same principles as conventional iron-core CTs [19]. The main difference with others CTs is that RC windings are wound over an air-core instead of an iron-core, resulting in a linear behaviour of the output voltage [19]. This current transformer (Figure 5.2) can be used to provide input signals for microprocessor-based devices with high input resistance, which measure the voltage across its output terminals [19]. The output voltage is proportional to the rate of change of measured current as given by the formula [19]:

𝑣𝑐𝑜𝑖𝑙= −𝑀 ∗ 𝑑

𝑑𝑖 𝑑𝑡;

Figure 5.1: Protection devices located in a real environment

(53)

5.3 Software component

29 With this concepts in mind, it is possible to understand how protection devices measure the current flowing in the feeder: considering L1 in Figure 5.1 for example, the current transformer output is directly connected with the input port of the IED. The latter processes the value of the voltage received (in the range of 𝑚𝑉), gaining in this way the actual value of the line current. A logical scheme of the entire process is presented in Figure 5.3:

Figure 5.3: Protection devices located in a real environment

5.3 Software component

This part of the setup was entirely developed with LabVIEW. LabVIEW is a system-design platform and development environment for a visual programming language from National Instrument [20]. It is a powerful program which offers a wide range of possibilities in programming.

The main purpose of the project was to simulate an LV distribution system (with utility grid, loads and a synchronous generator (SG) exploited as DG unit) during normal working conditions (Figure 5.4) and during fault conditions (Figure 5.5). In the next part of the chapter it is explained how the different components of the power system simulation have been developed in LabVIEW and which library were mostly exploited.

(54)

Chapter 5. HARDWARE-IN-THE-LOOP SETUP

30

Figure 5.5: Simulated power system during fault condition

5.3.1 Synchronous generator modelling and implementation

As previously mentioned, a mathematical model of an SG has been developed in order to represent the presence of a DG unit in the system. The model was reproduced taking as reference the one developed in Matlab Simulink [21]. Using LabVIEW Control and Simulation library, time parameters (initial time, step size time, etc.,) and the method for solving the differential equation of the model have been chosen and set up (Figure 5.6). Finally, a real-time simulation of a DG unit was achieved [22] [23].

(55)

5.3 Software component

31 Taking as input the technical parameters of the SG (stator resistance, stator reactance, etc.), mechanical torque and the excitation voltage and converting the reference frame of a 3-phase voltages to a 𝑑𝑞0 (direct - quadrature - zero) [34] with Park transformation matrix [21], LabVIEW performs a computation of windings flux linkages and stator windings currents, obtaining in the end the electrical torque with the following formula:

𝑇𝑒𝑚= 3 2 𝑃 2𝜔𝑏 (𝛹𝑑𝑖𝑞− 𝛹𝑞𝑖𝑑).

This value is subsequently compared with the input mechanical torque, using the following formula:

𝑇𝑚− 𝑇𝑒𝑚= 𝑇𝑎= 𝐽

𝑑2𝛿

𝑑𝑡2.

obtaining in this way the value of the variation of the angle stability of the machine in each moment of the simulation [24]. Therefore, when a fault occurs, voltage drop makes the electrical torque become smaller, creating an imbalance between this value and the mechanical torque and leading and, in some cases, leading to the generator’s instability. In this way, it is possible to control the dynamic behaviour of the machine during transient conditions (fault event, switching to island mode, load variation, etc.,).

In Figure 5.7 the graph of the angle stability during a fault event is shown:

Figure 5.7: Angle stability during fault event

In Figure 5.8 the SG control panel layout with all the input variables (excitation voltage, mechanical torque, etc.,) and the output variables (active power, reactive power, etc.,) is presented. In Figure 5.9, the overall script of the LabVIEW scheme.

Once the DG unit model has been developed, it is possible to develop the entire DN (line, loads, utility grid) in which the machine will be installed.

(56)

Chapter 5. HARDWARE-IN-THE-LOOP SETUP

32

The next step is the definition of the structure of the electrical system through the impedance matrix and the admittance matrix and of the power flow iteration model that, taking as input the values of the main grid voltage, loads and the results of the generator simulation, give as results the voltage profile and current profile of each bus and each branch in the electrical system.

Figure 5.8: Control panel of synchronous generator

(57)

5.3 Software component

33

5.3.2 Power flow implementation

For the development of the power flow calculation, the three busses of the system (Figure 5.10) were considered as follow:

• Bus 1: SLACK BUS; • BUS 2: PV BUS; • BUS 3: PQ BUS;

Figure 5.10: Model of the simulated power system

For the implementation of the power flow algorithm the library MathScript RT has been used. First, the Impedance matrix and the admittances matrix of the system was computed taking as value for the line impedances the values used in Matlab Simulink. As it is shown in figure 5.11, at the beginning of the procedure there are the values of the impedance matrix and of the admittances matrix for each configuration of the system (grid-connected mode with generator, grid-connected mode without generator and island mode) and the inputs values of each node.

(58)

Chapter 5. HARDWARE-IN-THE-LOOP SETUP

34

Figure 5.12: LabVIEW script of the power flow subVI (Gauss-Seidel iteration)

In figure 5.11 instead are presented the mathematical iterations for the computation of bus voltages and branch currents using Gauss-Seidel iterative method [26] with the following formula: 𝑉𝑖,𝑛𝑒𝑤= 1 𝑌𝑖𝑖[ 𝑃𝑖− 𝑗𝑄𝑖 𝑉𝑖,𝑜𝑙𝑑∗ − ∑ 𝑌𝑖,𝑘𝑉𝑘 𝑁 𝑘=1,𝐾≠𝑖 ].

Since loads values and power produced by the generator are inputs that could be controlled during the simulation, every time there is a change in these values also bus voltages and branch currents profiles change accordingly, and it is possible to observe a real-time variation of all parameters of the entire system.

Once this step is calculated, the PS is ready to be simulated. At this point, in order to simulate a fault condition and to test the protection algorithm described above, it is necessary to develop a dedicated code for this purpose, as it will be described in the next paragraph.

(59)

5.3 Software component

35

5.3.3 Short Circuit implementation

The purpose of this code is to perform the simulation of a three-phase fault in the PQ bus of the system (BUS 3). Even here, the library used was MathScript RT.

The formula for the computation of the short circuit current in case of a three-phase fault is [27]:

𝐼𝑖,𝑠𝑐=

𝑉𝑖

𝑍𝑖,𝑖

,

where 𝑉𝑖 is the pre-fault value of bus 𝑖 voltage and 𝑍𝑖,𝑖 the value of the (𝑖, 𝑖) element of the

impedance matrix [27] [35]. In this case, with the results obtained in the previous computations, it is easier to program the code. Figure 5.13 and figure 5.14 show the script programmed for this scope: taking the pre-fault values of the voltages computed by the power flow subVI and the impedance matrix, the model gives as output the short circuit current flowing in the branches and the bus voltages during the fault.

Figure 5.13: LabVIEW script of the short circuit subVI (input values)

(60)

Chapter 5. HARDWARE-IN-THE-LOOP SETUP

36

5.3.4 Modbus TCP/IP

As last step of the entire simulation, the Modbus TCP/IP code was programmed to establish a communication between the software part and the hardware part (i.e., protection devices EMAX 2).

IEC 61850 is a standard for the design of electrical substation automation and it exploits Modbus TCP/IP network to exchange information with other IEDs [28]. This particular version is a variant of the MODBUS family of simple, vendor-neutral communication protocols intended for supervision and control of automation equipment [29]. Specifically, it covers the use of MODBUS messaging in an ‘Intranet’ or ‘Internet’ environment using the TCP/IP protocols. The most common use of the protocols at this time are for of PLC’s, I/O modules and ‘gateways’ to other simple field buses or I/O networks [29].

Thanks to EMAX 2 Modbus TCP/IP module and LabVIEW Modbus library, it was achievable the realization of a network with a master-slave structure. The software (master) obtains all the main information from each IED (slave) and, accordingly, updates the topology of the electrical system, based on this exchanged information.

Riferimenti

Documenti correlati

a proportional change of all money prices and money income leaves the budget line unchanged.. money is

Animal Models of p63 -Linked Diseases Much of our current knowledge on the role of disease genes for ectoderm development and limb morphogenesis has been gathered via the generation

Therefore, the task of operational forecasting is considered in terms of filing applications on a balancing market for the next hour, as well as for obtaining reference data on

The optimization model has been developed as Mixed Integer Linear Programming (MILP) for minimizing the cost of buying energy from the utility grid, maximizing energy efficiency

The uncertainty, the ramp risks and all the issues related to the strong penetration of NP-RES, led grid operators to involve more balancing generation units with a consequence

Il presente lavoro di tesi si focalizza sull’analisi delle strategie di demand response in alta tensione, ossia in trasmissione, e sull’introduzione di una

It should be emphasized that until 2017 referen- dum that changed the system of government, Turkey was actually ruled by a de facto presidential system that President Erdoğan

Alluvial rivers in northern and central Italy underwent simi- lar channel adjustments over the past 200 yr: (1) narrowing and incision occurred in all 12 selected rivers from the