Protection of distributed generation interfaced networks

Transcription

1 Protection of distributed generation interfaced networks Manjula Dewadasa B.Sc (Hons) in Electrical Engineering A Thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy Faculty of Built Environment and Engineering School of Engineering Systems Queensland University of Technology Queensland, Australia July 2010

2

3 Keywords Distributed generation, Microgrids, Distributed generator protection, Converter interfaced distributed generators, Protective relays, Inverse time admittance relay, Relay coordination, Relay Grading, Islanded operation, Re-synchronisation, Reclosing, Fold back current control, Fault detection, Fault isolation, Arc extinction, System restoration. i

4

5 Abstract With the rapid increase in electrical energy demand, power generation in the form of distributed generation is becoming more important. However, the connections of distributed generators (DGs) to a distribution network or a microgrid can create several protection issues. The protection of these networks using protective devices based only on current is a challenging task due to the change in fault current levels and fault current direction. The isolation of a faulted segment from such networks will be difficult if converter interfaced DGs are connected as these DGs limit their output currents during the fault. Furthermore, if DG sources are intermittent, the current sensing protective relays are difficult to set since fault current changes with time depending on the availability of DG sources. The system restoration after a fault occurs is also a challenging protection issue in a converter interfaced DG connected distribution network or a microgrid. Usually, all the DGs will be disconnected immediately after a fault in the network. The safety of personnel and equipment of the distribution network, reclosing with DGs and arc extinction are the major reasons for these DG disconnections. In this thesis, an inverse time admittance (ITA) relay is proposed to protect a distribution network or a microgrid which has several converter interfaced DG connections. The ITA relay is capable of detecting faults and isolating a faulted segment from the network, allowing unfaulted segments to operate either in grid connected or islanded mode operations. The relay does not make the tripping decision based on only the fault current. It also uses the voltage at the relay location. Therefore, the ITA relay can be used effectively in a DG connected network in which fault current level is low or fault current level changes with time. Different case studies are considered to evaluate the performance of the ITA relays in comparison iii

6 to some of the existing protection schemes. The relay performance is evaluated in different types of distribution networks: radial, the IEEE 34 node test feeder and a mesh network. The results are validated through PSCAD simulations and MATLAB calculations. Several experimental tests are carried out to validate the numerical results in a laboratory test feeder by implementing the ITA relay in LabVIEW. Furthermore, a novel control strategy based on fold back current control is proposed for a converter interfaced DG to overcome the problems associated with the system restoration. The control strategy enables the self extinction of arc if the fault is a temporary arc fault. This also helps in self system restoration if DG capacity is sufficient to supply the load. The coordination with reclosers without disconnecting the DGs from the network is discussed. This results in increased reliability in the network by reduction of customer outages. iv

7 Table of Contents List of figures List of tables List of appendices List of symbols and abbreviations ix xiii xv xvii Chapter 1: Introduction Background Aims and objectives of the thesis Significance of research The original contributions of the research A novel relay characteristic for DG connected networks A new DG control strategy for fast system restoration Structure of the thesis... 5 Chapter 2: Literature review Introduction Protection issues and solutions Islanding operation and anti-islanding protection Coordination between protective devices Protection in the presence of current limited converters Reclosing, re-synchronization and arc faults Communication based protection Summary v

8 Chapter 3: Protective relay for DG connected networks Introduction ITA relay characteristics ITA relay reach settings Different ITA relay elements Earth elements Phase elements Directional elements Connection of ITA relays to a network Settings of ITA relays to detect resistive faults Zone-1 settings Zone-2 settings Zone-3 settings Practical issues for admittance calculation Summary Chapter 4: Evaluation of ITA relay performance Introduction Inverse time overcurrent relays Distance relays ITA relays ITA relay performance A radial feeder with DGs Effect of source impedance on relay response ITA relay response for different DG and load distribution profiles An application of ITA relays to IEEE 34 node test feeder ITA relays for mesh network protection Limitations of ITA relays Summary vi

9 Chapter 5: Fold back current control and system restoration Introduction Fold back current control characteristics Fold back during contingency Restoration process Coordination with reclosers DG protection Arc fault model selection for simulation Primary arc fault Secondary arc fault Arc extinction Simulation studies Results for permanent faults Results for Arc Faults Auto reclosing Summary Chapter 6: Experimental results Introduction Test feeder arrangement Relay performance evaluation Relay response for different fault locations Fault at BUS Fault at BUS Fault at BUS Fault at BUS Relay response for source impedance change Analysis of ITA relay degradation factors The effect of fault resistance and infeed vii

10 6.5.2 The effect of fundamental extraction Summary Chapter 7: Conclusions and recommendations Conclusions Recommendations for future research Consideration of rotary type DGs for protection Fold back type current control for rotary type DGs The effect of single phase converters References 135 Publications arising from the thesis 143 Appendix-A 145 Appendix-B 147 Appendix-C 153 viii

11 List of Figures Fig. 2.1 Different types of communication networks (Adapted from [55]) Fig. 3.1 A radial distribution feeder Fig. 3.2 The variation of normalised admittance Fig. 3.3 Relay tripping characteristic curve Fig. 3.4 A radial distribution feeder with relays Fig. 3.5 Relay protection zones and relay coordination Fig. 3.6 Relay settings based on different forward and reverse reach Fig. 3.7 Relay connection diagram to the system Fig. 3.8 Process of relay tripping decision making Fig. 3.9 Relay tripping characteristics of different zones Fig. 4.1 A radial distribution feeder with relays Fig. 4.2 Inverse time overcurrent relay grading Fig. 4.3 MHO relay characteristic Fig. 4.4 MHO relay zone settings and timing diagram Fig. 4.5 ITA relay grading Fig. 4.6 Faulted line with a relay Fig. 4.7 ITA relay characteristic in R-X diagram Fig. 4.8 Radial distribution feeder with DGs Fig. 4.9 OC and ITA relay grading Fig OC and ITA relay response when DG1 is connected Fig Distance and ITA relay response when DG1 is connected Fig OC and ITA relay time-current characteristic Fig ITA relay response in grid connected mode Fig ITA relay response in islanded mode Fig ITA relay response for SLG fault in islanded operation Fig System with two parallel transformers Fig Relay response for impedance change Fig Distribution feeder with DGs and loads Fig ITA relay response when fault resistance is 0.05 Ω ix

12 Fig Fault current seen by each ITA relay along the feeder Fig Random load and DG distribution profiles along the feeder Fig ITA relay response for random load and DG distribution profiles.. 66 Fig IEEE 34 node test feeder with ITA relays Fig ITA relay response for SLG fault at node Fig ITA relay response for SLG fault at node Fig ITA relay response for SLG fault at node Fig Mesh network under study Fig Equivalent representation of the faulted network Fig ITA relay response for different values of fault resistances and DG currents Fig. 5.1 Proposed fold back characteristics Fig. 5.2 System restoration Fig. 5.3 Simulated radial feeder with DGs Fig. 5.4 Calculated ITA relay response for a three phase fault Fig. 5.5 DG1 response (a) output voltage (b) output current (c) real power output Fig. 5.6 DG1 response (a) output voltage (b) output current (c) real power output Fig. 5.7 System behaviour for an arc fault (a) arc voltage (b) arc current (c) arc resistance (d) relay response Fig. 5.8 DG1 behaviour for an arc fault (a) output voltage (b) output current 99 Fig. 5.9 DG1 behaviour when downstream relay fails (a) output voltage (b) output current Fig DG1 response during fault and system restoration Fig DG1 terminal voltage and output current Fig. 6.1 Experimental test feeder Fig. 6.2 Single line diagram of experimental setup Fig. 6.3 NI PXI-1042Q chassis Fig. 6.4 ITA relay implementation on LabVIEW Fig. 6.5 Simplified single line diagram of the test feeder x

13 Fig. 6.6 Calculated relay response in different zones for bolted faults Fig. 6.7 The variation of voltage and current for SLG faults at BUS Fig. 6.8 The variation of voltage and current for SLG faults at BUS Fig. 6.9 The variation of voltage and current for SLG faults at BUS Fig The variation of voltage and current for SLG faults at BUS Fig Voltage and current for a fault at BUS Fig Voltage and current for a fault at BUS Fig Voltage and current for a fault at BUS Fig Voltage and current for a fault at BUS Fig Change of parameters during a resistive fault at BUS Fig Test feeder with an infeed Fig Change of parameters for a fault at BUS-2 with fault resistance and infeed Fig A SLG fault at synchronous generator connected feeder Fig Current and voltage during a SLG fault Fig Values of relay parameters during a SLG fault Fig Faulted current and voltage during a SLG fault Fig Values of calculated relay parameters during a SLG fault xi

14

15 List of Tables Table 3.1 Selection criterion of a directional element Table 4.1 System parameters Table 4.2 OC relay settings Table 4.3 Zone characteristics of ITA relay Table 4.4 System parameters Table 4.5 ITA relay forward and reverse reach settings Table 4.6 System parameters Table 4.7 Zone-3 grading of ITA relays Table 4.8 Fault clearing time of ITA relays Table 5.1 Simulated system data Table 5.2 Arc model parameters Table 6.1 System parameters of the experimental setup Table 6.2 Relay reach setting and tripping characteristic in each zone Table 6.3 ITA relay response for faults at BUS Table 6.4 ITA relay response for faults at BUS Table 6.5 ITA relay response for faults at BUS Table 6.6 ITA relay response for faults at BUS Table 6.7 ITA relay response for SLG faults with higher source impedance 118 Table 6.8 Relay parameters during a resistive fault Table 6.9 Change of relay parameters due to fault resistance and infeed xiii

16

17 List of Appendices Appendix-A Positive sequence admittance seen by ITA relay Appendix-B Converter structure and control 147 Appendix-C LabVIEW program xv

18

19 List of principle symbols and abbreviations A, ρ, k Relay tripping constants CB CT DFT DG FFT I DG I p I Ra, I Rb I r ITA l p l s M I OC PCC R 1, R 2, R 3 R f SLG TDS t p VSC Vs Circuit breaker Current transformer Discrete Fourier transform Distributed generator Fast Fourier transform Distributed generator current Pickup current Current in faulted phases A and B Rated current of converter Inverse time admittance Primary arc length Secondary arc length Multiple of pickup current Overcurrent Point of common coupling Protective relays Fault resistance Single line to ground Time dial settings Tripping time Voltage source converter Source voltage xvii

20 VT Y m Y r Y RK1 Y t Z dg Z LG Voltage transformer Measured admittance Normalised admittance Positive sequence measured admittance Total admittance Source impedance of distributed generator Apparent impedance xviii

21 Statement of original authorship The work contained in this thesis has not been previously submitted to meet requirements for an award at this or any other higher education institution. To the best of my knowledge and belief, this thesis contains no material previously published or written by another person except where due reference is made. Signature:. Date:. xix

22

23 Acknowledgements First and foremost, I would like to convey my sincerest and deepest thanks to my supervisors, Prof. Gerard Ledwich and Prof. Arindam Ghosh, for their incomparable guidance and endless encouragement throughout my doctoral research. It has been a great privilege for me to work under this supervision. I wish to express my thanks to the Faculty of Built Environment and Engineering, Queensland University of Technology (QUT) for providing me with financial support during my research candidature. I would also like to thank staff in the research portfolio office in QUT for their generous support and assistance throughout the candidature, and the staff in the School of Engineering Systems for providing such a helpful environment. Further, I am thankful to staff in the Power Engineering Group for their valuable advice. I would like to extend my appreciation to all the technical staff who supported me during the laboratory experiments. Without this support, experimental work would not have been successful. I would further like to thank to all of my friends for sharing valuables ideas, for supporting me during the experimental work, and for making the research period an enjoyable one. Also, I am grateful to my parents for encouraging me to pursue higher studies, and I thank them and my relatives for their constant support. Last but not least, I would like to express my heartiest appreciation to my beloved wife for her encouragement and support during the period of research in Australia. Also, I cannot forget my son who brings joy and happiness to our small nest. xxi

24

25 Chapter 1: Introduction 1.1 Background With the rapid increase in electrical energy demand, power utilities are seeking for more power generation capacity. However, environmental and right-of-way concerns make the addition of central generating stations and the erection of power transmission lines more difficult. Thus, newer technologies based on renewable energy are becoming more acceptable as alternative energy generators. This renewable energy push is starting to spread power generation over distribution networks in the form of distributed generation and will lead to a significant increase in the penetration level of distributed generation in the near future. It is expected that 20% of power generation will be through renewable sources by the year 2020 [1]. However, by that time, the penetration level of DGs is expected to be higher in many countries which are seeking accelerated deployment of renewable technologies. The DGs based on renewable energy sources will help in reducing greenhouse gas emissions. Moreover, these DGs can provide benefits for both utilities and consumers since they can reduce power loss, improve voltage profile and reduce transmission and distribution costs due to their location close to customers [2, 3]. A microgrid can be considered as an entirely DG based grid that contains both generators and loads. It is usually connected to the utility grid through a single point: the point of common coupling (PCC). To the utility grid, the microgrid behaves as a 1

26 Chapter 1: Introduction fully controllable load which at peak hours can even supply power back to the utility grid. A microgrid can operate in either (utility) grid connected mode or islanded mode and can seamlessly change between these modes. In an islanded mode, the DGs connected to the microgrid supply its loads, where a provision for load shedding exists if the load demand is higher than the total DG generation. Most of the existing distribution systems are radial where power flows from substation to the customers in a unidirectional manner. Overcurrent protection is used for such systems because of its simplicity and low cost [1, 4]. However, once a DG or a microgrid is connected within the main utility system, this pure radial nature is lost [2, 5, 6] and the existing protection devices may not respond in the fashion for which they were initially designed [4]. This change in response may be due to the change in parameters, such as source impedance, short circuit capacity level and change of fault currents and fault current directions at various locations. Solar photovoltaic cells produce power at dc voltage. Similarly, fuel cells and batteries also produce dc output power. These are then converted into ac voltage through dc-ac converters. Also, other sources such as wind and microturbines use a converter stage for grid interconnection. All the converters try to protect themselves by limiting their output currents. This becomes more crucial during faults. In general, fault current is usually limited to a value that is twice the converter rated current [7, 8]. As a result of current limiting, the overcurrent devices may not respond or may operate slowly. This is specifically true when an islanded system is supplied by current limited converters. The aim of this research was to identify and address the protection issues of distribution networks in the presence of the DGs and microgrids. New protection strategies are proposed to overcome the difficulties of the existing protection schemes. 2

27 Chapter 1: Introduction 1.2 Aims and objectives of the thesis The main objective of this thesis was to design and develop efficient protection strategies to achieve the fault detection, faulted segment isolation, system restoration and reclosing for both grid connected and islanded operations of a microgrid or a distribution network which mainly consists of current limited DGs. To achieve this goal, the aims of the research project were identified as: analysing the protection issues related to a microgrid and a distribution network in the presence of DGs determining the applicability of the existing protection strategies determining the new protection strategies that are required to achieve appropriate fault detection and protection of a network addressing the protection issues associated with system restoration, arc extinction and reclosing in the presence of converter interfaced DGs in a network While the main objective of the thesis was to propose a generic protection solution for DG connected distribution networks, the focus was limited to converter interfaced DGs. Moreover, the protection of DG connected distribution networks without communication was considered for a simple and cost effective solution. 1.3 Significance of research The penetration level of DGs in the power distribution network is expected to be very high in the near future. In the current climate change scenario, many renewable energy sources such as wind and solar are being connected very rapidly to the utility network. This research will help to identify the protection problems related to a distribution network or microgrid which consists of distributed generators and 3

28 Chapter 1: Introduction minimize the protection issues in implementation with the use of the proposed strategies. 1.4 The original contributions of the research The main objective of this research was to propose protection strategies to incorporate DGs into a micro grid or a distribution network by overcoming the identified protection issues. The main contributions of this research can be listed as follows A novel relay characteristic for DG connected networks An inverse time admittance (ITA) relay characteristic is proposed to overcome the deficiencies of the existing overcurrent relays. The ITA relay has the capability of detecting faults under different fault current levels which is the usual scenario that can be seen in a distribution network when several DGs are present. These relays can isolate the faulted segments and allow the unfaulted segments to operate either in grid connected or islanded mode. Moreover, the relay is capable of providing adequate protection for the islanded system which has several converter interfaced DGs A new DG control strategy for fast system restoration The arc extinction during an arc fault, reclosing for temporary faults and the system restoration after a fault is cleared are major protection issues in a DG connected distribution network. Therefore, to overcome these problems, a control strategy based on fold back current control is proposed for converter interfaced DGs. The proposed control has the capability to restore the system automatically if the generation is sufficient to supply the load demand in an islanded section. Self 4

29 Chapter 1: Introduction extinction of arc is achieved by reducing the output current of DGs. Furthermore, an effective method is proposed to coordinate the operations of reclosers and converter interfaced DGs in a network. The fold back control provides maximum benefits to customers by reducing outages since the DGs are not disconnected immediately when there is a fault in the system. The proposed ITA relay and fold back current control strategy for a converter interfaced DG provide a complete protection solution for a DG connected network. The relays detect and isolate faults effectively while the fold back current control helps in arc extinction, system restoration and recloser coordination with DGs. 1.5 Structure of the thesis This thesis is organised in seven chapters and three appendices. The research aims and objectives are outlined in Chapter 1. The need and justification for the research in this field are identified in Chapter 2. In this chapter, a literature review is carried out to identify the protection issues related to DG connected distribution networks and microgrids. Moreover, the deficiencies of the existing protection schemes are identified and some of the already proposed solutions to overcome these protection issues are analysed. As a result of identification of the protection issues and the deficiencies of the existing protection schemes in Chapter 2, a new ITA relay is proposed for DG connected networks in Chapter 3. The ITA relay characteristics and its features, which include relay reach settings and different relay elements, are discussed in this chapter. Moreover, practical implementation issues of ITA relays are also discussed. The proposed ITA relay performance is then evaluated in Chapter 4. The fundamentals of the existing overcurrent and distance relays are discussed and their 5

30 Chapter 1: Introduction features are then compared with the ITA relays. Different case studies are carried out to show the efficacy of the ITA relays. Moreover, simulation studies related to the ITA relays are also presented in this chapter. Applications of ITA relays for both radial and mesh networks are examined and their limitations are identified. A fold back current control characteristic for a converter interfaced DG is proposed in Chapter 5. The protection issues related to the system restoration, arc extinction and reclosing are also addressed in this chapter. Different case studies of both permanent and temporary faults were carried out and are presented here to show the efficacy of proposed fold back converter control. Chapter 6 presents the hardware results obtained through the experimental laboratory tests. The ITA relay characteristic is modelled using LabVIEW software and the relay performance is investigated for different fault locations and different system configurations. Conclusions drawn from this research and recommendations for future research are given in Chapter 7. The list of references and a list of publications arsing from the thesis are provided at the end of the last chapter. In Appendix-A, different types of relay elements are discussed, while Appendix-B give a detailed description of the converter structure and control used in simulation studies. The LabVIEW program used in ITA relay implementation is presented in Appendix-C. 6

31 Chapter 2: Literature review 2.1 Introduction The cost of transmission and distribution is rising with the rapid increases in the load demand. However, the costs of distribution generation technologies are falling [2]. So from a costing point of view, it is becoming more worthwhile to increase the generation at the distribution level by connecting a distributed generator (DG) to meet the load requirement without expanding the transmission and distribution infrastructure. In addition, there are several advantages of having DGs; short construction time, lower capital costs, reduction in gaseous emissions, reduced transmission power loss since generation is now closer to the load, improving voltage profile, enhancing reliability and diversification of energy sources [9-11]. A microgrid can be considered as a small grid based on DGs. Generally, the microgrid consists of renewable energy based DGs and combined heat and power plants. It can operate either grid connected or islanded mode. Most of the DGs are connected to the microgrid through power electronic based power converters which pose operational challenges [12]. The protection system of a microgrid should respond to faults within the microgrid irrespective of its grid connected and islanded operation. For a fault in the utility grid, the microgrid should disconnect immediately from PCC to maintain a continuous supply to the microgrid loads. On the other hand, 7

32 Chapter 2: Literature review the smallest possible set of faulted lines of the microgrid must be isolated for a fault within this grid. However, protection of a distribution network becomes more complicated and challenging once several DGs are connected (as in a microgrid). In this chapter, the complications in system protection arising due to the connection of DGs to a distribution network are discussed. Also some of the already proposed solutions are mentioned. 2.2 Protection issues and solutions The present practice is to disconnect the DGs from the network using an islanding detection method when there is a fault in the system [13, 14]. This is as per the IEEE recommended practice, standard 1547 [15]. This may work satisfactorily when the penetration of DGs in a distribution system is low. However, as the penetration levels increase or in the case of micro or mini-grid, the DGs will be expected to supply power even when the supply from the utility is lost and the DGs form a small island. This will prevent unnecessary customer power interruption. Thus, the benefits of DG installations can be maximized allowing the DGs to operate in both grid connected and islanded modes of operation, especially when the DG penetration level is high. Some of the issues in DG connected distribution networks or microgrids that need attention are bi-directional power flow, change of relay reach, coordination between protective devices, islanding, reclosing, protection in the presence of current limited converters and temporary arc faults. These are discussed below. 8

33 Chapter 2: Literature review Islanding operation and anti-islanding protection Islanding occurs when the main supply is disconnected and at least one generator in the disconnected system continues to operate. If a DG is allowed to operate in this islanding condition, it will bring benefits to customers by reducing outages [16]. However, if DGs are not designed to operate in islanded operation, this can cause a number of safety issues [17]. The point where the islanded system is created after the disconnection of the utility for a fault cannot be identified exactly. Therefore at the moment of islanding, the generation and load capacity may not be equal. When synchronous generators are present in the islanded region and if loads are larger than the generation then the generators tend to slow down which can lead to under frequency tripping of generators. In this case, a load shedding scheme should be implemented to maintain the stability in the islanded system. On the other hand if load capacity is less than the generation, generators could experience over frequency tripping and require a fast governor controller to respond and balance the power [18]. Thus there is a need to identify the islanding condition in an expanded islanded system which has the loads beyond the PCC. The type of prime mover and controller mode (i.e. droop control, constant power, etc) affect the response of the system at the event of the islanding. These responses have been described according to the type of generation in [18]. Also islanding may increase the risk for the user equipment and utility power apparatus due to the potential reduction in performance standards for voltage and frequency and the issues relating to phase mismatching when reconnecting the DG and utility [1]. It also can be a potential hazard to utility personnel working to rectify the faulted segment as some portion of it can be live due to power supplied by DGs. 9

34 Chapter 2: Literature review Power quality may not be guaranteed within the island and there could be abnormal conditions in voltage and frequency [19, 20]. In the islanded mode, short circuit levels may drop significantly upon disconnection from the utility [1, 4, 19]. These factors are the reason why anti-islanding protection is traditionally applied to achieve the safety of personnel and equipment of the distribution system. Under and over voltage relays, under and over frequency relays, vector shift and relays for detecting rate of change of frequency (ROCOF) can be used as devices to detect islanding [10, 19, 21]. The common practice is to disconnect the DGs before the first reclosing occurs after a fault in the system. Therefore anti-islanding protection devices should be appropriately coordinated with other protective devices such as reclosers in the system. From the reliability point of view, applying the anti-islanding protection to a microgrid is disadvantageous. An anti-islanding protection relay should detect the islanding condition within the required time (typically 200 to 400 ms) and should trip all the generators. On the other hand, it should not trip for small frequency variations in the system. A microprocessor based line tracking system is suggested for detecting islanding condition of a hydro power distributed generator (HPDG) using the changes of voltage, frequency, active power and reactive power [10]. This method can be used to detect the islanding condition of HPDG quickly and to isolate it from the main grid. ROCOF relay needs very sensitive settings for the fast islanding detection under a small imbalance of active power. However, it may cause to trip the anti-islanding relay for the small frequency variations. Usually frequency tripping requirements (i.e. under and over frequency) of a relay and islanding detection relay settings are analysed separately. As a result, two relays are required to perform the task; one for the under/over frequency protection of the generator and another for the islanding 10

35 Chapter 2: Literature review detection. However, these two relays are operated based on the system frequency. Reference [21] has proposed a graphical method based on application region of the frequency relay to determine the islanding requirements without disturbing the frequency tripping requirements. Further this paper outlines how to coordinate the operation of the islanding detection relay and standard frequency tripping relay. Reference [20] also provides a mathematical development to determine the application region of a frequency relay which satisfies both the islanding detection and frequency tripping requirements. It has been shown that the frequency relay can be replaced by an islanding detection vector shift relay if the proper settings are selected. Similarly, a method is suggested to find out the application region of a voltage relay to satisfy both the anti-islanding and voltage variation protection in [22]. After disconnecting the main utility, the loading effect on DG is suddenly changed. As a result, balance condition of loads and harmonic currents will change. Therefore Total Harmonic Distortion (THD) of current and voltage unbalance at the DG terminal have been introduced as two new monitoring parameters to detect the islanding condition with voltage magnitude in [23]. Test results have shown that this method can be used efficiently for improved performance. The DGs are expected to supply either an increase of load at grid connected operation or emergency loads at the islanding operation. Thus the islanding operation is important to ensure supply continuity to customers. Therefore, implementing an anti islanding operation every time a fault occurs reduces the reliability of the system. The authors in [6] proposed a method for a distribution network with high penetration of DGs to use the conventional protective devices without disconnecting the DGs from the system when a fault occurs. In this case, each DG should be connected to two feeders which operate in a loop. The DG is required to be isolated 11

36 Chapter 2: Literature review from the faulted feeder after the fault occurs and a micro-processor based line protection relay is used to implement the scheme. However this scheme may increase the fault clearing time which can affect the dynamic condition of the system. Voltage and frequency should be maintained in the desired range, in the presence of disturbances in the islanding system. Control strategies should be implemented considering over-generated and under-generated islanding conditions [10]. It has been mentioned that the only way to maintain the existing coordination system in the presence of arbitrary DG penetration level is to disconnect all DGs instantly in the case of a fault [2]. It would result in the DG disconnection for a temporary fault as well. Therefore it is clear that new protection strategies are required to investigate with the DG penetration to the utility. In addition, if the DG is not disconnected from the system at the event of a fault, the fault arc would not extinguish during an automatic recloser open time, since the source feeding the fault still remains. Thus a compromise solution between islanding operation and antiislanding protection needs to evolve Coordination between protective devices The coordination of protective devices based on current is relatively easy when the distribution network is radial. However, with the connection of microgrids or DGs to the utility, the radial nature no longer exists and it permits the power flow to be bi-directional rather than uni-directional [14, 24]. This may create a number of protection coordination issues. On the other hand, the protective devices should be coordinated in the distribution network considering reliability (correct operation), selectivity (minimum system disconnection), speed of operation (minimum fault duration), simplicity (having minimum protective equipment) and economics (maximum protection under minimum cost). These coordinated actions should be 12

37 Chapter 2: Literature review implemented fast enough to prevent personal hazards and equipment damage [25]. Generally, the protection of the distribution network is done using the current measurement based on the coordination of fuses, overcurrent relays, reclosers and sectionalisers [26]. It should consist of a primary and backup protection system which has proper time grading between each devices. As an example, tripping time increases towards the main utility source from the fault location and operation device sequences for a fault in a DG may be the first low voltage breaker, then the fuse, after that the line recloser, finally if fault still exits it should be cleared by the substation circuit breaker. The coordination based on the current is relatively easy in the unidirectional power flow networks, because the fault current reduces along the feeder [26]. However, with the growth of distributed generators, the system permits the power flow to be bi-directional rather than uni-directional [5, 27]. This may create a number of feeder protection issues. It causes relays to under-reach or over-reach [28]. The DG location in the distribution network influences the relay reach to reduce or increase. It has been shown that the reach of an overcurrent relay will reduce in the presence of a DG [29]. Among the protective devices currently used, reclosers and fuses usually do not have the directional sensing feature but a relay can easily be made to have that feature [2]. In addition to that, the DG can contribute by suppling short circuit currents to the neighbouring faulted feeder and operating the protective device in the healthy feeder [30]. The only possible way to coordinate the existing protection schemes is to disconnect all the DGs for every fault even for the temporary faults [2]. However, it has been mentioned before that this is not a desirable solution to this problem. 13

38 Chapter 2: Literature review An adaptive protection method is proposed for the distribution system with high DG penetration level in [2]. In this approach, several zones are formed with a reasonable balance of loads and DGs. Each breaker and recloser should have communication capability and each individual zone breaker should be available to check the synchronization function. At the beginning, load flow and short circuit analysis for all types of faults need to be carried out. After the changes of system configuration due to the loads or DGs, the load flow and short circuit analysis again have to be repeated. This will not be feasible when a larger number of plug and play DGs is connected /disconnected. Moreover, this adaptive method is complex as it is not easy to define zones with the fluctuation of loads and DG generation. However, protection is independent of DG size and location. The impact of DG capacity on relay operation and coordination in a radial distribution system has been studied in [31]. It has been shown that for a downstream fault from the connection point of a DG, the relay selectivity remains unchanged and sensitivity improves due to the increase in fault current. But there is a maximum capacity for the DG to keep the relay coordination. Further a method was suggested to find out the maximum value for the DG capacity. On the other hand for an upstream fault from the DG connection point, it has been shown that the misoperation can occur for a low capacity DG. Problems of protective devices coordination in a distribution network have been addressed in [5]. To achieve the coordination among fuses, total clearing (TC) time of a fuse should be less than the minimum melting (MM) time of the other fuse for a particular value of fault current. After a DG penetrates into the system, the fault current magnitude and direction can change and this initiates the problems in the coordination. Usually a recloser has a sequence of operations as it employs two operating characteristics curves called fast and slow. Most of the faults, which are 14

39 Chapter 2: Literature review around 80% of the total faults in the distribution system, are temporary. Therefore protective devices coordination should be done in an appropriate way when recloser and fuse are present in a distribution system. Moreover the recloser should operate fast enough to give a chance to clear the fault before the fuse [2, 5]. To achieve this fast characteristic, the recloser should lie below the MM curve of the fuse. The fuse should only operate for a permanent fault. This operation is obtained if the slow characteristic of recloser lies above the TC curve of the fuse within the considered minimum and maximum fault currents region. If DG is connected upstream to a recloser, the fault current seen by the recloser and further downstream fuses will increase. As a result the required margin between the fast characteristic of recloser and minimum melting curve will tend to reduce. Thus there is a probability of losing the coordination with any fuse further down to the recloser [32]. On the other hand, if a DG is connected between a recloser and a fuse, the fault current seen by a fuse increases and this may cause it to lose coordination. Before the DG connection, the recloser and fuse see the same fault current. However, after the connection, the fuse will see more current than the recloser and it responds before the recloser in the event of a fault downstream to the fuse location. The effect on coordination increases with DG capacity. Studies in [33] have shown that traditional reclosers are unable to keep the coordination with fuses in the presence of high DG penetration. Further this paper has proposed a microprocessor based recloser to perform the task under this system condition. When relays are present in the distribution system, time of operation of each relay and among relays is called Coordination time interval for the faults should be coordinated appropriately [5]. Overcurrent relays are the simplest and widely used in protection applications. They are used in the distribution system as the primary 15

40 Chapter 2: Literature review protection and in transmission as backup protection [34]. There are several types of overcurrent relays available to select from depending on the application. Instantaneous overcurrent relays are mostly used to protect sub-transmission lines while definite time relays are used in ungrounded or high impedance grounded systems. Moreover inverse time relays can easily coordinate with other protective devices and they are usually employed to protect distribution networks. A software model of a inverse time overcurrent relay has been developed to simulate in PSCAD [34]. High backup time for the minimum fault currents is a disadvantage of overcurrent relays. A method which proposes to find the time element function for an overcurrent relay to reduce the back-up time to a constant value independent of the fault current magnitude rather than in the conventional overcurrent relay is given in [35]. References [36] and [37] present the IEEE standard analytical equation for the different types of overcurrent relays (i.e. moderately inverse, very inverse, and extremely inverse) and operating and reset characteristics that can be taken for coordination purposes. Relays employed in the radial networks have both inverse time and instantaneous elements to achieve a quick response for the severe faults as well as the coordination among relays [26]. Also in the case of the islanded microgrid, the ratio between the source impedance and protected line is relatively high compared to the utility and this initiates a coordination problem since discrimination among relays are difficult at this time [26]. When the network with a large number of lines is fed by a single source station, the ground overcurrent relays have to be set with a large time delay period such as five seconds to maintain a good coordination in the system. As a result existing protection systems are upgraded with digital ground impedance 16

41 Chapter 2: Literature review elements to achieve high speed fault clearing [38]. Reference [39] shows a method to calculate directional overcurrent relays setting for both grid connected and microgrid which consists of synchronous generator based DGs. In this method, the Particle Swarm Optimization algorithm is used in the relay coordination problem to obtain the optimal settings for the directional overcurrent relays while maintaining the minimum operating time and coordination among relays. It has been shown that it is not possible to calculate a setting time for the relays in both the grid connected and islanded modes of operation. Hence a central control protection unit is required to change the setting according to the system configuration. However fault current seen by each device may change according to the location of microgrid connected to the utility and fault location. Hence attention to coordinate protective devices is essential. There are numerous papers which address the coordination issues with the presence of DGs in the distribution network. However, so far there is little attention to the coordination analysis of the current limited converter interfaced DGs Protection in the presence of current limited converters The fault current may change due to the presence of DGs in the network [2, 16, 19, 39, 40]. Its impact depends on the size, type, number of the DG, location of the DG [5, 31]. Basically three types of DGs exist with different properties; synchronous generators, induction generators and converter interface DGs [4]. Transient behaviour and short circuit current levels vary based on the type of the generation [41]. Microgrids which consist of synchronous generators tend to contribute an additional fault level in the system [31]. However generators can increase or decrease the fault current seen by protective devices depending on the location. The impact of synchronous DGs on coordination between voltage sag and overcurrent protection 17

42 Chapter 2: Literature review are studied in [40] considering the sensitive equipment response. Fault current behaviour and fault detection in a distribution network for different types of faults in the presence of an induction generator has been studied in [4]. The system which is not designed with DGs may not work properly with existing protective devices once several DGs are connected to the system [6]. In the presence of a generator within the network, the fault current detected by a protective device located at the beginning of the feeder can be reduced due to the rise of voltage drop over the feeder section between the generator and the fault [4]. Therefore the faults previously cleared in a very short time may now require a significant time to clear. Most of the distribution resources in the microgrid are connected through the power electronic converters [12]. For example, the dc power is generated by using the sources such as fuel cell, micro turbine, or a photovoltaic and converters are utilised to alter the dc power into ac power. These converter interface generators supplies the currents not much greater than the nominal load currents [26]. Basically the controller of the converter mainly consists of two control schemes named voltage control and current control and it regulates the output active and reactive power [42]. In the voltage control mode, the converter produces a three phase balanced ac voltage at the terminal. The current control scheme, which is explained in [42], uses two control loops, an inner loop for the current output and outer loop for the power output. It has been shown that the current control scheme responds slowly due to the outer power loop. However, converters do not supply sufficient current to operate current sensing devices in a fault condition because they have been designed to limit the fault current at a value that is not more than twice the rated current. As a result, overcurrent devices may not respond or take a long time to respond [8, 26, 43, 44]. Therefore 18

43 Chapter 2: Literature review protecting a converter dominated microgrid is a challenging technical issue under the current limited environment [25]. Moreover there is a requirement to find other protection techniques to solve this problem [7, 26, 27]. One possible approach which facilitates using the existing overcurrent protection is up-rating of converters to supply the required fault current. However this will be a costly process. Another approach that is proposed to overcome this problem is to use a flywheel energy storage system to obtain the necessary fault current in the event of a fault [44]. The flywheel supplies the required fault current to operate the overcurrent protective devices in the islanding operation. A stand-alone three phase four leg voltage source converter model has been studied to observe the fault behaviour of an islanded microgrid for different types of faults in [7]. During a fault, the converter works as a constant current source supplying the positive sequence current to the system. There are no active sources in the negative or zero sequence networks. So it has been shown that the microgrid is equivalent to a current source with parallel impedance which depends on the fault type. In this converter topology, large voltages can be seen in healthy phases for unbalanced faults. In [25], fault behaviour in a converter supplied microgrid has been presented considering different types of converter topologies and microgrid earthing systems. The paper concludes that the fault response strongly depends on the converter control strategy. Reference [26] suggests a voltage based fault detection scheme as a solution for the converter dominated microgrid which operates on low fault current levels. It describes three types of methods to detect the voltage in the faulted phase and compares the detection time of the methods. The paper concludes that the neutral grounding configuration affects fault detection. In addition, the paper further 19

44 Chapter 2: Literature review proposes an adaptive overcurrent scheme which selects the lower current threshold to operate the overcurrent device based on the value of voltage detection. In reference [45], abc-dq transformation of the voltage waveforms is used to identify if the short circuit condition is inside or outside a set zone in a microgrid. Voltage disturbance at each relay location is calculated by comparing the reference value with the obtained dc values in the d-q synchronous rotating frame. The tripping decision is made by selecting the location which has the highest mean average disturbance value with the help of a communication link among relays. A differential relay based protection scheme is proposed to protect a microgrid in either grid connected or islanded mode in [16]. In this, a central control unit is used to make decisions on control and protection devices. Line parameters of the two ends of a protected line have to be monitored by means of a wire connection if the line is short or by a pilot wire communication if the line is long. The need for communication channel is a disadvantage of the differential protection scheme. Moreover, the response of DGs places between two relays will affect their performance. Another approach for the protection of microgrid with converters in both islanded and grid connected operation is presented in [8]. A static switch has been designed to open the microgrid for all types of faults and faults should be cleared using techniques which do not rely on high fault currents within the microgrid. In [29], simulation results show that a converter based DG has a considerable effect on the detection of the fault current as seen by an overcurrent relay. The relay reach will be reduced with the DG connection due to the reduction in the fault current. An adaptive technique is proposed to set the pickup current of the overcurrent relay based on the amount of DG power injected to the system. The 20

45 Chapter 2: Literature review minimum pickup current of the relay is update depending on the fault type and location. Further, the response of an converter interfaced fuel cell under a fault conditions has been investigated and it has been shown that the fault will cause the voltage to drop below to a value such that the undervoltage relay would operate to trip the DG if the fault occurs near the DG. Therefore undervoltage relay can be used under a fault condition to determine the status of the DG. Furthermore, IEEE standard states that converters will sense a short circuit by voltage drop rather than sensing the short circuit current. Another option is to design the protective devices to operate for small fault currents. However, this may cause nuisance tripping [16, 19, 46]. Thus there is a need to assure that for both the microgrid itself and for the grid connected modes, the protection system is operating in an adequate fast, selective and reliable way to clear the faults [39] Reclosing, re-synchronization and arc faults Most of the faults (around 90%) in the power system are temporary arc faults (such as insulator failures, conductors clashing due to strong wind, animal contacts, lightning strikes, etc). These faults can be successfully cleared by de-energizing the line long enough such that the arc self extinguishes. Usually reclosers which open and close a few times successively are used to clear such faults without any large scale power interruption [47]. Maximum dead time of single phase reclosing in transmission lines are decided by the system stability requirements, where time exceeding s is not permissible [48]. While the consumers experience a shorter outage time due to automatic reclosing, these breakers cannot be used for permanent faults [47]. Reclosing a circuit in which the fault still exists can be harmful to the system components such as generators, transformers, etc. A proposal has been made 21

46 Chapter 2: Literature review based on artificial neural network algorithms to solve this problem by analysing the voltage of the open phase conductor during the recloser dead time interval [47]. Usually three phase reclosers are used in distribution networks. In a DG or microgrid connected distribution network, the reclosing should be done with proper synchronization since this will join two live systems. The maximum time available for automatic reclosing without losing synchronism should be considered. During the auto recloser open time, if the island and main grid undergo a phase mismatch, then it may lead to damage to the equipment and DGs in the microgrid [5]. However, if the DG is connected using a converter, the risk of damage to the DG is low as it has its own protection [49]. Dead line voltage relay and sync-check relay can be used to prevent out of phase reclosing [19]. In general, a DG is disconnected before the first reclosing occurs in the system. This requires that any anti-islanding protection should operate very quickly. As a result, the recloser should coordinate with the antiislanding protection, which is a challenging task [19]. A communication link can be established between the line recloser and the DG to transfer trip signal to disconnect the DG quickly [50]. An automatic synchronizing or synchronism check relay should be used at the PCC breaker when restoring the system after disconnection [18]. Resynchronizing can be done manually or automatically using synchronism check relay with a synchronous generator based DG. However for a converter interfaced DG, automatic re-synchronizing is preferred [46]. In the case of arc faults, sufficient time should be given to de-ionize the gas path during the recloser open condition. Otherwise the arc may reignite again and fault will not be clear [49]. Also, if DGs are kept connected to the system during recloser open time, they can sustain the arc. The arc self-extinction action depends not only on the fault current magnitude, but also on the transient recovery voltage 22

47 Chapter 2: Literature review rate after successful arc extinction at the current zero crossing [51]. Also the arc extinction time is proportional to the arc time constant [52]. On the other hand, the fault current magnitude of an arc fault is limited by the arc resistance. Sometimes it results in difficulties of detecting the fault [53]. Moreover, the arc voltage at the fault point is a source of errors in the fault locating process [54]. Therefore protection of distribution network and restoration under arc fault is a challenging task Communication based protection The distribution system protection will be complicated when the DGs are spread throughout the network. As a result new protection issues will arise for the traditional distribution networks. To address some of the issues, a protection based on a communication medium has been developed. Communication media including power line carrier (PLC), microwave and optical fibre have long been used for the transmission line applications. However, in nature, the distribution lines are different from transmission lines. These lines are shorter and they have numerous tapped loads. Therefore a particular communication method for a distribution system protection should be fast and reliable. Basically three types of communication networks can be identified as shown in the Fig. 2.1 [55]. In centralised networks, all nodes are connected to a central point, which is the acting agent for all communications. A network distributed across many nodes rather than centralized around a central point is known as a decentralised network. On the other hand, in a distributed network, if nodes are located on scattered way, they may still be capable of working either independently or jointly as required. The increase of implementation of renewable energy sources to the distribution system has changed the configuration from centralised to decentralised network. 23

48 Chapter 2: Literature review Fig. 2.1 Different types of communication networks (Adapted from [55]) The installation of a larger number of DGs can cause the loss of protection selectivity. Communication media may be the internet, PLC, wireless communication, etc. In [56], PLC based methods are proposed for the coordination of voltage control, islanding detection for a DG and controlling the interface devices at the PCC. The Islanding detection method is introduced to minimize the problems of traditional methods based on frequency and voltage measurements. High attenuation levels can be expected in distribution lines when their structure is complex and lines are long. To avoid such problems, repeaters need to be installed in this implementation. Application considerations of internet as the real time communication medium for providing the loss of mains protection of a DG has been studied in [55]. The distribution system becomes a multi-source when one after another DG gets connected at different locations. This change in system configuration will cause false tripping and relay coordination problems. As a solution for these problems, reference [57] has proposed a new current protection scheme based on communication to a multi-source distribution system. Wide area measurement is used to decide the appropriate protection actions to locate the fault with the use of communication channel. An adaptive method is proposed in [58] to set the relay settings in real time using wide area measurements based on communication. A multi 24

49 Chapter 2: Literature review agent approach based on communication is proposed in [17] to provide protection of the power system and coordination between the protective devices in the presence of DGs. A new method is proposed in [13] based on analysing the sign of wavelet coefficients of the fault current transient to locate and isolate a faulted segment. In this, relay agents are proposed to implant the proposed protection scheme. A fault location and fault isolation technique of a DG connected distribution network using neural networks is presented in [59]. In this, the system has different zones and the relay at substation communicates with zone breakers to take appropriate actions. With the use of communication, relay coordination has the ability to rapidly select the faulted region. However, installation of extensive communication will require time. Once the power system is smart grid ready, various smart relays can be installed. Till that time, protection without any or low levels of communication will be the most cost effective solution. 2.3 Summary In this chapter, a brief summary is presented based on the review of the previous published research work on the protection issues which arise after the connections of DGs and microgrids to distribution networks. There are several benefits available for both the network operator and customer by utilising DGs or DG based Microgrids. Reliability can increase if the islanded system can continue the supply to the loads rather than disconnecting all the DGs by anti-islanding protection schemes. Therefore within the islanded system, a protection scheme should work satisfactorily. Different types of protection issues have been addressed in the literatures and different solutions have also been proposed to overcome these issues. 25

50 Chapter 2: Literature review The proposed protection scheme should isolate the faulted segment as quickly as possible from the network. The DGs can then supply the power to unfaulted segments in the network if they have been designed to operate in islanded mode. To achieve that solution, several protection solutions have been proposed based on communication for DG connected networks. However, most of them need reliable communication medium for fast operation. Most of the time, current sensing protective devices have been used to detect the faults in the network. However, with the high penetration level of converter based DGs, protection of the system has been identified as a key challenging issue. Although different solutions have been proposed to solve this problem, further studies are still required to identify and improve the efficient fault detection methods. In the near future, when more DGs come into operation, protection will be a challenging task due to the network complexity. 26

51 Chapter 3: Protective relay for DG connected networks 3.1 Introduction In a high penetrative DG network, a small possible portion should be isolated during a fault allowing unfaulted segments to operate in either grid connected or islanded mode to increase the system reliability by maximizing the DG benefits. To achieve the faulted segment isolation, both upstream and downstream protective devices should detect and isolate the fault. However, with the connection of DGs to a distribution network or within a microgrid, fault current level can vary depending on the DG connectivity, DG type and DG location. It results in difficulty of coordinating existing overcurrent protective devices since network configuration changes. Moreover, settings of these overcurrent relays to incorporate DGs are not possible if DG power output changes with time or their connectivity is not consistent. Furthermore, protection will be a challenging task when using converter interfaced DGs because of the output current limiting during a fault in the network. As a result of current limiting and intermittent nature of DGs, the fault isolation from downstream side will be very difficult using the existing overcurrent relay which normally operates depending on the fault current levels. Therefore new protection schemes, which are not dependent on the fault current level of the network, are required to accomplish the protection challenges in the DG context. In this chapter, a 27

52 Chapter 3: Protective relay for DG connected networks novel Inverse Time Admittance (ITA) protective relay is proposed based on the measured admittance of the protected line to avoid deficiencies of existing protection schemes. The fundamentals of ITA relays are explained in this chapter. 3.2 ITA relay characteristics A radial distribution feeder as shown in Fig. 3.1 is considered to explain the ITA relay characteristics. It is assumed that the relay is located at node R and node K is an arbitrary point on the feeder. The total admittance of the protected line segment is denoted by Y t while the measured admittance between the nodes R and K is denoted by Y m. Then the normalised admittance (Y r ) can be defined in terms of Y t and Y m as Y Y m r = (3.1) Y t Fig. 3.1 A radial distribution feeder The variation of normalised admittance along a radial feeder is shown in Fig. 3.2 by assuming the feeder has a length of 3000m while the total feeder impedance is ( j ) Ω. It can be seen that normalised admittance decreases when measured point moves away from the relay location. 28

53 Chapter 3: Protective relay for DG connected networks Fig. 3.2 The variation of normalised admittance The change of normalised admittance along the feeder is used to obtain an inverse time tripping characteristic for the relay. The general form for the inverse time characteristic of the relay can be expressed as t A = ρ Y 1 p + r k (3.2) where A, ρ and k are constants, while the tripping time is denoted by t p. The values for these constants can be selected based on the relay location in a feeder and the protection requirements. The shape of the proposed inverse time tripping characteristic can be changed by varying the constants to obtain the required fault clearing time. When a network consists of different types of protective devices, these constants can be selected appropriately for coordination purpose. For example, the coordination between the relay and a fuse can be considered. In this case, these constants should be selected properly according to the tripping characteristic of the fuse. The relay tripping characteristic for A = , ρ = 0.08 and k = 0 is shown in Fig The magnitude of the normalized admittance (i.e. Y r ) becomes higher as the fault point moves towards the relay location. As a result, the relay gives a lower 29

54 Chapter 3: Protective relay for DG connected networks tripping time for a fault near to the relay. On the other hand, higher fault clearing time can be obtained when the fault is further away from the relay location. Fig. 3.3 Relay tripping characteristic curve It is to be noted that the normalized admittance in (3.2) should be greater than 1.0 for relay tripping. This implies that the measured admittance is greater than the total admittance as shown in (3.3). This constraint is used by the relay algorithm to detect a faulted condition in the network. Moreover, the relay algorithm checks this constraint continuously during the faulted condition until relay issues the trip command to avoid any unnecessary tripping due to the effect of transients. The tripping time is decided depending on the calculated value of measured admittance. Y Y m r > 1 > 1 Ym > Yt (3.3) Y t 3.3 ITA relay reach settings The ITA relay reach settings can be implemented by choosing a suitable value for the Y t. This is totally dependent on the protection requirements such as primary and backup protections. For a particular relay, different values of Y t can be assigned 30

55 Chapter 3: Protective relay for DG connected networks to generate a number of required zones of protection. In each zone, the relay has a unique tripping characteristic. It checks whether the measured admittance is greater than the total admittance of that particular zone before starting the relay tripping time calculation. A large coverage and minimum tripping time can be achieved by increasing the number of zones. It also leads to a good coordination amongst the relays in a feeder. Any upstream relay always provides the back up protection for the immediate downstream relay in the feeder. The radial feeder shown in Fig. 3.4 is considered to explain the relay reach settings. The relays are located at BUS-1, BUS-2 and BUS-3. It is assumed that each relay has two zones of protection. Zone-1 of each relay is selected to cover the whole line segment between two adjacent relays, while Zone-2 is selected to cover twice the length of the first line segment. The reach setting is set based on the positive sequence admittance of the considered line segments. Zone-1 and Zone-2 tripping characteristics are the same for all the relays. For example, relay tripping characteristic curves for two adjacent relays R 1 and R 2 are illustrated in Fig The locations of relays R 1 and R 2 and the tripping time of these relays against the distance to the fault from the relay locations are shown in the figure. Each zone has different values for the constants in (3.2) resulting different relay tripping characteristic curves. It can be seen from Fig. 3.5, Zone-2 of R 1 will provide a backup for the relay R 2. Another zone can be assigned with a different tripping characteristic, if required. Fig. 3.4 A radial distribution feeder with relays 31

56 Chapter 3: Protective relay for DG connected networks Fig. 3.5 Relay protection zones and relay coordination The proposed new relay does have the ability to isolate the faults occurring at either side of the relay in a radial feeder. This is because the absolute value is taken into consideration in admittance normalizing process. However, for the relay to operate for reverse faults there must be an infeed that is located downstream from the relay. If the distribution network consists of these relays located at equal distances, the same forward and reverse reach can be used to isolate forward and reverse faults. The value for the reach of a particular zone should be selected according to the requirement. However, the reach setting should be different for forward and reverse faults, when the relays are not placed equidistant from one another. In this case, each relay has the capability such that the forward and reverse reach settings can be set appropriately. For example, the reach setting of Zone-1 of the relay R 3 in Fig. 3.4 is considered. It is assumed that the lengths of the line segments 2-3 and 3-4 are not equal. The forward reach of R 3 is selected as 120% of line 3-4, while the reverse reach of R 3 is chosen as 100% of line 2-3 and 20 % of line 1-2. In this case, both the reverse and the forward reach should have different values since line length between 32

57 Chapter 3: Protective relay for DG connected networks relays are not equal. To accomplish forward and reverse reach in relays, the relay should sense the fault direction. Moreover, the relay has the capability to identify whether the fault is in the forward or reverse direction. Any method which will determine the fault direction can be used for this purpose. One possibility is to measure the relative difference of angle between the current and bus voltage. The fault current lags the bus voltage for a forward fault while for a reverse fault the fault current leads the bus voltage. In [60], relative phase angle between fault current and pre-fault voltage is used to determine the fault direction. Another possibility is to calculate the negative sequence impedance seen by the relay. Based on the calculated value, the relay identifies the fault direction to select the appropriate reach setting. This approach is only valid if the fault is unsymmetrical since negative sequence will not be present for symmetrical faults. The negative sequence impedance is always positive for the reverse faults and it is negative for the forward faults [61]. The positive sequence directional element proposed in [62] can be also used to identify the fault direction. After identifying the fault direction, the process of tripping time calculation can be implemented as shown in Fig Fig. 3.6 Relay settings based on different forward and reverse reach 33

58 Chapter 3: Protective relay for DG connected networks 3.4 Different ITA relay elements The ITA relay has different types of protection elements to detect different faults. All elements are designed to operate based on measured admittance of the protected line. These elements are explained below Earth elements These elements will respond for the line to ground faults. The number of elements varies depending on whether protection has been configured as directional or non-directional. If protection is directional, then there are two independent earth elements per phase. The positive sequence measured admittance; Y RK1 seen by this relay element is given by (3.4). The derivation of this formula is given in Appendix- A. Y RK1 = I R a + I Y Y V Ra0 Ra RK1 RK0 1 (3.4) where I Ra is the rms line fault current through the relay while I Ra0 is the zero sequence fault current seen by relay and V Ra is the faulted phase rms voltage. The line parameters are used to calculate the ratio of Y RK1 / Y RK0. The relay reach is set based on the positive sequence admittance of the protected line segment. This relay reach and calculated measured admittance in (3.4) are utilized to make the tripping decision and tripping time calculation Phase elements These elements will respond for the line to line faults in the network. Similar to the earth elements, the number of phase elements per phase varies depending on whether protection has been configured as directional or non-directional. For 34

59 Chapter 3: Protective relay for DG connected networks example, for phase A, two phase elements are employed, if protection is nondirectional, one for the faults between phase A and phase B and another for the faults between phase A and phase C. Measured admittance seen by a phase element for a line to line fault, between phase A and phase B, can be expressed as, Y RK1 IRa I = V V Ra Rb Rb (3.5) where I Ra and I Rb are rms phase currents in faulted phases and V Ra and V Rb are faulted phase rms voltages. This measured admittance in (3.5) is used by relay logic to detect a line to line fault in the network Directional elements The directional elements can be used to identify whether the fault is in forward or reverse direction from the relay. This will help to implement separate reach settings for each direction especially when a relay protects non-equidistant zones. The user has been given the facility to select the preference as listed in Table 3.1. Table 3.1 Selection criterion of a directional element Setting Operation Directional Each element has two settings to cover both forward and reverse direction faults Non-directional Each element operates regardless of the fault direction Directional blocking Each element only operate either forward or reverse direction (user can select the direction) 3.5 Connection of ITA relays to a network The basic connection diagram of the ITA relay is shown in Fig Voltage and current at the relay location are obtained using a voltage transformer (VT) and a 35

60 Chapter 3: Protective relay for DG connected networks current transformer (CT) respectively. The relay output is linked to the tripping coil of the circuit breaker (CB). The relay continuously monitors the input parameters and executes the relay logic to identify a faulted condition in the network. The process of making the tripping decision is shown in the Fig Based on the fundamental voltage and current, the admittance is calculated, which is the measured admittance of the relay point at a given time. The measured admittance and values for the relay reach settings are the inputs to the relay logic. This logic consists of normalized admittance calculation, relay characteristic equations, relay tripping time calculations, identification of fault direction and defined relay constraints. The faulted condition is detected by using the constraint in (3.3). Once fault is detected, the calculated tripping time based on measured admittance is fed through an integrator to obtain the tripping signal for the CB. Also relay checks whether the fault detection signal exists until relay issues the tripping command to avoid any nuisance tripping. Fig. 3.7 Relay connection diagram to the system Fig. 3.8 Process of relay tripping decision making 36

61 Chapter 3: Protective relay for DG connected networks 3.6 Settings of ITA relays to detect resistive faults Higher fault resistance can affect the operation of ITA relays. Therefore a method of relay settings is described to achieve successful relay operation in the presence of fault resistance. The relay carefully checks the constraint in (3.3) continuously, which is the comparison between the measured admittance (Y m ) and the total admittance (Y t ) setting of a particular zone. The relay detects a fault in the network when Y m becomes higher than Y t. For a fault within a particular zone, Y m is always greater than Y t, if fault resistance is zero. However, with the increase of.fault resistance, Y m can become less than Y t. Also the maximum fault resistance which allows the relay to operate depends on the fault location of the line. For example, the relay can operate for a higher resistive fault, if the fault is near the relay than when it is further away from the relay since a higher value of fault resistance can be compensated by each zone for near faults. Another protective zone is introduced to achieve the tripping operation of the relays under resistive faults. The maximum fault resistance which can be tolerated by the relay is decided based on the loads of the feeder. In this case, the relay operation can be obtained up to a pre-defined value of fault resistance. This method will not work for higher resistive faults, where fault currents are in same levels as load currents. The minimum equivalent impedance of loads (i.e. the maximum load condition in the system) is calculated based on the known system parameters under the normal operation. That will be the corresponding maximum fault resistance which can be tolerated by the relay under the faulted condition. If the relay reach settings are below the minimum equivalent impedance of the loads, the relay may trip under normal operation as identifying the load as a high resistive fault. However, a safety factor is introduced to avoid unnecessary relay operations. For example, 37

62 Chapter 3: Protective relay for DG connected networks effect of cold load inrush can be considered. Therefore, it is proposed to select one third of minimum equivalent impedance of loads as the fault resistance setting. A relay, if it has two zones, has two tripping characteristic curves. In this case, total admittance, Y t should be set separately for each zone depending on protection requirements. Instead of these two zone characteristics, another characteristic will be introduced to discriminate the high resistive faults as mentioned above. Hereinafter, this zone is denoted by Zone-3. In this case Y t consists of corresponding line impedance and the maximum fault resistance which is determined based on loading condition. A coordination time interval should be kept between adjacent two Zone-3s of relays to obtain the correct relay grading. Otherwise relay characteristic of Zone-3 in each relay will not show a considerable time difference for the faults with low fault resistance. Also the tripping time is set to a little higher value than the settings in the normal zone operations, since there is no requirement to isolate the faults with lower fault currents faster than the faults with higher fault currents. The radial network shown in Fig. 3.4 is considered again to illustrate the relay settings for all the zones Zone-1 settings Zone-1 reach setting is similar for all the relays if they are located equidistant. Therefore, Y t is set by assuming the Zone-1 will protect 120% of the first line, Z 12 being the impedance between two adjacent buses. Y 1 = (3.6) (1.2 Z ) t _ Zone 1 Tripping characteristic for Zone-1 can be given by, t p = (3.7) Y 1 r 38

63 Chapter 3: Protective relay for DG connected networks Zone-2 settings Y t is set by assuming Zone-2 will protect 200% of the first line. This setting is also similar for all the relays. The reach setting and tripping characteristic can be given by Y t _ Zone 2 1 = (3.8) (2 Z ) t p = (3.9) Y 1 r Zone-3 settings This zone represents a broader coverage of the protected line including the compensation for fault resistance. The value of Y t can be set using the allowable maximum fault resistance. The allowable maximum fault resistance is denoted by Z f after calculating the maximum load and adjusting it using the safety margin. It should be noted that Z f is the maximum fault resistance that can be handled by the relay when fault occurs in the far end of the protected zone. It is not the fault resistance in a particular fault condition. In this case, Y t for the Zone-3 can be set as, Y t _ Zone3 = 1 ( Z + Z ) (3.10) 12 f Zone-1 and Zone-2 tripping characteristics are same for all the relays. However, setting of Zone-3 will be different for each relay since coordination time margin should be kept among the relays. If relays are designed to operate for both forward and reverse faults, then the relay will have two different tripping characteristics in Zone-3; one for forward faults which can be denoted by Zone-3F and other one for reverse faults which can be given by Zone-3R. In this case, the directional feature should be added to the relay for differentiation of forward and 39

64 Chapter 3: Protective relay for DG connected networks reverse faults. If the relay detects the fault as forward, then forward tripping characteristic, t_ Zone3F is activated. On the other hand, reverse tripping characteristics, t_ Zone3R is activated if fault is detected by the relay as reverse. The Zone-3 tripping characteristic of each relay can be modified by assigning different constant values. A minimum tripping time characteristic should be selected for the furthest downstream relay to discriminate the forward faults. It can be then increased according to the coordination time interval between two adjacent relays. This Zone-3 grading is similar to the TDS setting of an overcurrent relay in a radial feeder. On the other hand, the minimum tripping time characteristic for reverse fault is selected to the furthest upstream relay. The settings of Zone-3 for the relays R 1, R 2, and R 3 in the radial feeder of Fig. 3.4 can be given as shown in (3.11)-(3.13) respectively. These settings can be changed according to the protection requirements. t t t t t t Zone3 F _ R = Yr Zone3 R _ R = Yr Zone3 F _ R = Yr Zone3 R _ R = Yr Zone3 F _ R = Yr Zone3 R _ R = Yr (3.11) (3.12) (3.13) The tripping characteristics of relays R 2 and R 3 in radial feeder of Fig. 3.4 are considered to illustrate the Zone-3 operation. The relay characteristics are shown in Fig In this case, only two relays are considered for clear illustration. Zone-3 40

65 Chapter 3: Protective relay for DG connected networks tripping characteristics of R 2 and R 3 are calculated using (3.12) and (3.13) respectively. The figure also shows the combined forward relay characteristics of Zone-1 and Zone-2. It can be seen that forward Zone-3 as well as reverse Zone-3 characteristics of each relay have been graded appropriately to achieve the backup protection. For example, it is assumed that a fault between BUS-1 and BUS-2 in the system cannot be detected by the primary zones (i.e., Zone-1 and Zone-2) of R 1 and R 2. Then, in this case, R 2 detects the fault in reverse Zone-3 to isolate the fault from downstream side while R 3 provides the backup protection as shown in Fig Fig. 3.9 Relay tripping characteristics of different zones 3.7 Practical issues for admittance calculation The process of fundamental voltage and current extraction is very important on tripping time calculation since the tripping time is decided based on the calculated measured admittance. Therefore the fundamental extraction methods and the factors which can influence on the fundamental extraction should be considered. Harmonics, current transients and decaying dc magnitude and time constant can be identified as the major challenges on fundamental extraction. The decaying dc component can 41

66 Chapter 3: Protective relay for DG connected networks usually appear in current signal. However, the decaying dc magnitude and time constant cannot be calculated before a fault occurs since it depends on the system configuration (X/R ratio), fault location and the value of fault resistance. Fast Fourier Transform (FFT) or Discrete Fourier Transform (DFT) can be used to extract the fundamental component from a sampled waveform. FFT is a fast way of calculating the DFT. FFT can accurately calculate the fundamental in the presence of harmonics and signal noises. However, it is not immune to decaying dc component. DFT is widely used in digital protective relays. DFT can extract the fundamental in the presence of harmonics, however it is also not immune to decaying dc component [63]. Some previous studies propose some interesting techniques to calculate the fundamental component accurately in the presence of decaying dc component. A method is proposed in [63] to eliminate the effect of decaying dc component by calculating the time constant of the faulted current waveform. In this algorithm, the dc magnitude is then calculated based on the calculated time constant for a cycle and subtract the dc magnitude from samples to obtain the DFT without decaying dc component. A DFT based filter algorithm is presented in [64] for digital distance relays to extract the fundamental accurately. In [65], a method is described to remove the decaying dc component for an application of protective relays. It can be applied either half cycle or full cycle of the sampled waveform to calculate the fundamental. In proposed ITA relays, the measured admittance is calculated using the faulted phase fundamental voltage and current in the relay location. The first coefficient is only sufficient for this calculation. One of the dc decaying removal algorithms as mentioned above can be used in ITA relay application to minimise the error in 42

67 Chapter 3: Protective relay for DG connected networks tripping time calculation. However, the speed of the calculation and burden on the processor should be carefully considered when selecting a particular algorithm to avoid the errors coming form decaying dc component on tripping time calculations. 3.8 Summary In this chapter, the basic features of the proposed ITA relay to protect a distribution network or a microgrid which has several DGs are discussed. The relay inverse time characteristic and relay reach setting have been explained. Furthermore, different relay elements and a method of relay setting to achieve fault detection under higher resistive faults have been explained. Finally, the challenges of implementing ITA relays and possible solutions to avoid them are identified. 43

68

69 Chapter 4: Evaluation of ITA relay performance 4.1 Introduction Faults can be usually identified by sensing the current level in an electrical power system since high currents can be seen during the faults. Overcurrent (OC) relays, fuses and moulded case circuit breakers (MCCBs) are the common type of current sensing protective devices used in distribution networks. The OC relays can be classified as definite current or instantaneous, definite time and inverse time based on the operating characteristics. In this chapter, features of inverse time OC relays are briefly considered since they are commonly used in distribution networks. On the other hand, distance type relays are commonly used to protect the transmission networks where speed of operation and reliability are very important. Fundamentally, MHO type distance relay are considered in this chapter. These existing OC and distance relay protection schemes are compared with the proposed ITA relay to demonstrate the performance of the ITA relay. The grading and coordination of relays are explained. Furthermore, the effects of DGs on relay operation are considered. The relay performances are validated by PSCAD/EMTDC simulations and MATLAB calculations. 45

70 Chapter 4: Evaluation of ITA relay performance 4.2 Inverse time overcurrent relays Most of the existing distribution networks are radial and they are employed with OC protective devices because of their simplicity and low cost [4, 30]. Coordination of such protective devices based on current is relatively easy in the radial networks. The IEEE Standard inverse time relay tripping characteristic is given by [37] A t p = + B TDS p M 1 (4.1) where the constants A, B and p are used to select the relay characteristic curve and time dial setting (i.e. TDS) is used for the coordination between several OC relays. M is the multiple of pickup current and it is defined by I f M = I (4.2) p where I f is the fault current seen by the relay and I p is the relay set current (i.e. pickup current). Three inverse time OC relay characteristic equations are given in the IEEE report [37]. They are moderately inverse, very inverse and extremely inverse. Each relay curve has different constants values in (4.1). To illustrate the grading of inverse time OC relays, a four bus bar radial feeder is considered as shown in Fig Relays are located at BUS-1, BUS-2 and BUS-3 and are denoted by R 1, R 2 and R 3 respectively. An upstream relay will provide the backup protection for the adjacent downstream relay. A time margin, called coordination time interval (discrimination), is kept between relays to achieve the relay coordination by setting the TDS of each relay. Inverse time tripping characteristic of the graded inverse OC relays for this radial system is shown in Fig. 46

71 Chapter 4: Evaluation of ITA relay performance 4.2. The relay tripping time t p is shown with the fault location. Coordination time intervals for R 1 -R 2 and R 2 -R 3 are denoted by t 12 and t 23 respectively. Lowest TDS value is set to the furthest downstream relay R 3. Then TDS for R 2 is selected appropriately such that coordination time interval between R 2 and R 3 to be t 23. Similarly TDS for R 1 is selected. The relay nearest to the source (i.e. R 1 ) will see the highest fault current in the feeder. However, as a result of grading, this relay (R 1 ) has a higher TDS compared to the relays R 2 and R 3. Therefore R 1 will take longer time to clear faults near BUS-1. This is a disadvantage of OC relay grading, because faults which have higher fault currents cannot not be cleared quickly. Fig. 4.1 A radial distribution feeder with relays Fig. 4.2 Inverse time overcurrent relay grading The effect on OC protection is considered once a DG or DGs are connected to a radial network. The fault behavior of the network will change considerably with the change of fault current and fault current direction. As a result, several protection issues can arise. Some of these issues are relay coordination, relay reach and relay 47

72 Chapter 4: Evaluation of ITA relay performance response in islanding operation. A DG can initiate problems related to the relay coordination in a distribution feeder depending on the DG size, type, and location [5]. Fault current seen by a relay at the beginning of a feeder will be reduced in the presence of DGs in the system. As a result, a fault can remain uncleared for a longer time. It has been shown that the OC relay reach can be reduced in the presence of DGs in the feeder [29]. The adverse effects on the reach increase when the penetration level of DGs increases. If the DGs are plug-and-play and may not be connected to the system at all times, then their power output fluctuates. As a result, the OC relays will respond differently. Moreover, in an islanded condition, the fault current levels will be low if the supply is from current limited converters. The OC relays may not operate satisfactorily under such a scenario. Therefore the protection of a distribution network with OC relays is a difficult task, especially when several DGs are connected to the system. The numerical comparison of performance between OC and ITA relays is given in Section Distance relays The basic operation and coordination of MHO type distance relay are explained in this sub-section. The reach setting of this relay is carried out based on the positive sequence impedance of the protected line. Each relay has a number of protection zones, where it measures the impedance to a fault and checks whether the measured value lies inside or outside the defined zones in order to make a tripping decision. The apparent impedance seen by an earth element of the relay for ground faults is calculated by using the standard equation [66] Z LG V phase = (4.3) I phase + KI 0 48

73 Chapter 4: Evaluation of ITA relay performance where I 0 is the zero sequence current, K is the residual compensation factor for apparent impedance (Z LG ) calculation and V phase and I phase are the respective phase voltage and current. Even though positive sequence impedance for a fault can be calculated using (4.3), in most of the real distance relay applications, comparison of two quantities called operating and polarizing are carried out to determine whether the fault exists inside or outside a particular zone. The MHO characteristic in R-X plane is shown in Fig. 4.3 assuming, for illustration purpose, the relay has two zones of protection. The reach setting for the zones can be established following the procedure described in [66]. For R 1 settings, Zone-1 covers 80 percent of the first protected line length and Zone-2 covers the whole first line plus 50 percent of the adjacent line length. A 20 percent portion is kept as a safety margin to mitigate errors caused by current and voltage transformers and impedance calculations when selecting the Zone-1 reach. Similarly, zone settings for the other relays can be performed. Zone-2 of R 1 provides the back up protection for R 2. Time delay is set between Zone-1 and zone-2 to enable correct discrimination as shown in the timing diagram in Fig As can be seen from the figure, the relay tripping is the same if the fault is in a particular zone. For example, if the fault is within the Zone-1, the relay will trip after a time period of t 1. It does not consider the distance to the fault for tripping within a particular zone (for example point A and B in Fig. 4.4). Unlike an OC relay, the distance relay does not follow any inverse time characteristic. The lack of inverse time characteristic is disadvantageous from the point of view of coordinating with other inverse time protective devices such as reclosers, fuses etc. 49

74 Chapter 4: Evaluation of ITA relay performance Fig. 4.3 MHO relay characteristic Fig. 4.4 MHO relay zone settings and timing diagram As mentioned before, distance relays are used primarily for transmission line protection. Distribution networks are different from transmission networks. Most of the distribution networks are radial and they have distributed three phase as well as single phase loads. Moreover, it is expected that different types of DGs may be connected at different locations along a radial distribution network. Therefore the suitability of protecting such a network using the conventional distance relays has to be analyzed. The preliminary investigation of distribution network protection using distance relays has been reported in [67]. It has been shown that distance relays having negative sequence directional feature can be used to protect a distribution feeder that has a single current limited converter connected DG at the beginning of the feeder. 50

75 Chapter 4: Evaluation of ITA relay performance This observation is valid both when this DG is operating either in islanded or gridconnected mode. However further studies are required to confirm the validity when feeder is operated with several current limited DGs at different locations. A numerical performance comparison between distance relay with proposed ITA relay is given in Section ITA relays The ITA relay characteristic is given in (3.2) and it is reproduced below. t A = ρ Y 1 p + r k (4.4) where A, ρ and k are constants, while the tripping time is denoted by t p. To illustrate the grading of ITA relays, the same four bus bar radial system shown in Fig. 4.1 is considered. The ITA relays do not have a TDS setting as in the case of OC relays to achieve coordination. These relays will make the tripping decision based on the measured admittance to the fault location. The tripping characteristics of graded ITA relays for the radial system are shown in Fig It has been assumed that each relay has two zones of protection. The combined tripping characteristic of Zone-1 and Zone-2 is shown in the figure. Coordination time intervals are denoted by t 12 and t 23. In OC protection, the relay near to the source takes longer time to operate. However in the case of ITA relays, the relay near to the source will take the same time to operate as other relays do. The measured admittance is the only parameter that will decide the relay tripping time. This is an advantage of the ITA relay over an OC relay. 51

76 Chapter 4: Evaluation of ITA relay performance Fig. 4.5 ITA relay grading To compare how the OC relay and ITA relay are different, a line segment as shown in Fig. 4.6 is considered. The node R represents the relay location, while fault point is denoted by node F, which is at a distance X from the relay. The current and voltage seen by the relay for the fault is given by I x and V x respectively. Fig. 4.6 Faulted line with a relay The pick up current setting of an OC relay for this feeder protection is usually calculated by taking the one third of minimum fault current at the end of feeder. The calculated pick up current is denoted by I p, while corresponding voltage at the relay point for the current of I p is represented by V p. For the fault at point F, the multiple of pick up current for the OC relay tripping characteristic can be calculated similar to (4.2) by using I M x I = (4.5) I p 52

77 Chapter 4: Evaluation of ITA relay performance To obtain the same sensitivity for the ITA relay, the total line admittance should be set corresponding to the admittance which is given by I p and V p. Therefore, the normalized admittance, which is the ratio between the measured admittance and total admittance, can be given for the fault at point F as Y ( / ) m ( I x / V ) I x I p Y x r = = = (4.6) Yt ( I p / Vp ) ( Vx / Vp ) From (4.5) and (4.6), normalised admittance Y r can be given by M I Vx Y r = where MV = (4.7) MV Vp Note that while M I is the multiple of the pick up current and M V can be defined as the multiple of pick up voltage since it is the ratio between faulted voltage to the pick up voltage. For a fault in the network, M I is greater than one. However, the magnitude of M V is less than one when a fault occurs in the network. It can be seen from (4.7) that ITA relay uses both current and voltage multiples instead of only current based multiple in the OC relay. As a result, the ITA relay has the ability to detect the faults effectively irrespective of the available fault current level in a network. This may also result in fault detection under low fault current level environment, specially the cases where current limited DGs are connected to the network. To show how the ITA relay can be related to a distance type relay, real imaginary (R-X) plane representation is considered with the relay tripping curve. Zone-1 of the ITA relay is considered for this illustration. The ITA relay characteristic can be represented in both distance-tripping time and R-X plane as shown in Fig The tripping time characteristic curve of ITA relay can be mapped into circles in R-X plane, in which each circle has a unique tripping time. For example, point D on tripping time curve is considered. This point is corresponding to 53

78 Chapter 4: Evaluation of ITA relay performance a fault with tripping time t 4. When the point D is mapped into R-X plane, the point becomes a circle which gives the same tripping time of t 4. In a similar manner, four more circles are shown in R-X plane corresponding to fault points A, B, C and E on tripping time curve of the relay. It can be concluded that infinite number of concentric constant tripping time circles exist for Zone-1 of the ITA relay. Therefore, the ITA relay can be identified as similar to a distance relay with infinite number of zones to give different tripping times depending on the location of the fault point. Fig. 4.7 ITA relay characteristic in R-X diagram 4.5 ITA relay performance A radial feeder with DGs In this section, the ITA relay performance is compared with the existing protection schemes in a radial distribution feeder considering DG connections. The major expectations of this study are to 54

79 Chapter 4: Evaluation of ITA relay performance show the relay grading performances show the effect of a DG or DGs on relay sensitivity illustrate the possibility of coordination with other protective devices investigate the relay response under converter connected DGs in grid connected and islanded mode operations. Fig. 4.8 shows a radial distribution network with DG1, DG2 and DG3 connected at BUS-2, BUS-3 and BUS-4 respectively. The relays are located at BUS- 1, BUS-2 and BUS-3. The parameters of the considered system are given in Table 4.1. Several case studies are considered depending on the system configuration and the type of protective devices employed. The OC relay settings for the given system are calculated based on the maximum and minimum fault current levels at each bus as illustrated in [68], assuming the relays have moderately inverse OC characteristic [37]. The discrimination time for the OC relays is chosen as 0.3 s. The calculated OC relay settings are listed in Table 4.2. Fig. 4.8 Radial distribution feeder with DGs Table 4.1 System parameters System data Value System frequency 50 Hz Source voltage (Vs) 11 kv rms (L-L) Source impedance Zs = j Ω Feeder impedance Z 12 = Z 23 = Z 34 = j Ω 55

80 Chapter 4: Evaluation of ITA relay performance Table 4.2 OC relay settings Parameter R 1 R 2 R 3 CT ratio 200:5 150:5 150:5 Pick-up current (A) TDS The constants for the ITA relay characteristic, given in (4.4), are chosen assuming that the relays have two zones of protection in which Zone-1 covers 120% of the first line length and Zone-2 covers twice the length of the first line. The selected constant values for Zone-1 and Zone-2 are given in Table 4.3. Table 4.3 Zone characteristics of ITA relay Zone number A p k Zone Zone The relay response is observed by creating three phase faults at different locations along the feeder. Results are obtained through MATLAB calculation while PSCAD simulation results are used to validate the calculations. Case Study-1: Relay grading in a radial feeder The grading of OC relays and ITA relays for the radial network of Fig. 4.8 is shown in Fig. 4.9 without considering the DGs. It is clear that in the case of OC relays, the relay near to the source takes long time to trip due to the discrimination time and higher TDS values whereas ITA relays clear the fault quickly by considering the distance to the fault (i.e. measured admittance). Therefore proposed ITA relay grading is superior compared to the OC relay. 56

81 Chapter 4: Evaluation of ITA relay performance Fig. 4.9 OC and ITA relay grading Case Study-2: Relay sensitivity For this case, it is assumed that only DG1 is connected to the radial feeder of Fig The ratio between utility (source) fault current to DG1 fault current is chosen as 5 (i.e. I s /I DG1 =5). The OC and ITA relay responses are shown in Fig In the case of OC relay, the sensitivity of R 2 and R 3 has increased while coordination time interval between the relays has reduced due to the fault current contribution of DG1. However for the ITA relays, the sensitivity of R 2 and R 3 remains the same with or without the DG. The two curves, for both these ITA relays, overlap and therefore they cannot be differentiated in Fig The only noticeable change for the ITA relay is for R 1 and this is due to the infeed from DG1. It can therefore be concluded that the ITA relay coordination between downstream relays from DG1 remains as same, irrespective of the fault current levels of the DG. The same system is considered to compare the ITA relays with conventional distance relays. It is considered here that distance relays have three zones which Zone-1 covers 80% of the protected line length while Zone-2 and Zone-3 cover 120% and 200% of the protected line length respectively. The relay response is 57

82 Chapter 4: Evaluation of ITA relay performance shown in Fig The sensitivity of R 2 and R 3 remain the same for both the relay types. However, the ITA relay shows an inverse time characteristic and it is an advantage when coordination aspects are concerned. Fig OC and ITA relay response when DG1 is connected Fig Distance and ITA relay response when DG1 is connected Case Study-3: Time-current relay characteristic The relay operating time for different levels of fault current is important when the relay is used for the coordination with other current sensing protective devices. Relay R 3 is considered in the system of Fig. 4.8 assuming that no DGs are present. 58

83 Chapter 4: Evaluation of ITA relay performance Fig shows the time-current characteristic for both OC and ITA relays. It is evident that the ITA relay also has an inverse time characteristic curve. Since the OC relays can be used for coordinating with other current sensing protective devices in the network, this result confirms that the ITA relay can also be used for the same purpose. Case Study-4: Grid protection with converter interfaced DGs In this study, it is considered that all the three DGs (DG1, DG2 and DG3) are connected and are current limited during faults. The ITA relay response is shown in Fig The fault current contribution of each DG is one 40th of the source fault current (i.e. I s /I DG = 40). Let there be a fault at point A, which is between buses 2 and 3. Both relays R 2 (in forward direction) and R 3 (in reverse direction due to DG3) will isolate the fault. Relay R 1 provides a backup for R 2. In this case, the protection system of DG2 will disconnect it after a defined time if the fault persists. In this manner, it can be seen that the ITA relays have the capability to isolate the faulted segment from both the sides allowing both grid connected and islanded operations. Fig OC and ITA relay time-current characteristic 59

84 Chapter 4: Evaluation of ITA relay performance Fig ITA relay response in grid connected mode Case Study-5: Islanded protection with converter interfaced DGs To investigate the relay response in islanded mode of operation, it is assumed that the utility source is isolated from the rest of the system (upstream from BUS-1) and the DGs together have the capability to supply the total load demand. All the DGs are considered to be converter interfaced. To discuss the results, two terms are introduced relay response and relay tripping characteristic. The latter is defined by (4.4) using only the line parameters. It does not consider that DGs are present in the system or the presence of any fault resistance. The relay response is calculated using (4.4) considering all loads, fault resistance and DGs present in the system. The relay response and the relay tripping characteristics are shown in Fig when all the DGs are connected to the system. The solid lines represent the forward relay tripping characteristic, while dotted lines indicate the relay response to faults. The relay response shows the capability of detecting the faults in islanded system under low fault current levels. Fig shows the relay response obtained from PSCAD simulation for a single line to ground fault (SLG) between BUS-2 and BUS- 3. The fault is created at 0.1 s and R 2 (forward isolation) and R 3 (reverse isolation) 60

85 Chapter 4: Evaluation of ITA relay performance operate after s and s respectively. It is to be noted that if OC relays are used, they will fail to detect this fault in islanded condition due to the low current level produced by the converters. Fig ITA relay response in islanded mode Fig ITA relay response for SLG fault in islanded operation Effect of source impedance on relay response In this study, the behavior of OC and ITA relays is compared when the source impedance changes. For this purpose, a system with two parallel transformers connected between buses A and B is considered as shown in Fig The supply feeder starts at BUS-B. The impedance of each transformer is taken as 0.3 p.u., while feeder impedance is chosen as 0.1 p.u. The relay response for faults between BUS-B 61

86 Chapter 4: Evaluation of ITA relay performance and BUS-C when only one transformer is connected and both transformers are connected is shown in Fig The impedance between buses A and B is 0.15 p.u. when both transformers are on, while it is 0.3 p.u. when only one is operating. Therefore the fault current level is higher when both transformers are one. Since an OC relay response time depends on the fault current levels, there are two different curves for OC relay in Fig However, the ITA relay shows the same response irrespective of the impedance change caused by transformer connections. Fig System with two parallel transformers Fig Relay response for impedance change ITA relay response for different DG and load distribution profiles A radial distribution feeder which has n number of buses as shown in Fig is considered. In this system, utility is connected at BUS-1 which does not have a load or a DG. Either a DG or a load or their combination is connected to each of the other buses. All the DGs considered in this study are connected through the current 62

87 Chapter 4: Evaluation of ITA relay performance limited converters. The ITA relays can be employed at different locations according to the protection requirements. Three phase faults are considered under this study at different locations. The DG penetration level is said to be 100% when the load demand is equal to the DG generation at normal operating condition. Moreover, it is assumed that DGs supply twice of their rated current in constant current control mode during a fault. Two types of load and DG distribution profiles are considered to investigate the relay response. These are uniform load and DG distribution and random load and DG distribution. Fig Distribution feeder with DGs and loads To demonstrate the operation of the ITA relays, a 45 bus radial feeder is considered. In this system, the ITA relays R 1, R 2, R 3 and R 4 are located at buses 1, 12, 23 and 34 respectively. It is to be noted that load demand is selected such that the voltage at the end of the feeder is slightly higher than 0.95 p.u. when all the loads are connected without any DG. This will be the maximum loading in the system. Results are obtained using MATLAB for different values of fault resistance. The parameters of the study system are listed in Table 4.4 assuming the DGs and loads are distributed uniformly along the feeder and DGs can supply the whole load demand (i.e. 100% of DG penetration). 63

88 Chapter 4: Evaluation of ITA relay performance Table 4.4 System parameters System data Value System frequency 50 Hz Source voltage (V s ) 11 kv rms (L-L) Source impedance Z s = j Ω Feeder impedance Z 12 = Z 23 = Z k(k+1) = j Ω Each load impedance Load1 =Load2= j Ω Each DG power output DG1=DG2= MVA A. Uniform load and DG distribution It is assumed that DGs and loads are distributed uniformly along the feeder. Therefore, 44 DGs and 44 loads are connected from BUS-2 to BUS-45, which implies a DG penetration level of 100%. The ITA relay response for three phase faults along the feeder is shown in Fig for fault resistance of 0.05 Ω. The relays respond to isolate both the forward and reverse faults effectively by allowing unfaulted segments to operate in islanded operation. Fig ITA relay response when fault resistance is 0.05 Ω The fault current seen by each relay for faults along the feeder is shown in Fig It can be seen that reverse fault current is significantly smaller compared to the 64

89 Chapter 4: Evaluation of ITA relay performance forward fault current. This is because the utility also feeds the fault current in the forward direction, while the DGs, which operate in current limit mode, only feed the reverse fault current. However, the ITA relays still have the ability to detect the faults under this lower fault current level. Fig Fault current seen by each ITA relay along the feeder B. Random load and DG distribution In this study, it is considered that the distribution of DGs and loads along the feeder is not uniform. They are connected randomly while maintaining the total DG penetration level as 100%. Different random distribution profiles are considered. However, results are presented for only one study. The considered random DG and load distribution profiles are shown in Fig The ITA relay response under these distribution profiles is shown in Fig It can be seen that the relays will respond both in the forward and reverse directions to isolate the faulted segment, thereby increasing the system reliability. 65

90 Chapter 4: Evaluation of ITA relay performance Fig Random load and DG distribution profiles along the feeder Fig ITA relay response for random load and DG distribution profiles An application of ITA relays to IEEE 34 node test feeder To illustrate the ITA relay reach settings, relay grading and relay performance on a realistic system, the IEEE 34 node test feeder is considered [69]. The test feeder, shown in Fig. 4.23, is modified by connecting three converter interfaced DGs at nodes 838,840 and 862. Three ITA relays R 1, R 2 and R 3 are assigned to the nodes 832, 834 and 860 respectively. In this example, two regions, Region-A and Region- B, are selected assuming that these regions have the capability to operate in islanded 66

91 Chapter 4: Evaluation of ITA relay performance mode with the aid of DGs. In the islanded mode, DG1 supplies the load demand of Region-A, while DG2 and DG3 supply the load demand of Region-B. Each DG capacity is chosen appropriately to enable the autonomous operation of each of these regions. Furthermore, it is assumed that all the DG converters operate in a current limited mode during the faults. Relay R 1 provides the primary protection up to the node 860 (covering nodes 858 and 864), while R 2 and R 3 provide the primary protection for the Region-A and Region-B respectively. This protection example (i.e., the protection downstream from node 832) is considered to illustrate the efficacy of the ITA relays. The test system is modeled and simulated in PSCAD, where the DGs are assumed to be ideal dc voltage sources. The converter structure and control is given in Appendix-B. The converters are switched in output feedback voltage control mode during normal operations and output feedback current control mode [70] during the faults to limit the faults currents. Fig IEEE 34 node test feeder with ITA relays The forward and reverse reach settings of ITA relays are calculated using the given system parameters of the IEEE 34 feeder and they are given in Table 4.5, in which the lengths and impedance values are given. The forward and reverse relay 67

92 Chapter 4: Evaluation of ITA relay performance reach are different since lines have different lengths. The relay tripping characteristics and zone settings are similar to those given in Sub-section Different types of fault are created at different locations to study the efficacy of proposed scheme. However, a few cases are presented here. It is to be noted that a DG should go into the current limiting mode as soon as a fault is detected to supply the fault current such that the ITA relays can detect and isolate the fault. However, if the fault persists for a longer period of time, the DG should isolate itself from the supply in order to protect its power electronic switches. Therefore, a time interval is defined for each DG for which it will supply the fault current. Table 4.5 ITA relay forward and reverse reach settings Parameters for forward reach settings Relay Primary protection Length (m) Z 1 (Ω) R 1 Node j R 2 Node j R 3 Node j Parameters for reverse reach settings Primary protection Length (m) Z 1 (Ω) R 1 Not considered R 2 Node j R 3 Node j Z 1 - positive sequence impedance A. Fault at Node 858 A SLG fault is created at 0.3 s. The relays R 1, R 2 and R 3 operate at s, s and s respectively to isolate the faulted segment. Note that R 2 and R 3 see this as a reverse fault. The ITA relay response is shown in Fig Note that, once the fault is cleared, DG1 in the Regions-A and DG2 and DG3 in Region-B continue to supply the loads, thereby increasing the reliability of the system. 68

93 Chapter 4: Evaluation of ITA relay performance Fig ITA relay response for SLG fault at node 858 B. Fault at Node 842 A SLG fault is initiated at 0.3 s. The relay R 2 responds at s as shown in Fig to isolate the fault. The rest of the system except Region-A operates in grid connected mode. Since there are no other protection devices between DG1 and the fault point, DG1 is disconnected after the defined time interval. Fig ITA relay response for SLG fault at node 842 C. Fault at Node 862 A SLG fault is initiated at 0.3 s and relay R 3 operates at s to isolate the fault. The ITA relay response for this fault is shown Fig The rest of the system except Region-B operates in grid connected mode. 69

94 Chapter 4: Evaluation of ITA relay performance Fig ITA relay response for SLG fault at node 862 The above results verify that the ITA relays have the ability to respond under low fault current levels. These relays isolate the faulted segment from the network allowing unfaulted segments to operate either in grid connected or islanded mode operations. Also it is to be noted that proposed ITA relays have the ability to detect the faults in islanded conditions as well. For example, if an islanded system is created by opening the circuit breaker at Node 832, the rest of system operates with adequate protection. However the fault current level will be significantly lower in such a case ITA relays for mesh network protection To demonstrate an application of ITA relays to a mesh network protection, a system shown in Fig is considered. This system has a partly mesh network containing BUS-1, BUS-2 and BUS-5. There are three DGs and three loads in this system. All the DGs are connected through voltage source converters (VSCs). As before, these VSCs limit their output current to twice of the rated current during a fault. Eight ITA relays are employed for secure and reliable operation of the system. The relay locations are shown in the figure. The one of the main aims of the ITA relays is to isolate the faulted segment quickly in the event of a fault allowing unfaulted sections to operate either in grid connected or islanded mode depending on 70

95 Chapter 4: Evaluation of ITA relay performance the fault location. In the case of an islanded mode operation, each DG or DGs in the islanded section can operate in autonomous mode if there is sufficient generation to supply the load demand. The system parameters are listed in Table 4.6. In this study, no communication between relays is considered for a simple and cost effective solution. Fig Mesh network under study Table 4.6 System parameters System parameter Value Voltage 11 kv L-L rms Frequency 50 HZ Source impedance ( j ) Ω Each feeder impedance ( j 4.335) Ω The relays R 12, R 21, R 15, R 51, R 52 and R 25 which are located in the mesh network have the directional blocking feature in which these relays only respond to forward faults. This results in proper relay coordination within the mesh network. For example, consider relay R 15. It protects the line segment between BUS-1 and BUS-5. Also it provides the backup protection for the line segment between BUS-5 71

96 Chapter 4: Evaluation of ITA relay performance and BUS-2. However, R 15 is blocked for the reverse faults since R 12 should operate for the faults between BUS-1 and BUS-2. The relays R 12 and R 52 cover the line segment between BUS-2 and BUS-3 in forward direction. On the other hand, the relay R 32 has the directional feature and thus it can detect faults in either sides of BUS-3. The relay R 43 is also a directional blocking relay which only responds for reverse faults since it is located at the end of the feeder. The relay reach settings and tripping characteristics of Zone-1 and Zone-2 are the same as given in Sub-Section The reach setting of Zone-3 is selected to cover fault resistance of 50 Ω. However, the grading of relays for Zone-3 is different as explained in Chapter 3. In this system, R 32, R 52, R 15 and R 12 in the forward direction and R 51, R 25, R 21, R 32 and R 43 in the reverse direction should be coordinated separately. When performing the ITA relay grading in Zone-3, tripping time for forward faults should be increased, while it should be decreased for reverse faults from downstream to upstream relays in the network. The graded Zone-3 tripping characteristics of ITA relays are given in Table 4.7. Table 4.7 Zone-3 grading of ITA relays Relay grading for forward faults Relay grading for reverse faults t Zone3 F _ R = t Zone3 R _ R = Y 1 Y 1 r t Zone3 F _ R = t Zone3 R _ R = Y 1 Y 1 r t Zone3 F _ R = t Zone3 R _ R = Y 1 Y 1 r t Zone3 F _ R = t Zone3 R _ R = Y 1 Y 1 r t r r r r Zone3 R _ R = Yr

97 Chapter 4: Evaluation of ITA relay performance The system is simulated in PSCAD. A SLG fault is created at different locations with different values of fault resistances at 0.2 s. The ITA relay fault clearing times are listed in Table 4.8. In each line segment, two fault locations are considered. As can be seen from the results, the relays respond to isolate the faulted segment effectively. For example, in the event of a fault between BUS-1 and BUS-2, the both relays R 12 and R 21 respond to isolate the faulted segment. In this case, the rest of the system operates in grid connected mode after the successful isolation of the faulted segment. Table 4.8 Fault clearing time of ITA relays Fault location Fault clearing time of respective relay (seconds) (between) R f = 0.05 Ω R f = 20 Ω 10% from BUS-1 R 12 =0.071,R 21 =0.137 R 12 =0.438,R 21 =0.774 BUS-1 and 90% from BUS-1 R 12 =0.137,R 21 =0.072 R 12 =0.443,R 21 =0.359 BUS-2 All the loads are supplied in grid connected mode without line Z 12 10% from BUS-1 R 15 =0.071,R 51 =0.137 R 15 =0.540,R 51 =0.774 BUS-1 and 90% from BUS-1 R 15 =0.136,R 51 =0.073 R 15 =0.544,R 51 =0.251 BUS-5 All loads are supplied in grid connected mode without line Z 15 10% from BUS-2 R 25 =0.072,R 52 =0.137 R 25 =0.348,R 52 =0.445 BUS-2 and 90% from BUS-2 R 25 =0.137,R 52 =0.073 R 25 =0.458,R 52 =0.443 BUS-5 All loads are supplied in grid connected mode without line Z 25 R 12 =0.150,R 52 =0.158, R 12 =0.459,R 52 = % from BUS-2 R 32 =0.082 R 32 =0.480 BUS-2 and R 12 =0.410,R 52 =0.427, R 12 =0.481,R 52 = % from BUS-2 BUS-3 R 32 =0.075 R 32 =0.509 Load3 is supplied in grid connected mode while Load1 and Load2 BUS-3 and BUS-4 supplied in islanded mode without line Z 23 10% from BUS-3 R 32 =0.074,R 43 =0.137 R 32 =0.345,R 43 = % from BUS-3 R 32 =0.139,R 43 =0.072 R 32 =0.349,R 43 =0.913 Load1 and Load3 are supplied in grid connected mode while Load2 is supplied in islanded mode without line Z 34 73

98 Chapter 4: Evaluation of ITA relay performance Higher fault clearing time can be experienced for resistive faults due to the relay grading and infeed effect of DGs. Within the mesh configuration, fault current seen by relays are coming from different directions. Limitations of relay operation due to the fault resistance and DG infeed are discussed in next sub-section. 4.6 Limitations of ITA relays The limitations of ITA relay operation due to the fault resistance and DG infeed are discussed in this sub-section. To explain the change of relay response with the increase of fault resistance, a fault between relays R 1 and R 2 is considered in the network shown in Fig The Thevenin equivalent line impedance between R 1 and fault point is denoted by Z f, while that between R 2 and fault point is denoted by Z r. The fault current that is fed from the source be denoted by I S, while the fault current fed from the downstream side DGs is denoted by I DG. Fig Equivalent representation of the faulted network The measured admittances seen by relays R 1 and R 2 in the presence of fault resistance and DGs can be respectively given by Ym( R1) = Z f Is Is + R f ( Is + I DG ) = Z f 1 + R f 1 + I DG Is (4.8) Ym( R2 ) = Zr I DG IDG + R f ( Is + I DG ) = Zr + R f Is I DG (4.9) 74

99 Chapter 4: Evaluation of ITA relay performance As per (4.8) and (4.9), an error occurs in the measured admittances due to the fault resistance. This is the second part of the impedance (one which is multiplied by R f ). If fault resistance is zero, the error becomes zero and relays can effectively detect the faults. When fault resistance is present, the ratio between source current and DG current (I S /I DG ) also affect the measured admittance. For fault detection, the measured admittance (Y m ) should be greater than the total admittance (Y t ). Generally, I S >I dg since I S is fed by a utility with higher capacity than a DG. Therefore, for a particular DG penetration level and a particular value of fault resistance, the error on the measured admittance R 2 (Y m(r2) ) is greater than that of R 1 (Y m(r1) ). Thus it can be seen that fault resistance will affect the downstream relay more than the upstream relay. However, once the forward relay isolates the fault, source current I s becomes zero and this will allow reverse relay to operate. As can be seen from (4.9), the error due to the source current will become zero. However, the error due to R f will remain. The ability of the ITA relay fault detection depends on the system configuration. The maximum value for a fault resistance which relay can reliably respond can be calculated based on the relay settings, line parameters and DG connections. One such study is carried out considering different fault resistance values and source current to DG current ratios (denoted by I S/DG ). The line impedance between two relays in Fig is taken as j Ω. The results are shown in Fig The total admittance setting of each relay (i.e. Yt) is shown assuming this zone covers three times the length of the line segment. The measured admittances seen by relays R 1 and R 2 are denoted by Y m(r1) and Y m(r2) respectively. The measured admittance change of R 2 due to the fault isolation by the forward relay R 1 is also shown in the figure. If the measured admittance of a relay is greater than total admittance, that relay can successfully detect the fault. 75

100 Chapter 4: Evaluation of ITA relay performance It can be seen from Fig that when fault resistance becomes higher, the measured admittance decreases reducing the relay ability of detecting faults. On the other hand, when I S/DG becomes higher (i.e. DGs have less capacity), the fault detection capability of forward relay has improved. The measured admittance of reverse relay R 2 is always below the total admittance. As can be seen from the figure, the measured admittance of R 2 becomes higher than Y t once forward relay isolates the fault in the forward direction. Therefore the reverse relay does not detect the faults until the forward relay operates. Fig ITA relay response for different values of fault resistances and DG currents The operation of the ITA relay may be affected by fault resistance and infeed of DGs located downstream. As a result, the relay may take more time to trip than expected. The values of line parameters, total admittance setting, fault location, fault 76

101 Chapter 4: Evaluation of ITA relay performance resistance and DG capacity will determine the amount of error on the measured admittance and hence the tripping time of the ITA relay. 4.7 Summary In this chapter, the ITA relay performances have been compared with the existing relays. The relay grading and coordination, the effect of DGs on the relay response and the effect of source impedance on the relay response for both ITA and OC relays are considered. The ITA relay performances are evaluated in a radial distribution feeder with DGs considering different DG and load distribution profiles. Moreover, an application of these relays to IEEE 34 test node feeder has been demonstrated. The protection of a mesh network which has several converter interfaced DGs and loads are considered using the ITA relays. Both resistive and non-resistive faults are simulated to see the efficacy of ITA relays. Finally, the limitations of ITA relay protection due to the fault resistance and DG connections have been addressed. According to the relay performance analysis, it is clear that ITA relays have the capability to isolate a faulted segment in a DG connected network allowing unfaulted segments to operate either in grid connected or islanded mode operations. This results in improving the reliability of a network since customer outages are reduced. 77

102

103 Chapter 5: Fold back current control and system restoration 5.1 Introduction Most of the faults around 80-90% in the power system are temporary and they can be successfully removed by performing reclosing [47]. However, when DGs are connected to the network, reclosing becomes more complicated. The current approach is to disconnect all the DGs as soon as a fault occurs. However, the disconnection of DGs even for temporary faults will reduce the system reliability. Again, most of the temporary faults are arcing type. As long as current is supplied to the fault, the arc is sustained. If a network contains DGs and these DGs are not disconnected during these arc faults, then the DGs will supply fault current thereby not allowing the arc to extinguish. Thus, reclosing and arc extinction can be identified as two major protection issues in a DG connected network. To overcome these problems, specifically in a converter based DG connected network, a novel control strategy for a converter interfaced DG is proposed using fold back current control characteristic. The main aim of this work is to restore the system after a temporary fault without disconnecting the converter interfaced DGs from the system while facilitating the reclosing effectively. Furthermore, DGs can restore the unfaulted segments even for permanent faults if employed line protection scheme can isolate the faulted segment effectively. The proposed fold back converter 79

104 Chapter 5: Fold back current control and system restoration control will be explained in this chapter. Arc model selection for system simulation with a DG is then considered. Simulation results of proposed converter control for temporary arc faults and permanent faults will be presented considering automatic reclosing. 5.2 Fold back current control characteristics A fold back current control for a converter interfaced DG is proposed in this section. This control strategy helps in the arc extinction and self restoration while facilitating reclosing effectively. Moreover, the converter control can maintain sufficient current level to aid fault detection if the protective devices in the network are designed to respond to a limited fault current level. The converter nominally operates in voltage control mode. Once a fault is detected, it switches to fold back current control mode discussed in Section The mode of operation is decided based on the value of converter terminal voltage. The description of the converter control characteristic is given below Fold back during contingency The converter control contains two separate characteristics; one for normal operation and contingency (faulted) conditions and the other one for system restoration. The voltage-current relation of the converter for normal and contingency operation is shown in Fig. 5.1(a). During normal operating condition, the VSC operates along the line segment AB under a voltage control mode. It is assumed that when the VSC output voltage reaches V 2, its output current reaches 2I r, I r being the rated current. The operation of the VSC shifts to current control mode once the output voltage falls below V 2. This is shown in Fig. 5.1 (b), in which the output current is gradually reduced with time along the line segment DE. This gradual 80

105 Chapter 5: Fold back current control and system restoration decrease of current can help to trigger the protective devices for isolating the faulted segment from the network. Especially, the protective devices which are located downstream from the fault location can respond due to this fault current. If the current is suddenly reduced, the protective devices may not have any information about the fault. From point E, the current is reduced rapidly along the segment EF until it reaches a value ni r, where n is a very small number. The number n is selected to be very small since terminal voltage of the VSC is maintained to a very lower value for a defined time period after the point F reaches. It is to be noted that the line segment BC cannot be represented exactly in Fig. 5.1 (a) since converter voltage, in current control mode, may change with the system parameters such as fault location and fault resistance. This is why the voltage current relation in this case is shown with a dotted line in the figure. The time period t 12 and t 23, shown in Fig. 5.1 (b), can be selected according to the protective devices employed and their requirements. Especially, the time, t 12 allows relays to detect the faults. Therefore it should be long enough for successful fault isolation. The selection can be done by calculating the maximum fault clearing time of a known relay characteristic. The VSC checks the terminal voltage continuously to identify whether the fault is cleared or still persists. If the fault is cleared during the current fold back period, the VSC will restore the system to its pre-faulted state by changing back to the voltage control mode. For example, if the fault is cleared during fold back at the point P shown in Fig. 5.1 (c), then the VSC restores through point M to the operating point O. The load line is shown assuming the DG capacity is sufficient to supply the load demand. 81

106 Chapter 5: Fold back current control and system restoration (a) During voltage control mode (b) During current control mode (c) During fault clearing Fig. 5.1 Proposed fold back characteristics After the fold back current time periods of t 12 and t 23, the VSC output current has reached to ni r. The DG is then kept connected to the network for a pre-defined time period in which it injects a very small current. This mode of converter operation 82

107 Chapter 5: Fold back current control and system restoration can be called sleep mode and it allows for the arc faults to self-extinguish. The required time period for sleep mode can be calculated based on the arc de-ionization time which is given by [71]. kv t = cycles (5.1) 34.5 where kv is the line to line rated voltage and the unit of time t is cycles. During this sleep time period, self extinction of arc faults is achieved without disconnecting DGs from the network Restoration process The self-recovery process of the DG starts after the defined de-ionization time for the network. In the restoration process, a VSC is not allowed to exceed the rated current I r. Recovery characteristic of the VSC is along the points CKL as shown in Fig The voltage-current characteristic of the VSC for the line segment CK can be given by V2 nv v = i 2 (5.2) I (2 n) (2 n) r where v and i are the rms voltage and current magnitudes, I r is the rated current, V 2 is the converter output voltage when it injects 2I r and n is a small number. It is to be noted that the rms magnitude is calculated by a moving average filter with a window of one cycle. During transients, the filter will produce a time-varying output. Therefore, from knowledge of the value of i at any instant, v can be calculated or vice versa. The VSC controller can calculate corresponding voltage/current for a particular value of current/voltage using (5.2) during the restoration process through the line CK. Three different cases are considered to explain the restoration process depending 83

108 Chapter 5: Fold back current control and system restoration on the network conditions. Fig. 5.2(a) shows the restoration process when the load demand is less than the DG capacity. At the start of the restoration process, the VSC controller calculates the voltage using (5.2) for current at point C. The operating point then shifts to point M on the load line based on the calculated value. At this instant, the controller calculates the amount of current corresponding to the voltage at point M. Then the rated current at point N is injected which will take the voltage to point L. Thereafter the controller switches to voltage control mode and the system operation will shift to point O on the load line. In this case, the DG has restored the system successfully thereby increasing the reliability. Next, the unsuccessful restoration processes of the DG are considered. These can occur if the fault is not cleared or if the load demand becomes higher due to some DG/utility tripping. Two such cases are considered to show the VSC response to safeguard the system. A possible fault line for a ground fault with a fault resistance is shown in Fig. 5.2(b), in which the voltage cannot rise above point O until the fault is cleared. In the case of higher load demand, the load line will shift as shown in Fig. 5.2(c). It is obvious that the operating point will reach point O through the path as shown in this figure. The VSC will still remain in the current control mode since the load demand is more than that the DG can supply. The restoration process is carried out by the VSC for a defined time interval. If the VSC has not been restored successfully within this time period, the DG will be disconnected from the system by tripping the circuit breaker connected at the output of the DG. 84

109 Chapter 5: Fold back current control and system restoration (a) For low load demand (b) For a fault (c) For high load demand Fig. 5.2 System restoration It is to be noted that only constant impedance type loads are considered for the illustration. For the low inertial loads, it is expected that loads such as induction motors have reached zero speed during the sleep time of the DG and they behave like 85

110 Chapter 5: Fold back current control and system restoration constant impedance loads once the restoration process is started. The effect of load mainly influences the recovery characteristic of the converter interfaced DG. The control strategy checks the possibility of automatic system restoration by checking the terminal voltage of the DG for particular injected current as mentioned. The rms value of the DG terminal voltage at that current is the measure which determines the possibility of system restoration. For example, if the load demand is high compared to the DG generation then restoration will be unsuccessful due to the inadequate terminal voltage level. Therefore, the shape of the load curve in voltage current graph only changes the restoration path and ultimate operating point Coordination with reclosers The total restoration time can be used to coordinate with the reclosers in the system. Two methods can be introduced to synchronise a DG with a recloser. In the first method, the DG takes the opportunity to restore the system before the operation of any auto recloser. This method is advantageous, if DG penetration level is significant and DGs have the ability to supply the load demand in autonomous operation. The restoration process of DGs can be successful or unsuccessful depending on the load demand and fault status as discussed earlier. After the defined time interval of the restoration process, the auto recloser activates and this can result in a live to live or live to dead reclosing depending on the result of DG restoration. The auto recloser should be capable of checking for synchronism and make sure that there is no phase mismatch, if it performs live to live reclosing. There is less possibility to have a phase mismatch since DGs are not fully shutdown and they maintain the original phases during the fault and after the restoration. However, the DG is safeguarded during network contingency conditions and it is discussed in the next sub-section. 86

111 Chapter 5: Fold back current control and system restoration In the second method, an opportunity is given to the recloser to restore the system before the DG start to restore the system. This method can be used for a system when the DG capacity is not enough to supply the load demand. In this case, the sleep time of the DG should be adjusted according to the recloser operating time. The recloser may restore the system depending on whether the fault still exists or it is cleared. If the system is successfully restored, then DG can start restoration process which will then be successful. This results in maximising the DG benefits to the customer. On the other hand, if reclosing fails to restore the system, then DG can start the restoration process. Again, depending on the fault status and load demand, the self restoration can be successful or unsuccessful. The DG will restore the system, if fault clears and load demand is less than the DG capacity. However, DG will disconnect from the network if restoration is unsuccessful due to higher load demand or uncleared fault. A load shedding scheme can be implemented to restore the system when load demand is high, however it is out of scope of this thesis DG protection It is important to consider the consequences of out of phase reclosing when DGs are not disconnected during the auto recloser open time. The risk of DG damage due to the out of phase reclosing is lower, if DG is connected through a converter [49]. In the proposed reclosing scheme, the recloser is capable of checking the synchronisation which ensures there is no phase mismatch when it performs live to live reclosing. From the point of DG protection, the DG should be protected itself. To achieve basic DG protection requirements, in the proposed method, a DG is employed with several protective elements; fold back current control, reverse power flow, over voltage and synchronism check. The proposed fold back control protects the DG 87

112 Chapter 5: Fold back current control and system restoration from excessive current injection and unsuccessful system restoration. The reverse power flow protection is activated to trip the DG when current flows towards the DG. The over voltage element responds when terminal voltage of the DG rises above a pre-defined limit. However, under voltage protection is incorporated with the proposed fold back current control since the DG is usually allowed to operate under the rated voltage in current control mode. The synchronised relay ensures a trouble free connection to the feeder when it is being reconnected after any disconnection. These protection schemes will minimise the DG safety risks associated with reclosing. 5.3 Arc fault model selection for simulation An accurate representation of an arc in simulation is difficult due to its random nature. However, for studying the effects of DG on arc faults, a realistic arc model is needed. The selected arc model should indicate whether the arc is sustained or extinguished. One possibility is to choose a current dependent arc resistance model, which represents the arc with a time varying resistance or a square wave voltage source in phase with the arc current [54]. This model is valid when the fault current level is high, i.e., when the utility is connected. However, once the utility is disconnected and fault current is supplied by the DGs, the arc parameters change. The new parameters are difficult to define accurately. Therefore, this model has not been used. Another possibility is to use both primary and secondary arc models for simulation in the presence of DGs in the network. Most line faults are single phase to ground and they are temporary arc faults. Therefore, the arc fault can be successfully removed by performing the single-pole reclosing in high voltage (HV) lines [72]. 88

113 Chapter 5: Fold back current control and system restoration The primary arc exists in HV lines before the circuit breaker opens and a secondary arc occurs due to hot plasma remaining from the primary arc after the circuit breaker opens. The secondary arc is sustained by the mutual coupling (capacitive and inductive) between the faulted phase and un-faulted phases [51]. However, reclosing is usually three-pole in medium and low voltage systems. Therefore, in a way similar to HV arcs, a secondary arc model can be used in the presence of the DGs to simulate arc faults after the disconnection of the utility supply, where the DGs will sustain the secondary arc. In [49], a similar arc fault study has been performed with a wind power plant, where the measured arc voltage waveforms are compared with the simulation results to validate the arc model. The primary and secondary arc models are used for the simulation studies to investigate the effect of DG on arc faults. The selected arc models are discussed in detail below. These models indicate the arc behaviour during the fault. However, a criterion to determine the arc extinction should be known and the selected criterion is also given Primary arc fault The theory of switching arc was recently proposed to model the long fault arcs in air, both primary and secondary [73]. Heavy fault current flows during the primary arc period. The arc column has a large cross sectional area since the system provides a high input electrical power to the arc. It can be assumed that there is no elongation of the arc length during this period. The dynamic arc characteristics can be written as [51, 52, 73], ( G g ) dg p = 1 p p (5.3) dt Tp 89

114 Chapter 5: Fold back current control and system restoration where, T p is the arc time constant, G p and g p are the stationary arc conductance and the instantaneous arc conductance respectively. G p and T p can be expressed as, Gp i α I p = ; Tp = V l (5.4) p l p p where, i is the absolute value of the primary arc current, V p is the arc voltage gradient, l p is the primary arc length, I p is the peak value of primary arc current and α is a constant Secondary arc fault The secondary arc is usually self extinguishing, however its duration can depend on many factors and it is mainly dependent on the secondary arc current [72]. The secondary arc length will vary over time. Wind velocity and the magnitude and duration of the primary arc current are the two factors which effect the elongation of the arc length. However, the total secondary arc voltage is practically proportional to the arc length [73]. The low current secondary arcs can be expressed as [51] dg dt s ( G g ) = 1 s s (5.5) Ts where, T s is the secondary arc time constant, G s is the stationary arc conductance and g s is the instantaneous secondary arc conductance. G s and T s can be given by Gs i β I 1.4 = ; T s s = ( ) ls ( tr ) (5.6) Vs ls tr where i is the absolute value of the secondary arc current, t r is the time from initiation of secondary arc, l s (t r ) is the time varying arc length, I s is the steady state peak secondary arc current and β is a constant. 90

115 Chapter 5: Fold back current control and system restoration Arc extinction Defining the arc extinction condition is a challenging task in arc modelling. The arc self-extinction action depends not only on the fault current magnitude, but also on the transient recovery voltage rate after successful arc extinction at current zero crossing. Furthermore, the arc extinction time is proportional to the arc time constant. In [51], the arc extinction is proposed based on dielectric breakdown. The arc model in (5.5) only considers the thermal re-ignition, while dielectric re-strikes are not considered. In [72], the secondary arc extinction is determined, if the derivative of arc resistance is higher than the value in (5.7) and the instantaneous conductance is lower than the value in (5.8). dr arc ' dt = 20 M Ω /( s m) (5.7) g' min = 50µ S m (5.8) However, this criterion only considers the thermal extinction of the arc and there is a probability of dielectric re-ignition of the arc. This has not been considered in this study. 5.4 Simulation studies Several case studies have been carried out to investigate the effect of DGs on both permanent and transient arc faults. For this purpose, a four bus radial distribution feeder is considered as shown in Fig All the DGs considered in this study are connected to the feeder through the converters and they all are employed with the proposed fold back current control strategy. The converter structure and basic control used in this simulation is given in Appendix-B. DG1, DG2 and DG3 are connected to the feeder at BUS-2, BUS-3 and BUS-4 respectively. Load1, Load2 and Load3 have the same power consumption. The System parameters of the 91

116 Chapter 5: Fold back current control and system restoration simulated system are given in Table 5.1. The capacity of each DG is selected such that it can supply the load connected to its bus in autonomous mode. Fig. 5.3 Simulated radial feeder with DGs Table 5.1 Simulated system data System data Value System frequency 50 Hz Source voltage (V s ) 11 kv rms (L-L) S base V base Utility source impedance (Z s ) DG source impedance (Z dg ) Feeder impedance (Z 12 =Z 23 =Z 34 ) Load impedance (Z L ) DG power output 10 MVA 11 kv j p.u j p.u j p.u j p.u. 0.5 MW The performance of proposed ITA relays in the presence of fold back current control DGs are also investigated for both permanent and arc faults. Therefore the ITA relays R 1, R 2 and R 3 are employed to protect the feeder as shown in the figure. The relays are assumed to have two zones of protection in which Zone-1 covers 120% of the first line and Zone-2 covers twice the length of the first line. The following constants are chosen for the tripping characteristics of the relays. For Zone-1: A = , p = 0.08, k =

117 Chapter 5: Fold back current control and system restoration For Zone-2: A = , p = 0.1, k = 0.01 The relays are located just before the BUS-1, BUS-2 and BUS-3. With this arrangement, the infeed effect on relays which are located downstream from the fault can be minimised. All the circuit breakers connected to the relays are considered to have reclosing capability, since part of this study is to demonstrate the compatibility of proposed protection and control strategy on reclosing Results for permanent faults Three phase permanent faults are simulated along the feeder of Fig. 5.3 to evaluate the relay response. Fig. 5.4 shows the ITA relay characteristic curves and relay response which have been obtained by MATLAB calculation. The combined Zone-1 and Zone-2 forward characteristics are shown for each relay in Fig It is assumed that each DG injects a constant current of 0.06 ka during the fault and there is 0.05 ohm of fault resistance at the fault point. Results show that ITA relays can respond to isolate faults from both the upstream and downstream sides of the feeder effectively. Each relay also provides backup protection for the adjacent relay. For example, for a fault at point A shown in the figure between BUS-2 and BUS-3, R 2 will trip first to isolate the fault from the upstream side while R 3 responds next to isolate the fault from downstream side. However, if R 2 fails to trip, R 1 will trip by providing the backup protection. Further backup protection for relay R 2 can be provided, if a 3-zone protection scheme is chosen for R 1. 93

118 Chapter 5: Fold back current control and system restoration Fig. 5.4 Calculated ITA relay response for a three phase fault The system shown in Fig. 5.3 is simulated in PSCAD for different system configurations since all the DGs and loads may not be connected at all the time. A SLG fault is created at s at different locations. Different case studies are considered by changing the DG and load connectivity to the network. Case study-1: When DG generation is more than load demand In this study, it is considered that all three DGs and three loads are connected to the feeder of Fig A SLG fault is created between BUS-1 and BUS-2, at a point that is 10% of the line length away from BUS-1. The DGs switches from voltage control mode to current control mode soon after the fault. The DGs then start to reduce the current gradually in current control mode according to the fold back characteristic. Relays R 1 and R 2 respond to isolate the faulted segment at 26 ms and 51 ms respectively after the initiation of fault (i.e. R 1 operates at s and R 2 operates at s). As a result of successful faulted segment isolation, the islanded system, beyond BUS-2, can operate in autonomous mode. The voltage, output 94

119 Chapter 5: Fold back current control and system restoration current and output real power of DG1 is shown in Fig Once R 2 opens at s, the DGs are switched back into the voltage control mode before they reach the sleep mode. This results in fast system restoration without disconnection of any of the DGs from the network. The response of the other two DGs is similar and is not shown here. In this case, faulted segment isolation and fast system restoration have been achieved by using the proposed ITA relay protection scheme and fold back current control of the DGs. These result in maximising the DG benefits to the customers by increasing the system reliability. On the other hand, if DGs are disconnected after the fault, customers beyond BUS-2 experience a power outage since this is a permanent fault. Fig. 5.5 DG1 response (a) output voltage (b) output current (c) real power output Case study-2: When DG generation is less than load demand To simulate this scenario, only DG1 and DG3 are assumed to be connected to the system with three loads. A fault is created at the same time mentioned in the case study-1. The relays R 1 and R 2 respond 26 ms and 51 ms respectively to isolate the 95

120 Chapter 5: Fold back current control and system restoration faulted segment. Once the relay R 2 responds to isolate the faulted segment, DG1 and DG2 supply the power to the loads in the islanded section beyond BUS-2. The output voltage, output current and real power of DG1 are shown in Fig At the moment R 2 opens, the DGs try to restore the system, which can be seen from the voltage transient at s. However, since load demand is higher than the power generation from DGs, the system restoration is not possible. Therefore, the system does not recover at the instant of fault clearing and DGs further decrease the output current until it reaches the sleep mode. The DGs remain in sleep mode for a defined time period (note that it is 100 ms in this simulation) without disconnecting from the system. After this time duration, the DG restoration process starts, during which the controller calculates the output voltage as per Fig. 5.2(c). This causes an increase in voltage as evident from Fig. 5.6(a). However, this voltage is insufficient to restore the system and after a further 50 ms, the DGs are disconnected due to unsuccessful restoration. The DG circuit breakers then open to isolate the DGs from the system. Load shedding on the basis of the restoration attempt is not addressed in this study. Moreover, only one attempt of DG restoration is considered. However, several attempts can be considered to restore the system since it may increase the system reliability for temporary faults. In Fig. 5.6, it can be seen that the voltage and current becomes zero at s, while the power exponentially reduces to zero. This apparently strange behaviour is caused due to the low-pass filter used in the power measuring circuit. This prevents power measurement to become zero instantaneously. However, since both voltage and current become zero at s, the DG output power also becomes zero at the same instant. 96

121 Chapter 5: Fold back current control and system restoration Fig. 5.6 DG1 response (a) output voltage (b) output current (c) real power output Results for Arc Faults The radial feeder of Fig. 5.3 is again considered to investigate the effect of fold back current on arc faults and the response of ITA relays. The simulated arc parameters are shown in Table 5.2. Table 5.2 Arc model parameters Arc parameters Value Primary arc model (including numerical values) dg p l p i = dt I p 15l p g p Secondary arc model (including numerical values) dg dt s l = s ( t r ) 3 primary arc length (l p ) 0.5m secondary arc length (l s ) 10 l p t (t is time) I 1.4 s 75I i 0.4 s l g t ) s( r s The arc fault is initiated at the peak of a voltage waveform. A high fault current flows in the beginning since both utility source and DGs feed the fault. At this stage, the arc is modelled as a primary arc. Once the relay responds to isolate the fault from 97

122 Chapter 5: Fold back current control and system restoration the utility side, the rest of the system becomes islanded and the fault current reduces due to the current limit applied by the VSC controllers. During the islanded operation, the arc is modelled as secondary arc. The simulated results are shown in the following sub-sections. Case study-3: Arc fault on the line between BUS-1 and BUS-2 An arc fault is created at s. The primary arc has high arc current, low arc voltage and low arc resistance as can be seen from Fig The relay R 1 responds at s as shown in Fig. 5.7(d) to isolate the arc fault from upstream (i.e. utility side). After the response of R 1, the arc resistance increases due to the secondary arcing, as evident from Fig. 5.7(c). The DGs start to reduce the output current gradually with the initiation of the fault. The response of DG1 is shown in Fig The relay R 2 responds at s to isolate the fault from the downstream side. The isolation of the faulted segment results in the islanded mode of operation containing all the three DGs. Thereafter, the system recovers successfully when DGs are switched to the voltage control mode. Fig. 5.8(b) verifies the sinusoidal current limiting and gradual current decrease in fold back characteristic. It is to be noted that a hardware current limiter is employed for the VSCs to limit the instantaneous peak output current to the value 100 A. The fold back current limit is applied once the hardware limit is reached. The results reveal that the proposed ITA relay scheme can isolate the faulted segment from the feeder. It leads to fast restoration of the system. The DGs restore the system without disconnecting from the feeder thereby increasing the reliability for even temporary faults. 98

123 Chapter 5: Fold back current control and system restoration Fig. 5.7 System behaviour for an arc fault (a) arc voltage (b) arc current (c) arc resistance (d) relay response Fig. 5.8 DG1 behaviour for an arc fault (a) output voltage (b) output current Case study-4: Arc fault on the line between BUS-1 and BUS-2, assuming relay R 2 fails to operate The same scenario as Case study-3 is considered here to illustrate the effect of DGs on arc extinction. However, in this case, it is assumed that the downstream relay R 2 fails to detect the fault. Once R 1 responds to the fault, DGs feed the arc in 99

124 Chapter 5: Fold back current control and system restoration secondary stage. The output voltage and current of DG1 are shown in Fig The DGs decrease the output currents according to the fold back characteristic as can be seen in the figure. Then DGs reach to sleep mode and inject very small currents. As a result of low current injection during the sleep mode, the arc extinguishes at s. The DGs then start to self-restore the system at s after remaining in the sleep mode for 100 ms. The system is restored completely at s since arc fault is cleared by the time DGs starts the restoration process. This study confirms that the DGs can self-restore the system without sustaining the arc. It is to be noted that only thermal arc extinction is considered in this case. However the DGs are kept in the sleep mode for 100 ms, a sufficient time required for dielectric arc extinction. Fig. 5.9 DG1 behaviour when downstream relay fails (a) output voltage (b) output current Auto reclosing Reclosing can be considered as one of the major protection issues when DGs are connected to distribution networks. Therefore an effective method is proposed to coordinate a recloser with a converter connected DG in the feeder. It is assumed that 100

125 Chapter 5: Fold back current control and system restoration the feeder is protected by either ITA relays or conventional overcurrent relays. Therefore two separate cases are considered to explain the reclosing possibilities based on the employed protection scheme. The radial feeder as shown in Fig. 5.3 is considered here as well. The circuit breakers associated with the relays have the reclosing capability. Moreover, synchronism check element is incorporated. Therefore the recloser element senses the voltages on the two sides of a breaker in exact synchronism before performing the reclosing operation. A. With proposed fold back current control and ITA relays The restoration of the faulted segment by coordinating the DGs and the reclosers in the system is performed based on the identification of fault direction. Reclosing opportunity is given to the relay which sees the fault as forward. For example, a temporary arc fault is considered on the line between BUS-1 and BUS-2. The fault occurs at s, which is subsequently cleared by relays R 1 and R 2 at s and s respectively. After the faulted segment isolation, DGs operates in autonomous mode supplying the load power. Now R 1 tries to close the circuit breaker first (live to dead reclosing) by identifying this fault as forward after a pre-defined delay-time period 0.3 s (this predefined time is denoted by T d ). The relay R 2 waits till the upstream side is restored before performing the reclosing operation. Therefore in this case, R 2 performs live to live reclosing after further time delay of 0.1 s. The DG1 response is shown in Fig The smooth transfer between utility and islanded feeder section at s validates the suitability of proposed reclosing scheme. The DG supplies the increase of real power demand during the islanded operation and after successful reclosing the real power output has reduced to the value before the fault as shown in Fig. 5.10(c). 101

126 Chapter 5: Fold back current control and system restoration Fig DG1 response during fault and system restoration If the fault is between BUS-2 and BUS-3, the relays R 2 and R 3 respond to clear the fault. Since DG1, still supplies the fault, it is disconnected from the system after a time period (this includes current fold-back time, sleep mode time and selfrestoration time of DG1), which is less than the delay-time period T d. Then, R 2 will try to reclose and if that is successful, R 3 will be connected once the system settles down. Note that, in this case, DG1 needs to be manually reconnected after fault is cleared since no automatic procedure is proposed for the DGs that are totally disconnected. B. With proposed fold back current control and conventional protection scheme In this sub-section, the coordination of reclosers and converter interfaced DGs are discussed assuming the relays employed in the radial feeder shown in Fig. 5.3 are overcurrent type. Therefore these relays only clear the forward faults from the feeder. To explain the sequence of operation, a fault on the line between BUS-1 and BUS-2 at s is considered. The forward relay R 1 detects the fault and isolates it from 102

127 Chapter 5: Fold back current control and system restoration utility side at s. However, in this case, DGs feed the arc fault since R 2 has not responded. As a result of fold back current control, DGs reduce the output current gradually and reach to the sleep mode at 0.39 s as shown in Fig It results in arc extinction at 0.41 s. The DGs then restore the system and supply the loads in autonomous mode. The reclosing of R 1 starts after the restoration time. The recloser successfully connects the utility and islanded section at s. The response of DG1 terminal voltage verifies the smooth transfer between islanded and grid connected mode operation. The output current of DG1 is more during the islanding mode than the grid connected mode since total load demand is supplied by the DGs. Fig DG1 terminal voltage and output current If there are motors and generators in the system, the first recloser may be time delayed. Circuit checking is needed before any time delayed recloser action to ensure that either synchronism exists or one circuit is dead. 103

128 Chapter 5: Fold back current control and system restoration 5.5 Summary A novel control strategy based on fold back current control is proposed for a converter interfaced DG to overcome the challenging protection issues in a DG connected distribution feeder. The control strategy enables the fast arc extinction, system restoration, and reclosing without disconnecting the DGs. The arc extinction is achieved by reducing the DG output current to a small value, while automatic system restoration is obtained if DG power generation is sufficient enough to supply the load demand. Recloser coordination with DGs is considered without any explicit communication. It can be seen that the reclosing is possible with converter connected DGs. The only requirement is to determine a sequence of operations with appropriate time delays between each recloser and DG depending on the system configuration. The results reveal that the DG benefits can be maximised by increasing the reliability of the system if fold back current control is employed with converter interfaced DGs. 104

129 Chapter 6: Experimental results 6.1 Introduction In the previous three chapters, the proposed ITA relay characteristic and its application on DG connected distribution networks are discussed. The discussions are presented in these chapters with the help of numerical results. However, practical implementation of the relay should be considered to verify the suitability of the relay for realistic protection applications. Therefore the main aim of this chapter is to present the experimental work which involves building a software prototype of the ITA relay to evaluate the performance in a distribution test feeder. The experimental results are then used to validate the calculated and simulated results. This chapter begins with a brief description of the test feeder arrangement used in the laboratory. Then relay performance is examined by changing the source impedance to illustrate the robust operation of the relay. Furthermore, relay deterioration factors are analysed. 6.2 Test feeder arrangement The ITA relay characteristic is examined in a laboratory test feeder to compare the experimental relay performance with the calculated and simulated results. A photograph of the test feeder is shown in Fig. 6.1 where each line segment consists of a resistor and inductor. A single line diagram of the experimental setup is given in 105

130 Chapter 6: Experimental results Fig In this test feeder, the source voltage can be controlled according to the requirement. The circuit breaker CB1 provides the protection for the entire circuit. Fig. 6.1 Experimental test feeder Fig. 6.2 Single line diagram of experimental setup The NI PXI-1042Q chassis shown in Fig. 6.3 is used to implement the relay algorithm and to acquire data. The NI chassis has a PXI-8187 windows XP Embedded card, two analog input cards and an output relay card. The voltage and current signals at the relay location are acquired to analog cards using the voltage and current transducers respectively. One of the switches in the output relay card is used to create faults. One typical fault location is shown in Fig The fault 106

131 Chapter 6: Experimental results location however is changed to study the relay performance. This switch can be fully controlled by the LabVIEW software. A SLG fault is created at different locations along the test feeder. The fault is created at a random time by closing the switch of the output relay card, while the ITA relay sends fault clearing signal to the same switch. It is to be noted that CB1 provides back up protection in case the switch does not operate to clear a fault. The system parameters of the test system are given in Table 6.1. Fig. 6.3 NI PXI-1042Q chassis The ITA relay characteristic is implemented on LabVIEW software. The detailed LabVIEW program is given in Appendix-C. A block diagram of the software model is shown in Fig The voltage and current signals are sampled and these samples are held in data buffers. The samples of one cycle are used to extract the fundamental component. Fast Fourier Transform (FFT) is used for this purpose. The measured admittance is then calculated using the extracted fundamental voltage and current. The measurement and control blocks used to calculate measured admittance on LabVIEW software is shown in Fig. C.1, Appendix-C. The relay reach setting is manually entered according to the line parameters. The relay algorithm issues the fault detection signal and calculates the tripping time when a fault occurs 107

132 Chapter 6: Experimental results in the network based on the measured admittance. Fault clearing signal is then obtained through an integrator. The LabVIEW implementation of ITA relay algorithm is shown in Fig. C.2, Appendix-C. Table 6.1 System parameters of the experimental setup System parameter Value Source V rms (L-G), 50Hz Feeder impedance R = 1.12 Ω L = 0.01H ( j 3.15 Ω) Load impedance (Z L ) 125 Ω CB1 rated current 1 A Hardware specifications NI chassis PXI-1042Q NI controller PXI-8187 windows XP Embedded Transducers Voltage Differential amplifier Current LEM LTSR 6-NP Data acquisition Analog inputs PXI-4070 FlexDMM Digital outputs PXI-2565 relay switches Sampling rate 3200 samples per seconds Fig. 6.4 ITA relay implementation on LabVIEW 108

133 Chapter 6: Experimental results 6.3 Relay performance evaluation A simplified single line diagram of the test feeder is shown in Fig The test feeder has five buses. A resistive load is connected at BUS-5. The ITA relay is located at BUS-1 and it has three zones of protection. In the experiment, it is assumed that the relay will protect the line segment from BUS-1 to BUS-5. Zone-1 covers the 120% of the line length between buses 1 and 5 while Zone-2 covers twice the line length. Zone-3 has been employed to achieve the relay operation in the presence of resistive faults. Therefore, the relay reach setting of Zone-3 is set to cover three times of the line length. The selected relay tripping characteristics and reach settings of each zone are given in Table 6.2. SLG faults are created at different buses to observe the ITA relay response. A digital oscilloscope is used to capture the faulted phase voltage at relay location and the current flowing through the relay. It is to be noted that two differential probes are used for the isolation purposes when measuring the voltages and currents by the oscilloscope. Fig. 6.5 Simplified single line diagram of the test feeder 109

134 Chapter 6: Experimental results Table 6.2 Relay reach setting and tripping characteristic in each zone Relay zone Tripping characteristic and reach Zone-1 1 Y t 1 = 1.2 ( t = Y r j12.6) Zone-2 1 Y t 1 = 2 ( j12.6) t = Y r Zone-3 1 Y t 1 = 3 ( j12.6) t = Y r 1 The calculated relay response using MATLAB for SLG faults along the feeder is shown in Fig In this calculation, the fault resistance is considered as zero. The calculated theoretical tripping times are used to validate the experimental results. As can be seen from Fig. 6.6, the relay tripping characteristics are selected appropriately for different zones. For example, consider Zone-1 and Zone-3 tripping characteristics. Zone-3 should always give higher tripping time than Zone-1 for faults if tripping characteristics of the zones are properly selected. Otherwise, Zone-3 can give lower tripping time than Zone-1 for bolted faults (i.e. zero resistive faults) if same tripping characteristics are selected for these zones, since reach setting of Zone- 3 is higher. 110

135 Chapter 6: Experimental results Fig. 6.6 Calculated relay response in different zones for bolted faults The ITA relay has been designed to operate for different fault current levels in a network. Specifically, when a converter interfaced DG is connected to the network, different voltage levels can be seen during the faulted condition. However, the ITA relay response should be same irrespective of the network fault current level/fault voltage level. Therefore, a number of test runs are carried out under different test feeder configurations to evaluate the relay performance. 6.4 Relay response for different fault locations In this sub-section, the ITA relay response for faults is investigated by changing the fault location and the source impedance to demonstrate the robustness of the relay operation. The relay should give higher tripping time when the fault point moves away from the relay location due to the inverse time characteristic. Also, the relay should respond in the same manner irrespective of source impedance, since the ITA relay tripping time does not change with the fault current level. To observe these features, several tests are carried out. The source rms voltage is adjusted to

136 Chapter 6: Experimental results V. Then SLG faults are created at different buses using the software control switch as mentioned earlier. The fault is initiated at a random time. Relay response time, which is the time relay algorithm calculates based on the measured admittance, is observed. The voltage and current seen by the relay are captured and actual fault clearing time is obtained using these two captured waveforms. It is to be noted that actual fault clearing time is longer than the relay response time. Once the relay issues the trip command, the hardware switch can take up to 10 ms to open [74]. The theoretical tripping time of the relay is calculated using Fig. 6.6 for comparison purposes. The test results are discussed in the subsequent sub-sections Fault at BUS-2 The relay response time for a SLG fault at BUS-2 is shown in Table 6.3. Several tests are carried out for the same fault location. However, the results of only three tests are shown here. Comparing the data listed in Table 6.3, it can be seen that the relay response time is close to the calculated theoretical values. The actual fault clearing time is slightly higher than the relay response time values due to the fault clearing time taken by the hardware switch. The captured voltage and current signals at relay location during the fault for test runs 1 and 2 using the digital oscilloscope are shown in Fig The source current lags the voltage before the fault. The current increases rapidly after the initiation of the fault while the voltage reduces. During the fault, the normalised admittance becomes higher than 1.0 and it causes the relay to initiate the trip signal. The tripping time is decided based on the value of the measured admittance. It is obvious from Fig. 6.7 (a) and (b) that the points on the voltage (or current) cycle at the faults occur are different. However, the relay response time is the same for test runs 1 and 2 (Table 6.3). 112

137 Chapter 6: Experimental results Table 6.3 ITA relay response for faults at BUS-2 Test Theoretical Relay response Actual fault run tripping time (ms) time (ms) clearing time (ms) (a) Test run-1 (b) Test run-2 Fig. 6.7 The variation of voltage and current for SLG faults at BUS Fault at BUS-3 The relay response time and the actual fault clearing time for a SLG fault at BUS-3 are given in Table 6.4. According to the theoretical fault calculations, the relay should clear the fault after 40 ms. The relay response time is very close to the 113

138 Chapter 6: Experimental results theoretical value as can be seen in the table. The variation of voltage and current for Test run-1 and Test run-2 is also shown in Fig. 6.8 to demonstrate the actual fault clearing time. The fault has created at different points on the current waveform and the relay isolates the fault approximately within two cycles. Table 6.4 ITA relay response for faults at BUS-3 Test Theoretical Relay response Actual fault run tripping time (ms) time (ms) clearing time (ms) (a) Test run-1 (b) Test run-2 Fig. 6.8 The variation of voltage and current for SLG faults at BUS-3 114

139 Chapter 6: Experimental results Fault at BUS-4 The fault point is further moved to downstream side along the feeder. A SLG fault is created at BUS-4. The main aim of changing the fault location is to check whether the tripping time follows the ITA tripping curve. For this fault location, calculated theoretical tripping time is 58 ms. The relay response time and fault clearing time are listed in Table 6.5. The variation of voltage and current for two of the tests is shown in Fig As can be seen from the figure, the fault current has reduced compared to the fault currents at BUS-2 and BUS-3. On the other hand, the faulted voltage has risen. However, the ITA relay response is accurate as expected from the calculation. Table 6.5 ITA relay response for faults at BUS-4 Test Theoretical Relay response Actual fault run tripping time (ms) time (ms) clearing time (ms) (a) Test run-1 115

140 Chapter 6: Experimental results (b) Test run-2 Fig. 6.9 The variation of voltage and current for SLG faults at BUS Fault at BUS-5 The fault is created at the end of the feeder. According to the calculations, the relay should issue the trip commend after 120 ms after the fault initiation. Table 6.6 shows the relay response results for this fault. Since the line impedance is significantly high between the relay location and fault location, the fault current is comparably low while the voltage is significantly high during the fault as shown in Fig However, the relay isolates the fault effectively. The results show the ability of ITA relay to detect and isolate the faults according to the designed inverse time tripping characteristic. Table 6.6 ITA relay response for faults at BUS-5 Test Theoretical Relay response Actual fault run tripping time (ms) time (ms) clearing time (ms)

141 Chapter 6: Experimental results (a) Test run-1 (b) Test run-2 Fig The variation of voltage and current for SLG faults at BUS-5 It can be seen that in all the cases presented above, the relay response time for the faults is lower than the theoretical tripping time. This error may be due to the voltage and current transducers. These transducers give an output which has a dc offset and lower magnitude compared to the input signals. Therefore, the relevant mathematical operations have been performed on acquired signals to match the values with original signals Relay response for source impedance change The source impedance of the existing line model has been changed by adding a series impedance of (4.6+ j 3.14). The main aim of increasing the source impedance 117

142 Chapter 6: Experimental results is to weaken the source which will result in lower fault current levels in the feeder. Therefore this study can be considered similar to the case where a DG is connected to the feeder. The experimental results for this study are shown Table 6.7. The actual fault clearing time for randomly selected tests at different fault locations is shown from Fig to Fig According to the figures, the fault inception angle is different for different fault locations. However, the fault clearing times are within the acceptable limits and they are very similar to the values which are obtained before adding the additional source impedance. The results reveal that the relay response is not affected by the value of the source impedance. Therefore it can be concluded that the ITA relay response is not sensitive to source impedance. Table 6.7 ITA relay response for SLG faults with higher source impedance Fault Theoretical Relay response Actual fault location tripping time time (ms) clearing time (ms) BUS BUS BUS BUS Fig Voltage and current for a fault at BUS-2 118

143 Chapter 6: Experimental results Fig Voltage and current for a fault at BUS-3 Fig Voltage and current for a fault at BUS-4 Fig Voltage and current for a fault at BUS-5 119

144 Chapter 6: Experimental results 6.5 Analysis of ITA relay degradation factors The ITA relay detects a fault in a feeder if the value of calculated normalised admittance of a particular zone increases beyond the value 1.0. The required tripping time is then decided based on the value of the normalised admittance. In the case of a three zone protection relay, all the zones of the relay may detect the fault. However, in such a case, the tripping time of each zone is different and the minimum tripping time is selected to issue the trip command. The calculated value of the normalised admittance for a particular zone will depend on the measured admittance since the total admittance setting for that zone is constant. Therefore, the calculated relay tripping time based on normalised admittance can deviate from the expected value if the calculated measured admittance is not accurate. Thus, the factors which can affect the measured admittance should be considered and they should be minimised to improve the relay performance. Typically, two types of errors which can affect the measured admittance are identified. The first type of error occurs due to the fault resistance and downstream sources (infeed). This is due to the network configuration and the nature of the fault. The maximum DG penetration level of a particular network can be known in advance. However, the fault resistance cannot be predicted. The calculated measured admittance errors due to the fundamental extraction can be considered as the second type of error. Current transients, harmonics, and decaying dc magnitude and time constant can cause errors in the fundamental extraction. Both the error types will be explained in following two sub-sections The effect of fault resistance and infeed The same experimental setup as shown in Fig. 6.2 is used to test the relay performance in the presence of fault resistance and infeed. Zone-3 has been 120

145 Chapter 6: Experimental results introduced into the ITA relay to compensate the fault resistance. The selected Zone-3 reach setting and tripping characteristic are the same as those shown in Table 6.2. In the first phase of the test, the resistive faults are considered without any infeed source. The ITA relay performance is tested for faults at different locations by inserting a 10 Ω resistor between the fault point and the ground. Both the calculated and the experimental results are given in Table 6.8 where measured admittance and normalised admittance are given by Y m and Y r respectively. The tripping time of the ITA relay is shown by t p. This table also includes these values calculated for a bolted fault (R f =0). According to the data in Table 6.8, the normalised admittance is always greater than 1.0. Also, the measured admittance decreases when fault point moves downstream along the feeder. The normalised admittance values for resistive faults are lower compared to the zero resistive faults. This results in higher relay tripping times for resistive faults as can be seen from Table 6.8. However, it can be concluded that experimental test results are very similar to that of the calculation results obtained from MATLAB. Table 6.8 Relay parameters during a resistive fault Fault Calculated results Experimental results location R f =0 Ω R f =10 Ω R f =10 Ω Y m Y r t p (ms) Y m Y r t p (ms) Y m Y r t p (ms) BUS BUS BUS BUS The variation of the measured admittance and the normalised admittance during the fault at BUS-2 is shown in Fig The measured admittance has increased with the initiation of the fault. It results in an increase of the normalised admittance during the fault as can be seen from Fig. 6.15(b). The value of the 121

146 Chapter 6: Experimental results normalised admittance will determine the relay tripping time. The calculated tripping time by the ITA relay algorithm for this fault is shown in Fig. 6.15(c), where the tripping time is 117 ms. (a) measured admittance (b) normalised admittance (c) tripping time Fig Change of parameters during a resistive fault at BUS-2 In the second phase of the test, the effect of both the fault resistance and infeed on relay operation is considered. The test feeder is modified by connecting another source at BUS-5 to represent the infeed as shown in Fig The source impedance (Z s2 ) of infeed is selected as 14 times greater than the main source impedance (Z s1 ). 122

147 Chapter 6: Experimental results This test feeder configuration is similar to a system where a DG is connected at BUS-5. The ITA relay response is obtained for the faults at BUS-2, BUS-3 and BUS- 4 with a fault resistance of 10 Ω. The calculated and the experimental relay parameters during the faults are given in Table 6.9. It can be seen that experimental results are close to the calculated results. However, the relay tripping time has further increased compared to Table 6.8 due to the infeed effect. Fig Test feeder with an infeed Table 6.9 Change of relay parameters due to fault resistance and infeed Fault Calculated results Experimental results location Y m Y r t p (ms) Y m Y r t p (ms) BUS BUS BUS The variations of measured admittance and normalised admittance, and calculated tripping time during the fault at BUS-2 are shown in Fig The measured admittance increases during the fault resulting normalised admittance to rise beyond the value 1.0. The relay algorithm calculates the tripping time as 138 ms based on the value of normalised admittance. 123

148 Chapter 6: Experimental results (a) measured admittance (b) normalised admittance (c) tripping time Fig Change of parameters for a fault at BUS-2 with fault resistance and infeed The effect of fundamental extraction In this sub-section, practical issues related to the admittance calculation are experimentally investigated. The ITA relay calculates the measured admittance based on the fundamental voltage and current at the relay location. Therefore, the process of fundamental extraction from the voltage and the current signals is very important 124

149 Chapter 6: Experimental results in the presence of transients, noises, harmonics and decaying dc component. In this experiment, FFT is used to calculate the fundamental rms magnitudes of sampled voltage and current signals. A 415 V three phase synchronous generator connected test feeder as shown in Fig is considered for this analysis. A SLG fault is created at the end of line segment as shown in the figure. Several tests are conducted to investigate the ITA relay behaviour during the fault. However, two results are presented below. Fig A SLG fault at synchronous generator connected feeder The faulted voltage and current captured during a SLG fault is shown in Fig It can be seen that the current waveform has decaying dc component during the transient period. Also the voltage has harmonics during the faulted period. The values of extracted current and voltage using FFT during each cycle are shown in Fig (a) and (b) respectively. The extracted rms current is higher than the steady state fault current during the first three cycles. It can be seen that the current has a decaying dc component that lasts for three cycles. However, the extracted rms voltage almost reaches the steady state within two cycles. Also note that depending on the instant at which a fault occurs, the first cycle of the rms calculation may contain a part of unfaulted voltage/current samples. Therefore it is always expected that the first cycle will have transient data. The measured admittance and the tripping time calculated in each cycle are shown in Fig. 6.20(c) and (d) respectively. It can be seen from Fig. 6.20(d) that the tripping time reduces slightly with the change of measured admittance. However, in 125

150 Chapter 6: Experimental results these tests, the fault duration is intentionally maintained for a defined time period to observe the relay parameters for few cycles. Otherwise, if relay is allowed to clear the fault, the change of parameters for few cycles cannot be investigated. For example, consider the test results shown in Fig The relay issues the trip command around 43ms after the fault. If relay clears the fault, approximately two cycles of fault period can be only seen. Fig Current and voltage during a SLG fault 126

151 Chapter 6: Experimental results Fig Values of relay parameters during a SLG fault 127

152 Chapter 6: Experimental results Fig shows another test results captured during a SLG fault. This test has a different fault inception angle compared to the previous test. The decaying dc component of the current can be seen clearly from the figure while harmonics associated with the faulted voltage can be also visible. The calculated fundamental components of current and voltage using LabVIEW FFT blocks and the measured admittance and the tripping time calculation in each cycle are shown in Fig The results show that the relay tripping time is slightly higher at the beginning of the fault while it reduces with the time. The FFT has successfully extracted the voltage fundamental in the presence of harmonics. Fig Faulted current and voltage during a SLG fault 128

153 Chapter 6: Experimental results Fig Values of calculated relay parameters during a SLG fault 129

154 Chapter 6: Experimental results According to the results, the FFT can extract fundamental in the presence of harmonics and signal noises. However, the FFT is not immune to the decaying dc component. Therefore, the calculated tripping time of the relay can vary slightly. The amount of error on fundamental extraction depends on the network configuration since the magnitude and time constant of the decaying dc component can vary. To improve the accuracy of the relay tripping time, one of the methods mentioned in Section 3.7, Chapter 3 can be used to accurately calculate the fundamental components. However, the speed of the calculation and burden on the processor should be considered when selecting a particular algorithm. 6.6 Summary Several experimental tests are carried out at different fault locations and different system configurations. The results demonstrate that the ITA relay follows the inverse time characteristic curve as designed based on the measured admittance. The relay response time closely matches the calculated results. The source voltage and source impedance do not affect the relay operation. Moreover, the relay can respond in the same manner with different fault current levels in the feeder. The fault resistance and infeed may cause delay in the relay operation. The decaying dc component can also change the relay tripping time slightly during the transient period of the fault. 130

155 Chapter 7: Conclusions and recommendations In this chapter, the general conclusions of the thesis and recommendations for future research are presented. 7.1 Conclusions The general conclusions of the thesis are: (1) The connections of DGs or microgrids to a distribution network are gaining importance with the increase of electrical power requirements and environmental concerns. However, these DG connections can create challenging protection issues. It is identified that new protection strategies are required to overcome these challenges. (2) Protective devices based on current sensing are usually used to detect the faults in distribution networks. The connections of DGs change the fault current level and fault current direction. Moreover, if DGs are connected through intermittent sources, the fault current contribution from DGs cannot be exactly identified to set the tripping parameters of current sensing protective devices. Furthermore, protection with converter interfaced DGs in islanded operation is difficult with current sensing protective devices due to the lower fault current levels. Unless viable solutions can be found for the protection issues, DGs have to be disconnected from the grid after a fault, thereby affecting reliability. 131

156 Chapter 7: Conclusions and recommendations (3) DGs should be kept connected to the unfaulted segments of a network once a fault occurs since they can supply the loads in unfaulted segments either in grid connected or islanded mode operations. This results in maximising DG benefits. However, the protection system should be capable of isolating the faulted segment and providing adequate protection for the islanded network. (4) A new relay - the ITA relay - has been designed to provide protection to a DG connected distribution network or microgrid. The relay is employed with different fault detection elements, such as earth element, phase element and directional element, to respond to different faults. The ITA relay senses both the current and voltage at the relay location. This results in effective fault detection irrespective of the network fault current level. The relay is also capable of isolating the faulted segment, thus allowing the unfaulted segments to operate either in grid connected or islanded mode operations supplying the load demand. In the islanded operation, these relays provide the adequate protection for the network even when current limited converters are connected. The fault resistance and infeed may affect the relay operation. However, the system configuration and relay settings will determine the maximum fault resistance for which the relay can effectively detect faults. (5) The reclosing can be identified as one of the major protection challenges in a DG connected distribution network. The DGs are usually disconnected before performing the reclosing in the network. The fault arc will not extinguish if DGs remain connected to the network. Also, the coordination of the recloser with DGs is a challenging task. Therefore, new control and protection strategies are required to overcome these problems. 132

157 Chapter 7: Conclusions and recommendations (6) In this thesis, a novel control strategy based on fold back current control has been proposed for a converter interfaced DG. The control strategy has the ability to extinguish the fault arc, to restore the system quickly if possible, and to perform the reclosing in a converter interfaced DG connected network. In this proposition, DGs are not required to disconnect immediately after a fault. The DGs are allowed to supply the load demand either in grid connected or islanded mode operations. The self extinction of arc is achieved by reducing the DG output current to a small value, while automatic system restoration is obtained if DG power generation is sufficient to supply the load demand. The coordination between reclosers and DGs in the network is obtained by appropriately defining a sequence of operations with suitable time delays. (7) In this research, implementation of protection and control strategies for DG connected distribution networks without communication is considered since the solutions based on communication are still expensive. However, with the development and spread out of cheaper communication methods in future, communication can be used to improve the protection of the DG connected distribution networks. The use of communication between DGs and protective devices in the network either centralized or decentralized manner can improve the efficacy of protection and control strategies. For example, if a relay has the DG connectivity information, it can then detect the faults minimising the error coming from infeed effect. Also, system reclosing can be performed much faster having the information of DG status. 133

158 Chapter 7: Conclusions and recommendations 7.2 Recommendations for future research The scope for future research are: Consideration of rotary type DGs for protection In this research, all the DGs are considered as converter interfaced DGs for the protection analysis. However both rotary and converter type DGs can be considered for fault detection, fault isolation and system restoration in future research Fold back type current control for rotary type DGs The fold back current control strategy is implemented for a converter interfaced DG in this research. This fold back strategy cannot be easily implemented for rotary type DGs which will continuously supply large fault currents. Therefore, suitable fold back type control strategies are required for these rotary DGs to achieve fast arc extinction and system restoration and will be the subject of future research and development The effect of single phase converters In this research, DGs connected to the network through three phase converters are considered. However, there may be single phase converter connected DGs which can cause system unbalance. The effect of these single phase converters on network protection and system restoration can be considered in future works. 134

159 References [1] J. C. Gómez and M. M. Morcos, "Coordination of Voltage Sag and Overcurrent Protection in DG Systems," IEEE Transactions on Power Delivery, vol. 20, pp , [2] S. M. Brahma and A. A. Girgis, "Development of Adaptive Protection Scheme for Distribution Systems With High Penetration of Distributed Generation," IEEE Transactions on Power Delivery, vol. 19, pp , [3] A. M. Azmy and I. Erlich, "Impact of Distributed Generation on the Stability of Electrical Power Systems," IEEE Power Engineering Society General Meeting, vol. 2, pp , [4] J. Driesen, P. Vermeyen, and R. Belmans, "Protection Issues in Microgrids with Multiple Distributed Generation Units," Power Conversion Conference - Nagoya, pp , [5] A. Girgis and S. Brahma, "Effect of Distributed Generation on Protective Device Coordination in Distribution System," Large Engineering Systems Conference on Power Engineering, pp , [6] F. A. Viawan, D. Karlsson, A. Sannino, and J. Daalder, "Protection Scheme for Meshed Distribution Systems with High Penetration of Distributed Generation," Power Systems Conference: Advanced Metering, Protection, Control, Communication, and Distributed Resources, pp , [7] M. Brucoli, T. C. Green, and J. D. F. McDonald, "Modelling and Analysis of Fault Behaviour of Inverter Microgrids to Aid Future Fault Detection," IEEE International Conference on System of Systems Engineering, pp. 1-6, [8] H. Nikkhajoei and R. H. Lasseter, "Microgrid Protection," IEEE Power Engineering Society General Meeting, pp. 1-6, [9] A. D. T. Le, M. A. Kashem, M. Negnevitsky, and G. Ledwich, "Distributed Generation Diversity Level for Optimal Investment Planning," Australian Universities Power Engineering Conference (AUPEC) Melbourne, Australia, [10] M. A. Kashem and G. Ledwich, "Anti-islanding Protection and Islanding Operation of Grid-connected Hydropower Distributed Generation," International Journal of Global Energy Issues, vol. 24, pp ,

160 [11] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac, Embedded Generation, The Institute of Electrical Engineers, London, [12] J. A. P. Lopes, C. L. Moreira, and A. G. Madureira, "Defining Control Strategies for MicroGrids Islanded Operation," IEEE Transactions on Power Systems, vol. 21, pp , [13] N. Perera, A. D. Rajapakse, and T. E. Buchholzer, "Isolation of Faults in Distribution Networks with Distributed Generators," IEEE Transactions on Power Delivery, vol. 23, pp , [14] S. P. Chowdhury, S. Chowdhury, C. F. Ten, and P. A. Crossley, "Islanding Protection of Distribution Systems with Distributed Generators A Comprehensive Survey Report," IEEE Power and Energy Society General Meeting, [15] IEEE Std. 1547, "IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems," [16] H. H. Zeineldin, E. F. El-Saadany, and M. M. A. Salama, "Distributed Generation Micro-Grid Operation: Control and Protection," IEEE Power Systems Conference: Advanced Metering, Protection, Control, Communication, and Distributed Resources, 2006, pp [17] H. Wan, K. P. Wong, and C. Y. Chung, "Multi-agent Application in Protection Coordination of Power System with Distributed Generations," IEEE Power and Energy Society General Meeting, [18] G. Dalke and et. al, "Application of Islanding Protection for Industrial and Commercial Generators An IEEE Industrial Application Society Working Group Report," 59th Annual Conference for Protective Relay Engineers, 2006, pp [19] L. K. Kumpulainen and K. T. Kauhaniemi, "Analysis of the Impact of Distributed Generation on Automatic Reclosing," Power Systems Conference and Exposition, vol. 1, pp , [20] J. C. M. Vieira, W. Freitas, X. Wilsun, and A. Morelato, "Performance of Frequency Relays for Distributed Generation Protection," IEEE Transactions on Power Delivery, vol. 21, pp ,

161 [21] J. C. M. Vieira, W. Freitas, X. Wilsun, and A. Morelato, "Efficient Coordination of ROCOF and Frequency Relays for Distributed Generation Protection by Using the Application Region," IEEE Transactions on Power Delivery, vol. 21, pp , [22] J. C. M. Vieira, W. Freitas, and D. Salles, "Characteristics of Voltage Relays for Embedded Synchronous Generators Protection," IET Generation, Transmission & Distribution, vol. 1, pp , [23] S. Jang and K.-H. Kim, "An Islanding Detection Method for Distributed Generations Using Voltage Unbalance and Total Harmonic Distortion of Current," IEEE Transactions on Power Delivery, vol. 19, pp , [24] L. Wang, H. Cheung, A. Hamlyn, C. Yang, and R. Cheung, "Network- Integrated Protection and Control Strategy for Power Distribution Systems," IEEE Large Engineering Systems Conference on Power Engineering, [25] M. Brucoli and T. C. Green, "Fault Response of Inverter Dominated Microgrids," Department of Electrical and Electronic Engineering, Imperial College. [26] R. M. Tumilty, M. Brucoli, G. M. Burt, and T. C. Green, "Approaches to Network Protection for Inverter Dominated Electrical Distribution Systems," The 3rd IET International Conference on Power Electronics, Machines and Drives, 2006, pp [27] H. Al-Nasseri, M. A. Redfern, and R. O Gorman, "Protecting Micro-grid Systems Containing Solid-State Converter Generation," International Conference on Future Power Systems, pp. 1-5, [28] T. Gallery, L. Martinez, and D. Klopotan, "Impact of Distributed Generation on Distribution Network Protection," ESBI Engineering & Facility Management, Ireland. [29] M. Baran and I. El-Markabi, "Adaptive Over Current Protection for Distribution Feeders with Distributed Generators," IEEE Power Systems Conference and Exposition, vol. 2, pp , [30] J. A. Martinez and J. Martin-Arnedo, "Impact of Distributed Generation on Distribution Protection and Power Quality " IEEE Power & Energy Society General Meeting,

162 [31] Y. Lu, L. Hua, J. Wu, G. Wu, and G. Xu, "A Study on Effect of Dispersed Generator Capacity on Power System Protection," IEEE Power Engineering Society General Meeting, pp. 1-6, [32] M. T. Doyle, "Reviewing the Impacts of Distributed Generation on Distribution System Protection," IEEE Power Engineering Society Summer Meeting, vol. 1, pp , [33] S. M. Brahma and A. A. Girgis, "Microprocessor-based Reclosing to Coordinate Fuse and Recloser in a System with High Penetration of Distributed Generation," IEEE Power Engineering Society Winter Meeting, vol. 1, pp , [34] J. C. Tan, P. G. McLaren, R. P. Jayasinghe, and P. L. Wilson, "Software Model for Inverse Time Overcurrent Relays Incorporating IEC and IEEE Standard Curves," IEEE Canadian Conference on Electrical and Computer Engineering, vol. 1, pp , [35] A. C. Enr ıquez and E. V. Mart ınez, "Enhanced Time Overcurrent Coordination," Electrical Power System Research, vol. 76, pp , [36] G. Benmouyal and et. al, "IEEE Standard Inverse-Time Characteristic Equations for Overcurrent Relays," IEEE Transactions on Power Delivery, vol. 14, pp , [37] Working Group of the Relay Standards Subcommittee, "IEEE Standard Inverse-Time Characteristic Equations for Overcurrent Relays," The Institute of Electrical and Electronics Engineers, Inc., pp. 1-13, [38] J. Hubertus, J. Mooney, and G. Alexander, "Application Considerations for Distance Relays on Impedance-Grounded Systems," 61st Annual Conference for Protective Relay Engineers, 2008, pp [39] H. H. Zeineldin, E. F. El-Saadany, and M. M. A. Salama, "Protective Relay Coordination for Micro-grid Operation Using Particle Swarm Optimization," Large Engineering Systems Conference on Power Engineering, pp , [40] F. A. Viawan and M. Reza, "The Impact of Synchronous Distributed Generation on Voltage Dip and Overcurrent Protection Coordination," International Conference on Future Power Systems, pp. 1-6,

163 [41] J.C. Gomez and M.M. Morcos, "Coordinating Overcurrent Protection and Voltage Sag in Distributed Generation Systems," IEEE Power Engineering Review, vol. 22, pp , [42] M. E. Baran and I. El-Markaby, "Fault Analysis on Distribution Feeders with Distributed Generators," IEEE Transactions on Power Systems, vol. 20, pp [43] R.H.Lasseter, "MicroGrids," IEEE Power Engineering Society Winter Meeting, vol. 1, pp , [44] N. Jayawarna, N. Jenkins, M. Barnes, M. Lorentzou, S. Papthanassiou, and N. Hatziagyriou, "Safety Analysis of a MicroGrid," International Conference on Future Power Systems, pp. 1-7, [45] H. Al-Nasseri, M. A. Redfern, and F. Li, "A Voltage Based Protection for Micro-grids Containing Power Electronic Converters," IEEE Power Engineering Society General Meeting, [46] W. E. Feero, D. C. Dawson, and J. Stevens, "Consortium for Electric Reliability Technology Solutions, White Paper on Protection Issues of the MicroGrid Concept," [47] K. J. Zoric, N. D. Jejina, and M. B. Djuric, "Secondary Arc Faults Detection and Determine Arc Extinction Time on Overhead Lines Using Neural Network " Electric Machines and Power Systems, vol. 28, pp , [48] G. Ban, L. Prikler, and G. Banfai, "Testing EHV Secondary Arcs," IEEE Porto Power Tech Conference, [49] L. Kumpulainen and K. Kauhaniemi, "Distributed Generation and Reclosing Coordination," Nordic Distribution and Asset Management Conference, [50] T. Seegers and et. al, "Impact of Distributed Resources on Distribution Relay Protection," IEEE power system relay committee [51] A. T. Johns, R. K. Aggarwal, and Y. H. Song, "Improved Techniques for Modelling Fault Arcs on Faulted EHV Transmission Systems," IET Generation, Transmission and Distribution, vol. 141, pp , [52] M. Kizilcay and P. L. Seta, "Digital Simulation of Fault Arcs in Medium Voltage Distribution Networks," 15th Power System Computation Conference,

164 [53] H. Cheung, A. Hamlyn, L. Wang, C. Yang, and R. Cheung, "Investigations of Impacts of Distributed Generations on Feeder Protections," IEEE Power & Energy Society General Meeting, [54] V. V. Terzija and H.-J. Koglin, "On the Modeling of Long Arc in Still Air and Arc Resistance Calculation," IEEE Transactions on Power Delivery, vol. 19, pp [55] D. Laverty, D. J. Morrow, and T. Littler, "Internet Based Loss-of-mains Detection for Distributed Generation," 42nd International Universities Power Engineering Conference, 2007, pp [56] R. Benato and R. Caldon, "Application of PLC for the Control and the Protection of Future Distribution Networks," IEEE International Symposium on Power Line Communications and Its Applications, 2007, pp [57] X. Wu, Y. Lu, and J. Du, "Wide-area Current Protection for Distribution Feeders with Distributed Generators," Third International Conference on Electric Utility Deregulation and Restructuring and Power Technologies, 2008, pp [58] S. Shaoqun, Z. Yongli, H. Min, and Y. Hong, "Multiagent and WAN based Adaptive Coordinated Protection System," Asia and Pacific Transmission and Distribution Conference and Exhibition, 2005 pp. 1-6 [59] S. A. M. Javadian, M. R. Haghifam, and N. Rezaei, "A Fault Location and Protection Scheme for Distribution Systems in Presence of DG Using MLP Neural Networks " IEEE Power & Energy Society General Meeting, [60] M. M. Eissa, "A New Digital Feed Circuit Protection Using Directional Element " IEEE Transactions on Power Delivery, vol. 24, pp , [61] B. Fleming, "Negative Sequence Impedance Directional Element," Schweitzer Engineering Laboratories, Inc., Pullman, WA USA [62] P. G. McLaren, G. W. Swift, Z. Zhang, E. Dirks, R. P. Jayasinghe, and I. Fernando, "A New Directional Element for Numerical Distance Relays," IEEE Transactions on Power Delivery, vol. 10, pp [63] Y. S. Cho, C. K. Lee, G. Jang, and H. J. Lee, "An Innovative Decaying DC Component Estimation Algorithm for Digital Relaying," IEEE Transactions on Power Delivery, vol. 24, pp ,

165 [64] J. C. Gu and S. L. Yu, "Removal of DC Offset in Current and Voltage Signals Using a Novel Fourier Filter Algorithm " IEEE Transactions on Power Delivery, vol. 15, pp , [65] J. F. M. Argüelles, M. A. Z. Arrieta, J. L. Domínguez, B. L. Jaurrieta, and M. S. Benito, "A New Method for Decaying DC Offset Removal for Digital Protective Relays," Electric Power Systems Research, vol. 76, pp , [66] J. M. Gers and E. J. Holmes, Protection of Electricity Distribution Networks,, 2nd ed., The Institution of Electrical Engineers, [67] M. Dewadasa, A. Ghosh, and G. Ledwich, "Line Protection in Inverter Supplied Networks," Australian Universities Power Engineering Conference (AUPEC), [68] W. D. Stevenson, Elements of Power System Analysis, 4th ed., McGraw-Hill Companies, [69] IEEE Distribution system analysis subcommittee, "Radial Test Feeders," available online at [70] A. Ghosh, A. K. Jindal, and A. Joshi, "Inverter Control Using Output Feedback for Power Compensating Devices," IEEE Asia-Pacific Region-10 Conference TENCON, [71] G. Kaur and M. Vaziri, "Effect of Distributed Generation (DG) Interconnection on Protection of Distribution Feeders," IEEE Power Engineering Society General Meeting, [72] L. Prikler, M. Kizilcay, G. Ban, and P. Handl, "Improved Secondary Arc Models based on Identification of Arc Parameters from Staged Fault Test Records," 14th Power System Computation Conference, [73] A. I. Megaheda, H. M. Jabrb, F. M. Abouelenina, and M. A. Elbakryb, "Arc Characteristics and a Single-pole Auto-reclosure Scheme for Alexandria HV Transmission System," Electric Power Systems Research, vol. 76, pp , [74] "NI PXI Channel High-Power General-Purpose Relay Switch," available online at 141

166

167 Publications arising from the thesis Conference papers (1) M. Dewadasa, A. Ghosh and G. Ledwich, Line protection in inverter supplied networks, Australian Universities Power Engineering Conference, Sydney, Australia, 2008 (2) M. Dewadasa, A. Ghosh and G. Ledwich, Distance protection solution for a converter controlled microgrid, National Power Systems Conference, Mumbai, India, 2008 (3) M. Dewadasa, A. Ghosh and G. Ledwich, Foldback current control for a DG to achieve fast arc extinction in a distribution network, 3rd biennial Smart Systems Postgraduate Student Conference, Queensland University of Technology, Brisbane, 2009 (4) M. Dewadasa, A. Ghosh and G. Ledwich, An inverse time admittance relay for fault detection in distribution networks containing DGs, IEEE Asia- Pacific Region-10 Conference TENCON 09, Singapore, (5) M. Dewadasa, R. Majumder, A. Ghosh and G. Ledwich, Control and protection of a microgrid with converter interfaced micro sources, Third International Conference on Power Systems, Kharagpur, India, Journal papers (1) M. Dewadasa, A. Ghosh and G. Ledwich, Fold back current control and admittance protection scheme for a distribution network containing DGs, IET Generation, Transmission and Distribution, Vol. 4, pp ,

168

169 Appendix-A Positive sequence admittance seen by ITA relay The faulted network for single line to ground (SLG) fault shown in Fig. A.1 is considered. The relay location is denoted by node R while fault point is at node k. The fault impedance is represented by Z f. The admittance between the relay and fault point is denoted by Y RK. The voltage and current seen by the relay in faulted phase A are taken as V Ra and I fa respectively. Negative and zero sequence source voltages are also considered assuming the source is unsymmetrical during the fault. Moreover, the zero, positive and negative sequence Thevenin admittances between the relay point and faulted point are considered as Y Rk0, Y Rk1 and Y Rk2 respectively. The sequence network for this SLG fault can be represented as shown in Fig. A.1(b). (a) Fig. A.1 SLG fault representation (a) faulted network (b) sequence network (b) 145

170 The sequence voltages at the relay point can be expressed as, V V V Ra1 Ra2 Ra0 = v 1 = v = v ( I Ra 1 / YRK1) ( IRa2 / YRK1) ( I / Y ) Ra0 RK 0 (A.1) Voltage seen by relay is given by V Ra = V + V + V Ra1 Ra2 Ra0 (A.2) Assuming fault resistance to be zero, v 1 + v 2 + v 0 = 3 Z f I fa1 = 0 (A.3) From (A.1), (A.2) and (A.3), ( Y ) + ( I / Y ) ( I Y ) V Ra = I Ra 1 / RK1 Ra2 RK1 + Ra0 / RK 0 (A.4) Also it can be written, YRK1 = YRK 2 IRa1 + IRa2 + IRa0 = IRa (A.5) Using (A.4) and (A.5), positive sequence measured admittance Y RK1 can be written as, Y Rk1 = I R a + I Y Y V Ra0 Ra RK1 RK 0 1 (A.6) where I Ra is the line fault current through the relay while I Ra0 is the zero sequence fault current seen by the relay and V Ra is the faulted phase rms voltage. 146

171 Appendix-B Converter structure and control (A). Converter structure The three phase structure of the converter which is used in PSCAD simulation is shown in Fig. B.1. It contains three single phase H-bridge converters that are supplied a common dc bus containing the DG. Three single-phase transformers are connected to the three converters to provide isolation and voltage boosting. In this figure, L f is the leakage reactance of the transformer, R f is the transformer losses and L 0 is the output inductance of the DG-converter system. The filter capacitor C f is used to bypass the switching harmonics. The advantage of the converter structure shown in Fig. B.1 is that each phase of the converter can be controlled independently. Fig. B.1 The converter structure (B) Converter control The converter has two control loops. It can either operate in voltage control or current control modes depending on the network operating conditions. During the normal operating condition, the converter is controlled using output feedback of the 147

172 converter output voltage maintaining the rated voltages at the terminals. The voltage reference is chosen such that the power demand can be met. The converter switches into output feedback current control mode during a fault in the network. In this case, the reference current is chosen to have a magnitude that is twice the rated current of the converter. Also in current control mode, the magnitude of the output current can be changed with time according to the user requirement. However, the defined currents are only injected in faulted phases. The voltage control mode of the converter is explained in detail below. The equivalent circuit of one phase of the converter is shown in Fig. B.2. In this, u V dc represents the converter output voltage, where u = ± 1. The main aim of the converter control is to generate u. Fig. B.2 Equivalent circuit of one phase of the converter given as, From the circuit of Fig. B.2, the state space description of the system can be x & = Ax + Bu c (B.1) where u c is the continuous time control input, based on which the switching function u is determined. The discrete-time equivalent of (B.1) can be given by x( k + 1) = Fx( k) + Gu ( k) (B.2) c 148

173 Let the output of the system given in (B.2) be v cf. The reference for this voltage is given by the instantaneous peak and phase angle of each phase. Let this be denoted by v*. The input-output relationship of the system in (B.2) can be written as, v cf ( z) u ( z) c = M ( z N( z 1 1 ) ) (B.3) The control is computed from u ( z) = c S( z R( z 1 1 ) ) * { v ( z) v ( z) } cf (B.4) The closed-loop transfer function of the system is then v 1 1 cf M ( z ) S( z ) * ( z) = v ( z) N( z ) R( z ) + M ( z ) S( z ) (B.5) The coefficients of the polynomials S and R can be chosen based on a pole placement strategy [B.1]. Once u c is computed from (B.5), the switching function u can be generated as if uc elseif > h then uc < h u = + 1 then u = 1 (B.6) where h is a small number. The control in (B. 4) is computed based on the reference voltage v and the feedback of the capacitor voltage v cf. The reference voltages are given by * va = Vnsin( ωt) * o vb = Vnsin( ωt 120 ) * o vc = Vnsin( ωt ) (B.7) where V n is the peak voltage magnitude. 149

174 A similar control strategy is also used for current control. The control in (B.4) is calculated based on the output current i oa and reference currents. The reference currents for faulted phases are chosen based on the converter rating. To illustrate how the switching pulses are selected in each mode of operation, Fig. B.3 is considered. The converter in both voltage and current modes is operated in output feedback pole shift voltage/current control mode to select R1, S1, R2 and S2 [B.2] as explained above. Each phase is controlled according to its output voltage (v cf ) or current (i oa ). These signals are sampled at 10 µs. The output sampled signals are then used in the discrete-time output feedback controllers shown in this figure. The controller mode change operations are also indicated in this figure. The switching pulses are generated either in voltage control mode or current control mode depending on the selected mode of operations. Fig. B.3 Single line diagram of the converter structure and control 150

175 [B.1] A. Ghosh, Performance study of two different compensating devices in a custom power park, Proc. IEE Generation, Transmission & Distribution, Vol. 152, No. 4, pp , [B.2] A. Ghosh, K. Jindal and A. Joshi, Inverter control using output feedback for power compensating devices, Proc. IEEE Asia-Pacific Region-10 Conference TENCON, Bangalore, 2003, pp

176 152

177 Appendix-C LabVIEW Program Fig. C.1 Measured admittance calculation on LabVIEW 153