- Email: [email protected]
A novel fault detection method for VSC-HVDC transmission system of offshore wind farm
Electrical Power and Energy Systems 73 (2015) 475–483
Contents lists available at ScienceDirect
Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes
A novel fault detection method for VSC-HVDC transmission system of offshore wind farm S.H. Ashrafi Niaki ⇑, H. Kazemi Karegar 1, M. Ghalei Monfared Faculty of Electrical and Computer Engineering, Shahid Beheshti University, Evin Street, Tehran 1983963113, Iran
a r t i c l e
i n f o
Article history: Received 4 January 2014 Received in revised form 18 January 2015 Accepted 25 April 2015
Keywords: Fault detection Offshore wind farm VSC-HVDC system DC fault Cable sheath
a b s t r a c t VSC-HVDC transmission system is going to become the most economical way of power delivery for large and remote offshore wind farms. An accurate and fast fault detection method is necessary to protect sensitive devices of these systems and maintain uninterrupted power delivery. This paper investigates an innovative technique for recognizing DC zone faults including HVDC cable faults and unbalancing of DC capacitor bank. Sheath voltage is presented as a novel criterion for detecting abnormal situations in the system. Transient voltage of cable sheath and Wavelet transform are used to identify different types of DC faults. Extensive simulation examples are performed on EMTDC–PSCAD platform and post-processed using MATLAB. The results illustrate that the proposed technique gives a robust performance and can be applied to protection scheme of offshore wind farms. Ó 2015 Elsevier Ltd. All rights reserved.
Introduction Wind energy has been presenting itself as an attractive renewable energy and is going to become highly penetrative energy source in the electrical power systems. In addition to land-based wind farms, Offshore Wind Farms (OWFs) have great potential to supply the main grid with the electrical power [1,2]. In fact, the electrical power which can be received from the OWFs is more than the land-based wind farms because of high winds at sea. From power transmission’s point of view, there are two types of connections, HVDC and AC connection. Economic matters play major role in selecting the best way to transmit the electrical power from offshore substation to onshore substation. Research studies have been mentioning two factors as the principal factors in this case: 1. Distance of the OWF from the onshore substation (the main grid) 2. Nominal active power of the OWF [3,4]. According to these studies, for large and remote OWFs, the HVDC transmission sounds more interesting way to deliver the electrical power to the main grid. Traditional HVDC systems use Line-Commutated Converters (LCCs) that can reverse power transmission direction by changing voltage polarity. From two previous decades, an interesting replacement for the LCCs has been presented that can alter the power transmission direction by reversing current direction called ⇑ Corresponding author. Tel.: +98 21 29904136; fax: +98 21 22431804. 1
E-mail address: amir_ashrafi[email protected] (S.H. Ashrafi Niaki). Tel.: +98 21 29904136; fax: +98 21 22431804.
http://dx.doi.org/10.1016/j.ijepes.2015.04.014 0142-0615/Ó 2015 Elsevier Ltd. All rights reserved.
Voltage Source Converter (VSC). The VSC-HVDC systems have some great advantages. They can remarkably control active and reactive power independently. Therefore, the system does not need reactive compensators which are necessary for the LCCs. Moreover, it becomes possible to use of inexpensive Cross Linked Polyethylene (XLPE) cables in the HVDC systems [5–7]. An undetected fault may have a catastrophic impact on the VSC-HVDC systems which connect the OWF to the main grid. It is vital to apply a fast and reliable fault detection method with regard to sensitivity and vulnerability of these systems. Most of studies presented in this field are related to the protection scheme of the system rather than focusing on DC fault detection. In [8], a protection scheme for HVDC line including cable has been proposed based on differential current methods. When a DC fault occurs in the VSC-HVDC transmission systems, their Insulated Gate Bipolar Transistors (IGBTs) which are not capable of clearing fault, will be by-passed and anti-parallel diodes conduct to feed the fault current. Economical fast DC switches have been used to isolate faulted DC line [9]. A protection strategy in [10] utilizes voltage chopper and overcurrent limitation controller to suppress overvoltage and overcurrent of the DC side of the VSCs. On the other hand, some faults can happen to converter itself and power electronic devices like IGBT misfiring. The behavior of overcurrent and differential relays has been investigated under different internal fault conditions [11]. Also, protection of high voltage capacitors has been studied in [12]. For protection of multi-terminal DC distribution systems, an overcurrent-based scheme has been suggested [13].
476
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
A DC fault current can be cleared by AC circuit breakers on the AC side of the VSC. This is called handshaking method. A handshaking method for locating and isolating faulted DC line without telecommunication is analyzed in [14]. Majority of the suggested protection schemes for the VSC-HVDC systems which are current-based methods are reviewed in [15]. Wavelet transform is also proposed to detect DC faults in multi-terminal VSC-HVDC system [16]. None of aforementioned protective studies used detailed XLPE cable model for HVDC connection in their studies. A protection design against DC faults for multi-terminal DC wind farm is offered and fault analysis is discussed in [17,18]. Although some useful discussions are represented in [18], the protection scheme is just carried out for a small-scale system. However, lack of a fast and reliable method for identifying DC faults and distinguishing them from AC faults can be perceived in these systems. In this paper, an innovative strategy based on using the sheath voltage is presented to diagnose wide range of DC faults consisting of positive cable to negative cable faults, positive cable ground faults, negative cable ground faults and DC capacitor unbalancing. Under normal operating condition, the magnitude of the sheath voltage is equal to zero while this magnitude increases under faulty conditions. Transient voltage of the cable sheath as new criterion and Wavelet analysis are used to detect the faulty situations. Simulation of a 400-MW OWF connected to the main grid via the XLPE HVDC cable is carried out. Obtained results show that the proposed method is effective and reliable. This paper is organized as follows: in section ‘VSC-HVDC transmission system for OWF and different faults scenarios’, model of studied system and possible AC and DC faults are explained. Section ‘The proposed concept’ describes the concept of the proposed method. In section ‘The performance evaluation’, performance of the proposed method is evaluated by the simulation results. Finally, conclusions are discussed in section ‘Conclusions’.
substation. Pulse Width Modulation (PWM) or Spatial Vector Pulse Width Modulation (SVPWM) techniques are used in order to control IGBTs in the VSCs. Two parameters can be controlled independently by applying the PWM technique to the VSCs. Those are the magnitude and the phase angle of the AC voltage generated on the AC side of the VSC. DC link power and AC system voltage are adjusted by phase shift and magnitude control respectively. In fact, DC link voltage magnitude and reactive power which are directly related to the DC link power and the AC system voltage will be controlled. Topologies of converters can be three-level neutral point clamped VSC or two-level VSC [19,20]. These kinds of the topologies have similar behaviors under faulted conditions [21]. A two-level topology is used in the simulated system. There are other topologies such as Modular Multilevel Converters (MMC) that have different behaviors in case of faults and can be more tolerant of faulted conditions [22,23]. Finally, the third section is the AC main grid. Usually, there is a power transformer to adjust the voltage magnitude of the inverter for connecting to the main grid which is called converter transformer. Faults may happen in each mentioned sections of the system. Types of faults may happen in the first and the third sections are AC faults. On the other hand, DC faults may occur in the second section. Moreover, converters inner faults that are related to power electronic devices can happen to the VSCs like the IGBT misfiring. The Power electronic devices have their own protections which are significantly faster than transmission system protections [24,25]. Detection of the converter inner faults is not included in this paper.
VSC-HVDC transmission system for OWF and different faults scenarios
1. 2. 3. 4. 5.
The area of the OWF transmission system can be categorized into three sections as shown in Fig. 1: (1) wind generation section (OWF), (2) HVDC transmission section, (3) the main grid section. The first section contains high power wind turbines at sea. Type of the wind generator used in the OWF can be Permanent Magnet Synchronous Generator (PMSG), Doubly Fed Induction Generator (DFIG) or Squirrel Cage Induction Generator (SCIG). The OWF transmits its power by the HVDC transmission system. The second section is the HVDC transmission section consisting of marine substation, long submarine HVDC cable and onshore
DC faults The DC faults in the second section can be classified as follows: Pole to pole fault (Positive cable to negative cable fault). Positive cable ground fault. Negative cable ground fault. Unbalancing of positive pole capacitor bank. Unbalancing of negative pole capacitor bank.
The submarine DC cables are almost immune to the pole to pole faults as insulation and conduit set the positive and negative cables apart. On the other hand, the cable ground faults are more common. This kind of the DC faults is due to the insulation failure (insulation breakdown). Cable aging and exposure to wet environment can be the reasons of the insulation failure. In this case, fault current will loop through grounding points of the system. In order
Fig. 1. Single line diagram of the simulated system.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
to mitigate imbalance between the positive and negative poles and provide a reference voltage in bipolar HVDC configurations, the midpoint of the DC capacitor is usually grounded [26,27]. Also, the sheath of the cable is usually grounded at both ends at substations. Therefore, fault currents can return through the sheath of the cable that is connected to the ground and other grounding points that can be including the midpoint of the DC capacitor and the neutral-ground link of the converter transformer [21]. The insulation failure can also occur with the DC capacitors in the capacitor bank of each pole. In addition, an improper installation is another reason for losing some parts of the capacitor bank that will lead to the capacitor unbalancing. AC faults The AC faults in the first section usually are submarine cable faults that can lead to loss of some wind turbines. In this paper, wind turbines are connected together through 33-kV AC cables. Also, the third section is liable to common AC line faults consisting of one, two or three phase to ground faults. The proposed concept In order to present a robust fault detection method, it is vital to have enough knowledge about the VSC-HVDC system and its behavior under different circumstances. It is useful to investigate transient state of the system when different faults happen. Detection of DC cable faults When a DC cable fault (pole to pole fault or cable ground fault) happens, DC current suddenly increases and contains two main
Fig. 2. Contribution of different currents under cable fault condition.
477
current: anti-parallel diodes currents (ivsc) and capacitor discharging current (ic) as shown in Fig. 2. The IGBTs will be blocked rapidly and anti-parallel diodes start to feed the fault current from the AC side. Paths of the fault current when the positive cable ground fault happens are depicted in Fig. 3. Noticeable point is that the first contributor of the fault current is ic because of small time constant of the DC capacitor [18]. XLPE HVDC cable as shown in Fig. 3 contains main conductor and two conductive layers: the sheath and armor separated by insulation. Under normal operating condition, no current passes through the sheath. On the contrary, under fault circumstance, the fault current passes through the cable sheath as can be seen from Fig. 3. As a result, the cable sheath encounters transient overvoltages. Sensors for measuring sheath voltage can be installed at each substation where the sheath connected to the ground electrode. The DC capacitor discharging leads to very fast and severe rising of the sheath voltage. Therefore, the rise time of the sheath voltage signal as a DC fault occurs, is smaller than the rise time of the sheath voltage signal as an AC fault happens. In other words, the signal of the sheath voltage in case of the DC faults has higher frequency spectrum than the same signal in case of the AC faults. This discrepancy can show itself in frequency content of the signal. In order to make this distinction clear, one of the best signal processing methods called Wavelet transform can be used.
Detection of DC capacitor unbalancing Fig. 4 shows combination of four equal capacitors that is considered as a DC capacitor bank for each positive and negative pole in the simulations. The major duty of the DC capacitor bank is voltage regulation. Effects of an unbalanced capacitor bank on the system are similar to the DC cable fault because in both cases, DC capacitor voltage and current are changed. Also, the frequency contents of voltage and current signals are similar in these two cases. However, there is delicate difference between them. For instance, if one out of four capacitor of the positive pole on the inverter side in Fig. 4 losses, a serious voltage dip will occur at this pole and the direction of the current passes through the cable sheath in Fig. 3 will be reversed. Consequently, the peak of sheath voltage gets a negative magnitude, while it is positive for the case of the positive cable ground fault. This feature is used for distinguishing the DC cable fault from the unbalanced DC capacitor fault. Generally, when a positive cable ground fault happens, sheath voltages of positive cable measured on both rectifier and inverter sides have the peaks with positive sign. On the contrary, signs of
Fig. 3. Current paths under positive cable ground fault condition.
478
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
Fig. 4. Current path when a negative capacitor unbalancing happens on the inverter side.
voltage peaks for negative cable sheath on two sides are negative for the case of the negative cable ground fault. When a DC capacitor unbalancing occurs, sign of the signal peak on rectifier side is different from inverter side. For example, the voltage peak of the negative cable sheath for the unbalancing case of the negative DC capacitor on inverter side has positive sign, while it is negative for the signal measured on rectifier side. The reverse is true for the case of positive capacitor unbalancing. Current path under the negative capacitor unbalancing condition on the inverter side is illustrated in Fig. 4. Wavelet transform Wavelet transform is a robust signal processing method which decomposes signal over dilated and translated wavelet functions. One of significant advantages of the Wavelet transform is that it can hold local time and frequency together. The Wavelet transform of signal x(t) at time s and scale s can be calculated by following equation [28,29]:
1 wtðs; sÞ ¼ pffiffi s
Z
þ1
xðtÞ W
1
ts dt s
ð1Þ
With W⁄ the complex conjugated of wavelet function W which itself can be represented as a family functions based on parameters of s and s as follow:
1 s
ts s 0; s
Wðs;sÞ ðtÞ ¼ pffiffi W
s2R
ð2Þ
In fact, this mathematical transform is a developed short-time Fourier transform. Different functions of wavelet can be appropriate for detecting frequency components of different signals based on correlation between them. The proposed algorithm The suggested algorithm is shown in Fig. 5. A certain band called detection band is considered for the sheath voltage. If the sheath voltage exceeds this band, an abnormal operating condition
Fig. 5. Proposed algorithm.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
479
(AC or DC faults) will be detected. Then, the sheath voltages of positive and negative cables will be analyzed by detail coefficients of Wavelet transform. A DC fault will be detected provided that the wavelet coefficients exceed a considered threshold called detection threshold, otherwise AC fault will be diagnosed. Finally, if the proposed algorithm diagnoses that a DC fault has occurred, the peak of the sheath voltage will be checked on both rectifier and inverter sides in order to distinguish the DC cable ground fault from the unbalanced DC capacitor fault. The performance evaluation Case study of an OWF connected to 400-kV main grid by VSC-HVDC system and XLPE cables is shown in Fig. 1. Nominal power of the OWF is 400 MW and the length of ±150-kV XLPE cable is 200 km. A 3.6 MW SCIG is considered as the generator of the wind turbines which its parameters are given in Appendix A. It would be extremely hard and time consuming to model each wind turbine solitarily, therefore four 100-MW aggregated wind turbines are used to simulate on EMTDC–PSCAD software. Fig. 6 depicts the wind turbine used in the 100-MW aggregated model. Frequency-dependent line model is used in the simulations for the HVDC cables. This accurate model is basically distributed RLC traveling wave model, which incorporates the frequency dependence of all parameters. The model is based on the theory proposed in [30,31]. Table 1 contains the parameters of the 150-kV XLPE cable which are given in Appendix B. In order to evaluate the performance of the proposed method, varieties of simulations are performed. Detail coefficients of DB10 wavelet at first level of decomposition is selected to analyze the signals because the DB10 wavelet has significant competency to detect high frequency components of signals. It has shown its excellent proficiency in previous studies on DC networks [32,33]. Normally, sheath voltage is equal to zero, but it is may affected by noises. Consequently, it is not exactly equal to zero. Therefore, limitation must be higher than zero to not trip under normal operating condition. Usually, it should be selected several times higher than noise magnitude. On the contrary, the limitations must not exceed magnitude of faults with high impedances (corresponding to minimum sheath voltage under fault condition). Selecting appropriate limitations for the proposed method depends on different factors like characteristics and configuration of grid, rating of voltage and current for HVDC system, values of reactors and capacitors. Therefore, it should be determined through simulations under different conditions. With investigating of an extensive set of simulations, it is found that 0.05 and 0.05 are appropriate values for upper and lower limitation of the detection band. Also the detection threshold is set to 0.02. Different AC and DC faults with resistance of 0 X, 10 X and 100 X are applied to the case study. HVDC cable and AC grid faults
Fig. 7. Sheath voltages for three-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and positive cable to ground fault with Rf = 0 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 8. Wavelet coefficients of the sheath voltages for the case of Fig. 7, (a) positive cable (b) negative cable.
Figs. 7 and 8 depict voltages of the cable sheath measured on rectifier side and their wavelet analyses when three phase to
Fig. 6. Wind turbine model used in the simulations.
ground fault happens to 400-kV AC grid at t = 2 s and the cable ground fault happens in the middle of the positive cable length at t = 4 s. At time t = 2 s, voltage magnitudes of the sheaths exceed the detection band, but their wavelet coefficients do not break the detection threshold. Therefore, the proposed algorithm properly announces occurrence of AC fault. On the other side, at time t = 4 s, the sheath voltages exceed the detection band and the wavelet coefficients of the positive sheath voltage break the detection threshold. Thus, the proposed algorithm properly detects occurrence of positive cable ground fault at this time. Exact time of fault
480
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
detection shown in Fig. 8 is equal to 0.3 ms when the sheath voltage reaches to the detection threshold. Totally, the proposed algorithm takes less than 1 ms to detect fault conditions through the simulation results which demonstrates fast performance of the method. From current rising’s point of view, it is found from the simulation results that the most serious AC fault in the main grid is two-phase fault. Therefore, two-phase fault with resistance of 0 X and the positive cable ground fault with resistance of 100 X can be the most difficult condition to distinguish the DC fault from the AC fault. Figs. 9 and 10 show the results for this case. As it can be seen from Fig. 10, the fault detection time for this case is equal
to 0.7 ms. Detection time depends on rise time of sheath voltage signal which itself depends on discharging time of DC capacitor. Larger value of fault resistance leads to more discharging time of capacitor and consequently the detection time will be increased. Figs. 7–10 illustrate correct performance of the proposed method. Also, proper selectivity of the proposed method with the negative cable ground fault is shown in Figs. 11 and 12. As expected, the sheath voltage of the negative cable has higher peak comparing to the sheath voltage of the positive cable. In Fig. 12(b), wavelet analysis depicts exceeding of the detection threshold by the negative cable sheath.
Fig. 9. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and positive cable to ground fault with Rf = 100 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 11. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and negative cable to ground fault with Rf = 10 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 10. Wavelet coefficients of the sheath voltages for the case of Fig. 9, (a) positive cable (b) negative cable.
Fig. 12. Wavelet coefficients of the sheath voltages for the case of Fig. 11, (a) positive cable (b) negative cable.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
Figs. 13 and 14 show the results for the case of pole to pole fault. The sheath voltage for both positive and negative cables at time t = 4 s goes beyond the detection band in Fig. 13. Also, the detection threshold is exceeded by both positive and negative cables in Fig. 14. Therefore, the proposed method diagnoses the fault condition as a pole to pole fault.
DC capacitor and AC cable faults If occurrence of the DC capacitor unbalancing is possible, the sheath voltages on both rectifier and inverter sides must be
481
checked to distinguish this kind of fault from the DC cable faults. On the other hand, AC faults can happen to the wind farm side through 33-kV AC cables. In order to test the performance of the suggested algorithm under DC capacitor unbalancing and AC cable fault conditions, one out of four capacitors of the positive pole on rectifier side is shorted and one-phase cable to ground fault on the wind farm side is applied. The results for these types of faults are depicted in Figs. 14 and 15 which show reliability of the proposed method. According to the suggested algorithm, AC and DC faults can be distinguished easily. For the DC fault, the Wavelet coefficients exceed the detection threshold in Fig. 16(a) and (c) at t = 8 s, while the coefficients do not pass this threshold in Fig. 16(b) and (d). The last stage of the algorithm is that which kind of the DC faults occurred, the cable ground fault or the DC capacitor unbalancing. Sign of the peak of the sheath voltage at t = 8 s in Fig. 15(a) is
Fig. 13. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and pole to pole fault with Rf = 100 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 14. Wavelet coefficients of the sheath voltages for the case of Fig. 13, (a) positive cable (b) negative cable.
Fig. 15. Sheath voltages for one-phase cable fault just before 33-kV collector bus with Rf = 0 X (t = 2 s) and positive capacitor unbalancing (t = 8 s), (a) rectifier side, positive cable (b) rectifier side, negative cable (c) inverter side, positive cable (d) inverter side, negative cable.
482
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483 Table 1 Parameters of 150-kV XLPE HVDC cable.
Conductor Insulator 1 Sheath Insulator 2 Armor Insulator 3
Outer Radius (m)
Resistivity (X m)
0.022 0.039 0.044 0.049 0.054 0.059
1.72e8
Relative permittivity
Relative permeability 1
2.3 2.2e7
1 2.3
1.8e7
100 2.25
The results illustrate that the suggested technique is capable of fast detection of different DC faults including the DC cable faults and the DC capacitor unbalancing and distinguishing them from the AC faults. Appendix A Parameters of SCIG used in the simulations Nominal active power: 3.6 MW, Terminal voltage: 0.69 kV, Frequency: 50 Hz, Stator resistance: 0.00488 p.u., Rotor resistance: 0.00549 p.u., Stator leakage reactance: 0.09241 p.u., Rotor leakage reactance: 0.09955 p.u., Magnetizing reactance: 4 p.u. Appendix B See Table 1. References
Fig. 16. Wavelet coefficients of the sheath voltages for the case of Fig. 15, (a) rectifier side, positive cable (b) rectifier side, negative cable (c) inverter side, positive cable (d) inverter side, negative cable.
negative while it is positive in Fig. 15(c). Therefore, the proposed algorithm will correctly announce the state of the positive DC capacitor unbalancing.
Conclusions Fast and reliable fault detection scheme is vital for the OWFs which transmit power using the VSC-HVDC system. A novel strategy to detect the DC faults in the VSC-HVDC transmission system for the OWF using transient voltage of the cable sheath is presented. Voltage signals of the sheaths for both the negative and positive cable and their Wavelet analyses were the criteria to identify the DC faults. Voltage of the cable sheath has demonstrated itself as straightforward and appropriate criterion to diagnose abnormal operating conditions in the results.
[1] Ackermann T. Transmission systems for offshore wind farms. IEEE Power Eng 2002:23–7. [2] Xu L, Andersen BR. Grid connection of large offshore wind farms using HVDC. Wind Energy 2006:371–82. [3] Kirby NM, Xu L, Luckett M, Siepman W. HVDC transmission for large offshore wind farms. IEE Power Eng 2002:135–41. [4] Morton AB, Cowdroy S, Hill JRA, Halliday M, Nicholson GD. AC or DC economics of grid connection design for offshore wind farms. In: Proc IEE int conf AC DC power transmission; 2006. p. 236–40. [5] Schettler F, Huang H, Christl N. HVDC transmission systems using voltage sourced converters—design and applications. In: Proc IEEE power eng soc summer meeting; 2000. p. 715–20. [6] Reidy A, Watson R. Comparison of VSC based HVDC and HVAC interconnections to a large offshore wind farm. In: IEEE power eng soc gen meeting; 2005. [7] Bahrman MP, Johansson JG, Nilsson BA. Voltage source converter transmission technologies: the right fit for the application. In: Proc IEEE power eng soc summer meeting; 2003. [8] Takeda H, Ayakawa H, Tsumenaga M, Sanpei M. New protection method for HVDC lines including cables. IEEE Trans Power Del 1995:2035–9. [9] Tang L, Ooi BT. Protection of VSC-multi-terminal HVDC against DC faults. In: Proc IEEE 33rd annual power electronics specialists conf; 2002. p. 719–24. [10] Liu H, Xu Z, Huang Y. Study of protection strategy for VSC based HVDC system. In: Proc transmission and distribution conference and exposition IEEE PES; 2003. [11] Darwish HA, Taalab AMI, Rahman MA. Performance of HVDC converter protection during internal faults. In: Proc IEEE power engineering and society, general meeting; 2006. p. 18–22. [12] Nian M, Yinhong L, Xianzhong D, Jiang Y, Chuang F. Study on high voltage capacitor unbalance protection in HVDC projects. In: Proc power and energy engineering conference, APPEEC; 2009. [13] Baran ME, Mahajan NR. Overcurrent protection on voltage-source-converterbased multiterminal DC distribution systems. IEEE Trans Power Del 2007:406–12. [14] Tang L, Ooi BT. Locating and isolating DC faults in multi-terminal DC systems. IEEE Trans Power Del 2007:1877–84. [15] Candelaria J, Park Jae-Do. VSC-HVDC system protection: a review of current methods. In: Power Systems Conference and Exposition (PSCE); 2011. [16] De Kerf K, Srivastava K, Reza M, Bekaert D, Cole S, Van Hertem D, et al. Wavelet-based protection strategy for DC faults in multi-terminal VSC HVDC systems. Gener Transm Distrib, IET 2011:496–503. [17] Yang J, Fletcher JE, O’Reilly J. Protection scheme design for meshed VSC-HVDC transmission systems for large-scale wind farms. In: The 9th international conference on AC and DC power transmission, IET; 2010.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483 [18] Yang J, Fletcher JE, O’Reilly J. Multiterminal dc wind farm collection grid internal fault analysis and protection scheme design. IEEE Trans Power Del 2010:2308–18. [19] Andersen BR, Xu L, Horton PJ, Cartwright P. Topologies for VSC transmission. Power Eng J 2002:142–50. [20] Rodriguez J, Lai JS, Peng FZ. Multilevel inverters: a survey of topologies, controls, and applications. IEEE Tran Ind Electr 2002:724–38. [21] Yang J, Fletcher JE, O’Reilly J. Short-circuit and ground fault analysis and location in VSC-based DC network cables. IEEE Trans Ind Electr 2012:3827–37. [22] Marquardt R. Modular multilevel converter: an universal concept for HVDCnetworks and extended DC-Bus-applications. In: Proc IEEE IPEC, Sapporo, Japan; 2010. p. 502–7. [23] Arrillaga J, Liu YH, Watson NR, Murray NJ. Self-commutating converters for high power applications. Wiley; 2009. [24] Lu B, Sharma SK. A literature review of IGBT fault diagnostic and protection methods for power inverters. IEEE Trans Ind Appl 2009:1770–7. [25] Wang F, Lai R, Yuan X, Luo F, Burgos R, Boroyevich D. Failure-mode analysis and protection of three-level neutral-point-clamped PWM voltage source converters. IEEE Trans Ind Appl 2010;46:866–74. [26] Behrman MP, Johnson BK. The ABCs of HVDC transmission technologies. IEEE Power Energy Mag 2007;5:32–44.
483
[27] Baran ME, Mahajan NR. DC distribution for industrial systems: opportunities and challenges. IEEE Trans Ind Appl 2003;39:1596–601. [28] Mallat SG. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans Pattern Anal Mach Int 1989;11:674–93. [29] Robertson D, Camps O, Mayer J, Gish W. Wavelets and electromagnetic power system transients. IEEE Trans Power Del 1996;11:1050–8. [30] Marti J. Accurate modeling of frequency dependent transmission lines in electromagnetic transients simulation. IEEE Trans Power Ap Syst 1982;101:147–55. [31] Morched A, Gustavsen B, Tartibi M. A universal model for accurate calculation of electromagnetic transients on overhead lines and underground cables. IEEE Trans Power Del 1999;14:1032–8. [32] Weilin L, Luo M, Monti A, Ponci F. Wavelet based method for fault detection in medium voltage DC shipboard power systems. In: Instrumentation and measurement technology conference; 2012. [33] Zhao J, Hu W, Gu X, Yang P. The application in the fault of high voltage electric power measurement system based on wavelet analysis with the improved threshold algorithm. In: 3rd International workshop on intelligent systems and applications; 2011.
Contents lists available at ScienceDirect
Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes
A novel fault detection method for VSC-HVDC transmission system of offshore wind farm S.H. Ashrafi Niaki ⇑, H. Kazemi Karegar 1, M. Ghalei Monfared Faculty of Electrical and Computer Engineering, Shahid Beheshti University, Evin Street, Tehran 1983963113, Iran
a r t i c l e
i n f o
Article history: Received 4 January 2014 Received in revised form 18 January 2015 Accepted 25 April 2015
Keywords: Fault detection Offshore wind farm VSC-HVDC system DC fault Cable sheath
a b s t r a c t VSC-HVDC transmission system is going to become the most economical way of power delivery for large and remote offshore wind farms. An accurate and fast fault detection method is necessary to protect sensitive devices of these systems and maintain uninterrupted power delivery. This paper investigates an innovative technique for recognizing DC zone faults including HVDC cable faults and unbalancing of DC capacitor bank. Sheath voltage is presented as a novel criterion for detecting abnormal situations in the system. Transient voltage of cable sheath and Wavelet transform are used to identify different types of DC faults. Extensive simulation examples are performed on EMTDC–PSCAD platform and post-processed using MATLAB. The results illustrate that the proposed technique gives a robust performance and can be applied to protection scheme of offshore wind farms. Ó 2015 Elsevier Ltd. All rights reserved.
Introduction Wind energy has been presenting itself as an attractive renewable energy and is going to become highly penetrative energy source in the electrical power systems. In addition to land-based wind farms, Offshore Wind Farms (OWFs) have great potential to supply the main grid with the electrical power [1,2]. In fact, the electrical power which can be received from the OWFs is more than the land-based wind farms because of high winds at sea. From power transmission’s point of view, there are two types of connections, HVDC and AC connection. Economic matters play major role in selecting the best way to transmit the electrical power from offshore substation to onshore substation. Research studies have been mentioning two factors as the principal factors in this case: 1. Distance of the OWF from the onshore substation (the main grid) 2. Nominal active power of the OWF [3,4]. According to these studies, for large and remote OWFs, the HVDC transmission sounds more interesting way to deliver the electrical power to the main grid. Traditional HVDC systems use Line-Commutated Converters (LCCs) that can reverse power transmission direction by changing voltage polarity. From two previous decades, an interesting replacement for the LCCs has been presented that can alter the power transmission direction by reversing current direction called ⇑ Corresponding author. Tel.: +98 21 29904136; fax: +98 21 22431804. 1
E-mail address: amir_ashrafi[email protected] (S.H. Ashrafi Niaki). Tel.: +98 21 29904136; fax: +98 21 22431804.
http://dx.doi.org/10.1016/j.ijepes.2015.04.014 0142-0615/Ó 2015 Elsevier Ltd. All rights reserved.
Voltage Source Converter (VSC). The VSC-HVDC systems have some great advantages. They can remarkably control active and reactive power independently. Therefore, the system does not need reactive compensators which are necessary for the LCCs. Moreover, it becomes possible to use of inexpensive Cross Linked Polyethylene (XLPE) cables in the HVDC systems [5–7]. An undetected fault may have a catastrophic impact on the VSC-HVDC systems which connect the OWF to the main grid. It is vital to apply a fast and reliable fault detection method with regard to sensitivity and vulnerability of these systems. Most of studies presented in this field are related to the protection scheme of the system rather than focusing on DC fault detection. In [8], a protection scheme for HVDC line including cable has been proposed based on differential current methods. When a DC fault occurs in the VSC-HVDC transmission systems, their Insulated Gate Bipolar Transistors (IGBTs) which are not capable of clearing fault, will be by-passed and anti-parallel diodes conduct to feed the fault current. Economical fast DC switches have been used to isolate faulted DC line [9]. A protection strategy in [10] utilizes voltage chopper and overcurrent limitation controller to suppress overvoltage and overcurrent of the DC side of the VSCs. On the other hand, some faults can happen to converter itself and power electronic devices like IGBT misfiring. The behavior of overcurrent and differential relays has been investigated under different internal fault conditions [11]. Also, protection of high voltage capacitors has been studied in [12]. For protection of multi-terminal DC distribution systems, an overcurrent-based scheme has been suggested [13].
476
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
A DC fault current can be cleared by AC circuit breakers on the AC side of the VSC. This is called handshaking method. A handshaking method for locating and isolating faulted DC line without telecommunication is analyzed in [14]. Majority of the suggested protection schemes for the VSC-HVDC systems which are current-based methods are reviewed in [15]. Wavelet transform is also proposed to detect DC faults in multi-terminal VSC-HVDC system [16]. None of aforementioned protective studies used detailed XLPE cable model for HVDC connection in their studies. A protection design against DC faults for multi-terminal DC wind farm is offered and fault analysis is discussed in [17,18]. Although some useful discussions are represented in [18], the protection scheme is just carried out for a small-scale system. However, lack of a fast and reliable method for identifying DC faults and distinguishing them from AC faults can be perceived in these systems. In this paper, an innovative strategy based on using the sheath voltage is presented to diagnose wide range of DC faults consisting of positive cable to negative cable faults, positive cable ground faults, negative cable ground faults and DC capacitor unbalancing. Under normal operating condition, the magnitude of the sheath voltage is equal to zero while this magnitude increases under faulty conditions. Transient voltage of the cable sheath as new criterion and Wavelet analysis are used to detect the faulty situations. Simulation of a 400-MW OWF connected to the main grid via the XLPE HVDC cable is carried out. Obtained results show that the proposed method is effective and reliable. This paper is organized as follows: in section ‘VSC-HVDC transmission system for OWF and different faults scenarios’, model of studied system and possible AC and DC faults are explained. Section ‘The proposed concept’ describes the concept of the proposed method. In section ‘The performance evaluation’, performance of the proposed method is evaluated by the simulation results. Finally, conclusions are discussed in section ‘Conclusions’.
substation. Pulse Width Modulation (PWM) or Spatial Vector Pulse Width Modulation (SVPWM) techniques are used in order to control IGBTs in the VSCs. Two parameters can be controlled independently by applying the PWM technique to the VSCs. Those are the magnitude and the phase angle of the AC voltage generated on the AC side of the VSC. DC link power and AC system voltage are adjusted by phase shift and magnitude control respectively. In fact, DC link voltage magnitude and reactive power which are directly related to the DC link power and the AC system voltage will be controlled. Topologies of converters can be three-level neutral point clamped VSC or two-level VSC [19,20]. These kinds of the topologies have similar behaviors under faulted conditions [21]. A two-level topology is used in the simulated system. There are other topologies such as Modular Multilevel Converters (MMC) that have different behaviors in case of faults and can be more tolerant of faulted conditions [22,23]. Finally, the third section is the AC main grid. Usually, there is a power transformer to adjust the voltage magnitude of the inverter for connecting to the main grid which is called converter transformer. Faults may happen in each mentioned sections of the system. Types of faults may happen in the first and the third sections are AC faults. On the other hand, DC faults may occur in the second section. Moreover, converters inner faults that are related to power electronic devices can happen to the VSCs like the IGBT misfiring. The Power electronic devices have their own protections which are significantly faster than transmission system protections [24,25]. Detection of the converter inner faults is not included in this paper.
VSC-HVDC transmission system for OWF and different faults scenarios
1. 2. 3. 4. 5.
The area of the OWF transmission system can be categorized into three sections as shown in Fig. 1: (1) wind generation section (OWF), (2) HVDC transmission section, (3) the main grid section. The first section contains high power wind turbines at sea. Type of the wind generator used in the OWF can be Permanent Magnet Synchronous Generator (PMSG), Doubly Fed Induction Generator (DFIG) or Squirrel Cage Induction Generator (SCIG). The OWF transmits its power by the HVDC transmission system. The second section is the HVDC transmission section consisting of marine substation, long submarine HVDC cable and onshore
DC faults The DC faults in the second section can be classified as follows: Pole to pole fault (Positive cable to negative cable fault). Positive cable ground fault. Negative cable ground fault. Unbalancing of positive pole capacitor bank. Unbalancing of negative pole capacitor bank.
The submarine DC cables are almost immune to the pole to pole faults as insulation and conduit set the positive and negative cables apart. On the other hand, the cable ground faults are more common. This kind of the DC faults is due to the insulation failure (insulation breakdown). Cable aging and exposure to wet environment can be the reasons of the insulation failure. In this case, fault current will loop through grounding points of the system. In order
Fig. 1. Single line diagram of the simulated system.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
to mitigate imbalance between the positive and negative poles and provide a reference voltage in bipolar HVDC configurations, the midpoint of the DC capacitor is usually grounded [26,27]. Also, the sheath of the cable is usually grounded at both ends at substations. Therefore, fault currents can return through the sheath of the cable that is connected to the ground and other grounding points that can be including the midpoint of the DC capacitor and the neutral-ground link of the converter transformer [21]. The insulation failure can also occur with the DC capacitors in the capacitor bank of each pole. In addition, an improper installation is another reason for losing some parts of the capacitor bank that will lead to the capacitor unbalancing. AC faults The AC faults in the first section usually are submarine cable faults that can lead to loss of some wind turbines. In this paper, wind turbines are connected together through 33-kV AC cables. Also, the third section is liable to common AC line faults consisting of one, two or three phase to ground faults. The proposed concept In order to present a robust fault detection method, it is vital to have enough knowledge about the VSC-HVDC system and its behavior under different circumstances. It is useful to investigate transient state of the system when different faults happen. Detection of DC cable faults When a DC cable fault (pole to pole fault or cable ground fault) happens, DC current suddenly increases and contains two main
Fig. 2. Contribution of different currents under cable fault condition.
477
current: anti-parallel diodes currents (ivsc) and capacitor discharging current (ic) as shown in Fig. 2. The IGBTs will be blocked rapidly and anti-parallel diodes start to feed the fault current from the AC side. Paths of the fault current when the positive cable ground fault happens are depicted in Fig. 3. Noticeable point is that the first contributor of the fault current is ic because of small time constant of the DC capacitor [18]. XLPE HVDC cable as shown in Fig. 3 contains main conductor and two conductive layers: the sheath and armor separated by insulation. Under normal operating condition, no current passes through the sheath. On the contrary, under fault circumstance, the fault current passes through the cable sheath as can be seen from Fig. 3. As a result, the cable sheath encounters transient overvoltages. Sensors for measuring sheath voltage can be installed at each substation where the sheath connected to the ground electrode. The DC capacitor discharging leads to very fast and severe rising of the sheath voltage. Therefore, the rise time of the sheath voltage signal as a DC fault occurs, is smaller than the rise time of the sheath voltage signal as an AC fault happens. In other words, the signal of the sheath voltage in case of the DC faults has higher frequency spectrum than the same signal in case of the AC faults. This discrepancy can show itself in frequency content of the signal. In order to make this distinction clear, one of the best signal processing methods called Wavelet transform can be used.
Detection of DC capacitor unbalancing Fig. 4 shows combination of four equal capacitors that is considered as a DC capacitor bank for each positive and negative pole in the simulations. The major duty of the DC capacitor bank is voltage regulation. Effects of an unbalanced capacitor bank on the system are similar to the DC cable fault because in both cases, DC capacitor voltage and current are changed. Also, the frequency contents of voltage and current signals are similar in these two cases. However, there is delicate difference between them. For instance, if one out of four capacitor of the positive pole on the inverter side in Fig. 4 losses, a serious voltage dip will occur at this pole and the direction of the current passes through the cable sheath in Fig. 3 will be reversed. Consequently, the peak of sheath voltage gets a negative magnitude, while it is positive for the case of the positive cable ground fault. This feature is used for distinguishing the DC cable fault from the unbalanced DC capacitor fault. Generally, when a positive cable ground fault happens, sheath voltages of positive cable measured on both rectifier and inverter sides have the peaks with positive sign. On the contrary, signs of
Fig. 3. Current paths under positive cable ground fault condition.
478
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
Fig. 4. Current path when a negative capacitor unbalancing happens on the inverter side.
voltage peaks for negative cable sheath on two sides are negative for the case of the negative cable ground fault. When a DC capacitor unbalancing occurs, sign of the signal peak on rectifier side is different from inverter side. For example, the voltage peak of the negative cable sheath for the unbalancing case of the negative DC capacitor on inverter side has positive sign, while it is negative for the signal measured on rectifier side. The reverse is true for the case of positive capacitor unbalancing. Current path under the negative capacitor unbalancing condition on the inverter side is illustrated in Fig. 4. Wavelet transform Wavelet transform is a robust signal processing method which decomposes signal over dilated and translated wavelet functions. One of significant advantages of the Wavelet transform is that it can hold local time and frequency together. The Wavelet transform of signal x(t) at time s and scale s can be calculated by following equation [28,29]:
1 wtðs; sÞ ¼ pffiffi s
Z
þ1
xðtÞ W
1
ts dt s
ð1Þ
With W⁄ the complex conjugated of wavelet function W which itself can be represented as a family functions based on parameters of s and s as follow:
1 s
ts s 0; s
Wðs;sÞ ðtÞ ¼ pffiffi W
s2R
ð2Þ
In fact, this mathematical transform is a developed short-time Fourier transform. Different functions of wavelet can be appropriate for detecting frequency components of different signals based on correlation between them. The proposed algorithm The suggested algorithm is shown in Fig. 5. A certain band called detection band is considered for the sheath voltage. If the sheath voltage exceeds this band, an abnormal operating condition
Fig. 5. Proposed algorithm.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
479
(AC or DC faults) will be detected. Then, the sheath voltages of positive and negative cables will be analyzed by detail coefficients of Wavelet transform. A DC fault will be detected provided that the wavelet coefficients exceed a considered threshold called detection threshold, otherwise AC fault will be diagnosed. Finally, if the proposed algorithm diagnoses that a DC fault has occurred, the peak of the sheath voltage will be checked on both rectifier and inverter sides in order to distinguish the DC cable ground fault from the unbalanced DC capacitor fault. The performance evaluation Case study of an OWF connected to 400-kV main grid by VSC-HVDC system and XLPE cables is shown in Fig. 1. Nominal power of the OWF is 400 MW and the length of ±150-kV XLPE cable is 200 km. A 3.6 MW SCIG is considered as the generator of the wind turbines which its parameters are given in Appendix A. It would be extremely hard and time consuming to model each wind turbine solitarily, therefore four 100-MW aggregated wind turbines are used to simulate on EMTDC–PSCAD software. Fig. 6 depicts the wind turbine used in the 100-MW aggregated model. Frequency-dependent line model is used in the simulations for the HVDC cables. This accurate model is basically distributed RLC traveling wave model, which incorporates the frequency dependence of all parameters. The model is based on the theory proposed in [30,31]. Table 1 contains the parameters of the 150-kV XLPE cable which are given in Appendix B. In order to evaluate the performance of the proposed method, varieties of simulations are performed. Detail coefficients of DB10 wavelet at first level of decomposition is selected to analyze the signals because the DB10 wavelet has significant competency to detect high frequency components of signals. It has shown its excellent proficiency in previous studies on DC networks [32,33]. Normally, sheath voltage is equal to zero, but it is may affected by noises. Consequently, it is not exactly equal to zero. Therefore, limitation must be higher than zero to not trip under normal operating condition. Usually, it should be selected several times higher than noise magnitude. On the contrary, the limitations must not exceed magnitude of faults with high impedances (corresponding to minimum sheath voltage under fault condition). Selecting appropriate limitations for the proposed method depends on different factors like characteristics and configuration of grid, rating of voltage and current for HVDC system, values of reactors and capacitors. Therefore, it should be determined through simulations under different conditions. With investigating of an extensive set of simulations, it is found that 0.05 and 0.05 are appropriate values for upper and lower limitation of the detection band. Also the detection threshold is set to 0.02. Different AC and DC faults with resistance of 0 X, 10 X and 100 X are applied to the case study. HVDC cable and AC grid faults
Fig. 7. Sheath voltages for three-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and positive cable to ground fault with Rf = 0 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 8. Wavelet coefficients of the sheath voltages for the case of Fig. 7, (a) positive cable (b) negative cable.
Figs. 7 and 8 depict voltages of the cable sheath measured on rectifier side and their wavelet analyses when three phase to
Fig. 6. Wind turbine model used in the simulations.
ground fault happens to 400-kV AC grid at t = 2 s and the cable ground fault happens in the middle of the positive cable length at t = 4 s. At time t = 2 s, voltage magnitudes of the sheaths exceed the detection band, but their wavelet coefficients do not break the detection threshold. Therefore, the proposed algorithm properly announces occurrence of AC fault. On the other side, at time t = 4 s, the sheath voltages exceed the detection band and the wavelet coefficients of the positive sheath voltage break the detection threshold. Thus, the proposed algorithm properly detects occurrence of positive cable ground fault at this time. Exact time of fault
480
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
detection shown in Fig. 8 is equal to 0.3 ms when the sheath voltage reaches to the detection threshold. Totally, the proposed algorithm takes less than 1 ms to detect fault conditions through the simulation results which demonstrates fast performance of the method. From current rising’s point of view, it is found from the simulation results that the most serious AC fault in the main grid is two-phase fault. Therefore, two-phase fault with resistance of 0 X and the positive cable ground fault with resistance of 100 X can be the most difficult condition to distinguish the DC fault from the AC fault. Figs. 9 and 10 show the results for this case. As it can be seen from Fig. 10, the fault detection time for this case is equal
to 0.7 ms. Detection time depends on rise time of sheath voltage signal which itself depends on discharging time of DC capacitor. Larger value of fault resistance leads to more discharging time of capacitor and consequently the detection time will be increased. Figs. 7–10 illustrate correct performance of the proposed method. Also, proper selectivity of the proposed method with the negative cable ground fault is shown in Figs. 11 and 12. As expected, the sheath voltage of the negative cable has higher peak comparing to the sheath voltage of the positive cable. In Fig. 12(b), wavelet analysis depicts exceeding of the detection threshold by the negative cable sheath.
Fig. 9. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and positive cable to ground fault with Rf = 100 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 11. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and negative cable to ground fault with Rf = 10 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 10. Wavelet coefficients of the sheath voltages for the case of Fig. 9, (a) positive cable (b) negative cable.
Fig. 12. Wavelet coefficients of the sheath voltages for the case of Fig. 11, (a) positive cable (b) negative cable.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483
Figs. 13 and 14 show the results for the case of pole to pole fault. The sheath voltage for both positive and negative cables at time t = 4 s goes beyond the detection band in Fig. 13. Also, the detection threshold is exceeded by both positive and negative cables in Fig. 14. Therefore, the proposed method diagnoses the fault condition as a pole to pole fault.
DC capacitor and AC cable faults If occurrence of the DC capacitor unbalancing is possible, the sheath voltages on both rectifier and inverter sides must be
481
checked to distinguish this kind of fault from the DC cable faults. On the other hand, AC faults can happen to the wind farm side through 33-kV AC cables. In order to test the performance of the suggested algorithm under DC capacitor unbalancing and AC cable fault conditions, one out of four capacitors of the positive pole on rectifier side is shorted and one-phase cable to ground fault on the wind farm side is applied. The results for these types of faults are depicted in Figs. 14 and 15 which show reliability of the proposed method. According to the suggested algorithm, AC and DC faults can be distinguished easily. For the DC fault, the Wavelet coefficients exceed the detection threshold in Fig. 16(a) and (c) at t = 8 s, while the coefficients do not pass this threshold in Fig. 16(b) and (d). The last stage of the algorithm is that which kind of the DC faults occurred, the cable ground fault or the DC capacitor unbalancing. Sign of the peak of the sheath voltage at t = 8 s in Fig. 15(a) is
Fig. 13. Sheath voltages for two-phase to ground fault in 400 kV main grid with Rf = 0 X (t = 2 s) and pole to pole fault with Rf = 100 X (t = 4 s), (a) positive cable (b) negative cable.
Fig. 14. Wavelet coefficients of the sheath voltages for the case of Fig. 13, (a) positive cable (b) negative cable.
Fig. 15. Sheath voltages for one-phase cable fault just before 33-kV collector bus with Rf = 0 X (t = 2 s) and positive capacitor unbalancing (t = 8 s), (a) rectifier side, positive cable (b) rectifier side, negative cable (c) inverter side, positive cable (d) inverter side, negative cable.
482
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483 Table 1 Parameters of 150-kV XLPE HVDC cable.
Conductor Insulator 1 Sheath Insulator 2 Armor Insulator 3
Outer Radius (m)
Resistivity (X m)
0.022 0.039 0.044 0.049 0.054 0.059
1.72e8
Relative permittivity
Relative permeability 1
2.3 2.2e7
1 2.3
1.8e7
100 2.25
The results illustrate that the suggested technique is capable of fast detection of different DC faults including the DC cable faults and the DC capacitor unbalancing and distinguishing them from the AC faults. Appendix A Parameters of SCIG used in the simulations Nominal active power: 3.6 MW, Terminal voltage: 0.69 kV, Frequency: 50 Hz, Stator resistance: 0.00488 p.u., Rotor resistance: 0.00549 p.u., Stator leakage reactance: 0.09241 p.u., Rotor leakage reactance: 0.09955 p.u., Magnetizing reactance: 4 p.u. Appendix B See Table 1. References
Fig. 16. Wavelet coefficients of the sheath voltages for the case of Fig. 15, (a) rectifier side, positive cable (b) rectifier side, negative cable (c) inverter side, positive cable (d) inverter side, negative cable.
negative while it is positive in Fig. 15(c). Therefore, the proposed algorithm will correctly announce the state of the positive DC capacitor unbalancing.
Conclusions Fast and reliable fault detection scheme is vital for the OWFs which transmit power using the VSC-HVDC system. A novel strategy to detect the DC faults in the VSC-HVDC transmission system for the OWF using transient voltage of the cable sheath is presented. Voltage signals of the sheaths for both the negative and positive cable and their Wavelet analyses were the criteria to identify the DC faults. Voltage of the cable sheath has demonstrated itself as straightforward and appropriate criterion to diagnose abnormal operating conditions in the results.
[1] Ackermann T. Transmission systems for offshore wind farms. IEEE Power Eng 2002:23–7. [2] Xu L, Andersen BR. Grid connection of large offshore wind farms using HVDC. Wind Energy 2006:371–82. [3] Kirby NM, Xu L, Luckett M, Siepman W. HVDC transmission for large offshore wind farms. IEE Power Eng 2002:135–41. [4] Morton AB, Cowdroy S, Hill JRA, Halliday M, Nicholson GD. AC or DC economics of grid connection design for offshore wind farms. In: Proc IEE int conf AC DC power transmission; 2006. p. 236–40. [5] Schettler F, Huang H, Christl N. HVDC transmission systems using voltage sourced converters—design and applications. In: Proc IEEE power eng soc summer meeting; 2000. p. 715–20. [6] Reidy A, Watson R. Comparison of VSC based HVDC and HVAC interconnections to a large offshore wind farm. In: IEEE power eng soc gen meeting; 2005. [7] Bahrman MP, Johansson JG, Nilsson BA. Voltage source converter transmission technologies: the right fit for the application. In: Proc IEEE power eng soc summer meeting; 2003. [8] Takeda H, Ayakawa H, Tsumenaga M, Sanpei M. New protection method for HVDC lines including cables. IEEE Trans Power Del 1995:2035–9. [9] Tang L, Ooi BT. Protection of VSC-multi-terminal HVDC against DC faults. In: Proc IEEE 33rd annual power electronics specialists conf; 2002. p. 719–24. [10] Liu H, Xu Z, Huang Y. Study of protection strategy for VSC based HVDC system. In: Proc transmission and distribution conference and exposition IEEE PES; 2003. [11] Darwish HA, Taalab AMI, Rahman MA. Performance of HVDC converter protection during internal faults. In: Proc IEEE power engineering and society, general meeting; 2006. p. 18–22. [12] Nian M, Yinhong L, Xianzhong D, Jiang Y, Chuang F. Study on high voltage capacitor unbalance protection in HVDC projects. In: Proc power and energy engineering conference, APPEEC; 2009. [13] Baran ME, Mahajan NR. Overcurrent protection on voltage-source-converterbased multiterminal DC distribution systems. IEEE Trans Power Del 2007:406–12. [14] Tang L, Ooi BT. Locating and isolating DC faults in multi-terminal DC systems. IEEE Trans Power Del 2007:1877–84. [15] Candelaria J, Park Jae-Do. VSC-HVDC system protection: a review of current methods. In: Power Systems Conference and Exposition (PSCE); 2011. [16] De Kerf K, Srivastava K, Reza M, Bekaert D, Cole S, Van Hertem D, et al. Wavelet-based protection strategy for DC faults in multi-terminal VSC HVDC systems. Gener Transm Distrib, IET 2011:496–503. [17] Yang J, Fletcher JE, O’Reilly J. Protection scheme design for meshed VSC-HVDC transmission systems for large-scale wind farms. In: The 9th international conference on AC and DC power transmission, IET; 2010.
S.H. Ashrafi Niaki et al. / Electrical Power and Energy Systems 73 (2015) 475–483 [18] Yang J, Fletcher JE, O’Reilly J. Multiterminal dc wind farm collection grid internal fault analysis and protection scheme design. IEEE Trans Power Del 2010:2308–18. [19] Andersen BR, Xu L, Horton PJ, Cartwright P. Topologies for VSC transmission. Power Eng J 2002:142–50. [20] Rodriguez J, Lai JS, Peng FZ. Multilevel inverters: a survey of topologies, controls, and applications. IEEE Tran Ind Electr 2002:724–38. [21] Yang J, Fletcher JE, O’Reilly J. Short-circuit and ground fault analysis and location in VSC-based DC network cables. IEEE Trans Ind Electr 2012:3827–37. [22] Marquardt R. Modular multilevel converter: an universal concept for HVDCnetworks and extended DC-Bus-applications. In: Proc IEEE IPEC, Sapporo, Japan; 2010. p. 502–7. [23] Arrillaga J, Liu YH, Watson NR, Murray NJ. Self-commutating converters for high power applications. Wiley; 2009. [24] Lu B, Sharma SK. A literature review of IGBT fault diagnostic and protection methods for power inverters. IEEE Trans Ind Appl 2009:1770–7. [25] Wang F, Lai R, Yuan X, Luo F, Burgos R, Boroyevich D. Failure-mode analysis and protection of three-level neutral-point-clamped PWM voltage source converters. IEEE Trans Ind Appl 2010;46:866–74. [26] Behrman MP, Johnson BK. The ABCs of HVDC transmission technologies. IEEE Power Energy Mag 2007;5:32–44.
483
[27] Baran ME, Mahajan NR. DC distribution for industrial systems: opportunities and challenges. IEEE Trans Ind Appl 2003;39:1596–601. [28] Mallat SG. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans Pattern Anal Mach Int 1989;11:674–93. [29] Robertson D, Camps O, Mayer J, Gish W. Wavelets and electromagnetic power system transients. IEEE Trans Power Del 1996;11:1050–8. [30] Marti J. Accurate modeling of frequency dependent transmission lines in electromagnetic transients simulation. IEEE Trans Power Ap Syst 1982;101:147–55. [31] Morched A, Gustavsen B, Tartibi M. A universal model for accurate calculation of electromagnetic transients on overhead lines and underground cables. IEEE Trans Power Del 1999;14:1032–8. [32] Weilin L, Luo M, Monti A, Ponci F. Wavelet based method for fault detection in medium voltage DC shipboard power systems. In: Instrumentation and measurement technology conference; 2012. [33] Zhao J, Hu W, Gu X, Yang P. The application in the fault of high voltage electric power measurement system based on wavelet analysis with the improved threshold algorithm. In: 3rd International workshop on intelligent systems and applications; 2011.