Selective overcurrent protection for large MV installations

Transcription

1 Selective overcurrent protection for large MV installations FRANCESCO MUZI Department of Industrial and Information Engineering and Economics University of L Aquila Monteluco di Roio, L Aquila ITALY francesco.muzi@univaq.it Abstract: - The coordination of protection systems may prove very difficult at large installations, especially when many circuits are present and several hierarchical levels must be satisfied, involving both protection devices and automatic control systems. In this context, protection selectivity plays an important role as concerns the continuity level of power supply and the general power quality of the whole electrical system. The issue was investigated at an important underground laboratory of the Italian National Institute for Nuclear Physics (INFN). A preliminary analysis was performed paying special attention to the phenomenon of inrush currents, which was investigated also through laboratory tests consisting of a sequence of energizations of a power transformer. In order to replicate the behavior of a large installation, a digital simulator was implemented. Moreover, in this paper a case study is reported where this same simulator was used to analyze a portion of the plant under investigation, assuming as reference a number of significant scenarios involving both normal operating conditions and fault contingencies. The results of the experimental tests, the simulation analysis and possible strategies for a selective protection plan are finally examined and compared, with an aim to supply suggestions to correctly protect very large, sensitive electrical installations. Key-Words: - selectivity and protection, large electrical installations, power system protection. 1 Introduction The requirements to be met by a protective system are manifold. The most important are reliability, speed, and selectivity but also cost-effectiveness and sensitivity play important roles. Reliability refers to the probability that a protection system will operate correctly under certain conditions. In other words, the system must step in when required (dependability) and not step in when not required (security). Selectivity means to eliminate fault situations in the shortest possible time by disconnecting the minimum number of circuits in the electrical system. A high speed will not only facilitate the fulfillment of selectivity, but also reduce any fault-related damages, preserve system stability and positively influence the power quality of the electrical plant [20], [21]. The sensitivity of a protection system is linked to the capacity of identifying the minimum fault level in the protected zone. A further, particularly important concept, is the so-called coordination of protective devices, which involves a respect of selectivity. That is to say, the protection closest to the fault must step in first, while other, more upstream protections will act as backup in case downstream protections fail. Usually, this last concept also involves certain relay setting procedures [12], [16]. The coordination of protections can be achieved in different ways, corresponding to the different types of selectivity, whether amperometric or chronometric, energetic, logic, zonal, etc. Protective relays often measure the relevant quantities directly at their points of installation. Modern protection systems, however, are increasingly using communication systems and channels to exchange information measured and processed at different points in the electrical system [3]. In this case, the intelligence of the protection system can be distributed and/or centralized. In the most complex applications, digital systems offer a number of advantages [17], such as: - Protection, measurement and control; - Simple integration with other digital systems; - Friendly programmability; - Reduced load on CTs and VTs; - Fault recording; - Self-testing; - Adaptive protection; - No Maintenance; - Low costs. ISBN:

2 With reference to a power system, it is well known that a transformer energization generates a transient with peak currents that may be many times larger than the transformer rated current [8]. In case of large MV-cable installations, this phenomenon may be amplified by the presence of line capacitances that in this situation will rise to considerable values [10], [13]. As a matter of fact, when the total cable length exceeds dozens of kilometers at 20 kv, the first peaks of a transformer inrush current can cause a number of problems, among which a particularly worrying onset of the general overcurrent protection, which will create serious problems in the power supply continuity of a user s installation. The present paper was inspired by a study carried out for a real MV installation located in central Italy, namely at the Gran Sasso National Laboratory (LNGS) of the INFN. Due to its underground location, the MV distribution system, extending for about 50 kilometers, is made entirely of 20 kv cables. The considerations developed in this paper, although related to a specific installation, are general and can be applied to any large MV-cable installation. 2 The transformer inrush current An important phenomenon investigated in this study is the electromagnetic transient following the first transformer energization (inrush current), [15]. During the transient following the closure of a circuit breaker, a transformer acts as an unloaded inductance absorbing a very high current with a transient that can last up to 2s for high power transformers. The value of the current is progressively reduced, thanks to the dissipative phenomena linked to the resistance of the winding conductors, until a steady-state value is reached. Both the maximum value of the inrush current and the duration of the transient depend on a number of parameters: The rated power of the transformer: that is to say, the construction features of the machine since the associated factors affect the value of the inductance to be loaded and damping resistances. The short circuit power of the supply network: the transient will get much longer and heavier with an increase in the ratio between the power of the network and that of the transformer. The residual flux: the presence of a residual flux in the transformer s magnetic core will increase the maximum value of inrush current. The instant of energization: this is the most variable among the parameters considered, since the maximum value of the current (10 to 20 times higher than the rated value) will be reached when the closure of the circuit breaker coincides with the zero voltage crossing, while the minimum value (about twice the rated current) will occur when the closure of the circuit breaker coincides with the passage of the maximum value of the voltage waveform. In order to estimate the value of the inrush current in the worst case, since a direct experimentation on the installation was out of the question for service continuity needs, there were two viable alternatives: Refer to the manufacturer's data, that indicates the relationship between the maximum current peak and the rms value of the rated current of the transformer and the value of the time constant. Analyze the behavior of the electrical system during an energization by means of a specific software simulation. The former option, involving an evaluation performed with the data supplied by the manufacturer, appeared conservative since no circuit breaker intervention was observed on field. Therefore, the latter alternative was followed: a 20 kv network was modeled and afterwards its behavior was studied during transients caused by transformer energizations. 3 Experimental and simulation tests In order to develop a reliable simulator, experimental tests were carried out on a three-phase resin encapsulated transformer. The test involved a measurement of the peak values of inrush currents. The performed research work was set according to the following schedule: 1. A calculation of an equivalent circuit based on the test report and testing certificate supplied by the transformer s manufacturer. 2. A reproduction of the experimental test through the implementation of a Matlab- Simulink software procedure. 3. A comparison between the results of the experimental test and those of the simulation. The rated data of the tested transformer are shown in Table 1. The repeated energizations ISBN:

3 performed during the experimental tests were carried out by closing a circuit breaker that was set so as to close in the first test whenever the voltage crossed zero, then to close with a 30 phase shift in each subsequent test [8], [2]. The results obtained are reported in Table 2. Table 1. Data of the tested transformer. The transformer s rated data Primary Secondary Power [kva] 1500 Voltage [V] Current [A] Connection Delta Wye Vcc% 6% The transformer was tested with a three-phase power supply applied to the primary having 100MVA short-circuit power at 11 kv; the secondary was maintained unloaded. The laboratory experimentation consisted of 12 tests with a duration of 9 seconds; for each test the closing angle of the Vrs-line voltage was varied of 30 and the peak values of the inrush current were recorded. According to the layout of the measuring system, a Matlab simulation model was developed, first for each individual block, then for the entire power system. Of course, particular attention was paid to the model of a three-phase saturable transformer. Fig. 1 shows, as an example, the results of a simulation carried out with the system simulator. The performed simulation tests, showed also the presence of overvoltages during plant energization. Maximum peak values ranged from 50 kv, registered at an external receiving substation, to 61 kv recorded at the main underground transforming substation; the transient duration was about 40 milliseconds. The overvoltage phenomena are of considerable importance since they could lead to an insulation failure in electrical equipment and at the system weak points. Table 2. Inrush currents from experimental tests. V n =11 kv A n =100 MVA V rs closure angle [ ] I r [A] I s [A] I t [A] The mentioned overvoltages are of internal origin and oscillatory character, with a waveform determined by the superposition of a sine wave at operating frequency and one or more components of considerably higher oscillatory frequency, caused by complex transient phenomena due to energy exchanges between the various components, especially by charging and discharging of capacitances and inductances and the damping of resistive parameters. Performed simulations show also that high switching overvoltages occurred at the ends of the longest cables during the energization of an unloaded line. The presence of non-efficient circuit breakers may amplify the amount of this kind of overvoltage [13]. Fig. 1 An example of the inrush current estimated by the implemented simulator. ISBN:

4 4 The LNGS case study As concerns the selectivity of the protection system, the simulation results obtained allowed to establish the following points: 1. With regard to overload, only a chronometric selectivity can be adopted and only on the delayed threshold (short delay). 2. With regard to the short circuit, overcurrent protections are not selective. 3. The medium voltage cables are fully protected against both overloads and short circuits. 4. A three-phase short circuit in the general switchgear bars involves the opening of the protection placed upstream the transformer, causing a total blackout of the LNGS electrical system. 5. Protections are not selective for ground faults. The above observations show that the selectivity of the adopted protections, in the absence of interventions on the actual system, is not verified. For this reason, in order to increase the degree of selectivity of the protection system, the following possible solutions are suggested [1], [4], [5], [6], [7], [9], [11], [18]: 1. Avoid parallel operation of substation transformers. 2. Request the supply company to adopt appropriate exceptions for line faults. 3. Adopt a logic selectivity for both the line fault and ground fault. 4. Install directional protections in the rest of the medium voltage distribution system in order to improve the selectivity towards the ground fault. 5 Logic selectivity In order to solve the main problems concerning the protection coordination, the adoption of the logic selectivity was suggested. More precisely, this type of selectivity was developed to overcome the drawbacks inherent in chronometric selectivity. The method is normally used when a fault is to be eliminated in a short time. An example of logic selectivity is shown in Fig 2. A. Mode of operation The exchange of logic information between subsequent protections allows the suppression of selectivity intervals, and will therefore considerably reduce the delay of intervention of the circuit breakers located closest to the supply source. As a matter of fact, in a radial network the protections located upstream the fault point are activated, whereas those downstream are not; this allows to unambiguously locate the fault point and the circuit breaker that must step in. Each protection enabled by the fault will send: an order of logic block to the upstream protection (block of the opening order of the circuit breaker); an opening order of the associated circuit breaker if it has not received a logical block from the downstream protection. Phase-to-phase fault Fig. 2 An example of logic selectivity. For the breaker-failure protection, a timed opening is expected; this principle is shown in Fig. 3 where it can be noted that, when a fault happens downstream of B, the protection located at B stops the protection at A and, therefore, only the B protection causes the opening after the TB time. Using a logic selectivity it is possible to reduce trip times in all cases to the minimum value, typically 0.1 s, which is necessary for an information exchange between the installed protections. The lock signal exchanged between the various units is of a digital kind (closing or opening contact) and therefore only needs a simple electrical connection between the various units. The duration of the logical block order for the A protection is limited to TB+T3 with T3 greater than the opening time of the circuit B breaker (typically 200ms), so that in case of non-breaker opening at B, the protection at A drives its circuit breaker at TB+T3; in ISBN:

5 the presence of a fault between A and B the A protection drives the opening after TA. Sometimes it can be useful to adopt the so-called combined selectivity, which provides additional benefits in comparison with simple selectivity. With reference to the performed study, a combined selectivity that appeared of particular interest was the logic/chronometric selectivity exemplified in Fig 4, where: logic selectivity is inside each switchgear (A-B and C-D); chronometric selectivity is between the two switchgears QA (protection B) and QB (protection C) with TB = TC + ΔT. Fig. 3 Example of timed opening in case of breakerfailure protection. B. Advantages Opening time is independent from the fault position, the levels of selectivity and the number of cascading protections. In this way, it is possible to achieve selectivity between an upstream, short-timed protection and a downstream high-timed protection; in other words, timing can be reduced at the protection closest to the supply source with respect to downstream protections. In addition, this system will integrate the breaker-failure protection. Chronometric selectivity Logic selectivity Logic selectivity 6 Further remarks An electrical system may be in faulted conditions if the functioning of one or more elements is altered by causes that can be of different nature. Protection systems must not only put a failed component out of service quickly, but also provide the staff managing the electrical system with the appropriate information necessary to implement suitable measures aimed at eliminating the causes leading to the fault condition. A protection system must fulfill the following functions: Preventive function: to identify the very first occurrence of a fault and/or more serious disturbances [19]; Healing function: to eliminate a fault fast and selectively, in order to contain its spread in both temporal and spatial terms; Containment function: to try to mitigate the effects once the disturbance has taken place (e.g. shedding of non-priority loads in underfrequency conditions, [14]); Return to normality function: to correctly operate the components of the electrical system in the early stages of recovery; protection systems must not hinder the return of the electrical system to normal conditions. Fig. 4 An example of mixed logic-chronometric selectivity. In this way, it is not necessary to install a transmission link for the signal logic block between two switchgears that may be distant. The time necessary to eliminate a fault is reduced when compared to a simple time-based selectivity implemented at all the distribution levels (1.1 s for a fault downstream the A protection). 7 Conclusion Logic selectivity In order to protect large MV installations correctly, a series of solutions are proposed. For phase faults, it is important to avoid parallel operation of MV/LV transformers and set the architecture of the protection system on the concept of logic selectivity. For phaseground faults, the recommendation is to adopt directional protections and implement the logic selectivity. In order to investigate this issue, reference was made to a real underground large installation. In the paper a general survey of the ISBN:

6 protection philosophies is offered, focusing on the issue of selectivity coordination. Moreover, possible solutions aimed at devising a proper selectivity plan for the protection system were proposed and discussed. References [1] G. Fazio, V. Lauropoli, F. Muzi, G. Sacerdoti, Variable-window algorithm for ultra-high-speed distance protection, IEEE Transactions on Power Delivery - Vol. n. 18, NO. 2, April [2] Y. Cui, S. G. Abdulsalam, S. Chen, W. Xu, A Sequential Phase Energization Technique for Transformer Inrush Current Reduction, IEEE Transactions on Power Delivery, VOL. 20, No. 2, April [3] C. Bartoletti, G. Fazio, F. Muzi, S. Ricci, G. Sacerdoti, Diagnostics of Electric Power Components: an improvement on signal discrimination, WSEAS Transactions on circuits and systems, Issue7, Vol. 4, July [4] F. Muzi, A filtering procedure based on least squares and Kalman algorithm for parameter estimate in distance protection, International Journal of Circuits, Systems and Signal Processing, Issue 1, Vol. 1, [5] F. Muzi, Real-time Voltage Control to Improve Automation and Quality in Power Distribution, WSEAS Transactions on Circuits and Systems, Issue 4, Volume 7, April [6] K. P. Basu, A. Asghar, Reduction of magnetizing inrush current in a delta connected transformer, Power and Energy Conference, 2008, PECon [7] F. Muzi, Distance relays in conjunction with a new control algorithm of inverters for smart grid protection, 2011 CIGRE International Symposium, The electric Power System of the future Integrating Supergrids and Microgrids, Bologna, Italy, September 13-15, [8] M. Gong, X. Zhang, Z. Gong, W. Xia; J. Wu, C. Lv, Study on a new method to identify inrush current of transformer based on wavelet neural network, Electrical and Control Engineering (ICECE), [9] F. Muzi, A real-time harmonic monitoring aimed at improving smart grid power quality, 2011 IEEE International Conference on Smart Measurements for Future Grids (SMFG), Bologna, Italy, November 14-16, [10] R. Dashti, S. Afsharnia, Demand response regulation modeling based on distribution system asset efficiency, Electric Power Syst. Res. 81 (2011) [11] F. Muzi, A. De Sanctis, P. Palumbo, Distance protection for smart grids with massive generation from renewable sources, The 6th IASME/WSEAS International Conference on Energy & Environment (EE'11), Cambridge (UK), [12] F. Muzi, C. Buccione, S. Mautone, A new architecture for systems supplying essential loads in the Italian High-Speed Railway (HSR). WSEAS Transactions on Circuits and Systems, Issue 8, Volume 5, August [13] F. Muzi, R. Dercosi Persichini, An analysis of overvoltages in large MV-Cable installation, 15th IEEE-ICHQP International Conference, June 2012, Hong Kong. [14] Y.Y. Hong, Y.Z. Lai, M.C. Hsiao, Y.R. Chang, Y.D. Lee and H.C. Huang, Studies on Operation Modes for the First Outdoor Micro-grid Test Bed in Taiwan, IEEE PowerCon, Oct 30 - Nov 2, 2012, Auckland New Zealand. [15] F. Muzi, The transformer inrush currents in large MV-cable installations, 12 th WSEAS International Conference on Electric Power Systems, High Voltages, Electric Machines (POWER 12), Prague, Czech Republic, September 24-26, [16] M. M. A. S. Mahmoud, Electrical short circuit finding in MV network using fuzzy clustering techniques, WSEAS International Conference, Prague, Czech Republic, Sept , [17] F. Muzi, Computer relaying for smart grid protection, WSEAS International Conference on Information Technology and Computer networks (ITCN 12), Vienna, Austria, November 10-12, [18] F. Muzi, Symmetrical components and digital signal processors for smart grid protection, 2th IASTED Int. Conference on Power and Energy Systems and Applications (PESA 2012), November 12-14, 2012, Las Vegas, USA. [19] M. Cerullo, G. Fazio, M. Fabbri, F. Muzi, G. Sacerdoti, Acoustic signal processing to diagnose transiting electric-trains, IEEE Transactions on Intelligent Transportation Systems, Vol. 6, No. 2 June [20] F. Muzi, L. Passacantando, Improvements in power quality and efficiency with a new AC/DC high current converter, WSEAS Transactions on Circuits and Systems, Issue 5, Volume 7, May [21] F. Muzi, Supergrids and the new challenges to face. The Sixth IASTED Asian Conference on Power and Energy Systems (AsiaPES 2013), Phuket, Thailand. April 10 12, ISBN: