Impact of limiting reactors for voltage sag mitigation in distribution utilities

Transcription

1 Elektrotehniški vestnik 71(5): 31-37, 24 Electrotechnical Review; Ljubljana, lovenija Impact of limiting reactors for voltage sag mitigation in distribution utilities tefano Quaia, imone Castellan Department of Electrical Engineering, Electronics and Computer cience, University of Trieste, Via A.Valerio 1, Trieste (Italy). Abstract. The paper deals with the issue of Power Quality (PQ) in distribution utilities. Effects of Limiting Reactors () connected at the beginning of the MV feeders are investigated. They allow for effective mitigation of voltage dips and are considered a trivial solution, attractive for the reason of a very low cost, simplicity and reliability. The main con are voltage drops, which can be limited by taking measures described in the paper. Effects of on the system operation, including harmonic distortion, induction motor starting, stability of synchronous machines, are analysed. With regards to the increasing usage of Distributed Generators (DG), combined effects of local generators and are investigated, too. Key-words: distribution systems, power quality, limiting reactors, voltage sags. 1 Introduction The number of MV customers sensitive to transient voltage disturbances is constantly increasing. To limit the related often high costs, a very high quality standard of the electrical supply is required. To meet sensitive users specific expectations, an interesting possibility is to group them in areas, named industrial parks, for which high-quality (premium grade) electricity is delivered. This is of course done at higher prices than the normal supply. Implementation of this option requires changes and/or installation of certain devices in the distribution utility, for which investment funds must be provided. As seen from the technical literature, great attention is currently paid to Custom Power devices, in particular to the Dynamic Voltage Restorer (DVR). Though these devices allow for compensation for most of the incoming voltage disturbances and provide an outstanding PQ improvement, they are not yet widely used. This is mainly because of their high costs. It is thus of a practical interest to investigate intermediate solutions that require lower investment costs on the utility network, yet still providing a significant enhancement of the quality of the electric energy delivered to customers. As reported by several surveys, the most dangerous of all voltage dips are voltage disturbances affecting MV industrial users. These users can experience several tens of process interruptions per year resulting from voltage dips, although this number can widely vary from site to site and from year to year. According to the world-wide experience, most of voltage dips are caused by shortcircuits on MV overhead lines, and considerably less by faults in the HV system. It follows that arrangements capable to mitigate the voltage dips taking place in the distribution systems can significantly improve PQ. Tough underground MV cables efficiently reduce short circuits in the distribution system, this solution is quite expensive. The mean cost (material + labour) of a 2 kv underground cable is estimated at some 6-7 k$/km; this is roughly twice the cost/km of a corresponding overhead line. In typical distribution utilities with a radial topology, like the one shown in Fig. 1, a similar or even better voltage dip mitigation can be assured by installing devices at the beginning of MV overhead lines. HV 2 kv bus Figure 1. Installation of Limiting Reactors for voltage sag mitigation on a typical 2 kv distribution system. They are listed below in order of their decreasing complexity: Received 13. April, 24 Accepted 2. June, 24

2 32 Quaia, Castellan 1) Fault Current Limiters (FCL). They consist of a series impedance capable to change very quickly from a low to a high value, and suitable to limit the short-circuit current and along with it also the magnitude of voltage sags at the MV busbar. everal FCL have been proposed in the literature. Most of them require power electronic devices and related control circuits, but all of them have significant drawbacks. They can be shortly summarised as too complex, expensive and leading to eventual operation problems decreasing the distribution system reliability. 2) olid-tate Breakers (B) instead of (or in combination with) traditional electromechanical line circuit breakers. B limit to a fraction of a period the duration of the voltage sag at the busbar in case of short circuits in the distribution system. The cost of a 2 kv thyristor-based B, including the cooling system and control equipment, is roughly twice the cost of a traditional line circuit breaker [1]. The total cost of a utility station with 1 line feeders is some 1-2 k$ higher than when using the traditional circuit breakers. 3) MV fuses instead of (or combined with) traditional line circuit breakers. Compared to the previous solution, this arrangement provides similar results for voltage sags but at a lower cost. The back of the medal is an increased annual interruption time because of fuse substitutions. 4) Limiting Reactors () permanently connected [2]. This trivial solution can be used to reduce below a certain value voltage sags at the busbar (without limiting its duration) at a very low cost and with no effect in the distribution system reliability. On the other hand, permanently change the system. Their various effects, and particularly voltage drops, are investigated in the next sections. 2 Limiting Reactors consist of an adequate number of turns (in air) at the beginning of each MV line. While they are attractive for their simplicity, low cost and reliability, they increase voltage drops, reduce the short-circuit power and have several other effects that should be investigated. The following aspects need to be dealt with: - voltage drop and, more generally, voltage regulation, - harmonic distortion, - induction motors starting, - stability of synchronous machines. These aspects are studied in the next sections with reference to the Italian typical 2 kv distribution utility, whose lay-out is depicted in Fig. 1. Its characteristics are as follows: - radial topology with 8-1 MV feeders departing from the busbar, - 4 MVA HV/MV transformer with V C % = 13% and on-load tap changer ±5x1.5%, - short-circuit power on the HV (132 kv) side: 24 MVA, - MV line power rating: 5 MVA, - MV line impedance: X=.35 /km, R=.22 /km. The value of the limiting reactance X R is a compromise between two opposite requirements: effective voltage sag mitigation and the lowest possible impact on the system operation. X R depends on the maximum acceptable sag V at the MV busbar of the utility station. In the worst case, i.e. at a three-phase short circuit occurring immediately downstream of, the p.u. busbar voltage V B is: ( X X ) V B = X R R + (1) where resistances are neglected and X is the shortcircuit reactance at the busbar. Being V B = 1-V, it follows: X R ( 1 V ) V = X (2) X R can be reduced by raising the short-circuit power at the MV busbar (lower X ), and/or accepting a less effective voltage sag compensation (higher V). In practice, at any given site, X can be reduced using high power HV/MV transformers with low short-circuit voltage. To choose a proper value for V, one has to consider that a three-phase fault close to the busbar is a quite rare event. Unbalanced faults are much more frequent and cause less severe sags [3]. A further small contribution to sag mitigation is given by the (variable) impedance of the line between the busbar and the fault location. Then, V should be set reasonably around.15 p.u.. The above reference system has X = Assuming V =.15 p.u., it follows X R = 8.3 (L R = 26.5 mh). This value will be used throughout the paper. 3 Voltage regulation The main drawback of is the voltage drop introduced, which can lead to a poor voltage regulation in the distribution system. aid X R and R R the reactance and resistance (R R can be assumed.1x R ), the voltage drop introduced by the can be calculated as: 3(RR Icos + X R Isin ) 9V% (3) V The total voltage drop between the MV busbar of the utility station and the customer delivery point is the sum of the voltage drops on the and the line length concerned. For example, let us consider the above line parameters and a line load of 1 A (3.5 MVA, 7% of

3 Impact of limiting reactors for voltage sag mitigation in distribution utilities 33 the line rating) at.9 PF. Using (3), the voltage drop on the is about 3.2%, while at the distance of 6 km from the station the voltage drop grows up to 5% (assuming constant line current) instead of 1.8% without. The voltage regulation can be significantly improved through the on-load tap changer of the utility station HV/MV transformer: a proper tap selection can compensate for the mean voltage drop on. This requires a local voltage measurement for each line (downstream of ), instead of voltage measurement at the busbar. The relevant increase of complexity and cost of the utility station is very low. However, since voltage drops on depend on the load of each line and can be quite different, this action alone may be not sufficient to get a good voltage regulation in all operative situations. ince R R <<X R, from (3) it appears that voltage drops introduced by can be limited raising the load PF. A significant voltage drop limitation is obtained with PF in the range [2]. For instance, with the above data and PF =.97, the voltage drop on is limited to 1.8%. Of course, raising PF also reduces voltage drops on the MV line impedance, and has further evident advantages for the utility HV network. The combined effect of these two measures (PF raise and proper control of the utility station transformer) allows an effective limitation, in any operating state, of voltage drops at the customer delivery points. adopted for industrial parks, it does not appear practical for distribution systems. In distribution systems, proper control of the utility station transformer is thus the only practical measure to counterbalance voltage drops introduced by. Their installation may lead to a slightly less efficient voltage regulation, for which reason they should be evaluated in each case separately. 4 Harmonic distortion The wide use of power electronic converters in civilian and industrial applications is the main cause of the increased distortion of the current and voltage in power distribution systems. As an example, in the electric drives where an induction motor is supplied by a voltage-fed PWM inverter, the line-side converters are diode rectifiers and are loaded with a capacitor. They draw a current containing harmonic components of high amplitude that produce, in turn, a harmonic voltage drop across the equivalent power system impedance. introduce a further impedance, whose value linearly increases with the harmonic order, and so modifies the voltage distortion within the distribution system. To investigate the effects of on harmonic distortion, reference was made to the system shown in Fig. 3, where only one non-linear load (harmonic source) is included. V [kv] HV kv bus-bar [km] Figure 2. imulated voltage profile for a 2 kv 15 km long line, loaded with 3 MW,.97 PF. Note the voltage drop on. In the simulation reported in Fig. 2, the voltage drop between the busbar and the line end is limited to about 3.6%. The transformer control raises the busbar voltage by 2.5% over the nominal voltage and reduces the voltage drop at the line extremity to about 1.1%. To get a line PF increase, all customers should increase their own PF. The investment cost to improve PF from the usual.9 to is moderate, usually of the order of a 5% increase of the standard capacitor cost [2]. Nevertheless, this cost is justified only for those users that need a high-quality power supply and agree to pay for it. Therefore, though this measure can be easily Figure 3. One-line sketch of the system used to study harmonic distortion. The arrows represent linear loads. The equivalent impedance of the system, as seen from the distorting load, has a reduced resonance frequency because of the higher line inductance and the higher capacitance needed to improve the load PF. ince the distorting load can be considered a harmonic current source, according to Fig. 4 the introduction of causes an amplification of the harmonic voltages whose frequency is lower than f o and an attenuation of the harmonic voltages whose frequency is higher than f o. Then the voltage distortion on the line feeding the nonlinear load can be both amplified or reduced by the insertion of. Different cases were simulated. The

4 34 Quaia, Castellan results showed that even when some harmonic voltages (in particular the 5 th ) are magnified, the harmonic voltage distortion is considerably lower than the limits imposed by the European tandard EN-516. Figure 4. Hypothetical equivalent impedance as seen from the distorting load: without (dashed line) and with (solid line). Anyway, these results cannot be assumed to be generally valid. Therefore, when evaluating introduction of in a real net, it is advisable to check that each harmonic voltage and the THD do not exceed the imposed limits. An amplification of the harmonic voltages is possible also on the lines without non-linear loads. This can be explained as follows. If the busbar voltage V B contains a harmonic component of order h, the harmonic voltage of order h, downstream of, V Lh is: Reqh + jx eqh VLh = VBh (4) Reqh + j( X eqh + X Rh ) where the subscripts h, R and eq mean respectively a generic harmonic order, the impedance and the equivalent impedance downstream of (line and loads). ince X eqh includes also the PF correction capacitors, it can be negative for a certain harmonic order. In this case, if X eqh >X Rh /2, the ratio in (4) is greater than 1 and the harmonic voltage at the beginning of the line is amplified by. V Lh/V Bh frequency [Hz] Figure 5. Plot of the function V Lh /V Bh vs. frequency. Fig. 5 refers to a specific case in which the ratio V Lh /V Bh (i.e. the voltage divider (4)) has a maximum much greater than 1 for f 25 Hz. This means that the causes an amplification of the 5 th harmonic voltage V l5, at the beginning of the line. Usually, the harmonic components of the busbar voltage are very small and therefore, as in the case study, the voltage distortion at the beginning of the feeders without distorting loads does not exceed the allowed limits. Nevertheless, the depicted amplification phenomenon should be taken into account in any practical case. 5 Induction motors starting tarting is the most critical operating condition for large induction motors fed from the network. As known, starting is easier when the short-circuit power at the motor bus is high, and vice versa. aid I and = 3 VI respectively the motor starting current and apparent power, and C the short-circuit power at the motor bus, the voltage drop during starting can be approximately calculated as: 9V% 1 (5) + C For acceptable starting, the voltage drop must be properly limited, usually below 15-2%. Once a maximum acceptable voltage drop V max % has been fixed, equation (5) allows calculation of the maximum motor rating P: I Vmax % I P = cos C cos (6) I 1 V % I max Equation (6) points out that the maximum acceptable motor rating P is proportional to the short-circuit power C. ince reduce the short-circuit power, the maximum motor power is reduced by the same percentage. For example, consider a 4 V motor fed through a 1.6 MVA 2/.4 kv transformer with short-circuit voltage V C % = 6%. Using the same values already used in the previous sections and neglecting the impedance of the MV feeder, we have C = 24.3 MVA without, reduced to C = 16.1 MVA with. Assuming V max = 2%, I/I = 1/6, cos =.84, =.92, (6) provides a maximum power P 782 kw without, that becomes P 52 kw (33% reduction) with. However, (6) provides a theoretical power, that must be reduced to account for the 2% voltage drop during starting. o, the maximum motor rating can be assumed about 625 kw without and about 416 kw with. The reduction of the maximum acceptable motor rating caused by is then consistent; nevertheless this is a minor problem because so large 4 V induction motors are extremely rare: LV motors are usually used up to

5 Impact of limiting reactors for voltage sag mitigation in distribution utilities 35 about 2 kw, while MV motors are used for higher powers. Fig. 6 compares the maximum motor power P with and without, vs. the natural (i.e. without ) short circuit power at the 4 V motor bus. max motor rating [kw] hort circuit power at motor bus, without [MVA] Figure 6. Maximum motor rating without (upper straight line) and with (lower curve). If we consider a 6 kv motor fed through a 2 MVA 2/6 kv transformer with V C % = 6%, repeating the same calculation with the new transformer reactance and assuming I/I = 1/5 we obtain a maximum power P of about 57 kw with, instead of about 92 kw without. o large induction motors are extremely rare in plants fed at 2 kv, whose overall power demand is often limited to hundreds kw or a few MW. In conclusion, in most practical cases the limitation introduced by is not a problem. Anyway, should the maximum motor rating be exceeded in some particular case, the problem could be avoided by a proper soft starting technique. 6 tability of synchronous machines ynchronous machines in MV distribution systems are local generators or more rarely large motors. Their most important practical drawback is the loss of synchronism due to voltage dips (on the contrary, induction motors usually ride through the disturbance with only a small speed reduction). Then, voltage dip mitigation through can greatly benefit synchronous machines. In the same time, increase the series reactance, thus reducing the electrical stiffness of the synchronous machines connected to the distribution network. These aspects are studied in the present section with reference to the system shown in Fig. 7. 1) teady-state stability. Consider a small generator (rated up to some MVA) injecting power into the distribution system. Assume that this power flows backward to the utility busbar to be delivered to the users connected to other MV feeders. The steady-state stability of this system can be evaluated using the well-known equation: EVB P = sin = Pmax sin (7) X being P the power delivered, E the generator air gap voltage, V B the utility busbar voltage, X the equivalent reactance from the machine to the busbar and the angle between E and V B. HV 2 kv bus-bar 4 V bus Figure 7. One-line sketch of the system used for stability study. Assuming that the generator operates at unity PF, the air gap voltage E, roughly constant due to the action of the automatic voltage regulator, is only slightly higher than the voltage at the generator terminals. Then, for a rough calculation, we can assume V B E 4 V introducing a very small error. For example, consider a 1.8 MVA, 4 V generator having 2% transient reactance, and a 2 MVA, 2/.4 kv transformer having short-circuit voltage V C % = 6%. Neglecting the small line impedance, the equivalent reactance X is about 23 m (referred to 4 V level) which rises to 26.3 m with the. Accordingly, the steady-state stability limit P max decreases from 7. MW to 6.1 MW with the, and the angle corresponding to the full generator power P = 1.8 MW increases from about 15 to about 17. Note that, in practice, only a part of this power flows backward, the remaining being consumed by the loads connected to the local MV line. The reduction of the electrical stiffness R, calculated through (8), is around 1% and, therefore, negligible. dp R = = Pmax cos (8) d 2) Transient stability. It can be evaluated through the classical equal-area graphic criterion. The worst operating condition is a three-phase short circuit close to the 2 kv bus (see Fig. 7), causing the busbar voltage V B to drop nearly to zero (without ). In this case, considering the former generator-transformer group and unity PF operation, stability is kept if the faulted line is disconnected before the angle reaches the critical value c 19, as illustrated in Fig. 8.

6 36 Quaia, Castellan Using the equations reported in the literature (see for instance [4]) and average values of the generator inertia constant, one can check that the angle c is reached after about 1 ms. This time is the typical voltage sag duration determined by MV line circuit breakers. This confirms the practical experience that severe voltage dips can cause the generator loss of synchronism. P [MW] c [ ] Figure 8. Distribution system without : equal-area criterion for a 1.8 MVA local generator operated at unity PF. P [MW] [ ] Figure 9. Distribution system including : equal-area criterion for a 1.8 MVA local generator operated at unity PF. If are connected, the same short circuit causes the busbar voltage to drop to an acceptable value (about.85 p.u.). It follows that, as illustrated in Fig. 9, the rotor acceleration is very limited and stability is easily maintained even without disconnecting the faulted line. 7 Distributed Generation (DG) DG is expected to find a wide diffusion in the near future, changing distribution systems from passive to active systems. This will lead to important changes in their project, operation and control, and poses some problems for a correct operation [5]. This section deals with this new scenario, pointing out some aspects relevant to the connection of in a distribution system characterised by high DG penetration. 1) Voltage sag compensation It is known that a generator mitigates sags near its terminals. But the contribution of a local generator (rated up to some MVA) on the 2 kv level is fully negligible. Conversely, introduces a separation among the 2 kv feeders which makes not negligible the contribution of a local generator to voltage dips mitigation on the local 2 kv line [6]. aid X DG the reactance of the generator-transformer group, the voltage sags on the local 2 kv line are reduced by a factor: X R (9) X R + X DG For example, set X R = 8.3, a 3 MVA generatortransformer group with a total reactance X DG 35 provides a sag mitigation factor.2, meaning that voltage sags at the utility busbar are reduced by 2% on the local MV line. Note that this also holds for sags coming from the HV network. 2) hort- circuit currents DG increases the short-circuit currents in the distribution systems. One possible consequence is the substitution of the utility MV line breakers due to insufficient breaking capacity. For example, with the above system values, a 3 MVA generator-transformer group located close to the utility station rises by almost 5% the short-circuit current for a fault close to the MV busbar. Of course, the increase will be higher if more local generators are present. This drawback of the DG is fully avoided if are used, because of the consequent drastic reduction of the short-circuit currents. 3) Induction motors starting In a distribution system including, DG increases the short circuit power on the MV lines feeding local generators. This helps starting large induction motors, reducing the above discussed limitation due to. For example, using the above reported numerical values, a 1.8 MVA generator-transformer group with transient reactance 23 m (referred to 4 V level) increases by about 7% the short-circuit power at the motor 4 V bus. According to (6), the maximum motor size for acceptable starting is increased by the same percentage. 4) Voltage profiles DG changes the power flows in the distribution system and poses some problems concerning voltage regulation [7, 8]. A synchronous generator can strongly modify the voltage profile of the local 2 kv line, depending on its size, operation and position along the line. The voltage profile rises at the generator location, especially when the generator is far from the utility station. This can lead to significant differences among the voltage profiles of

7 Impact of limiting reactors for voltage sag mitigation in distribution utilities 37 the 2 kv lines. In order to limit the voltage rise at its terminals, the generator should be operated at nearly unity PF (that is the usual operating condition of local synchronous generators) or, especially when the generator is far from the utility station, under-excitation operation may be requested. However, under-excitation poses new problems since the generator owner usually has no direct interest in the power quality of the distribution network [8]. In a distribution utility where are installed, if the load PF is raised as above discussed, the reactive power flows are significantly reduced. It follows that, as shown in Fig. 1, the voltage profiles are flatter and the voltage regulation problems connected with DG are reduced. In usual distribution utilities, raising the load PF is unpractical and a proper control of the HV/MV transformer is the only available measure to limit voltage drops introduced by. can be regarded as a practical solution especially in industrial parks, where voltage dips mitigation is of a great importance and the load PF can be easily raised. The extra-costs for the customers should include a moderate investment in PF improvement and a very low energy cost increase to make up for the low utility expense. These extra costs are expected to be by far justified by a large decrease in production interruptions and the related costs. 9 References V [kv] [km] Figure 1. Voltage profiles of two lines: one has a 1.8 MW DG at its extremity (solid line) and the other has no DG (dashed line). 8 Conclusions Installation of at the beginning of the MV lines allows a drastic mitigation of voltage dips and requires simple and very low-cost changes on the utility net. Further advantages, that can be especially significant in distribution utilities characterised by high DG penetration, are short-circuit currents limitation, reduction of the breaking capacity required for MV circuit breakers, and increased stability of synchronous machines. The back of the medal is given by: voltage drop and voltage regulation problems caused by reduction of induction motors maximum power for acceptable starting. have also effects on voltage distortion, that however are not predictable in general and have to be evaluated in each practical case separately. A fully acceptable voltage regulation can be assured through proper control of the HV/MV transformer and raising the load PF from the normal.9 to [1] F. Tosato,. Quaia, Equipment Fault-Clearing Time Reduction: an Approach to Utility Voltage ag Mitigation, Electrotechnical Review, Vol. 67, No. 5, 2, pp [2] F. Tosato,. Quaia, Power quality enhancing in distribution utilities, Electrotechnical Review, Vol.68, No.5, 21, pp [3] M. H. J. Bollen, Understanding Power Quality Problems; voltage sags and interruptions. New York: IEEE Press eries on Power Engineering, IEEE, 2, chapter 7. [4] E. W. Kimbark, Power ystem tability, Vol. III. New York: IEEE Press Power ystem Engineering eries, 1995, pp [5] P. P. Barker, R.W. de Mello, Determining the impact of Distributed Generation on Power ystems: part 1 radial distribution systems, IEEE PE ummer Meeting, 2, Vol. 3, pp [6]. Quaia, F. Tosato, Limiting Reactors and Dispersed Generators Combined Effects in Voltage Quality Improvement, Proceedings of IEEE ERK 2, Portoroz (lovenia), 22, invited paper, pp [7]. J. van Zyl, C. T. Gaunt, Control strategies for distributed generators operating on weak distributions networks, Proc. of IEEE Power Tech 23, Bologna (Italy), June 23. [8] N. Jenkins, G. trbac, Effects of small embedded generation on power quality, IEE Colloquium on Issues in Power Quality, Warwick (UK), 1995, pp. 6/1-4. tefano Quaia (1962) graduated in Electrical Engineering at the University of Trieste/Italy in In 199 he joined the Dept. of Electrical Engineering, Electronics and Computer cience of the University of Trieste, where he presently works as Associate Professor in the area of Power ystems. He is a member of AEI and his main research interest includes Power Quality, Power Electronics Application, Power Delivery and Cathodic Protection. imone Castellan (1971) graduated in Electrical Engineering at the University of Padova/Italy in In 2 he joined the Department of Electrical Engineering, Electronics and Computer cience of the University of Trieste. In 22 he received his Ph.D. degree in Electrical Engineering from the University of Padova. His research interests are in the field of power quality and active compensators.