* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download 1 Introduction
Opto-isolator wikipedia , lookup
Solar micro-inverter wikipedia , lookup
Pulse-width modulation wikipedia , lookup
Power factor wikipedia , lookup
Power over Ethernet wikipedia , lookup
Variable-frequency drive wikipedia , lookup
Audio power wikipedia , lookup
Power inverter wikipedia , lookup
Electric power system wikipedia , lookup
Three-phase electric power wikipedia , lookup
Electric power transmission wikipedia , lookup
Stray voltage wikipedia , lookup
Surge protector wikipedia , lookup
Power MOSFET wikipedia , lookup
Two-port network wikipedia , lookup
Electrification wikipedia , lookup
Power electronics wikipedia , lookup
Buck converter wikipedia , lookup
Voltage optimisation wikipedia , lookup
Electrical substation wikipedia , lookup
Distributed generation wikipedia , lookup
Switched-mode power supply wikipedia , lookup
History of electric power transmission wikipedia , lookup
Power engineering wikipedia , lookup
Alternating current wikipedia , lookup
DISTRIBUTEDPOWERGENERATION UNITS ANDTHEIRIMPACTTOTHE POWERNETWORK CHRISTOPH HÄDERLI, ALEC STOTHERT, ALICE PIAZZESI, MARCO SUTER ABB Switzerland Ltd., Im Segelhof 1, 5405 Baden-Daettwil, Switzerland [email protected], Tel +41 58 586 83 18, Fax +41 58 586 73 12 UWE PRAUSE ABB Calor Emag Schaltanlagen AG, Käfertaler Strasse 250, 68128 Mannheim, Germany [email protected], Tel +49 621 386 27 96, Fax +49 621 386 27 85 ABSTRACT Introducing high concentrations of Distributed Power Generation units (DGs or DGs) into existing electrical networks can compromise, amongst others, grid stability, protection concepts, selectivity, and voltage profiles. In this paper, case studies focusing on the introduction of multiple DGs into a small European town are presented. Two typical network configurations based on real data have been chosen for this study; a meshed residential network and a residential network in a station to station configuration. The DGs have been virtually introduced into the network to calculate the impact on network parameters. Different criteria (e.g., voltage profile, equipment loading voltage rise, etc.) are chosen to compare the two configurations and to support the decision process for the placing of DG. In addition, a detailed model of a power electronic interface and techniques based on robust stability analysis of systems with multiplicative model uncertainty are used to show that both the frequency response of the installed units, in particular resonant peaks, and the distance between MV/LV transformers and the installed DGs are critical. Based on the results of the studies, the feasibility of a “plug and power” concept is evaluated. Keywords: protection, modeling, network analysis, stability 1 INTRODUCTION Interest in and application of Distributed Power Generation (DG) has been growing in recent years for a number of reasons. Emerging technologies have made small power generation more flexible, scalable and economical. Combined with the enhanced efficiency and environmental friendliness (usage of heat and power, low emissions, low noise) DG is in a very good position to provide power generation for the future and large growth is expected through a wide range of applications [1]. Small generators have been installed in large numbers for a long time, but predominantly in emergency power and UPS applications. Emergency systems have different requirements (different operating hours, etc.) and do not have to comply with the same regulations as grid connected systems. As they are not permanently operating, their impact on the distribution network under normal operating conditions is small. Installing DG suitable for regular (and backup) power instead of pure “stand by”- power asks for different schemes of interconnection as the distribution network has not been designed for large-scale installation of DG. In the past, the advantages (economies of scale, available technologies, geographic location) of large centralized power plants outweighed the disadvantages (need for strong transmission network, transmission losses, limited usage of waste heat) and made them the optimal cost effective solution for generation of electricity. Accordingly, the public grid has been designed in a top down manner with a relatively small number of very large generators feeding a very large number of relatively small loads. When applying DG in a large scale, the existing network will be operated with different power flows. The network must not be negatively disturbed; there will be constraints on the placing and operation of DG and/or alterations in the network may be necessary. The interconnection of DG units with the grid has to be done without negatively impacting safety, reliability and quality of supply. It has to be non-disruptive and economical, particularly when applied to smaller units. Furthermore, regulatory issues have to be considered (see also [2]). DG has many different faces and can be defined quite differently [1]. Characteristically, a DG unit is a small source of electric power generation or storage that is not a part of a large central power source and is located close to the load. These can be either grid connected or operate independently of the grid. Those connected to the grid are typically interfaced to the distribution system, and thus dispersed across the utility’s electric network rather than concentrated in a single location. Different definitions of DG refer to power, proximity to load, mode of dispatching, mode of planning (centrally or not) and other criteria. We can further distinguish between the types of networks that are used for interconnection (voltage level, industrial network, public network, etc.). Larger DG (which are connected at MV levels) and industrial DG usually allow for detailed planning, whereas small residential units should be connected with a scheme as simple as possible. No extensive planning (such as an individual network study for each unit) is possible on this level, as the costs for a single unit are too high. For this reason, new local sources on the LV-level should be introduced with a “plug and power” concept: all the possible problems and restrictions should be checked on a broad level during a planning phase and no further actions should be required by the end-users when introducing individual units. The focus of this paper is on the technical issues for interconnecting DG units in the range below 1MW to the public LV-grid. 2 Protection for DG depends largely on the topology of the network and the existing protection schemes. NETWORK ANALYSIS A network analysis is needed to determine the impacts of DG on the network in a given case: Changes in power flow, changes of electrical parameters of the network (voltage profiles, etc.), equipment loading, requirements for protection etc. Different protection schemes at both low voltage and medium voltage level will be required [6]. These will have to allow for reverse power flow while assuring tripping in case of a fault on the MV side. Stability of the network can be compromised through the interaction of DG and the grid. A network analysis is common practice on the MV level, where wind turbines, industrial DG or larger CHP plants may be connected. However, small DG units shall be interconnected on the low voltage level on an almost random basis. A “plug and power” scheme for small DG is highly desirable. Yet practical experience with extensive deployment of DG on the low voltage level is scarce. 2.1 Parts of a typical German city were chosen to perform the studies (Area of ca. 2 km2). Real network data (transformers, lines etc.) was used and hypothetical DGs were placed in different locations. Different sizes and types of DG were used (50 kW to 500 kW, synchronous generator and inverter type DG). A small part of the chosen area can be seen in Figure 1. The virtually introduced DG are marked with the -Icon. Figure 2: Meshed LV-network There are a number of different network structures, each with a different behavior [3],[4]. For comparative reasons, two different network topologies were considered in this study. A meshed low voltage grid with three redundant medium voltage lines (Figure 2) and the same grid supplied on a station to station basis (achieved by opening the connectors in selected cable distribution cabinets between the different parts of the grid, marked KD in Figure 1). CASE STUDIES Case studies with real data of an existing small town were performed to study the effect of installing a large number of DGs on the low voltage network. The simulations have been carried out with ABB’s network analysis tool CALPOS [14]. Figure 3: LV-network in station to station configuration Figure 1: Part of small town with DG (<-> ca. 250 m) In the station to station topology (Figure 3), there is no connection between different transformers on the low voltage level. Every low voltage grid has its dedicated transformer. In both cases, there were load profiles with a maximum power consumption of 5MVA assumed (corresponds with the real consumption of the considered grid). A DG generation profile with a maximum power generation of 7 MVA was applied. The effect of the DG on the electrical parameters of the network was determined by short circuit and load flow calculations. All calculations were carried out for a reference case without DG and for the cases with the DG. No alterations on the network to improve the performance were made at this point (as to check the impact of random placing of DG on an existing network). The following summary of results is based on the application of load profiles for all loads in the network. 140 (according to the voltage rise constraint) in a given point. Nodes closer to a substation have a higher maximum allowable power generation than remote nodes on long lines, as the short circuit power is much higher close to a transformer. 120 number of nodes out of profile 100 35 80 30 60 25 40 20 Units 20 15 non meshed meshed 0 DG off 1 0.9 10 5 0.8 Cos(phi) 0 20 50 100 Non meshed Meshed 200 500 DG power (kW) 1000 Figure 4: Voltage range violation Voltage range constraints (-10/+6% for low voltage in Germany) may be violated depending on power generation, reactive power and grid configuration. It can be seen from Figure 4, that regarding the voltage profile, the meshed configuration can generally handle more DG. The generation of reactive power has a negative impact on the profile; many more node voltages are out of the allowable range at cos()=0.8. The voltage rise caused by a single unit is a function (Equation 1, [13]) of DG power (SDG) and short circuit power of the grid in the point of common coupling (S”kPCC). Larger single units may violate the voltage profile or constraints of maximum voltage rise even if the aggregated power in the grid is low. Δu 1 S j ψ DG S DG S e kPCC 1 1 2 "DG cosψ kPCC DG "DG " S kPCC S kPCC S kPCC 2 1 Equation 1: Voltage rise in function of SDG and S”kPCC du [%] 10,00 9,00 8,00 7,00 6,00 5,00 4,00 3,00 2,00 1,00 0,00 0 DG SDG = 0.5 MW S”kPCC= 4.2 MVA SDG = 0.1 MW S”kPCC= 1.2 MVA SDG = 0.5 MW S”kPCC= 5.5 MVA Figure 6: Units according to voltage rise constraint Figure 6 shows the number of nodes where certain DG (maximum admissible power according to the voltage rise constraint) can be placed (for all main nodes in the grid). Note that in the meshed configuration, the majority of nodes can handle more power (200kW – 500kW) than in the station to station configuration (~100kW), as the meshed configuration has generally a higher short circuit power. The short circuit power in the grid rises because of the DG short circuit current contribution. This may result in unacceptable short circuit levels. One cable distribution cabinet in the meshed configuration exceeded the admissible value (50kA in this particular case). In the station to station configuration, short circuit currents are much lower. No point in the grid exceeded the admissible value. Synchronous generators were assumed for these calculations. With inverter based system, the short circuit current contribution is negligible. Cables in the LV network may overload due to changes in power flow. It can be seen from Figure 7 that this is mainly a problem in the station to station configuration where some cables are overloaded even in the case of cos()=1 for all the DG. In the meshed configuration, no cables are overloaded. Additional simulation show, that even if one of the three MV strands fails (meshed configuration), the loading remains within the admissible limits for all cables. SDG = 0.25 MW S”kPCC = 5.5 MVA 14 12 0,02 0,04 0,06 0,08 S DG /S"kPC C 0,1 0,12 Figure 5: Voltage rise in station to station network Figure 5 shows the analysis of the grid (station to station topology) with the inserted DG. It can be seen, that a number of the randomly placed generators violate the voltage rise constraint (3% voltage rise in Germany [5]). When calculating the short circuit power in all the nodes of the network, a map can be drawn with the maximum allowable power generation 10 number of overloaded cables 8 6 4 2 Non meshed Meshed 0 DG off 1 Cos(phi) 0.9 0.8 Figure 7: Overloaded cables Losses in the distribution network can be reduced with the implementation of DG. Figure 8 shows, that for both configurations, losses are lowest for cos()=1 for all DG. This value could be further optimized by determining the reactive power individually for each DG. The avoided losses of around 40kW (through the usage of DG) correspond with approximately 1% of the power consumption in the considered case. 250 200 150 Losses 100 (kW) 50 Non meshed Meshed 0 DG off 1 0.9 Cos(phi) 0.8 Figure 8: Losses The protection of the grid (MV and LV) and DG is affected in several ways. If distance relays (impedance measurement) are used (instead of DTO or ITO), parameter settings are affected by the infeed effect [6]. The determined impedance is changed due to current input along the protected line. This may shift the selective tripping schedule and selectivity may not be achieved. A re-setting of the distance relay parameters is necessary whenever a significant power is introduced into the MV-network. The selectivity of short circuit current indicators may be compromised. In case of a fault, the indicators may not only trip from the feeder side, but also from the dead end side of the line. This may happen if there is a significant short circuit current contribution by DG (possible with synchronous generators, but not usually with inverter based systems). The fusing scheme on the LV level may have to be adjusted to prevent overloading of lines. If a DG is introduced along a line, the line may be overloaded between the DG and a load on the same line without blowing the fuses at the end of the line. In this case additional fuses are needed. Reverse power relays (these relays are located at each transformer in a meshed LV grid and trip in case of a fault in the corresponding MV line) in meshed systems may have to be replaced or readjusted, as they may trip under normal working conditions when DG on the low voltage level are delivering power to the medium voltage level. An individual analysis has to be made to determine, if the nominal current differs sufficiently from the fault current to achieve selectivity. Additional protection devices are needed for DG interconnection, as required from regulations (e.g. loss of mains detection). It can be concluded that selectivity can be compromised both on LV and MV level. Accordingly, measures have to be taken (adjustment of protection devices, changes in protection schemes). The two cases considered behave quite differently. Both have their advantages for the placing of DG. Meshed distribution networks have a high short circuit power. Their advantage is a relatively balanced voltage profile and high reliability through redundancy. These networks can usually handle more aggregated DG power, if the short circuit power stays within allowable limits. Station to station based networks have a lower short circuit power, are relatively simple to design, but the voltage profile is more vulnerable to load steps. The introduction of DG usually has a higher impact on the voltage profile in these configurations (Depending on the location in the network). Station to station based networks do not need reverse power relays and therefore do not need any alternative solution for these relays when introducing DG. A meshed grid can usually be converted to a station to station configuration by opening the disconnectors in the according cable distribution cabinets. This could be necessary, when the short circuit current in a meshed system increases above allowable values (according to the installed equipment) by introducing DG. The splitting of a meshed network reduces the short circuit power in any given point in the grid. 3 STABILITY ANALYSIS Most of the DG sources are connected to the distribution grid via power electronic converters. Various authors have noted [7],[8] that mass introduction of power electronic devices to a distribution network could result in stability, protection, power flow, and harmonics difficulties. A well-documented case is the Swiss national railway shut down [9] caused by interaction of newly introduced converter driven locomotives. This section investigates the effect of connecting concentrations of DG-inverters to a low voltage bus-bar (A specific 100kW inverter is chosen). In this study, the unit is locally controlled and no coordinating control is allowed. Similar models can be developed for different DGs, with the appropriate Power Electronics interface. Prime Mover Rectifier Converter ~ = Grid EMC filter ~ = Generator ~ DC link Figure 9: Inverter interconnection scheme The rectifier, the generator, and the prime mover are modeled as constant DC Thevenin equivalent, i.e. a constant voltage source and a resistance. With the exception of the converter, all elements shown in Figure 9 are modeled as linear time invariant systems in a rotating dq0 coordinate system [10]. The converter is however inherently non-linear and for analysis purposes a suitable linear model around a specific operating point must be identified. By assuming ideal switching, and no energy loss or storage in the converter a linearized model can be derived [11]. By combining this model with models of the other components illustrated in [Figure 9] a transfer function for the DG can be computed. In particular for grid stability analysis a voltage controlled current source model is required: I l,dq H Vl,dq Equation 2 The model gives an indication of the inverter currents produced when grid voltage fluctuations occur. Equation 2 is a general description of various DGs while values of the matrix [H] depend on the specific system. 3.1 ANALYSIS METHOD Figure 10 illustrates a low voltage network with a connected DG. Note that the figure represents a single bus (node Vl) connected via a transmission line (Zl) to a strong grid. By adjusting the transmission line characteristics a weak grid condition can be simulated. Vs Transmission line Vl result from [12] which gives an upper bound on the size of delta that guarantees loop stability. 3.2 RESULTS The models and analysis techniques presented above were tested on the scenario of the small town in Germany presented in paragraph 2.1. The permissible concentrations of DG units at various nodes in the low voltage network were calculated. It must be noted that: Only the DG system model described above is considered, the matrix [H], parameters of the chosen DG inverter system (100kW). All the distributed generators are connected to the same grid node and they feed energy into the grid. There are no disturbance loads connected to that node. For analysis, the transmission lines (Zl in Figure 10) are modeled as series L, R impedances with equivalent grid values calculated at specific nodes using the data from the small town case study. 5 4 DG-Inverter 1 Il I l f Vl 3 Units 2 Zl 1 Stiff generator DG-Inverter 2 Il I l f Vl Figure 10: LV equivalent network with DGs (considered as load) The question under investigation is how many DGs can be connected to Vl without causing stability problems. For our purposes, this implies that if there is a bounded disturbance in Vs the current through the transmission line (or alternatively the voltage at Vl) remains bounded. This can be addressed by considering Figure 10 in a control theoretic framework as shown in Figure 11. Vs Vl + - + H e Il + Zl Figure 11: Negative feedback of the network impedance In this case, delta is a scalar (representing the no. of additional DGs) and the stability question reduces to identifying the bounds within which delta can vary without resulting in an unstable system. One approach is to use the robust stability 0 0 75 150 285 385 475 Distance from MV/LV transformer (m) Figure 12: Number of units in safe operation at increased distance from the MV node. Figure 12 shows the effect of the line inductance on the number of DG in safe operation. Connecting the units directly on the MV transformer (considering the inductance of the cables negligible) the number of possible parallel units is 5, while at 500m away from the transformer, this number is reduced to two units. It is evident that an increased inductance (both transformer and cables) leads to a reduced stability of the system. These numbers depend strongly on the DG control parameters and the filter values (L, R and C). They apply only for this specific inverter. Additional DG via transmission lines can be added to the node of interest (Figure 13) in order to make more realistic the simple configuration of Figure 10. The results obtained are similar to Figure 12 in case of short lines (<1km), while for lines longer than 1km the interaction of the new unit can be neglected. Figure 13: Additional connection of sources via transmission lines 4 CONCLUSIONS The presented case studies based on real data have shown that distributed power generation in the same power range as the load in the given network is feasible, specifically: When the generated power in the grid is significantly smaller than the grid power (e.g. several units in the low kW-range from PV adding up to 25% of the load), a “plug and power” scheme without detailed network analysis can be applied. At intermediate levels of power generation (up to 100% of the load), care has to be taken with the placing of the DG units. Depending on the network configuration, this may still be possible without alterations in the grid. Elevated levels of power generation (above 100% of the load) may still be possible. Alterations or reinforcement of the grid may be necessary depending on the existing configuration. A random placing of units at an elevated power level as in the presented case studies cannot be done (generated power equaled 125% of load). Several constraints were violated, yet they could all be met with some planning effort and optimization of the network. A large number of aggregated DG power on the LV network will affect all three areas that were discussed in this paper (network parameters, protection, stability). The impact on the distribution network depends largely on the aggregated power level, size and type of DG (synchronous generator/ inverter based), configuration of network, modes of operation (local supply/ selling power) and distribution of DG units. If the aggregated generated power is restricted to below 100% of the load in a low voltage – “microgrid”, requirements are simplified, as there is no reverse power flow to medium voltage networks. Based on a specific location, a network analysis may lead to general constraints that allow a “plug and power” scenario within certain limits (limitation of the aggregated power as well as a limitation of the DG size, requirements for cos(), restriction for reverse power flow and constraints on the proximity of DG). If any of these constraints cannot be met, detailed planning is needed. In such a case, some modern planning tools provide specific functions [14] to optimize distribution networks with DG (e.g., optimization of transformer tap settings, MV-separation points, reactive power, etc.). Modern DG units with power electronics can provide flexible cos() and can have a positive impact on voltage profile (ev. including local U/Q control) and losses. Regarding stability, the study has demonstrated that the number of units that can be connected on the same node without causing instabilities is strongly dependent on the network impedance at the point of common coupling as well as on the filter design and the control of the unit. This may limit a “plug and power” scheme. DG suppliers usually do not guarantee parallel operation with other inverter-based equipment, as stability can be compromised. If the DG do not feed back power to the utility but only serve as local supply support, there is no negative impact on the operation of the grid (But care has to be taken with short circuit current contribution, protection and stability). The customer can supply a large part of his own load without feeding back to the grid; the loading of the network is decreased. In this way, a significantly large portion of the power generation can be provided by DG. 5 REFERENCES [1] Peter Dondi et al., Network integration of distributed power generation, Journal of Power Sources, Special Issue Grove Fuel Cell Symposium, 2002 [2] Isabel Alvrez Ortega, Marco Suter , Interconnection of distributed Power Generation Resources to the European Low Voltage Electrical Grid, IASTED International Conference on Power and Energy Systems, 2001 [3] E. Lakervi and E.J. Holmes, Electricity distribution network design (London, Peregrinus Ltd.,2nd ed, 1995) [4] Planung und Betrieb städtischer Niederspannungsnetze, VDEW, Frankfurt, 1984 [5] Parallelbetrieb mit dem Niederspannungsnetz, VDEW, Frankfurt, 3. Edition 1991/ Reprint 1996 [6] Walter A. Elmore, Protective Relaying, Theory and Applications (New York, ABB Power T&D Inc, 1994) [7] N. Hadjsaid, J. Canard, F. Dumas, Dispersed Generation Impact on Distribution Networks, IEEE Computer Applications in Power, April 1999, 23–28 [8] CIGRE Working Group 37.23, Impact of increasing contribution of dispersed generation on the power system, Cigre report no. 137, February 1999 [9] M. Meyer and Jürg Schöning, Netzstabilität in grossen Bahnnetzen., Eisenbahn Revue No 7-8, 1999, 312–317 [10] P. Kundur, Power system stability and Control (McGraw Hill, 1994) [11] A. Stothert, E. Möllerstedt, A Model of a MicroTurbine Line Side Converter, IEEE-PES / CSEE International Conference on Power System Technology, Perth Australia, December 2000 [12] M. Green, D. Limebeer, Linear Robust Control (Prentice-Hall, 1995) [13] Audring, Netzbeeinflussung von BrennstoffzellenHeizkraftwerken zur Energieversorgung von Wohngebäuden, 2001 [14] Optimal Distribution Network in CALPOS User guide v4 (Mannheim, ABB Calor Emag, 1999), 7-1 to 7-5