CIRED
18th International Conference on Electricity Distribution
Turin, 6-9 June 2005
DISTRIBUTION NETWORK MODELS FOR STUDYING THE EFFECTS OF DISTRIBUTED GENERATION Henry LAGLAND, Kimmo KAUHANIEMI University of Vaasa - Finland
[email protected],
[email protected]
INTRODUCTION In this paper a collection of distribution network models is introduced. These simulation models are especially developed in order to enable detailed studies of future distribution systems where there is a relatively large amount of distributed generation (DG). By using computer simulations the effects of DG can be analysed before any installations. Also, with accurate simulation models the performance of existing and new apparatus and technical solutions in distribution networks can be trustworthy examined in advance with moderate costs. OBJECTIVES AND METHODS OF THE STUDY Electricity market liberalisation increases competition and the need to utilize distribution networks more efficiently. The increase of distributed generation transforms passive networks to active networks. However, the practices relating to the network are well established and considerable financial benefits would be needed if these practices are going to be changed. In order to develop DG and related solutions for global markets their suitability for local distribution networks must be verified. One way to do this is to apply simulation models tuned to represent distribution networks in various countries. This paper presents results from a study where the main aim has been to develop such models. The study is a part of a larger research project where both the University of Vaasa and VTT Technical Research Centre of Finland are participating.
paper the basic principles applied in the modelling and an example network with some simulation results are introduced. TYPICAL PARAMETERS OF MV NETWORKS Basing on the performed questionnaire and a literature survey the typical values of the key parameters were determined in [1]. A summary of the results is given in Table 1 (next page). These parameters define the set of network models and they are presented in details in the next chapter with some references to the survey results. Applying the gathered information a suitable classification for the MV networks was also developed. A similar classification for low voltage distribution networks can be found in [2], where they are classified in accordance to structure, operation, area type and network type. Figure 1 presents the proposal for the classification of medium voltage distribution networks based on the network configuration. Link arrangement systems, radial, open ring and satellite networks are operated radially while closed ring networks and primary network systems are operated meshed. In rural areas overhead line or underground cable link arrangement systems, radial, open ring and satellite networks are used while mostly underground cable open ring networks and primary network systems are used in urban areas. CONFIGURATION
In this paper the focus is especially on medium voltage (MV) networks. MV distribution network structures differ locally and from country to country. In order to develop an internationally covering set of distribution network models, an international questionnaire on MV network structures, applied techniques and methods in different countries has been performed. Nine countries from different continents responded to the questionnaire. The structure of MV rural and urban networks in different countries including the feeding network, the primary substation, grounding arrangements, network configuration, protection and distributed generation was reviewed. An extensive literature search was also made about the topic.
Figure 1. Proposal for medium voltage network classification.
Based on the fundamental characteristics of the network, this paper presents and defines a set of generic medium voltage network models. At first the typical parameters of medium voltage distribution network and their typical values in different countries are presented. After that the key parameters of the defined set of network models are reviewed. At the end of the
According to the questionnaire made all configurations except closed loop networks are used in both rural and urban areas. Closed loop networks use sectionalizing circuit breakers and need communications for network protection. Closed loop networks are rather complicated and expensive and are therefore expected to be used in the future in active networks.
CIRED2005 Session No 5
Radial network
Open ring network
Link arrangement system
Satellite network
Closed ring network
NETWORK OPERATION
Radial
Meshed
LOAD DENSITY
Rural
Urban
TYPE OF NETWORK
Overhead line
Underground cable
Primary network system
18th International Conference on Electricity Distribution
CIRED
Turin, 6-9 June 2005
Table 1. Typical network parameters used for modelling [1]. Values
Component
Parameter
Feeding network
Voltage Short-circuit power R/X Rated power Relative short-circuit impedance Secondary voltage Vector group Number Protection
Primary power transformer
MV distribution network
Earthing
Network wire system Network configuration
Feeder protection
Protection zones Short-circuit protection Earth-fault protection
Feeders: – number/substation – length – power – cross-section Secondary subNumber/ feeder station Power
Urban networks
Rural networks
66, 110, 132, 145 kV 5000 MVA 0,1 10–63 MVA 10–12 % 11, 13,8, 20, 33 kV wye-delta, wye-wye and delta-wye 2 Short-circuit Earth fault Differential Isolated Resistance earthing Resonant earthed Solid earthed Three wire, four wire Radial Open ring Link arrangement system Closed ring Primary network system Satellite Single zone Multiple zones Over current Directional over current Earth-fault Directional earth fault
66, 110, 132, 145 kV 1000 MVA 0,2 5–40 MVA 10–12 % 11, 13,8, 20, 33 kV wye-delta, wye-wye and delta-wye 1
10 5 km 5 MVA 240 mm2 Al 20 630 kVA
5 20 km 3 MVA 70 mm2 Al 20 160 kVA
THE NETWORK MODELS In the models the feeding high voltage (HV) network is considered only as an equivalent circuit seen from the primary side of the main transformer. The parameters of the feeding network are voltage, fault level and R/X ratio. The HV network voltages vary between 66 kV and 145 kV. Selected voltages for the network models were 66 kV, 110 kV, 132 kV and 145 kV. The fault level of the feeding network varies according to country, power company and load density. Selected fault levels for rural networks were 1000 MVA and 5000 MVA for urban networks. For the R/X ratio the value 0.2 was selected for rural networks and 0.1 for urban networks. The main transformer parameters are short-circuit impedance, vector group, rated power, primary and secondary voltage. Also the number of transformers in a primary substation is considered. The size of the main transformer varies according to country, power company and load density. In urban areas the power of the main transformer varies between 10 MVA and 75 MVA and in rural areas between 5 MVA and 40 MVA. In the models main transformers between 10 and 63 MVA are used. Corresponding short-circuit impedance varies between 10 % and 12 %. In rural areas one main transformer is common and CIRED2005 Session No 5
Three wire, four wire Radial Open ring Link arrangement system Satellite
Single zone Multiple zones Over current Earth-fault
in urban areas two or more. Common vector groups are wyedelta, wye-wye and delta-wye. The voltage of the MV network is usually between 6 kV and 33 kV. In most countries several medium voltages are used although there are some countries that use only one medium voltage. In rural areas higher voltage levels are used than in urban areas. Medium voltages selected for modelling were 11 kV, 13.8 kV, 20 kV and 33 kV. In most countries at least two earthing methods are used. Changes in environment, new knowledge and the development of earthing technology have challenged the original selections of earthing methods. Thus in the same utility more than one earthing method may be used. The earthing method of the network models can be selected as isolated from earth, resistance earthed, solid earthed or compensated. In the compensated network models the compensation degree and the value of the resistance connected parallel with the Petersen coil can also be set to desired values. Because in the rural areas networks are usually based on overhead lines and in urban areas there are mainly underground cables both of these network types are used in the models. The four wire system is used in wide area countries like South-
CIRED
18th International Conference on Electricity Distribution
Africa, India, Canada, Columbia, Pakistan, Tunisia and USA. Croatia is an exception of small countries using four wire systems. Three wire systems again are used in small area countries like Belgium, Spain, Italy, Portugal, France, Sweden, Germany, Finland and Switzerland. In Australia and Ireland SWER (Single Wire Earth Return) wire systems are used. In the network models both three wire and four wire systems are applied. In addition to the earthing method the network configuration influences distribution reliability and network protection. The radial and the open ring configurations are commonly used in rural networks, while the open ring configuration is most common in urban networks. The link arrangement system is used among others in India, China, Croatia, Norway, Poland and Finland. Satellite networks are used at least in Scandinavia, South-Africa, New Zealand and China. The chosen network configurations of the medium voltage distribution network models are the radial network, the open ring network, the link arrangement system network and the satellite network for urban and rural areas, the closed ring network for urban areas and the radial 4-wire network for rural areas. According to the questionnaire in all countries except in SouthAfrica definite time current relays were used for the MV feeder protection. According to the survey inverse time current relays were not so common. Directional over current relays were used in South-Africa, Sweden and India. In addition to the residual current based earth-fault relays also the directional earth-fault relays are used in isolated and compensated networks. DEVELOPMENT OF THE MODELS About the Simulation Tool The models are developed with the transient simulation tool PSCAD/EMTDC. This tool is especially suitable for studying the performance of relay protection which is one of the problematic issues relating to the wider application of DG. The simulations can be done either for producing realistic current and voltage waveforms in various fault situations or for detailed analysis of the behavior of specific relay functions. The generated waveforms can be applied for protection relay testing with a real time playback system. In the early stage of the project new functions for the protection relays were developed. The developed algorithms can be implemented as an integrated part of the simulation models and tested in a virtual environment. In addition to relay protection issues the transient simulation tool is applicable also when studying other events related to DG. These may be, e.g., the start-up or shut-down of a production unit. Also the spreading of harmonics generated from certain types of DG can be analysed accurately with this type of simulation tool. Principles Applied in the Modelling The aim in this project was to create a set of network models that represent well enough the typical real world networks. Since the models are primarily applied for studying the fault CIRED2005 Session No 5
Turin, 6-9 June 2005
situations it is important that the all the impedances in the fault path are included accurately. This means that in a radial network the lateral lines or cables and customer loads are less important from the modelling point of view. In practice this means that all or the main part of the laterals can be ignored and the main part of the loads can be represented as lumped models. The basic approach applied in this project was to model only one of the feeders in a substation with more details. In addition to this “Feeder 1” there is always “Feeder 2” where a lumped load model is applied. The full length of the feeder is included in the model but the total load of the feeder is modelled as a single one equivalent load at the end of the feeder. The idea is that in the simulations the focus is on the Feeder 1 but in case we want to study a case where the fault is in some other feeder, the fault can be made in the Feeder 2 at certain distance from the substation. The other feeders of the substation can be represented by a single equivalent circuit. This equivalent circuit includes the total load of all the other feeders and the total earth capacitance of the lines in these feeders. The latter is needed for correct representation of earth fault situation in earth isolated and resonant earthed systems. These principles applied in the modeling can be also seen in the example network presented in the next chapter. A way to limit the number of different network models was the “configurability” of the models. This means that some of the characteristics of the network model can be easily changed applying the user interface of the simulation tool. While the components of the system are fixed the user is allowed to decide which one of them are connected to the system. For this purpose there are several user interface components included in the model forming a “control center”. The user may, e.g., change the length of the Feeder 1 or the system earthing method. EXAMPLE MODEL As an example the 11 kV MV rural network model connected to a 66 kV HV system is described in the following together with a test simulation. The network configuration in this case is the open ring as shown in Figure 2. The fault level of the feeding network in this model is 1000 MVA with the R/X ratio of 0.2. There is only one main transformer at the primary substation. It is a 10 MVA delta-wye connected transformer with typical characteristics. The medium voltage network is compensated with a compensation degree of 1.0 and the resistive earth-fault current is set to 5 A. The Feeder 1 is loaded by 18 secondary distribution transformers with a rating of 160 kVA each. The main transformer and the secondary transformers are loaded up to 75 % of their rated power. The medium voltage feeder protection has three protection zones. For medium voltage bus infeed and network over current protection IDMT relays with normal inverse characteristics are used. For the earth-fault protection directional earthfault protection relays are used. The total number of feeders per substation is five and the length of the overhead line feeders is 20 km (132 mm2 aluminium conductors).
18th International Conference on Electricity Distribution
CIRED
Turin, 6-9 June 2005
HV network
R0 R1 Feeder 1
MV
R4 Background network
Feeder 2
C0
R2 MV load
LV load
MV load
R3
N/O Figure 2. Open ring MV network with multiple protection zones in Feeder 1.
In a test simulation a high resistance (500 ) phase-to-ground fault was initiated at the end of Feeder 1. In Figure 3 it can be seen that immediately after fault initiation the earth-fault relay is started and 0.2 s after fault initiation the first earth-fault relay upstream trips. According to similar test simulations performed with different fault locations the IDMT over-current and directional earthfault relays operated properly in all cases. It should be noted that there was no DG connected to the network in these test simulations. Adding various types of DG unit models in these network models makes it possible, e.g., to analyse the impact of the DG on the network protection as has been already done in [3], where only typical Finnish network models were applied. CONCLUSION In this paper the characteristic applied in a set of MV network models were presented. The aim has been to develop models that represent well enough typical networks in various countries. The developed network models will serve as a valuable toolbox when developing solutions for future distribution networks with large share of distributed generation
CIRED2005 Session No 5
Figure 3. Graphs from the earth-fault simulation. Above: residual currents and zero sequence voltage. Below: start and trip-signals from earth-fault relays. Earth-fault is applied at t = 0.4 s.
REFERENCES [1] Lagland H., 2004, Keskijänniteverkkojen analyysi mallintamista varten (Analysis of medium voltage networks for modelling purposes), M.Sc. thesis, University of Vaasa, (In Finnish) [2] A. Bertani, C. Bossi, B. Delfino, N. Lewald, S. Massucco, E. Metten, T. Meyer, F. Silvestro, I. Wasiak, 2003, “Electrical energy distribution networks: actual situation and perspectives for distributed generation” CIRED 17th International Conference on Electricity Distribution, Session 5, Paper No 65 [3] K. Kauhaniemi, L. Kumpulainen, 2004, “Impact of distributed generation on the protection of distribution networks”, Eighth IEE International Conference on Developments in Power System Protection, Amsterdam, Netherlands, 5-8 April 2004, pp. 315 - 318
