CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0690
EVALUATION OF ALTERNATIVE DISTRIBUTION NETWORK DESIGN STRATEGIES Chin Kim GAN Imperial College London – UK
[email protected]
Nuno SILVA Imperial College London – UK
[email protected]
Danny PUDJIANTO Imperial College London – UK
[email protected]
Goran STRBAC Imperial College London – UK
[email protected]
Robert FERRIS Central Networks Plc – UK
[email protected]
Ian FOSTER Central Networks Plc – UK
[email protected]
Martin ATEN E.ON Engineering Ltd – UK
[email protected]
A statistical distribution network design tool has been developed to quantitatively assess the impact of alternative distribution network investment plans in terms of capital costs, losses and reliability. In contrast to idealistic and network specific design approaches, this paper presents a statistical assessment approach in which the optimal network design policies are determined by evaluating the costs and benefits of different designs applied on a number of networks with similar properties such as consumer distributions, types, numbers and load density. Simulations have been carried out on multiple voltage levels (LV, MV and HV).
distribution network, as well as the optimal investment required depending on the number of substations used. For this purpose, a statistical tool has been specifically developed to generate different network layouts. Observations have been made concerning the optimal choice of underground cables, overhead lines, number of substations and transformer’s capacity depending on the network topology. The developed methodologies are applied in the attempt to define, develop and demonstrate design strategies that will ensure the most cost efficient solution that can provide reliable, cost effective and economic operation of distribution networks. Case studies examining the impact and possible benefits of direct 132/11kV transformation when compared to four voltage level design constituted by 132/33kV followed by 33/11kV transformation for typical urban cases are presented.
INTRODUCTION
ALTERNATIVE NETWORK DESIGN
Most of today’s electricity distribution networks in European countries were developed about 50 years ago to meet the fast growing industrialisation needs during that time. Much of the ageing plant in the network infrastructure is now approaching the end of its service life and eventually needs to be replaced [1]. In light of this, Distribution Network Operators (DNOs) in the UK have already started to invest in network replacement projects and this investment is expected to continue and increase over the next decade. This network infrastructure renewal process should be viewed as the precious opportunity to critically analyse the fundamentals of network design policy. In response to the climate change challenge and the government’s objective for a low carbon economy, the UK energy industry regulator OFGEM has envisaged that the policies set out in the latest Distribution Price Control Review 5 (DPCR5, 2010-2015) will encourage DNOs to reduce the carbon footprint. At present, DNOs’ activities make up approximately 1.3% of total UK greenhouse gas (GHG), with 97% of total DNO GHG emissions caused by electricity losses in the distribution networks [2]. Therefore, reducing the distribution electricity losses has a great potential in cutting down DNOs’ GHG emissions. In this work, the circuit and transformer ratings are optimised using a minimum life-cycle cost methodology [3] in order to estimate the optimal amount of losses in
Network design practices are an evolving process and are greatly influenced by customer distribution, geographical layout and load density at the time system is developed. Moreover, different regions adopted different network design and planning philosophies, such as 11kV voltage level versus 20kV, four versus three voltage levels design and etc. However, all these design philosophies meet the British Engineering Recommendation (ER) P2/6 which defines the minimum security of supply for various demand-groups that the system must meet [4]. The question arises as to what the optimal voltages and number of voltage transformation levels are for given network characteristics. Apart from that, the optimal number of substations, the types and ratings of the circuits and transformers used, network losses and reliability performance are amongst the important considerations in designing cost effective and sustainable electrical networks. The developed tool can be used to quantitatively evaluate alternative network design strategies with different network topologies and characteristics. Key network performance indicators such as network cost, losses and reliability are observed and analysed. Table 1 shows the typical voltage levels used in different European countries. Clearly, three voltage levels distribution networks are the preferred option amongst those countries.
ABSTRACT
CIRED2009 Session 5
Paper No 0690
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0690
Table 1: Typical distribution network voltage levels used by some of the European countries
Countries UK Germany France Finland Greece
transformers was kept at MV network. Figure 2 shows how different LV networks can be ‘entered’ into a MV network.
Typical voltage levels used 132 / 33 / 11 / 0.4 (kV) 110 / 20 / 0.4 (kV) 225 / 20 / 0.4 (kV) 110 / 20 / 0.4 (kV) 150 / 20 / 0.4 (kV)
DISTRIBUTION NETWORK MODELLING In order to evaluate network design strategies with different voltage levels, low voltage (LV), medium voltage (MV) and high voltage (HV) network models were developed.
LV network model The number of consumers, the types of consumer and how they are distributed in an area dictate the types and capacity of equipment used to supply them. Hence, previous research that used geometric models to represent the consumer point with assumed similar spacing between them may not be representative [5]. In order to better resemble the actual settlement of consumers with different layouts (urban, semiurban, rural, etc) and different connection path length, the LV network was created based on fractal science [6]. The way consumers are allocated can be controlled by setting the branching rate, consumer’s separation and other parameters accordingly. In addition, a realistic typical consumer type break down was used and it is divided into four major categories, namely (1) domestic customer without heating, (2) domestic customer with heating, (3) commercial and (4) industrial. The statistical similar networks are produced by setting different random sets of consumers with the same controlled parameters. The generated statistically similar networks will have different network layout, but the network characteristics in terms of load density, substation density and overall network lengths are maintained. By evaluating different network design strategies using many statistically similar networks enables meaningful conclusions to be drawn. Figure 1 shows a typical fractal model urban network with load density 5MVA/km2 and substation density 20sub/km2.
Figure 1: Fractal model of urban LV network
Figure 2: MV network grid-matrix
The MV customers are then connected with a controllable branching rate. Figure 3 shows a MV network filled with 65 LV networks. It is then connected with 69 percent of branching rate. The input LV networks have load density ranging from 5MVA/km2 to 25MVA/km2. The small ‘dots’ are MV/LV transformers and the ‘red stars’ being HV/MV transformers.
MV network model Essentially the network characteristics of MV networks are driven by the LV networks. The MV network will supply the MV/LV transformers as well as some industrial customers. It is also important to note that the load density of LV networks within the MV network vary from region to region. Thus, to address this situation, the MV network was modelled by inputting different sets of LV networks which can have different load and substation density into a gridmatrix. The location of MV/LV transformers and their annual loading profiles for each of the MV/LV transformers are recorded in the LV networks and become the input parameter of the MV network. By doing so, the loading characteristics and the distances between the MV/LV CIRED2009 Session 5
Paper No 0690
Figure 3: MV network (200,000 LV consumers, 300MVA-peak, 6MVA/km2, 0.3sub/km2) with 65 LV networks input
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0690
HV network model The HV network is simply modelled with loads (HV/MV transformer) assumed to be lumped at the end of the feeder. In order to satisfy the N-1 reliability requirement of ER P2/6, double circuit radial and double transformers arrangement are used as shown in Figure 4. The total HV network length was assumed to be at about 20 percent of the total MV network length and this can be adjusted accordingly.
primary substations are needed. Network losses could be reduced as a result of lower transformer fixed losses. Nevertheless, the transformer fixed loss reduction might be offset by the increase in circuit losses due to the fact that longer feeders are needed to supply customers with fewer substations. The three voltage levels design was approached by replacing the 132/33kV substations with 132/11kV substations. The 11kV feeders will be longer and the 132/11kV substation capacity will be of equivalent to 132/33kV. In order to have statistically meaningful conclusions, ten different networks with similar network characteristics were simulated for each of the network design strategy. In any case, if the design violates thermal or statutory voltage limit, more substation(s) will have to be added. The primary substation’s transformers are modelled with on-load-tap-changer (OLTC) capability, whereas 11/0.4kV transformers operate on fixed tap.
Results for urban case Figure 4: HV network model
NETWORK DESIGN METHODOLOGIES Hourly balanced three-phase load flow calculations were performed throughout the year using typical characteristic load profiles for four different customer types. Once the network loading is known after the load flow calculation, network sizing was done based on the minimum life-cycle methodology [3, 7]. It minimises the annualised investment and maintenance cost of equipment, together with the yearly cost of electricity losses to determine the optimal capacity of the circuits. The investment cost was assumed to be over 30 years with 7% rate of return. The optimal circuit sizing updates the network with new circuits and transformer parameters. The total network losses comprising circuit losses and transformer losses (fixed and variable) were computed next. Fault level calculation was also performed to make sure the circuits meet short circuit thermal limit in the worst case scenario. The annual loading for the supply points were recorded to become the input to the higher voltage levels. All designs meet statutory voltage requirement and minimum level of security supply required by ER P2/6 in the UK [4]. The reliability evaluation approach used was based on the failure event oriented technique [8], applicable to radial networks.
CASE STUDY Four versus three voltage levels design The UK electricity distribution system has typically four voltage levels (132/33/11/0.4kV) design. However, it is interesting to explore the possibilities of three voltage levels distribution design (phasing out 33kV level) as practised by other European countries. Avoidance of one voltage level has the potential to reduce network investment cost as fewer CIRED2009 Session 5
Paper No 0690
By phasing out the 33kV network, larger capacity of 132/11kV (compared to 33/11kV) substations with longer 11kV feeders will be needed to supply the same number of customers (11/0.4kV substations) as shown in Table 2. Table 2: Example of different transformer sizes used to supply 11kV urban networks Design strategies Transformer sizes [MVA] 15 24 32 45 60 90 Four voltage levels 2 4 8 0 0 0 Three voltage levels 0 0 0 2 2 2
As a consequence, the 11kV network with fewer substations will be more heavily loaded which in turn requires cables with higher capacity. Figure 5 shows that the 11kV network with three voltage levels design requires much longer 300mm2 cables to supply the customer.
Figure 5: Breakdown of 11kV cable cross-section
For the system considered, even though phasing out the 33kV network causes 11kV feeders to become longer and more heavily loaded, the voltage profiles are still well regulated by the 132/11kV OLTC transformers, because of the generally shorter feeders in urban areas. Simulation results show that all the 11kV buses are well maintained above the statutory voltage limit of ±6% [9]. Figure 6 shows the network cost and network losses of 11kV network supplied by 132/11kV substations for ten
CIRED
20th International Conference on Electricity Distribution
Prague, 8-11 June 2009 Paper 0690
different statistically similar networks. The standard deviation for network cost and network losses is six and seven percent respectively, which highlighted the robustness of the results obtained.
Figure 6: Network cost and network losses for different statistical similar networks
Table 3 shows the key network performance indicators, namely, cost, losses and reliability for both network design strategies (average value of ten statistical similar networks). Table 3: Overall result summary for urban network adopting two different design strategies
Transformers 11/0.4kV 33/11kV 132/33kV 132/11kV Losses 11kV circuits 33kV circuits 33/11kV
Three voltage levels Four voltage levels design design Number of substations used 724 724 0 14 0 4 6 0 Network losses (%) 0.45 0.22 0 0.11
(Fixed + variable)
132/33kV (Fixed + variable)
132/11kV (Fixed + variable)
Total Losses Incentive £48/MWh
0
0.45
0
0.30
0.35
0
0.80 1.08 Regulatory losses incentive (million £/year) 0.17 0
Reliability (11kV network)
CI CML Network cost 11kV 33kV Total
CI (int/100cust.year), CML (min/year) 39 17 161 72 Annualised network cost (£/kWpeak) 14.4 13.4 0 8.5 14.4 21.9
Owing to the removed 33kV network, the transformer losses for three voltage levels design are lower compared to four voltage levels design. However, due to longer 11kV feeders in the three voltage levels design, the circuit losses are roughly doubled. Due to the savings in transformer losses, the total network losses are still lower for the three voltage level design strategy. Apart from that, longer feeders have also led to poorer reliability performance in the three voltage levels design. Nevertheless, the reliability CIRED2009 Session 5
Paper No 0690
performance can be improved by some means of feeder control and automation. In the presented studies, the use of fewer primary substations in the three voltage levels design for typical urban areas has resulted in more cost effective network investment which has a cost saving of about £7.5/kWpeak a year over the next 30 years. The model could be easily extended to evaluate the network performance of typical rural areas and other types of network typology for a given design strategy. Note also that the UK regulatory losses incentive used in this work was based on DPCR4 [10], and this value should be updated based on the next price control review. The developed model can be adapted for any future changes with this regard.
CONCLUSIONS A software tool was created to allow evaluation of alternative distribution systems planning strategies for LV, MV and HV networks using many similar realistic consumer settlements and networks. Optimum design of the networks was determined by minimizing costs of equipment, maintenance and losses, while meeting all the technical and statutory constrains. The results suggest that network investment into a three voltage levels design for typical urban areas is likely to be more cost effective and has the potential to reduce network losses. However, the reliability performance for such design must be improved. This is consistent with the network design strategies adopted by other European countries. REFERENCES [1] E. Van Geert, 1997 "Towards new challenges for distribution system planners," CIRED 97, Birmingham, UK. [2] OFGEM, December 2008, "Electricity Distribution Price Control Review Policy Paper". [3] S. Curcic, G. Strbac, and X. P. Zhang, 2001, "Effect of losses in design of distribution circuits," Generation, Transmission and Distribution, IEE Proceedings-, vol. 148, pp. 343-349. [4] Engineering Recommendation P2/6, "Security of Supply," Energy Networks Association 2006. [5] T. Gonen, 1986, Electric Power Distribution System Engineering, McGraw-Hill series in electrical engineering, USA. [6] Peitgen and Heinz-Otto, 1988, The Science of Fractal Images: Springer-Verlag. [7] D. Melovic, 2005, "Optimal Distribution Network Design Policy." PhD Thesis, UMIST. [8] R. Billinton and R. N. Allan, 1996, Reliability Evaluation of Power Systems, 2nd ed., Plenum Press, New York. [9] The Electricity Supply (Amendment) (No. 2) Regulations 1994. [10] OFGEM, 2004 "Electricity Distribution Price Control Review (DPCR4)".
