EXAMINATION OF THE IMPACT OF POSSIBLE DISTRIBUTION ...

CIRED

20th International Conference on Electricity Distribution

Prague, 8-11 June 2009 Paper 0950

EXAMINATION OF THE IMPACT OF POSSIBLE DISTRIBUTION NETWORK DESIGN ON NETWORK LOSSES Ding-Mei CAO Danny PUDJIANTO Imperial College and EDF Energy-U.K. Imperial College London-U.K. [email protected] [email protected] Robert FERRIS Ian FOSTER Central Networks Plc-U.K. Central Networks Plc-U.K. [email protected] [email protected]

• installing reactive compensators at different voltage levels at rural and urban networks.

ABSTRACT In the present climate where reducing CO2 emissions is of increasing importance, reducing losses is becoming one of the important factors considered in distribution network design and operational strategies. This paper simulates various potential network design scenarios using generic distribution network models and evaluates their impact on network losses. INTRODUCTION To meet its obligations for CO2 reduction under the Kyoto protocols, the UK is making considerable efforts to improve energy efficiency. The gas and electricity regulator Ofgem introduced a new value to losses in the loss adjustment factor in its 2005-2010 Distribution Price Control Review and encouraged Distribution Network Operators (DNOs) to apply energy efficient equipment in their networks from March 2005. The concern for losses will increasingly be an important factor in distribution network design. A large proportion of UK distribution network assets are entering the end of their useful life and need to be replaced since they were installed in the 50’s60’s. Thus we now have a rare opportunity to review the present design of distribution networks to see whether the traditional network design is still suitable for the future. Various possible design strategies have been proposed, however, how these designs will impact the losses have not yet been investigated. In this context, this paper develops preliminary studies to assess and quantify impacts on the distribution network losses from a number of proposed future development scenarios using developed models, taking into account annual load, network structure characteristics as well as network technical constraints. The studies also investigated the use of reactive power compensation, as a means of increasing utilisation and efficiency of distribution networks which has not been widely applied in the UK. The scenarios include: • phasing out 33kV voltage level networks, • increasing the number of 33/11kV substations and shortening the length of 11kV feeders, • increasing the conductor sizes and hence changing the utilisation of circuits for 11kV networks,

CIRED2009 Session 5

Goran STRBAC Imperial College London-U.K. [email protected] Martin ATEN E.ON Engineering Ltd-U.K. [email protected]

Paper No 0950

GENERIC DISTRIBUTION NETWORK MODEL Considering the characteristics of the distribution network, the following common features can be derived for distribution network modelling: • A variety of distribution circuits are modelled based on the type, impedance and the degree of tapering. • A variety of transformers are modelled based on the rating, impedance and losses. • The loads are also distributed in various ways including linearly increasing, linearly decreasing or uniform distribution along the feeders [1,2] or lumped at the end of the feeders. According to those parameters, linearised optimal power flow equations and hence the voltage calculations for the different voltage levels, are developed. Loss and reactive power calculations can also be included. This modelling approach is applied to a generic UK-like Grid Supply Point (GSP) based radial distribution network model [3,4] shown in Figure 1. 400kV or 275kV 132kV

132kV

33kV

11kV 11kV

0.4kV 0.4

Load

Figure 1 Generic distribution network model

This model was primarily designed to reduce the complexity of modelling the whole distribution network but also to preserve all the major characteristics of the UK distribution networks in terms of length, capacity, tapering and number of circuits at each voltage level, and of substations’ capacities and electrical parameters. The model is also flexible enough to simulate reactive compensators to be connected at different voltage levels in the network. The model was then configured to represent average, rural and urban GSP networks, which respectively capture the current average, rural and urban characteristics with regard to the parameters of feeders,

CIRED



transformers and load at each voltage level in order to allow different case studies to be investigated. Hourly demand profiles for each season considering various types of customers were used to model the load connected to the distribution networks. IMPACT OF PHASING OUT 33KV ON LOSSES Direct transformation between 132kV and 11kV is currently the common approach taken in areas of high load density. This design strategy, however, could also be applied to other areas in order to reduce the amount of transformation necessary to distribute electricity from the GSP to the low voltage consumers. Therefore some voltage levels such as 33kV and 66kV are being considered for removal by some of DNOs by having a direct 132/11kV transformation. This is also in agreement with strategy for dealing with projected future load increases. Furthermore replacing ‘old’ 132/33kV and 33/11kV transformers with modern 132/11kV transformers could deliver an additional reduced losses benefit due to the fact that modern spec transformers with laser-etched grain oriented steel cores have much lower per unit losses (especially Fe losses) than typical 1950’s hot rolled steel core transformers of similar capacity.

The results of these two case studies show that the total distribution losses of the present networks and of the future networks with the 33kV has been phased out are 6.12% (present), 4.85% (case 1) and 5.33% (case 2) respectively. For case 1, the losses are reduced by 1.27 % or about 14.4 GWh/year. For case 2, the losses are reduced by 0.8% or 9GWh/year. Assuming there are 200 GSPs in the UK distribution networks, up to 2878.6GWh and 1799.4GWh losses can be saved each year for the new network structures of case one and two respectively. A comparison of losses between the original network design and the two proposed design cases broken down by voltage level is shown in Figure 2. About 1.2% of the losses can be saved by removing 33kV networks (including 132/33kV substations and 33kV circuits) at the expense of the increase in losses in other voltage levels. In case one, the 132kV was extended to cover the length previously covered by the 33kV network. The increase in losses in the 132kV network is negligible compared with the saving brought about by removal of the 33kV network. 1.8 1.6

CIRED2009 Session 5

Paper No 0950

1.4 1.2

Losses(%)

To assess the removal of 33kV voltage level networks on the whole distribution system losses, two case studies were carried out on a new network model formed by phasing out the 33kV networks from the generic GSP average network model. The new network design is also based on the standard equipment information which is available from [5,6] In the first case (case 1), it is assumed that 132kV networks will be extended by 6 – 18 km (typical 33kV network lengths) to reach the existing 11kV networks. 11kV parameters stay the same except that some of circuit capacities were adjusted to satisfy the statutory voltage standard. The second study (case 2) uses an opposite assumption. Instead of extending 132kV feeders, 11kV networks were extended by 6-18 km. When 11kV circuits are set longer, this situation causes more voltage drop problems in heavy loading and long circuits and large size conductors are required to maintain the statutory voltage level. For both cases, 0.4kV network models remain the same, which means the losses at 0.4kV are unchanged. Two 132/11kV substations with 2*60 MVA three-winding transformers and four 132/11kV substations with 2*30MVA two-winding transformers are selected to replace the original 33/11kV substations. It should be noted that to maintain the fault level within switchgear ratings, three winding transformers in the direct transformation sites from 132kV to 11kV are operated independently. This is in contrast to the two winding transformers normally operated in parallel pairs [6]. The loads which were previously connected to the removed 33kV networks are transferred to the closest updated 132/11kV substation.

1.0 0.8 0.6 0.4 0.2 0.0 0.4kV circuits

11/0.4kV transformers

11kV circuits

Original average network New network without 33kV system case 2

33(132)/11kV transformers

33kV circuits

132/33kV transformers

132kV circuits

New network without 33kV system case 1

Figure 2 Comparison of losses in various network components between current network and networks phasing out 33kV system

Losses in the 11kV circuits slightly decrease because the re-allocation of 11kV network loads to the 132/11kV substation requires a few 11kV circuit reinforcements to avoid voltage problems. Thus, the level of 11kV circuit losses is slightly smaller than the one in the original case. In case two, the 11kV circuit extension makes the losses in this asset increase by 0.43% compared with the losses in the present 11kV system. The increase in losses is not excessive since the conductor size of 11kV circuits needed to be increased to avoid voltage problems. IMPACT OF THE LENGTH OF 11KV FEEDERS AND NUMBER OF 33/11KV SUBSTATIONS ON LOSSES Considering that losses in 11kV circuits are about 13% to 32% of the overall distribution losses [3], shortening the 11kV circuits might be considered as one possible future solution to reduce the losses in the distribution network.

CIRED


Prague, 8-11 June 2009

In order to examine the impact of this strategy, a study was carried out with the following scenario: the 11kV circuit lengths were reduced by half and the number of 33/11kV substations was doubled compared with the present system. Since the supply area is reduced, the circuit and substation capacity was set to half of previous values. As 11kV circuits were shortened, the 33kV circuits were extended by the average reduced value of the 11kV circuits to maintain network length sufficient to reach to the customers. All the design practices follow the security standard. The component losses in each network voltage level are shown in Figure 3. 1.6 1.4

Losses(%)

1.2 1.0 0.8 0.6 0.4

Losses (%)

Paper 0950

10 9 8 7 6 5 4 3 2 1 0

0%

-5%

-10%

As expected, increasing conductor size by a given factor reduces utilization and losses by the same factor. It is important to bear in mind that the increase in conductor size will increase the cost of investment and hence, such network design strategy to use larger conductor sizes with the objective to reduce losses has to be justified using cost benefit analysis. IMPACT ON LOSSES OF REACTIVE COMPENSATORS

0.2

-15%

Reduction in utilization of 11kV network Figure 4 Rural network losses with different network utilisation reducing rate

INSTALLING

0.0 0.4kV circuits

11/0.4kV transformers

11kV circuits

Original network


33kV circuits


132kV circuits

New network with half 11kV circuit length

Figure 3 Different network assets losses of present network and the network with shortening 11kV circuits

It was found that shortening the 11kV feeders reduces the losses by only 0.16%. The majority of loss reduction occurs in 11kV feeders, which is half of the initial value. However, it is almost offset by the increased losses in 33kV circuits due to the increase in their length. Hence the overall benefit is relatively small. IMPACT ON LOSSES OF INCREASED CONDUCTOR SIZE OF 11KV CIRCUITS Another possible future scenario is to increase the conductor size of 11kV circuits while maintaining the present distribution network architecture. By increasing the conductor sizes, the electrical resistance of the circuits will drop while the reactance is similar. Thus, the losses can be expected to be smaller at the expense of lower utilisation of the assets.

One of the important functions of reactive compensation is to reduce the network losses by reducing reactive power flow in networks above the compensator connection point. This study evaluates the value of this service. Figure 5 and Figure 6 respectively show the break down of the annual losses in the generic urban and rural distribution networks at each voltage levels with various power factors. The reduction of losses at various voltage levels due to the power factor improvement can be derived and is shown as the chart of Figure 5. Reduction of 0.4kV losses by improving power factor from 0.85 lagging to unity power factor

Figure 5 Percentage of losses in the generic urban network

A number of case studies were carried out using the generic rural distribution network model. The largest impact will be on the level of losses of 11kV networks. Losses at 33kV and beyond are only slightly affected (there is a small change because 11kV losses are a load for 33kV and beyond). The losses at 0.4kV remain the same. The result of the study is illustrated in Figure 4, which shows the changes in losses and utilisation factor of 11kV circuits as the conductor sizes of the circuits are increased.

CIRED2009 Session 5

Paper No 0950

Figure 6 Percentage of losses in the generic rural network

The possibility of excessive use of reactive compensation (shunt capacitance) leading to harmonic resonance problems is not considered in this study. Figure 7 illustrates the total amount of reactive power

CIRED



compensation that needs to be installed in the generic distribution network (for one GSP) to improve the load power factor to unity.

Figure 7 Required capacity of reactive compensators for power factor improvement

This study adopted a simplified approach to determine the reduction in energy losses as a function of the voltage level at which reactive compensation devices are connected. This simplification will tend to produce optimistic results as it was assumed that the full benefits can be obtained at the compensators connected voltage level and all upstream voltage levels. The losses reduction (for one GSP) due to reactive compensators installed at various voltage levels are shown in Figure 8 and Figure 9 for generic urban and rural networks.

factor improvements are presented in Table 1. Table 1 Capitalised cost of losses savings in urban and rural networks due to the installation of reactive compensator at various voltage levels

Power factor improvement 0.85 lag to 1.0 pf 0.90 lag to 1.0 pf 0.95 lag to 1.0 pf

0.4 kV 58.7 46.3 32.4

Capitalised losses saving ( '000 £/MVAr) Urban network Rural network 11 33 132 0.4 11 33 kV kV kV kV kV kV 26.1 12.6 2.4 140.1 90.1 39.8 20.9 10.4 2.0 110.8 71.7 32.7 15.0 8.0 1.6 78.1 51.5 25.1

132 kV 9.8 8.2 6.5

Depending on the network characteristics, voltage level connection and level of power factor improvement, the value varies significantly between £1,616/MVar to £140,128/MVar. The value for rural networks is higher than for urban networks due to the bigger loss reduction in the rural networks (as shown in Figure 8 and Figure 9). As alluded in the assumptions, the accumulated losses saving and therefore the value for installing reactive compensators at lower voltage levels becomes higher. Also, it is clear that increased power factor improvement brings increased savings in losses. Note that these savings must be compared to the capital cost of equipment and it is unlikely that with the present cost of reactive compensators this would be cost effective to achieve power factor of unity. ACKNOWLEDGMENTS The authors gratefully acknowledge the valuable comments from Dave Openshaw, Jonathan Purdy and Oliver Day on this paper. CONCLUSION

Figure 8 Total losses reduction (1 GSP) due to reactive compensator installed at various voltage levels in generic urban networks

This paper described the impact of various distribution network design scenarios on network losses. It was found that phasing out the 33kV voltage level network is expected to be more beneficial for urban systems which have shorter 11kV circuits. In the case of increasing the number of 33/11kV substations and shortening the length of 11kV feeders, it was observed that the change in overall losses was small. Also, the value of installing reactive power compensation on active loss reduction is much larger in rural networks than in urban networks. REFERENCES

Figure 9 Total losses reduction (1 GSP) due to reactive compensator installed at various voltage levels in generic rural network

The benefit of a reactive compensator in active loss reduction at each voltage level can be calculated using the value of the total energy savings due to loss reduction (at and above that voltage level) over the life time of the equipment and the required capacity of the compensators. It is assumed that the average lifetime of equipment is 20 years and considered that the Ofgem losses incentive rate is £48/MWh and the interest rate is 6.9% [7]. Using the value obtained from Figure 7 to Figure 9, the present value of capitalised losses savings for various power

CIRED2009 Session 5

Paper No 0950

[1] Liew. S, September 2002, "Technical and economic assessments of active distribution networks", PhD Thesis, UMIST, UK. [2] Conti. S, Raiti. S, Tina. G, Vagliasindi. U, September 2001, "Study of the impact of PV generation on voltage profile in LV distribution networks", Proc. IEEE Porto Power Tech Conference, Porto, Portugal. [3] Grenard. S, 2005, "Optimal Investment in Distribution Systems Operating in a Competitive Environment", PhD Thesis, UMIST, UK. [4] Grenard. S, Pudjianto. D, Strbac. G, June 2005, "Benefit of active management of distribution network in the UK", Proc. CIRED 18th International Conference on Electricity Distribution, Turin, Italy. [5] Aquila Networks plc, November 2002, "Long term development statement 2002-2007", UK. [6] EDF Energy, 2005, "Long Term Development Statement Summary". [7] Ofgem, November 2004, "Electricity Distribution Price Control Review, Final Proposals", UK.