Selected paper presentation (and update) 5.68

LV NETWORK SIZING IN ELECTRIFICATION PROJECTS - REPLACING A DETERMINISTIC METHOD WITH A PROBABLISTIC METHOD

Ian Ferguson, C T Gaunt - South Africa

A large scale national electrification programme provided new access to electricity for 30% of the households in the country. As part of the programme the national electricity utility, Eskom, connected nearly two million new households between 1994 and 2001. Many of the new customers are rural households and many more still need to be connected. They are small consumers, with gradual load growth, and the costs of connection are unlikely to be recovered in the standard tariff. Therefore, the capital costs of the connections must be the minimum that will provide for five years, and the networks can be reinforced later when the need develops.

The process needs:

  • suitable load models, and
  • calculation methods that are a good representation of operational conditions.

Our paper describes the change that was implemented during the electrification programme: from a deterministic approach to calculating the critical parameter of voltage regulation, to a method based on a probability approach.

The difference between the two approaches is fundamental. The deterministic approach considers that the load of the typical domestic customer is a representative value, called the after diversity maximum demand (admd), to which factors can be applied to give an adequate design. The probabilistic approach considers, instead, that the load is not accurately known but can be represented by a probability model. Further, a right answer for the voltage conditions on the network also cannot be known, but can be represented by a probability distribution. The network designer then makes choices based on the understanding of the probabilities.

The probability-based load models, and the collection of the data on which they are based, were reported at CIRED in 1991 and 1999. The parameters of the Beta probability model of customers’ loads are linked to the socio-economic character of a community.

The main concern was what effect a change in approach would have on the sizing of feeders, the costs and the business implications. Redesign and re-costing of a recently completed electrification project was used to test the concerns.

A locally developed software package capable of calculating both design approaches was used. The actual network was loaded in the software until voltage regulation or transformer loading conditions exceeded the limits used in the design. The extent of reinforcement was recorded and the cost estimated.

It was quickly evident that the different approaches to calculating the voltage conditions on the feeders led to different feeder sizes and costs. So a business decision had to be taken.

The probability based approach was adopted as a business standard in January 2000, supported by suitable software tools. Although business “factors” had been considered in the adoption plan, there were several surprises requiring specific responses, including:

  • Resistance to change from the design groups, which was addressed by training and providing indices to check the acceptability of their designs.
  • Various design tools (software packages with different engines) gave differing results, so benchmarks were prepared to test them.
  • The existing load forecasting tool needed to be revised to improve the parameters generated for low-income rural communities.
  • The new approach provided much better control of turnkey contracts, and suitable quality control procedures were adopted.

Overall, the designers became more knowledgeable and comfortable with the new tools.

The use of the new approach appeared to lead to a 5% increase in the initial cost of feeders, but ‘savings’ provided by other packages are false and lost as soon as the loads develop. The new approach reduces the risks associated with the other approaches.

The introduction of a new method of sizing low voltage feeders appears to be largely a technical issue. However, the actual implementation showed that choosing the size of the feeder conductors is only one part of a complex business system needing attention. It also demonstrated how a change in technology can be a catalyst to improving other business elements, including the knowledge of the users.

Development of better approaches continues and papers 5.27 and 5.28 indicate some of the further technical advances. Continuing load measurements provide important input on how communities are developing as electricity customers. Most of all, the methods, data and understanding were essential for recent research into introducing a social tariff to support poor households. The new tariff being considered would be very risky without the knowledge that has developed of load modelling and network analysis based on probabilistic methods.

GAUNT_ZA_author_ALPHA5_BLOCK2_fPage 1