CIRED 2003
Author update to Session 5 Paper 56
“Using GIS data to assess network models and performance”
Leather P., Clarkson, P.L., and Hodgkins W.R.
One aspect of the paper deals with the use of GIS data to determine the most appropriate models for fault rates when estimating the reliability of the MV underground network. The availability of joint data from the GIS along with historic fault data for each circuit enables the incidence of faults to be correlated both with cable length and with the number of joints. When the paper was written the results of the study were not complete and the intention was stated to report further at the conference.
Although GIS data for the whole of the United Utilities distribution network is incomplete, it was possible to use MV network data for some circuits in one region where there was a relatively large incidence of recorded underground faults. The existing reliability model assumes that fault rate in each isolation section is proportional to cable length in that section. This model was compared with one where the fault rate was assumed to be proportional to the number of joints in the isolation section. Using the Maximum Likelihood method the latter gave a better fit. However, there was uncertainty in deciding on the exact location of the fault for a significant proportion of the fault data using the grid reference or other recorded information. Furthermore, there was a strong correlation between the number of joints and cable length as the maximum length of cable is normally only about 300m. When estimates of the Reliability Indices based on the two models were compared, the results fell well within the natural statistical variation. The analysis did show a tendency for repetitive faults to occur within the same section. This would have no effect on the estimates of average performance but would increase the probability of multiple interruptions. The results emphasise that data needs to be integrated and sufficient information needs to be recorded to enable the faults and other incidents to be located accurately on the GIS network. Making full use of a GIS may demand substantial improvements to other data systems and this evaluation should not be deferred until the GIS is complete.