Evaluation of the uncertainty of the length of energy distribution networks

Abstract

A statistical evaluation of the uncertainty of the length of a utility network is frequently omitted in current geographic information system (GIS) applications, although this information is important for management purposes. Nevertheless, the spatial database storing geographic information about the network might have been set up through a complex process (ground surveying, digitising of analogue maps, integration of different data sources), resulting in a dataset with unclassified uncertainty. This paper proposes a practical but ‘rigorous’ approach to evaluate the influence of the uncertainty in the determination of the length of a network. Starting from the definition of length uncertainty in terms of simple geometric elements (arc, pair of arcs, open and closed path), some rules for its evaluation in the case of complex networks have been developed. The statistical concept applied for the evaluation of uncertainty is quite standard because it is based on the covariance propagation theory. On the other hand, the influence of correlations between the lengths of adjacent elements has been considered as well. The application of this method only requires the topological structure of the network that can be built in a GIS environment. Furthermore, an alternative ‘approximate’ method is proposed to find an approximation of the uncertainty of the network length with a simple formula, which requires neither any particular processing step nor the network topology. Both methods have been tested and validated on simulated and real datasets.