Graph based method for critical pipe analysis in urban drainage networks and the effect of loop degree

Abstract

[EN] Urban drainage network (UDN) is an important lifeline for urban areas, without which living conditions as we know it cannot be imagined. Internal and external pressures on these systems like structural failures, climate change and urbanization, require efficient modelling strategies for its management and maintenance. For proper functioning of the system as a whole, some elements of UDN infrastructure hold more importance than other elements. Identifying these critical elements in UDNs is of utmost importance for forming efficient management strategies. In this study, a graph-based method based on “runoff edge betweenness centrality” is presented for determining the critical elements (pipes) of a UDN. In contrast to conventional hydrodynamic modelling method, proposed graph-based method does not rely on iterative hydraulic simulations. Instead, it takes a mathematical graph representing the structure of the network as starting point and incorporates hydraulic factors to mimic the hydraulic behavior of UDN. Effect of loop degree on this graph-based method is studied by employing the method for a fully branched network, partially looped and fully looped network. A fully branched network is employed as a case study comprising of 59 circular pipes, draining an area of 175 hectares. Loops are manually created in the same network to study the impact of different number of loops on the accuracy of graph-based method. The results from the graph-based method are compared to the hydrodynamic modelling results from SWMM model in terms of accuracy as well as computational time and effort required. Results show that the graph-based method is very accurate (99%) in case of a fully branched network but the accuracy decreases with the increase of loop degree in the network. This is due to the complex hydrodynamic effects like back-water phenomenon which cannot yet be modelled using graph-based approach. The method is hence least accurate (93%) in case of fully looped network though still very useful, because the computational time and effort required is much less than hydrodynamic modelling method. The proposed method is also less data intensive as compared to HMM and hence can be used by utilities where quick analysis is required.