Lifecycle operational resilience assessment of urban water distribution networks

https://doi.org/10.1016/j.ress.2020.106859Get rights and content

Highlights

  • We study the lifecycle operational resilience framework of water distribution networks.

  • The proposed framework considers daily accidents, pipe deterioration and restoration.

  • The disruptions are simulated by leakages and bursts based pipe accidents.

  • The proposed framework is applied to study a real system in city Mianzhu.

  • Six resilience strategies are investigated to conclude resilience enhancement guidance.

Abstract

This paper proposes a lifecycle operational resilience assessment framework of urban water distribution networks (WDNs), taking accidents, pipe deterioration, and restoration into account. First, the accident occurrence rates for three common types of pipes, namely, cast iron pipes (CIPs), ductile iron pipes (DIPs), and steel pipes (SPs), are fitted using the maintenance data provided by a water administrative sector. Second, for the two most common accidents, i.e., leakages and bursts, the accident pipes in the former case are simulated by reduced flow, while the accident pipes as well as the influenced ones in the latter case are isolated completely by closing valves, which are determined by a depth-first search method. Third, two restoration strategies are considered, including plugging and replacement, and the corresponding recovery time by days are given based on the maintenance data. Meanwhile, the aging induced pipe deterioration is modeled in the operation process. Finally, the operational resilience of the WDN in Mianzhu city is evaluated by above framework and six resilience strategies are investigated and compared. Results show that replacing accident pipes with new DIPs once burst occurs improve the operational resilience of WDNs most obviously. In addition, replacing CIPs with DIPs ahead of the design working life or adding valves also help to improve the resilience level. Based on the findings, some feasible and practicable resilience enhancement suggestions are concluded to provide guidance for local decision makers.

Introduction

As an indispensable lifeline engineering system, urban water distribution networks (WDNs) provide domestic and industrial water for modern cities [1]. However, due to the widespread distribution and complex operation environment, the WDNs inevitably suffer from various operational accidents throughout the lifecycle, such as leakages and bursts. Unlike natural disasters, these accidents result from deterioration, climate change, third-party destruction or other man-made factors and hence occur frequently in both developed and less developed areas. According to China's statistics in 2000–2003, the number of operational accidents in water peak period, including bursts, leaks or water pollution, is more than 20,000 with the influenced people of 38 million. In addition, more than 20 million people suffer from seriously insufficient water head (i.e., water pressure with the unit of m) for long, which are intensified by even small accidents due to the unreasonable topology [2]. These accidents can lead to huge repair cost and secondary accidents such as traffic congestion, which impacts government image and restricts urban development. Therefore, maintaining, repairing, and improving the WDNs based on the operational conditions are concerned by urban decision-makers.

Previous researches of above problem focus on the reliability of WDNs [3], but now emphasis is transiting to the resilience. Since resilience was introduced into academic fields by Holling in 1973 [4], it has become a new study direction in various disciplines, such as ecological, social, organizational and economic systems [5]. Since 1980s, resilience has been widely studied in lifeline engineering systems, including electrical, transportation, and water systems. In last decades, various academic conferences, organizations, and plans related to resilience emerge in large numbers [6], [7], [8], [9], which boosts the development of this discipline. However, different from reliability with a clear definition, no consensus has been reached on defining and quantifying system resilience to date. In 2003, a general resilience framework is presented by Bruneau et al. [10]. Based on this, resilience is described by four dimensions, including technical, organizational, social, and economic dimensions, and characterized by four properties, including robustness, redundancy, resourcefulness, and rapidity. Now, this general framework is accepted by an increasing number of scholars and also referred by us in this paper.

In recent 20 years, the resilience of WDNs has been emphasized by more and more scholars. Todini [11] first studied the resilience of WDNs in 2000 and viewed it as the energy surplus, namely, the intrinsic capacity to absorb adverse impacts of various events. Herein, the total energy of a WDN is defined as the summary of water discharge multiplied by head for all reservoirs. Therefore, more energy than required should be provided to enhance the system resilience level in the case of increasing demands or other failures. However, network redundancy is not considered in this method. For two WDNs with the same energy surplus, the resilience of the WDN with many loops is obviously much higher than the branched one with many vulnerable downstream consumers. To overcome this drawback, some scholars modified this method from different aspects. For example, Prasad and Park [12] proposed a network resilience index to measure the uniformity of pipe diameters connected to the generic nodes. Creaco et al. [13] introduced a loop diameter uniformity index to reflect the influence of pipe uniformity in the generic loop and thus branched pipes were removed by this method. Jayaram and Srinivasan [14] proposed a modified resilience index by incorporating the influence of multiple sources that were not considered in literature [11]. In addition, some articles [15], [16], [17] investigate the difference of various energy-related indices (including resilience index in [11], network resilience index in [12,13], modified resilience index in [14], flow entropy index in [18]) under various failure conditions such as over-demand and pipe failure. Surplus energy is a feasible way to evaluate the resilience of WDNs, but it focuses on the overall network performance rather than specific consumer's demand. For example, although the obtained head in some areas may be low in operation, it tends to be neglected when the surplus head in the network level is high. In addition, the energy theory is not applicable to the disaster resilience evaluation because the energy reserve of most WDNs are usually enough, but they still suffer from serious damages after disasters such as earthquakes.

Except for energy perspective, resilience is also widely studied by graph-based methods due to the network characteristics of WDNs. These graph-related metrics [19,20] are suitable for large WDNs because they do not rely on complex hydraulic flow analysis. However, pure graph metrics are difficult to reflect the real operational state of WDNs such as head loss and pipe pressure. For this consideration, some scholars evaluate the system resilience by combining graph theory with hydraulic flow analysis. For example, Herrera et al. [21] used graph-based parameters such as closeness centrality or K-weighted shortest routes to measure resilience, in which the head loss was calculated by hydraulic analysis. Similarly, Soldi et al. [22] studied the resilience and vulnerability of WDNs by connectivity metrics while the probability of pipe breakage was solved by hydraulic simulations. Two challenges exist for graph-based theories. First, real operation states of WDNs should be simulated, including the nodal head/flow, loop head loss, deterioration, elevation, and so on. This problem also involves the balance between refined model and efficient algorithm. Second, the restoration process is difficult to be simulated. For example, after a WDN suffers from an earthquake, the network characteristics will change during restoration, including the topology, pipe material, diameter, and even surrounding condition. Instead of a static graph, the WDN in restoration is a dynamic one, which is a challenge for the current theory.

Different from static evaluation by energy or graph methods, the effects of various recovery strategies are analyzed in several articles. For example, Chang and Shinozuka [23] compared two update programs of pumping stations from technical, organizational, and economic dimensions. In each dimension, robustness and rapidity are evaluated in different seismic magnitudes. Cimellaro et al. [24] analyzed the effects of three seismic restoration plans, i.e., tank closure, use of maximum available flow from pump station, combination of the two strategies in improving resilience. In this method, the final resilience index is the product of three indices corresponding to three dimensions (technical, social, and environmental). In addition, similar multidimensional resilience analysis is also studied in [25,26] and different restoration strategies are compared. However, these articles focus more on natural disasters (i.e., earthquakes) whereas less on daily operational accidents, which influence the system resilience more from the lifecycle perspective [27]. Moreover, among various components of WDNs, pipes are less studied than tanks and pumps.

Based on the above literature review and existing drawbacks, a lifecycle-based operational resilience analysis framework of WDNs is proposed in this paper. Herein, four models are considered, including the accident occurrence model, accident simulation model, pipe deterioration model, and restoration model. Based on the framework, both system performance curve during the lifecycle period and resilience level in each operation period are calculated, which provides us a comprehensive lens on the lifecycle operational resilience of the WDNs. The rest of this paper is organized as follows. First, the demand-based resilience index is introduced briefly in Section 2.1. Then, the accident occurrence models for cast iron pipes (CIPs), ductile iron pipes (DIPs), and steel pipes (SPs), the accident simulation models for leakages and bursts, the pipe deterioration model and restoration models for plugging and replacement are introduced in Section 2.2–2.5. In addition, the resilience simulation process is introduced in Section 2.6 to provide a clear clue of the whole process. Section 3 is a case study of the WDN in Mianzhu city to illustrate the above framework. Moreover, six cases improved by different strategies are compared to provide guidance for the local decision makers with the goal of enhancing system operational resilience.

Section snippets

Demand-based resilience index of WDNs

A conceptual description of system performance level (SPL) before, during, and after an accident is shown in Fig. 1. Three key time points are t0 (accident occurs), t1 (recovery activities begin), and t2 (recovery activities finish). Herein, four assumptions are usually used in the SPL curve for simplification [10]. (1) The SPL in normal operation stage is a constant value; (2) It drops to a low level immediately after the accident occurs; (3) It keeps unchanged in the period between t0 and t1,

Case study

With an aim at illustrating above framework, the WDN in Mianzhu, a city located in Sichuan, China, is simulated in this section. In daily operational condition, about 35,000 tons water is supplied per day by four water factories located in the city's northern part (Fig. 11). Three pipe types are CIP, DIP, and SP with the corresponding length are 40.20, 1.05, and 2.80 km, respectively. After simplification, 107 main pipes (96 CIPs, 6 DIPs, 5 SPs), 82 nodes, and 92 valves are included in the

Conclusions

This paper presents a lifecycle operational resilience assessment framework of WDNs. Four models are incorporated, including the accident occurrence model (CIPs, DIPs, and SPs), the accident simulation model (leakages and bursts), the pipe deterioration model, and the restoration model (plugging and replacement). The Monte Carlo simulation is used to consider the effects of randomness resulting from the accident occurrence, the accident type, and the recovery time. In order to illustrate the

CRediT authorship contribution statement

Wei Liu: Conceptualization, Methodology, Supervision, Writing - review & editing. Zhaoyang Song: Data curation, Software, Writing - original draft. Min Ouyang: Methodology, Supervision, Writing - review & editing, Validation.

Declaration of Competing Interest

None.

Acknowledgments

This material is based upon work supported in part by National Natural Science Foundation of China (Grant No. 51720105005), and also by China State Grid Corp headquarters science and technology project titled “Research on Regional Integrated Energy Supply Systems Modeling and Planning Techniques”. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsors.

References (43)

  • H. Baroud et al.

    Importance measures for inland waterway network resilience

    Transp Res Part E

    (2014)
  • J Li

    Lifeline earthquake engineering-basic methods and applications

    (2005)
  • China city water statistics yearbook 2003

    (2004)
  • A. Gheisi et al.

    Water distribution systems reliability: a review of research literature

    J Water Res Plan Man

    (2016)
  • C.S. Holling

    Resilience and stability of ecological systems

    Annu Rev Ecol Syst

    (1973)
  • Center for Risk-based Community Resilience Planning. http://resilience.colostate.edu....
  • 100 Resilient Cities. http://www.100resilientcities.org/cities....
  • Resilience Alliance. https://www.resalliance.org/....
  • S.A. Cauffman

    Community resilience planning guide for buildings and infrastructure systems

    (2016)
  • M. Bruneau et al.

    A framework to quantitatively assess and enhance the seismic resilience of communities

    Earthq Spectra

    (2003)
  • T.D. Prasad et al.

    Multiobjective genetic algorithms for design of water distribution networks

    J Water Res Plan Man

    (2004)
  • Cited by (0)

    View full text