Graduation Year

2021

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Civil and Environmental Engineering

Major Professor

Qiong Zhang, Ph.D.

Committee Member

Mahmood Nachabe, Ph.D.

Committee Member

Hadi Charkhgard, Ph.D.

Committee Member

E. Christian Wells, Ph.D.

Committee Member

Brian Cook, P.L.A.

Keywords

Complex Network Analysis, Complex Systems, Criticality, Infrastructure Interdependency, Urban Water Resilience

Abstract

This dissertation examines structure-function relationships of water distribution networks (WDNs), with focus on improving the decision-making process for water utilities. WDNs are lifeline infrastructure systems, necessary for the functioning of cities. Yet water utilities around the world struggle with emerging challenges that include but are not limited to aging infrastructure components, limited budgets, increased interdependencies between infrastructure systems, increased rates of urbanization, and increasing extreme weather conditions. This is why enhancing the performance of WDNs is a priority for water utilities around the world. This research begins to delineate a system architecture for WDNs and evaluates the use of various representations of structure to enhance the decision-making process for water utilities. From a network science perspective, this dissertation investigated three major questions: how to represent, quantify, and analyze the structure of WDNs; how do indicators of structure relate to WDN performance; and can bio-inspired design enhance the performance of WDNs?

The physical, logical, and hierarchical facets of WDN structure are represented as network models. The physical network model can be represented as a pipe-junction, or segment-valve network. The pipe-junction network indicates that junctions are represented as nodes and pipes as the edges connecting them. The segment-valve network indicates that segments are represented as nodes and valves are represented as the edges connecting them. Logical network models can be constructed from hydraulic simulation results and contain nodes that can represent any WDN component (e.g., pipe, segment, valve) and edges that represent influence (i.e., nodes that are connected to each other have some impact on the performance of each other). The logical network model can also contain nodes representing components from other infrastructure systems (e.g., roads), where edges represent some type of interdependency. The hierarchical structure of loops can be represented as a network model where nodes represent loops and edges represent loops nestedness. The hierarchy of pipe diameters is represented as a statistical distribution of diameter changes along flow paths.

For each representation various indicators that are related to hydraulic performance are measured. The segment-valve representation of physical structure is used to measure an importance index for WDN segments. The importance index is found to correlate strongly with hydraulic simulation-based criticality. The hierarchical structure is analyzed by quantifying parameters of the degree distribution of the network model of loop nestedness and the distribution of pipe diameter gradation along flow paths. The two indicators correlate with both path and energy redundancy. Pipe diameter gradation is found to be an important driver of WDN energy redundancy. Based on measurements of the two indicators, hierarchical structure significantly differs from the physical topology of the tested WDNs.

Various network models are generated for the logical facet of structure: one using the pipe-junction representation as input, another using the segment-valve representation as input, and another to represent the interface between pipes and roads. The logical network models also show different properties than those of the physical network models. The logical structure of WDNs displays some level of hierarchy in how components influence each other. The tested WDNs also display small-world connectivity, suggesting that seemingly small failures in WDNs can propagate to have larger impacts on the network. The logical network models are used with multiple criteria optimization to identify component and segment criticality. In particular, optimization frameworks accounting for social vulnerability and WDN interdependency with roads use logical network models as input. Using the logical network models generated based on the pipe-junction representation and the interface between pipes and roads does in some cases change the identification of vulnerable and critical components.

With the gained insight on WDN structure, bio-inspired design is investigated for its potential to enhance WDN performance for a case study in Tampa, FL USA. Various performance indicators are measured for the empirical, a physarum inspired diameters model, and a bio-inspired diameters and topology model. The bio-inspired diameters and topology model is based on: physarum to select diameter sizes and a generic growth model to design the topology. For all measured performance indicators, the bio-inspired diameters and topology model outperformed the empirical WDN. In particular, both bio-inspired models are more robust to changes in demand and their piping is more economical than the empirical WDN.

This research advances our understanding of WDN structure and how it relates to the emergent performance of WDNs and intervenes in the long-standing paradigm that siloes individual infrastructure systems. By reducing their complexity to various facets of structure, individual infrastructure systems can be more readily analyzed while accounting for their interdependencies, the communities they serve, and the landscape in which they are situated. In addition, appropriate bio-inspired growth models can be applied to enhance WDN performance. The findings highlight the necessity to continue delineating a system architecture for WDNs, allowing the complexity of evaluating criticality and vulnerability and optimizing WDN structure to be reduced.

Download

Share

COinS