Complex Network Analysis of the Washoe County Water Distribution System

Presentation By: Eric Klukovich

Date: 11/13/2014

Overview

Previous Work

Distribution Systems

The Dataset

Creating the Network

Network Analysis

Conclusion

Previous Presentation

Discussed an overview of what studies have been done in energy, water, and environment complex networks.

Topic was vague.

Changed topic to focus in on the Washoe County water distribution network. Combines natural and engineered water sources in the

network. Has not been done before. Focus on understanding the relationship between them.

Previous Work Water distribution is infrastructure that must

always be available.

Can analyze the efficiency, vulnerability, and create plans for alternative resources.

Rivers can also be modeled Monitor the water flow. Take protective action if the river is being depleted.

Study 1 - Water Distribution Analysis

Modeled four different water distribution networks Nodes represented source, control, and

storage/processing facilities. Edges were the pipes.

Study 1 - Water Distribution Analysis

Each network’s density was calculated All networks were sparse and resemble the urban areas.

The efficiency of the water distribution was measured Distance between the supply node and the demand

source Construction and cost has a major factor on how the

network is created.

The robustness was measured by random removal of nodes.

Study 2 - Modeling River Networks

Modeled the Haihe Basin River network in China 565 nodes (319 natural and 246 engineered nodes)

Natural and engineered nodes

Edges - natural or artificial water channels.

The degree distribution was calculated to categorize the different nodes. Could identify the sources, the sea nodes.

Study 2 - Modeling River Networks

River network

River Node/Edge Example

Water Distribution Systems

Water distribution systems are a critical part of everyday life.

Provides fresh drinking water to entire cities and towns.

Can grow to be very complex. Pipe layouts that guarantee water availability.

Complex networks can model and analyze the water distribution network.

Can calculate the efficiency and vulnerability of the system.

Improve the system by added or removing edges or nodes.

Understand what the impacts are if a node in the network failed.

The Dataset

Need a complete dataset in order to create a realistic complex network. Types of nodes Pipe diameter and length How they are connected

Need data for the engineered and natural nodes Comes from two different sources.

Water distribution in Washoe County is controlled by the Truckee Meadows Water Authority. The natural sources of water come from the lakes,

rivers, wells, and ground water that surround the Reno area.

Engineered Node Dataset

The data available to the public was vague and did not provide a complete dataset.

Contacted the Washoe County Community Services Department to get credentials to their Geographical Information System (GIS) Contained a complete dataset for all the water utilities in

Reno GIS Maps contains data for the

different neighborhoods Arrow Creek Double Diamond Steamboat

GIS Maps

Provides information on the different nodes Water mains and water values Service points Wells Storage tanks Water treatment facilities Fire hydrants

Provides information for the edges (pipes) Pipe length Diameter

GIS Maps – Overview

GIS Maps – Double Diamond

GIS Maps – Double Diamond

Natural Node Dataset

Data for the natural sources of water was found through the Truckee Meadows Water Authority. Lakes Rivers Creeks Reservoirs

Information for the flow of the rivers andcreeks were provided.

Lake and reservoir capacity was given.

Creating the Network - Nodes

Two types of nodes in the network, natural nodes and engineered nodes.

The natural nodes can be divided into two types Surface water sources, such as lakes and rivers that are

on the surface of the Earth. Groundwater sources that are located under the Earth’s

surface and requires wells to extract the water.

Engineered nodes Structures that humans created in order to process and

transport large amounts water to different areas. Water mains, booster pump stations, pressure regulator

stations, storage tanks, treatment facilities, and reservoirs.

Creating the Network - Edges

Two types of edges - natural and engineered.

The natural edges are the channels between two natural nodes. Rivers and creeks tend to go from one large body of water

to another. The river flow will be used for the weights.

Creating the Network - Edges

The engineered edges are the artificial channels and pipes.

The pipes are used within urban areas to provide water to all of the different areas within the city.

The pipe diameter and

pipe length data will be

used for the weights.

Natural to engineered

Engineered to engineered

Creating the Network

The network will be arranged based on the location of where the nodes and edges are located. Creates a more realistic network. Allows for an easier comparison to maps.

The dataset shows the water distribution system follows closely to the surface streets. The network will be created to represent this layout.

Network Analysis - Patterns

The network can be analyzed to find patterns within the data.

Can determine what factors impacted the network layout.

Ideally, a system should be efficient as possible, but the cost would be unrealistic.

Systems of this size have a limited budget and resources and impacts how the system is laid out.

The distribution network will be analyzed to determine if the construction costs and resources had an impact.

Will compare the network to a minimum spanning tree graph, with the same vertices.

If the two graphs differ, then cost and resources impacted the layout.

Network Analysis – Comparison

Can compare the natural water network to the man-made water network.

The degree distribution for each network can be calculated to compare the hubs.

Other complex network metrics can be calculated to determine if there are any similarities

Betweenness

Closeness

Clustering Coefficient

Network Analysis - Efficiency

Can also determine the efficiency of the system. Cost and energy to transport water can be substantial.

If the water distribution network is inefficient, then it could lead to extra structures that could have been avoided.

Pumping stations to move the water.

Pressurizing stations to guarantee water pressure.

The efficiency will be measured by taking the Euclidean distance from the source to destination .

This value will then be compared to the amount of pipe that was used to move the water to the same destination.

Network Analysis - Robustness

Robustness of the network can be calculated.

A system failure can have a major impact on the people using it.

Shows the amount of redundancy in the distribution system. There should be enough redundancy so the entire

system will not fail if a few nodes fail.

Robustness will be measured by removing nodes randomly and by highest degree first.

The nodes will continue to be removed until the system completely fails.

The number of removed nodes that caused the failure can be determined to conclude if the system is robust or not.

Network Analysis - Clustering

The complex network can also show clustering within the data. In Reno there are many different housing developments

throughout the city.

The clustering coefficient in the network can be calculated.

Can determine if the water distribution system also creates clusters to supply water.

The network should be sparse because the nodes do not connect to every other node.

Conclusion

Water distribution is a crucial part of todays infrastructure and cannot afford to be disrupted because many people depend on it.

A new approach is taken that combines natural and man-made water sources into one network.

The data to create the network primary comes from the GIS maps that contain data for the water utilities in Reno.

The efficiency, robustness, and other metrics will be calculated to understand different aspects within the network.

Questions