ABSTRACT The network of pipes in potable water distribution systems (WDS) are comprised of thousands of pipes made of various materials including PVC, concrete, cast iron, and steel, among several others. The pipes are subjected to internal and external conditions that lead to their failure. Stress conditions include, but are not limited to internal pressures, traffic loading, and corrosion. The deterioration of a pipe decreases its mechanical strength which results in an increase of its probability of failure. Failures lead to loss of service which translates to loss of money due to the cost of repairs and buildup of traffic caused by street closures. The focus of this study is the pipe network underneath cities that make it possible for communities to have access to potable water. The objective of this analysis is to evaluate the physical conditions of each pipe in a water distribution system in order to assess its probability of failure and ultimately calculate the risk associated with each pipe in the case that it were to fail. This model focuses only on the pipes of the WDS and does not take into consideration fittings, pumps, and other network components. This model assesses pipe age, material, diameter, internal pressure, traffic loading (industrial or residential), and length to determine the probability of failure. It then utilizes several economic factors such as material cost, customer criticality, demand, traffic impact, and land use to calculate the risk associated with each pipe. The risk associated with each pipe can then be used as a ranking system to identify the most vulnerable pipes, those with the highest economic impact upon failure. Identifying the pipes with the highest risk allows municipalities to better allocate funds for maintenance or replacement of pipes. It highlights the most critical pipes within a network of thousands. In order to check its functionality, this model applied to the WDS of the City of Arroyo Grande, California. Information on the City’s distribution system was analyzed using Bentley’s WaterCAD, ESRI’s ArcGIS, MathWorks’ MATLAB and Microsoft’s Excel software to perform the analysis. The risk analysis model provided 3 pipes within the distribution system made of cast iron as having a high probability of failure and a critical level of risk. A critical level of risk is defined as falling within the highest range of risk within this study. Considering that only 3 pipe segments were highlighted as having a Critical Risk, 4 as High Risk, and 6 as Medium Risk, in a system of 3572 pipes indicates that the model functions properly. This model was compared to a method developed by Jan C. Devera in his thesis “Risk Assessment Model for Pipe Rehabilitation and Replacement in a Water Distribution System” (2013), which was also applied to the City of Arroyo Grande’s distribution system. Results provided by this analysis prove that both models are functional due to similar results. The current study utilizes the concepts of random variables and conditional assessment to run various Monte Carlo Simulations as the means of calculating the probability of failure of a pipe. Mr. Devera’s model utilizes simplistic approach that does not involve intensive calculations, but results for both models turned out to be similar when looking at the Arroyo Grande distribution system. This risk assessment model demonstrates that a risk assessment model can provide a framework to prioritize pipes based on risk. The approach can help create a schedule for a city’s pipe distribution network for maintenance and repair. It is important to note that it is not a predictive model. This study may be employed to better allocate funds for the rehabilitation and replacement of a city’s existing pipe network to promote optimal operating conditions and service to the public.

A water distribution system is composed of thousands of pipes of varying materials, sizes, and ages. These pipes experience physical, environmental, and operational factors that cause deterioration and ultimately lead to their failure. Pipe deterioration results in increased break rates, decreased hydraulic capacity, and adverse effects on water quality. Pipe failures result in economic losses to the governing municipality due to loss of service, cost of pipe repair/replacement, damage incurred due to flooding, and disruptions to normal business operations. Inspecting the entire water distribution system for deterioration is difficult and economically unfeasible; therefore, it benefits municipalities to utilize a risk assessment model to identify the most critical components of the system and develop an effective rehabilitation or replacement schedule. This study compared two risk assessment models, a statistically complex model and a simplified model. Based on the physical, environmental, and operational conditions of each pipe, these models estimate the probability of failure, quantify the consequences of a failure, and ultimately determine the risk of failure of a pipe. The models differ in their calculation of the probability of failure. The statistically complex model calculates the probability of failure based on pipe material, diameter, length, internal pressure, land use, and age. The simplified model only accounts for pipe material and age in its calculation of probability of failure. Consequences of a pipe failure include the cost to replace the pipe, service interruption, traffic impact, and customer criticality impact. The risk of failure of a pipe is determined as the combination of the probability of failure and the consequences of a failure. Based on the risk of failure of each pipe within the water distribution system, a ranking system is developed, which identifies the pipes with the most critical risk. Utilization of this ranking system allows municipalities to effectively allocate funds for rehabilitation. This study analyzed the 628-pipe water distribution system in the City of Buellton, California. Four analyses were completed on the system, an original analysis and three sensitivity analyses. The sensitivity analyses displayed the worst-case scenarios for the water distribution system for each assumed variable. The results of the four analyses are provided below. Risk Analysis Simplified Model Complex Model Original Analysis All pipes were low risk All pipes were low risk Sensitivity Analysis: Older Pipe Age Identified 2 medium risk pipes Identified 2 medium risk pipes Sensitivity Analysis: Lower Anticipated Service Life Identified 2 medium risk pipes Identified 9 high risk pipes and 283 medium risk pipes Sensitivity Analysis: Older Pipe Age and Lower Anticipated Service Life Identified 1 high risk pipe and 330 medium risk pipes Identified 111 critical risk pipes, 149 high risk pipes, and 137 medium risk pipes Although the results appeared similar in the original analysis, it was clear that the statistically complex model incorporated additional deterioration factors into its analysis, which increased the probability of failure and ultimately the risk of failure of each pipe. With sufficient data, it is recommended that the complex model be utilized to more accurately account for the factors that cause pipe failures. This study proved that a risk assessment model is effective in identifying critical components and developing a pipe maintenance schedule. Utilization of a risk assessment model will allow municipalities to effectively allocate funds and optimize their water distribution system. Keywords: Water Distribution System/Network, Risk of Failure, Monte Carlo Simulation, Normal Random Variable, Conditional Assessment, Sensitivity Analysis.