ORCID Identifier(s)


Graduation Semester and Year




Document Type


Degree Name

Doctor of Philosophy in Civil Engineering


Civil Engineering

First Advisor

Mohammad Najafi


Utility managers and owners have challenges when addressing appropriate intervals for inspection of gravity sanitary sewer pipelines and other underground pipeline systems. Frequent inspection of sewer network is not cost-effective due to large inventory of pipes and high cost of inspections, such as using closed-circuit television (CCTV) surveys. Therefore, it would be more beneficial to first predict critical sewers most likely needing maintenance and then perform inspections to optimize use of their limited budgets and target pipelines most in need of repairs, rehabilitation or renewal. Development of sewer condition prediction models is extremely vital for utilities to evaluate the short-term and long-term behavior of their pipe network considering different uncertainties. However, providing a prediction model is difficult due to lack of adequate datasets. The primary objective of this dissertation is to develop prediction models that can predict future conditions of sanitary sewer pipes. The outcomes of the models can be used to prioritize inspection and renewal needs of sanitary sewer pipes. In addition, this dissertation identifies significant factors that affect deterioration of sanitary sewers. To achieve these objectives, three different statistical and artificial intelligence models, namely logistic regression, gradient boosting tree and K-nearest neighbors were developed in successive steps. Data collected from City of Tampa (Florida) was used to demonstrate the applicability of the developed models. Thirteen independent variables including pipes age, material, diameter, flow rate, pipe segment length, depth, slope, soil type, pH, and sulfate content, and water table, soil hydraulic group and soil corrosivity were used to build these prediction models. The results of this dissertation show that performance of all three developed models were acceptable; however gradient boosting tree achieved a higher accuracy during validation process. Additionally, pipe age, length, diameter, material and water table are found to be significant variables influencing deterioration of sanitary sewers. This dissertation contributes to body of knowledge by developing condition prediction models that can be used as part of a comprehensive asset management system of sanitary sewers.


Pipe prediction model, Deterioration model, Sewer condition prediction, Sewer deterioration, Asset management, Condition assessment, Gradient boosting tree


Civil and Environmental Engineering | Civil Engineering | Engineering


Degree granted by The University of Texas at Arlington