Expert system for prioritizing the inspection of sewers: Knowledge base formulation and evaluation

This paper describes the development of the knowledge base expert system denoted as Sewer Cataloging, Retrieval and Prioritization System. This computer support system prioritizes sewer pipeline inspections used to target critical areas within a sewer drainage system. This system addresses a growing need of municipalities. The sewer infrastructure of many cities is in a state of disrepair due to budgetary constraints, a history of neglect and, often most importantly, a lack of critical information about the aging and complex system of sewers that convey wastewater for 75% of the population. The knowledge base was assembled with input from a national group of experts from both the public and private sectors. Input from the experts assesses the overall need to inspect based on both the line's consequence and likelihood of failure. In turn, consequence and likelihood of failure are based on six mechanisms describing failure and two mechanisms predicting the impact of failure. Prioritization is accomplished using a Bayesian belief network that allows the uncertainty of the expert's beliefs to be propagated through the decision process. The knowledge base is evaluated with a series of case studies and is shown to be effective at mimicking the knowledge of experts.