Knowledge Acquisition and Validation of an Expert System for Prioritizing the Inspection of Sewers

The sewer infrastructure of the US is in a state of disrepair due to municipal budget constraints for utilities, a history of neglect and most importantly a lack of critical information about the aging and complex sewer networks that convey wastewater for 75% of the US population. This paper describes an expert system. Sewer Cataloging, Retrieval and Prioritization System (SCRAPS), that prioritizes sewer inspections that target the collection of information of critical areas sewer network. The knowledge base of this expert system was assembled with input from a national group of experts, drawn from both the public and private sectors. Input from the experts developed the system's logic, which assesses the overall need to inspect based on the line's consequence and likelihood of failure. The inference engine is based on Bayesian belief network theory, which allows the uncertainty in the experts' beliefs to be propagated through the system. The tool was developed with a rapid prototype application process. The validation of the tool shows that the tool assesses pipeline case studies similar to other sewer infrastructure experts.