Case-based reasoning approach for managing sewerage assets

Abstract

Statistical, deterministic and heuristic approaches to modelling sewer performance and condition are briefly examined and a case-based reasoning (CBR) approach is proposed that can operate within the prevailing data environment of UK water utility companies. This is a technique that solves new problems by adapting previously successful solutions to similar problems. The use of CBR for other types of infrastructure assets is noted, and a method for its application to sewers is proposed. The methodology was tested on sewer data drawn from two water company regions. Validation trials, based on inferring known outcomes on 20 pipe lengths, show that a range of performance and management information and maintenance prioritisation can effectively be inferred from a case library of a small number of pipes for which good information is available. The paper concludes this is a pragmatic and flexible approach that utilises the industry's own experience and avoids the need for widespread calibration of more statistically or deterministically oriented models.