Diagnosis of Geotechnical Failure Causes Using Bayesian Networks

Failure of a complex geotechnical system may be caused by many possible mechanisms. This paper introduces a Bayesian network-based method for the diagnosis of geotechnical distress mechanisms considering uncertainties in these mechanisms and interrelationships among these mechanisms. The methodology is illustrated through the diagnosis of a distressed dam with seepage problems. First, common patterns and causes of dam distresses are identified based on the information in a database of 993 distressed in-service embankment dams in China and used as the prior information. The interrelations among the dam distresses and their causes are quantified using conditional probabilities determined based on historical frequencies from the database. The observed leakage rates, seepage exit locations, and boundary conditions of a specific distressed dam are used as project-specific evidences. Bayesian networks are used to diagnose the distressed dam by combining the prior information based on the database and the project-specific evidence in a systematic way. Based on results of the diagnosis, key distress factors for the dam can be identified and suitable remedial measures can be suggested.