Eliciting Corrosion Knowledge Through the Fault-Tree Eyeglass

Abstract

While the adoption of higher performance materials and better understanding of the fundamental corrosion degradation mechanisms is helping to improve the corrosion control and prevention of metallic materials, long term improvement of the efficiency and cost-effectiveness of corrosion prevention measures could be greatly improved by the adoption of methodologies which could account for the complex interaction between materials, environments and people. In this context, the introduction of expert systems (ESs) or tools that could assist in their introduction has great potential. Ideal ESs should reproduce the functions normally performed by the source expert(s) and constantly deliver a reliable and accessible degree of expertise, but an important impediment toward these goals is the process of knowledge elicitation which remains a serious problem. Since the difficulties in developing ESs have made their introduction in the workplace very precarious, any developmental problem can be catastrophic. In reality knowledge is marked with individuality and while knowledge and experience can be shared by a group of people, expertise is uniquely tinted by experts’ individuality. This paper describes how the construction of an elicitation shell based on Fault Tree architecture could normalise and even facilitate the transfer of information during the production of ESs for the management of corrosion problems.