Two expert systems, that diagnose the sensitivity of environmentally assisted cracking (EXENAC; EXpert system for ENvironmentally Assisted Cracking)and the causes of failure of environmentally assisted cracking (EAC), based upon fractography have been developed using the OPS83 programming language.
The former system is composed of three independent subsystems, which respectively diagnose the sensitivity of ⅰ)stress corrosion cracking of austenitic stainless steels in a low- or high-concentration chloride solution and/or high-temperature water, ⅱ)hydrogen embrittlement of high-strength steels in various aqueous solutions and gaseous environments, and ⅲ)hydrogen attack of steels in high-temperature high-pressure hydrogen gases. The knowledge base covers the rules relating to environments, materials, loading conditions, heat treatments, material processing including welding and their interaction as well as fracture surface morphologies. The knowledge base consists of production rules, which have certainty factors to represent the uncertainty (contribution)of each rule to environmentally assisted cracking. The adoption of certainty factors enabled the uncertain but widely applicable knowledge to be built in a knowledge base, which in turn raised the flexibility of diagnoses made by the expert system. Applying the developed expert system to the case histories the diagnoses were accurately made. We discussed the characteristics of the individual knowledge base, related diagnosis results, and the suitable knowledge base structures for diagnoses in the area of environmentally assisted cracking. It is desirable that the inference is made through the plural affirmative and negative rules being executed. We also proposed the quantitative measure to evaluate the importance of each rule in the diagnosis.
The latter is composed of three independent subsystems, which respectively deal with EACs of high-strength or high-tensile strength steel, aluminum alloy, and stainless steel in dry and humidified air, water, and aqueous solutions containing Cl, Br or I ions. The concerned EAC issues cover stress corrosion cracking (SCC), hydrogen embrittlement, cyclic SCC, dynamic SCC and corrosion fatigue as well as fatigue and overloaded fracture. The knowledge base covers the rules relating to not only environments, materials, and loading conditions, but also macroscopic and microscopic fracture surface morphology. In order to deal with vague expressions of fracture surface morphology, Fuzzy theory is used in the system, and the description of rules about vague fracture surface appearance is thereby possible. Applying the developed expert system to case histories, accurate diagnoses were made. We discussed the related diagnosis results and usefulness of the developed system.
View full abstract