A Neural Based Fuzzy Logic Model to Determine Corrosion Rate for Carbon Steel Subject to Corrosion under Insulation

• Published in: Applied mechanics and materials Vol. 789-790; no. Manufacturing Science and Technology VI; pp. 526 - 530
• Main Authors: Khan, Muhammad Mohsin, Mokhtar, Ainul Akmar, Hussin, Hilmi
• Format: Journal Article
• Language: English
• Published: Zurich Trans Tech Publications Ltd 02.09.2015
• Subjects: API; Carbon steels; Corrosion; Corrosion rate; Deviation; Fuzzy logic; Insulation; Mathematical models; Neural networks; American Petroleum Institute Standard; Quantitative Approach; API 581; Risk-Based Inspection
• ISBN: 3038355437; 9783038355434
• ISSN: 1660-9336; 1662-7482; 1662-7482
• DOI: 10.4028/www.scientific.net/AMM.789-790.526

One of the most common external corrosion failures in petroleum and power industry is due to corrosion under insulation (CUI). The difficulty in corrosion monitoring has contributed to the scarcity of corrosion rate data to be used in Risk-Based Inspection (RBI) analysis for degradation mechanism due to CUI. Limited data for CUI presented in American Petroleum Institute standard, (API 581) reflected some uncertainty for both stainless steels and carbon steels which limits the use of the data for quantitative RBI analysis. The objective of this paper is to present an adaptive neural based fuzzy model to estimate CUI corrosion rate of carbon steel based on the API data. The simulation reveals that the model successfully predict the corrosion rates against the values given by API 581 with a mean absolute deviation ( MAD ) value of 0.0005, within that the model is also providing its outcomes for those values even for which API 581 has not given its results. The results from this model would provide the engineers to do necessary inferences in a more quantitative approach.