Computational based automated pipeline corrosion data assessment

Abstract

Corrosion is a complex process influenced by the surrounding environment and operational systems which cannot be interpreted by deterministic approach as in the industry codes and standards. The advancement of structural inspection technologies and tools has produced a huge amount of corrosion data. Unfortunately, available corrosion data are still under-utilized. Complicated assessment code, and manual analysis which is tedious and error prone has overburdened pipeline operators. Moreover, the current practices produce a negative corrosion growth data defying the nature of corrosion progress, and consuming a lot of computational time during the reliability assessment. Therefore, this research proposes a computational based automated pipeline corrosion data assessment that provides complete assessment in terms of statistical and computational. The purpose is to improve the quality of corrosion data as well as performance of reliability simulation. To accomplish this, .Net framework and Hypertext Preprocessor (PHP) language is used for an automated matching procedure. The alleviation of deterministic value in corrosion data is gained by using statistical analysis. The corrosion growth rate prediction and comparison is utilized using an Artificial Neural Network (ANN) and Support Vector Machine (SVM) model. Artificial Chemical Reaction Optimization Algorithm (ACROA), Particle Swarm Optimization (PSO), and Differential Evolution (DE) model is used to improve the reliability simulation based on the matched and predicted corrosion data. A computational based automated pipeline corrosion data assessment is successfully experimented using multiple In-Line Inspection (ILI) data from the same pipeline structure. The corrosion data sampling produced by the automated matching is consistent compared to manual sampling with the advantage of timeliness and elimination of tedious process. The computational corrosion growth prediction manages to reduce uncertainties and negative rate in corrosion data with SVM prediction is superior compared to A ^N . The performance value of reliability simulation by ACROA outperformed the PSO and DE models which show an applicability of computational optimization models in pipeline reliability assessment. Contributions from this research are a step forward in the realization of computational structural reliability assessment.