Corrosion Prediction in Offshore Pipelines using Statistical Techniques

Studies on external corrosions in offshore pipelines have not received much attention as compared to the internal corrosions. The prediction models for internal corrosions, however, are mostly derived using the knowledge of traditional statistical methods such as the Analysis of Variance (ANOVA). Unfortunately, the nature of corrosion data which are random may not always comply with the assumptions made in ANOVA. This paper then proposes another statistical approach in estimating external corrosions (d, unit %) in offshore pipelines by using the Exploratory Data Analysis (EDA) techniques. The EDA techniques are simple and are not based on strict assumptions which are rarely fulfilled in real world situations. The median polish model is an alternative to Analysis of Variance (ANOVA) and is used in the analysis in this paper. The proposed model was developed based on a recent discovery which conforms that external corrosion are not uniform along the circumference of the pipe but has got specific influence on the o’clock orientations. A prediction table was also developed to guide users on the estimate of external corrosion in offshore pipelines with regards to the o’clock orientations of the pipe circumference.