Application of Extreme Value Statistics for Backside Corrosion of Tank Bottom Plates (2nd Report)

Abstract

The propriety to apply an extreme value statistics to the data of pitting corrosion which took place on oil storage tank bottom plates was discussed by the 1st report. The maximum pit depth from each sample fitted in with a doubly exponential distribution. And it was also confirmed that the result of statistical analysis was affected by the sample number and dimension.
Therefore this report investigates the sampling method to estimate the X_max (the maximum value of pit depth expected on all bottom plates) exactly and effectively. The result was that sampling from the weld joint and the stratified sampling were effective to improve the precision of the estimation in this tank.
Ultrasonic thickness gauge was used to get the data of the maximum pit depth nondestructively. Measurement pitch was varied from 20mm to 5mm. Even though the measurement pitch was decreased to 5mm, most of the maximum pit depth obtained by the ultrasonic thickness gauge were smaller than those obtained by the depth gauge. About 1mm pitch is desired to obtain the maximum pit depth by use of the ultrasonic thickness gauge. In this case, it takes a lot of time to obtain the maximum pit depth from each sample, so this method is not practical.