|This article is part 2 of a 2-part series.|
|Part 1 | Part 2|
Inspection data analysis tools, like risk-based inspection, help us to focus on quantitative reliability targets. When considering thinning mechanisms, there is a certain probability that a piece of equipment will reach retirement thickness before or at the next inspection or the next turnaround. Statistical techniques can help us understand and control the probability of early retirement, allowing us to make better remaining life and reinspection decisions.
In Part 1 of this series, we saw how probability plots can help us visualize and analyze ultrasonic thickness data. We used these plots to identify the presence of localized corrosion and make accurate, minimum thickness estimates.
The example outlined in Part 1 was a relatively simple statistical application, as the data set was a single grid survey from piping components having the same nominal thickness. When the equipment contains a number of different components and/or when multiple thickness surveys are available, the problem becomes more challenging but the basic approach is the same; plot the data, examine the “fit”, look for evidence of localized corrosion and make a projection.
In this article, we will look at two statistical techniques for making remaining life estimates based on multiple thickness surveys. The first technique uses thickness probability plots, similar to those reviewed in Part 1. The second uses corrosion rate probability plots.