APPLYING A STATISTICAL METHOD FOR INSPECTION
Every year it feels like a new standard, a new recommended practice, a new internal initiative comes around and must be executed by the boots on the ground. We know that Corrosion Under Insulation (CUI) is one of our worst actors, and if we could devote time and effort to it, we are pretty sure we could make a difference.
Unfortunately, we have a plethora of initiatives we shall comply to, meaning we can’t focus on CUI as much as we would like to. Now this isn’t all bad, we need to have a well-rounded mechanical integrity program covering a broad range of programs, but if we could just spend a little more time on the ones we really know are troubling, we might feel like we are making a dent in discovering issues proactively.
A Shift in Strategy
Typically, compliance requirements, and internal processes cause owner-operators to manage a multitude of mechanical integrity and reliability programs. With so many programs, a shift in conventional strategy could provide some additional benefits, specifically to internal corrosion. Many in the petrochemical industry acknowledge that internal corrosion is not a very prevalent damage mechanism; however, maintenance teams are still required to inspect for corrosion. Unfortunately, this effort becomes one of the largest expenses within an inspection group and one of the most labor-intensive tasks. If plants can implement a system to control the internal corrosion program, it would be possible to focus inspection efforts onto programs like CUI.
How can plants use the codes and standards to its advantage to achieve confidence in a thickness management program and focus on CUI? The first step is understanding the current data. At many plants, an Inspection Data Management System (IDMS) is filled with thickness readings that may or may not be useful. Often, maintenance personnel have corrosion monitoring locations (CMLs) that they have been observing for years but have yet to figure out if what they see is poor repeatability or if it is actually corrosion.
IDMS Data Review & Validation
A powerful statistical method that can be used for IDMS data review and clean-up is Bayesian analysis. Thickness measurements can be affected by a range of factors and a Bayesian analysis incorporates prior knowledge thereby allowing factors, such as calibration or measurement errors and other variables to be accounted for and assessed. In the context of thickness data, prior knowledge can include information about the equipment being measured, such as the expected range of thickness values or the historical trends in thickness over time. Combining thickness measurement data with prior knowledge will improve the overall accuracy and reliability of the IDMS data. In addition, after defining the statistical model, a Bayesian analysis can estimate the acceptable limits of the thickness readings inside the IDMS, or when new readings are input. It is recommended to conduct this initial clean up if there are three or more readings in the CMLs’ history. Using the updated data will make more accurate and reliable predictions and assessments of the recorded thickness measurements.
Once the IDMS has been verified and cleaned, the data is now ready to calculate existing evaluations. Accurate and reliable thickness data allows companies to use the Bayesian analysis to identify the physical areas of the plant that are at a higher risk of corrosion and then prioritize inspections and maintenance activities accordingly. Applying the statistical method will remove the broad spectrum and random inspection methods currently in place at many facilities.
CUI is a particularly challenging damage mechanism because it is difficult to detect and can cause significant damage to equipment. Utilizing new technologies, such as a Bayesian analysis, companies can streamline CML-level decision making and shift their focus towards identifying and mitigating CUI. Analyzing piping thickness data will help maintenance personnel identify the high corrosion areas and build an inspection and maintenance strategy that is only focused on the high-risk equipment. Bayesian analysis will help plant personnel improve the accuracy and reliability of the corrosion management program and reduce the time spent chasing false positives or non-existent corrosion in internal corrosion programs.
The use of advanced data analysis tools can help companies optimize their overall corrosion management programs, resulting in further cost savings and improved asset reliability. The savings can be used to create a comprehensive strategy to manage the effects of CUI and invest in new technologies or techniques to detect and mitigate CUI, such as insulation systems or rust-resistant coatings. Developing corrosion models and predictions based on data analysis, companies can proactively anticipate corrosion-related problems to achieve reduced downtime, lower maintenance costs, and improved equipment performance.
Applying advanced analysis of piping thickness data and new technology to internal corrosion management programs can result in significant efficiency gains for a company’s mechanical integrity and reliability programs. This practice of proactively addressing potential internal corrosion-related issues allows companies to develop effective strategies to prevent costly equipment failures and downtime. Once an internal corrosion program is in an optimized state, companies can shift their focus towards detecting and mitigating CUI. Therefore, the use of data analysis in thickness management programs provides savings that can be used to create a comprehensive approach to managing the effects of CUI, resulting in further optimization of corrosion management programs and a safer and more reliable plant.
BY R. BRANDEN STUCKY, CORRSOLUTIONS TEAM LEADER, SENIOR ENGINEER I
This article was originally published in MTI CONNECT 2023, Issue 1.