Abstract
High temperature hydrogen attack (HTHA) has been a known failure mechanism for many years, with the Nelson Curves (ref. API RP 941) being almost 60 years old. While research and industry learning has been ongoing, failures below the Nelson Curves for non-post weld heat treated carbon steel (CS) have occurred in the last 10 years, and as such, the learning continues. These more recent failures have spurred multiple joint industry projects that are still ongoing. While we believe testing is extremely valuable, and look forward to continually testing our models as new information becomes available, we also believe that there are contributions that can be made right now. The gap is not in HTHA’s critical factors, which are well understood by material experts, nor is it in the technical feasibility of the Nelson Curves that has been repeatedly demonstrated (ref. API TR 941-A). Rather, the biggest gaps are:
- relating time to failure,
- incorporating varying operation data, and
- treatment of welds.
This article stands on the shoulder of giants to tie the Nelson Curves, which are the foundation of our industry’s HTHA programs, to mechanistic models from several sources that will allow us to recreate Nelson curves for different operational histories and time durations. By using this work, married with the recent advances in nondestructive examination (NDE) (e.g., new API 941 Appendix E guidelines) we believe that managing equipment with damage is now both possible and reasonable.
Introduction
Becht published an article discussing our Pono screening method for HTHA (“Pono HTHA Evaluation Method”) over five years ago in Inspectioneering Journal,[1] with the goal of handling the gaps in the Nelson Curves[2] from a high-level perspective. This method incorporated known critical factors such as exposure time, steel chemistry/quality, confidence in operating conditions, upsets, and operating stress levels to provide owner-operators with a higher confidence screening method.
Our goal with this article is to continue giving owner-operators higher confidence by extending our practical method to be more quantitative/less empirical. We do this by taking the results of decades of development on void growth and using them to produce the standard Nelson Curve. In our experience, the data and models essentially taken directly from the literature do this remarkably well for the carbon steel (CS) test case presented here. The benefits of describing the Nelson Curves with an analytical model vs. curve fitting (as the Pono Method did previously) are large; time is explicitly a part of the model and critical factors such as external stress and metallurgical factors can be directly incorporated.
Comments and Discussion
Add a Comment
Please log in or register to participate in comments and discussions.