Advanced Modeler Gives Rise To Improved Corrosion Control

Pipeline & Gas Journal, Feb, 2001 by Christian Bucherie, Sergio Saldanha

Monitoring and managing the condition of plant and pipelines is not an easy task. This is particularly true of instances of pipeline corrosion. Closure for maintenance is costly. So are the consequences of a malfunction or a spill if the corrosion is not identified correctly as posing a threat. The plant manager's art is achieving a balance between the two - initiating repairs when necessary - leaving alone when they are not. But how do you succeed in obtaining and maintaining this balance without wasting company money on unnecessary repairs, or leaking valuable product?

Until now, plant managers have necessarily had to err on the side of caution when considering the acceptability of corrosion. In instances of pipeline corrosion, they have relied on the calculations done by maintenance engineers using comparatively simple analytical formulae such as ANSI/B31G, RSTRENG, and, more recently, BS7910. Because they are based on simple material data and an approximation of the geometry of the defect, they have a high degree of conservatism built into them. Consequently, they are likely to produce a result that will prompt closure and maintenance when it is not yet required. In short, they are a "rather safe than sorry" solution to the problem of pipeline maintenance.

In some instances, managers do conduct more rigorous calculations to determine the acceptability of corrosion, based on finite element analysis - a complex, but accurate method of determining the condition of a material. In the past, this has been a time-consuming and expensive procedure, typically requiring a plant's technical department to invest three to four man-weeks to produce a calculation for each defect. As a result, this was only done in special cases, where the corrosion was in a sensitive area or where the traditional analytical formulae would not provide an adequate answer.

In an ideal world, managers would be able to conduct such finite analysis on all areas of corrosion, regardless of their location and perceived significance for the plant or pipeline. That ideal is now within the reach of everyone as a result of work undertaken by Bureau Veritas' Energy and Process business line in France in conjunction with SOFREGAZ, a subsidiary of Tecnimont and Gaz de France, as part of an ongoing pipeline maintenance research and development initiative.

Over the last 18 months, Bureau Veritas has been developing a finite element modeler based on the FORTRAN computer programming language that produces corrosion data that is compatible with ABAQUS, a computer-based application developed by Hibbitt, Karlsson & Sorensen Inc. used throughout the world for stress, heat transfer, and other types of analysis in mechanical, structural, civil, biomedical, and related engineering fields. Put simply, the modeler generates models of internal and external pipeline corrosion defects that can be fed into ABAQUS for analysis.

The finite element model to be fed into the ABAQUS software is produced with the help of a series of variables, such as pipeline diameter and thickness, material mechanical properties (including elasticity and plasticity), loading cases (including pressures, axial forces, and bending moments), and data on the geometry of the defect (length, width and depth). With the help of these parameters, the new BV application swiftly generates a 3-D model of the corroded surface using mesh generation technology based on 20-node, quadratic element adapted features to eliminate possible distortions, and ellipsoidal surfaces or a combination of ellipsoidal and cylindrical surfaces. This 3-D model can then be used for non-linear finite element analysis with the general purpose ABAQUS/STANDARD application. The result is a highly accurate calculation of the remaining strength of the corroded section of piping.

The result of the analysis is measured against the plastic strain value based on the Ultimate Tensile Stress of the corresponding piping material. In other words, if the segment proves to be too weak, it is failed and earmarked for immediate replacement.

The great benefit of the new pre-ABAQUS modeler is that it reduces the pre-processing time significantly. Whereas pre-processing would normally take around two to three weeks, this is now reduced to minutes for each individual corrosion defect. This cuts analysis costs and man-hours, and allows maintenance engineers and managers to assess more accurately the acceptability of all instances of corrosion.

There is more. The new method also reduces the conservatism inherent in the old way of doing things. Faced with the less sophisticated results generated by the ANSI/B31G, RSTRENG and BS7910 formulae, managers were forced to play it safe, leading to shutdowns and maintenance where none was yet required. This meant a loss of revenues and increased maintenance costs. A rare instance of conservatism encouraging waste. From an operational point of view, the new software reduces the plant expenditure significantly.

The new modeler also allows managers to see what the problem is and form a visual imprint of how significant the corrosion is. It provides an image of the location, depth and significance of the defect (see illustrations). While few conclusions can be drawn from a merely visual representation (if that were so, a simple visual inspection of a corrosion incident would be adequate in all cases), it helps put the defect in a less abstract light. If pictures are worth a thousand words, they are worth a few abstract numbers, too.