Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 80
PROCEEDINGS OF THE FOURTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY
Edited by: B.H.V. Topping and C.A. Mota Soares
Paper 139

A Study of Pipe Interacting Corrosion Defects using the FEM and Neural Networks

R.C.C. Silva, J.N.C. Guerreiro and A.F.D. Loula

National Laboratory for Scientific Computing, LNCC, Petrópolis, Rio de Janeiro, Brazil

Full Bibliographic Reference for this paper
R.C.C. Silva, J.N.C. Guerreiro, A.F.D. Loula, "A Study of Pipe Interacting Corrosion Defects using the FEM and Neural Networks", in B.H.V. Topping, C.A. Mota Soares, (Editors), "Proceedings of the Fourth International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 139, 2004. doi:10.4203/ccp.80.139
Keywords: corroded pipe assessment, interacting defects, interaction rules, finite element analysis, artificial neural networks.

Summary
Corrosion defects occurring in groups or clusters are a recurrent problem and have become of major concern in maintaining pipeline integrity. The assessment of the remaining strength is essential to ensure the continued safe operation, safeguarding the life and the environment. Since 1970s, considerable research has been dedicated to evaluate the serviceability of corroded pipe to provide reliable criteria for its maintenance, repair or removal. One of the most industry-accepted codes for the assessment of corroded pipelines is ANSI/ASME B31G [1]. The semi-empirical method, suggested in this code, is addressed to single corrosion pits, providing no guidance for dealing with the effects of interaction between closely spaced corrosion defects.

Aiming a more realistic representation of the metal loss, new criteria for evaluating corroded pipe, such as the RSTRENG Effective Area [2] and DNV RP-F101 [3], have been proposed. These methods estimate the remaining strength of a pipe with multiple defects considering an equivalent corroded profile obtained by the projection into a longitudinal projection line of all the defects supposed to interact.

Adjacent corrosion defects are supposed to interact when they lead to a failure pressure lower than that occurring in pipes with individual or single defects. As the distance between the defects increases, the interaction will vanish and the failure pressure tends to be the same as that of the most severe corrosion pit. In this context, two distinct topics have to be investigated. One is related to the measurement of the reduction in the burst pressure caused by the defects interaction, leading to an assessment criterion. The other is associated with the minimum distance between defects, which is required to avoid interaction, establishing interaction rules.

Even for isolated defects, the available assessment criteria are embodied with a certain level of conservatism, and each method predicts a different value for the failure pressure [4]. The statement of interaction rules seems also not to be in agreement. Kiefner and Vieth [5], based on experiments, suggested that defects separated circumferentially by a distance greater than six times the pipe wall thickness and longitudinally spaced by more than one inch are not expected to interact. The CSA [6] adopts one defect length or width, depending on the corrosion configuration, as the critical distance between defects. The DNV [3] recommends a criterion that takes into account the pipe external diameter and wall thickness. The inconsistency on these criteria makes the investigation of adjacent defects still necessary.

In this paper, we propose an alternative methodology for the assessment of interacting defects using the artificial neural network technology. Initially, a finite element database containing information about the interaction effects between two equally shaped defects of 80x32 mm is produced and used in the training and testing phases of the neural network. The neural network achieves knowledge from the training samples and is able to give appropriate response for new interacting situations, i.e., for any other defect depth and spacing, allowing the development of assessment criteria and the establishment of interaction rules.

Neural networks have been applied successfully as a method to evaluate the failure bending moment and the failure pressure [7] of pipes with single defects. It is not our intention to propose a general assessment criterion for interacting defects due to the restricted database adopted for its validation, but to show the viability and to emphasise the necessity of further research, especially to investigate the circumferential interaction.

References
1
ASME, "ASME-B31G- Manual for Determining the Remaining Strength of Corroded Pipelines - A Supplement to ANSI/ASME B31 Code for Pressure Piping", The American Society of Mechanical Engineers, New York, 1991.
2
Kiefner, J.F. and Vieth, P.H., "A Modified Criterion for Evaluating the Remaining Strength of Corroded Pipe", Final Report on Project PR 3-805, Pipeline Research Committee, American Gas Association, 1989.
3
DNV, "Corroded Pipelines - Recommended Practice RP-F101", Det Norske Veritas, Norway, 1999.
4
Stephens, D.R. and Francini, R.B., "A Review and Evaluation of Remaining Strength Criteria for Corrosion Defects in Transmission Pipelines", ETCE/OMAE2000 Joint Conference, Paper ETCE2000/OGPT-10255, 2000.
5
Kiefner, J.F. and Vieth, P.H., "PC Program Speeds New Criterion for Evaluating Corroded Pipe", Oil & Gas Journal, 91-93, 1990.
6
CSA Standard Z184, "Gas Transmission and Distribution Piping Systems", Canadian Standards Association, Mar. 18, 1968.
7
Han, L., Han, L. and Liu, C., "Neural Network Applied to Prediction of the Failure Stress for a Pressurized Cylinder Containing Defects", International Journal of Pressure Vessels and Piping, 76, 215-219, 1999. doi:10.1016/S0308-0161(98)00129-X

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)