The Use of Probabilistic Fracture Assessment Procedures in Design of Pipelines Subjected to Large Strains

Agrell, Christian (DNV GL Pipelines and Materials) | østby, Erling (DNV GL Pipelines and Materials) | Levold, Erik (Statoil ASA) | Hauge, Mons (Statoil ASA) | Bjerke, Steinar Lindberg (DNV GL Pipelines and Materials)

OnePetro 

Abstract

Design against fracture in pipeline girth welds deals with hypothetical defects. A logical step would be to apply probabilistic assessment procedures including an assumed distribution of weld flaws. The major part of fracture assessments carried out today is semi-deterministic. Even if such an approach might be applicable, it is proposed that it may not always lead to optimum result in terms of combination of safety level and cost-effectiveness. The risk of girth weld fracture is influenced by several parameters like weld defect size, tearing resistance, weld metal mismatch, misalignment, yield to tensile strength ratio, etc. The likelihood of the most detrimental values of each distribution occurring at the same position is usually very low. Further, fracture assessment of pipelines often requires that the different steps in the loading history are considered in the analysis. In this paper we illustrate how probabilistic assessment can be used to obtain the overall reliability of pipelines wrt girth weld fracture. We discuss statistical representation of the different key input parameters and how the analysis can be used to determine characteristic values to be used in design assessment to meet target reliability levels. Further, we illustrate how the procedure can be used to investigate the significance of previous load steps, such as the effect of installation loads on the strain capacity during operation.

INTRODUCTION

The field of strain-based design of pipelines, with special focus on tensile strain capacity due to potential defects, has received considerable attention the last 10-15 years. The ISOPE SBD symposium, having its 10th anniversary at this year’s conference, has been one of the main arenas for exchange of ideas in this respect. In this period there has been made significant contributions and developments within the field. An improved understanding of the main physical features controlling pipeline failure has been established, and several models for estimation of tensile strain capacity have been proposed, (see e.g. Østby (2007), Fairchild et al. (2014), Tang et al. (2014), Wang et al. (2012)). As a part of this development, increased use of FEA, both w.r.t. obtaining new knowledge and to development of quantitative models, has taken place (see review paper by Østby (2015)). There have also been several large-scale testing campaigns to validate models, and in general the models have been found to very well reproduce the average trend in the experimental results. It is underlined that there are still uncertain areas, e.g. embedded defects, where there is a need for a better understanding of the phenomena and how to include this in models. However, in general it is fair to say that we are in possession of quite accurate models for assessing tensile strain capacity of pipelines with defects.