PSI - Issue 13

Letícia dos Santos Pereira et al. / Procedia Structural Integrity 13 (2018) 1985–1992 Author name / Structural Integrity Procedia 00 (2018) 000–000

1986

2

initial porosity critical porosity

f 0 f c f N l y � � � � �

fraction of particles where new voids can nucleate

� , �

, �

element height

coefficients obtained empirically

mean strain

damage acceleration factor

cohesive energy

maximum cohesive stress

von Mises stress hydrostatic stress standard deviation

1. Introduction Accurate and safe structural integrity assessments of gas pipelines are of paramount relevance, since failures can lead to financial and human losses. Considering pipes containing crack-like defects, the unstable propagation is critical and, depending on the material’s toughness, the related fracture micromechanism can be from brittle to remarkably ductile. In this context, several design and integrity assessment protocols have been developed during the last decades, usually based on empirical or semi-empirical models calibrated by real full-scale pipeline burst tests (Leis, 2015; Scheider et al., 2014). Such models have been employed successfully for predicting crack arrestability in low-to medium toughness steels. However, considering the interest of this work on modern, high-resistance and high toughness steels (e.g.: API-5L X80), the quantification of materials’ ability to slow down the propagation of a running crack (crack arrest) is of central relevance, but is not always accurately predicted by available methods. Consequently, it is useful to briefly discuss the phenomenological basis and limitations of the available techniques to predict crack arrest in gas pipelines and how the increase in toughness of structural steels demands revisions and corrections to take larger amounts of plasticity into account. One of the most widely employed methods is the Battelle Two Curve Method (BTCM) formulated in 1970 at the Battelle Institute. This model computes the necessary energy to promote the crack arrest using two independent expressions: i) one to describe the material resistance, based on the speed of the propagating shear fracture; ii) and other to model the gas decompression speed in the vicinity of the growing crack (Leis, 2015). In simple terms, this model quantifies if the decompression speed is higher than the speed of the ductile crack propagation – if it happens, crack driving force is reduced and the desired crack arrest is predicted. For low toughness materials, predictions based on the original model present good experimental agreement and the relationship between this energy and the one obtained from Charpy impact tests (ISO 148-1, 2011, ASTM E23, 2013) is linear, making this small-scale test laboratory of high interest - once the BTCM model has been calibrated by full-scale burst tests, Charpy tests have been widely employed to assess the arrestability of varying steels and applications. However, for medium-to-high toughness steels (e.g.: whose absorbed Charpy energy is higher than ~ 90 J ), the linear relationship is violated and several corrections emerged - additional details can be found in Maxey (1974), Zhu (2015) and Leis (2015). In the case of Leis, for example, a semi-empirical correction factor was proposed and extended the applicability of the BTCM. Nevertheless, the corrected model is not accurate to predict the arrestability of modern high toughness steels applicable to gas pipelines, as discussed below. The phenomenological reasons for such limitations can be discussed based on Fig. 1(a) , despite several assumptions were considered by the author – here, the total Charpy energy obtained from several steels tested in instrumented impact pendulums were stratified in three energies: i) one for crack initiation; ii) one for the deformation of the specimen and; iii) one for crack propagation. It is possible to observe that until ~ 90 J , around 70% of total energy is related to crack propagation and less than 5% to deformation, making BTCM valid. For higher energies, the aforementioned corrections can help, but one can realize that around 250 J deformation energy becomes more pronounced than propagation, which represents less than 30% of the total energy. For ~ 350 J , all absorbed energy is consumed for initiation and deformation, which means that such impact test is not useful to characterize and quantify