CO2SAFE-ARREST: A Full-Scale Burst Test Research Program for Carbon Dioxide Pipelines — Part 2: Is the BTCM Out of Touch With Dense-Phase CO2?

The CO2SAFE-ARREST joint industry project (JIP) aims to (1) investigate the fracture propagation and arrest characteristics of steel pipelines carrying anthropogenic CO2, and (2) to investigate the dispersion of CO2 following its release into the atmosphere. The project is supported by two full-scale burst tests, each based on a layout of eight X65 grade 24″ line pipes filled with a dense-phase CO2-N2 mixture. The tests were conducted over the 2017–2018 period at the DNV GL testing site at Spadeadam, UK. An overview of both the CO2SAFE-ARREST JIP and the first full-scale burst test is provided in a companion paper (IPC2018-78517). The dispersion aspect is covered in another companion paper (IPC2018-78530). This paper presents the material properties, the design layout and the results of the first full-scale burst test. Material characterisation of the pipes available to the project and the motivation leading to the design of the layout are first presented. Six pipes had a nominal wall thickness of 13.5 mm and the remaining two pipes had a nominal wall thickness of 14.5 mm. Laboratory testing was conducted on the material at the end of each pipe section. The testing consisted of Charpy impact and Drop Weight Tear tests, capturing the upper shelf fracture energy, load-displacement curves and an assessment of the fracture surfaces. Charpy and Drop Weight Tear test energies as well as strength data are provided. The layout reflects the research focus of the project with both conventional and less conventional pipe arrangements. The test was primarily designed around 13.5 mm nominal wall thickness pipes with a 1m depth backfill and laid East-West. The design was telescopic and introduced an asymmetry with respect to the mid-point by arranging pipe sections with increasing Charpy toughness on one side and increasing yield strength on the opposite side. The fracture was initiated at half-length, across the girth weld between the ‘west’ and ‘east’ initiation pipes. A running ductile fracture ensued, followed by an arrest in the third pipe on either side of the test section. Experimental data relevant to fracture velocity, decompression wave speed of the CO2-N2 mixture and pressure at the crack tip are presented. The discussion is driven from the perspective of traditional running ductile fracture control technology applied to dense-phase CO2 carrying pipelines. Emphasis is put on the analysis of the fracture velocity and transient pressure data relative to the properties of the material and CO2 mixture. The limitations of the Battelle Two-Curve Method (BTCM) traditionally used in the analysis of running ductile fracture are discussed. The design of this test was different from that used in the three full-scale burst tests conducted as part of the COOLTRANS project. The conclusions drawn here support those from the COOLTRANS project and apply to larger D/t ratios. The first CO2SAFE-ARREST test provides additional evidence that the original Battelle Two-Curve Model is not applicable to dense-phase CO2 carrying pipelines. A shift in prediction tool technology is called for.

Analysis of Data From a Full-Scale Burst Test on 1219 mm OD Grade 550 Pipe: Implications for the Prediction of Fracture Velocity

In recent years, considerable doubt has arisen over the prediction of the level of toughness required to arrest a propagating fracture in higher-strength line pipe. It has been clear for many years that the most widely used traditional approach, the Two-Curve Method (TCM) developed at Battelle in the early 1970s, could not be applied directly when the required toughness, expressed as full-size Charpy energy, exceeded about 80–90 J. Initially, this issue was addressed by the adoption of empirical correction factors, but more recently, there have been indications that this approach is no longer effective for modern, high-strength materials. Additional information, which in general can only be derived from well-characterized burst tests, is essential to furthering understanding of the fracture arrest problem under conditions that are typical of modern, long-distance, large-diameter pipeline design. In the context of the Coastal GasLink (CGL) project, TransCanada has carried out a program of full-scale burst testing at the Spadeadam test site of DNV GL. The tests were supported by LNG Canada and the TransCanada Technology Management Program. These tests are described in another paper at this conference [1]. Though most of the testing was directed towards the assessment of different crack arrestor designs, one half of one test contained a run of four pipes of progressively increasing Charpy energy, up to a very high level (over 450 J). The fracture was observed to run through all four pipes, before being arrested by a crack arrestor fitted to a fifth pipe having lower toughness. Nearly all approaches to determining requirements for fracture arrest depend, directly or indirectly, on relationships between fracture velocity (for given levels of fracture resistance) and the driving force, generally considered to be directly related to the pressure in the plane of the crack tip. By comparing measured fracture velocity with the crack tip pressure determined either directly at pressure transducer locations or by comparison with propagation velocities within the expansion wave, conclusions can be drawn regarding the accuracy of existing relationships. Most previous work regarding correction factors has been based simply on discrepancies between predicted and observed propagation and arrest behaviour. Direct comparisons of observed and predicted fracture speed potentially provide much more data and focus more clearly on where model deficiencies may lie. The current analysis focuses on comparisons with the predictions of the traditional TCM and those of a transient model developed by JFE. While data from the present work are clearly limited, this approach appears to present a way of recalibrating fracture velocity formulations that may extend the range over which traditional, Charpy-based approaches can be applied. For the future, the incorporation of additional results from other recent, well-characterized burst tests would be extremely valuable in this respect.

STUDIES ON THE DYNAMIC FRACTURE PROPERTIES AND FAILURE MODES OF A PBX

Polymer-bonded explosives (PBXs) are particulate composite materials composed of crystalline explosive grains bound in a relatively soft polymeric binder. It is important to optimize the fracture properties, while still maintaining the low sensitiveness and high explosiveness of PBX. This paper describes a study on the fracture properties and failure modes of a PBX by adopting a newly proposed dynamic fracture experimentation method — notched semi-circular bend (NSCB) specimen loaded with split Hopkinson pressure bar (SHPB) which was used in this study. This method offers the advantage of simultaneously determining the fracture initiation toughness, fracture energy, fracture propagation toughness and fracture velocity. The crack propagation is monitored by using a synchronous high-speed camera, which allows the observation of strain field history via digital image correlation process. The experimental results indicate that both the initiation toughness and the propagation toughness linearly increase with loading rate. The propagation fracture toughness is found to increase with fracture velocity, and a limiting fracture velocity is obtained. The failure modes are interpreted by using various theoretical models. Results suggest that the debonding strength of the binder is much smaller than the crystal fracture strength. The tensile strength is similar to the debonding strength, while the compression strength is somewhere intermediate between them.

Correlation of plume morphologies on joint surfaces with their fracture mechanic implications

The fractography and conditions of propagation of joints that cut Devonian siltstones in the Appalachian Plateau, New York, and Eocene chalks from the Beer Sheva Syncline, Israel, are investigated. The joints cutting the siltstones are marked by S-type and C-type plumes, and the joints cutting the Lower Eocene and Middle Eocene chalks are marked by coarse and delicate plumes, respectively. The four plume types propagated under sub-critical (slow propagation) conditions. On the semi-quantitative fracture velocity (v) versus the tensile stress intensity (KI) curves, the S and C plume types fall in the KI=0.073–0.79 MPa m1/2 and v=2×10−4–10−2 m/s and KI=0.073–0.79 MPa m1/2 and v=10−6–10−4 m/s ranges respectively. The coarse and delicate plumes fall in the KI=0.03–0.17 MPa m1/2 and v=10−6–4×10−5 m/s and KI=0.03–0.17 MPa m1/2 and v=10−4–5×10−3 m/s ranges, respectively. Generally, slow plumes are relatively short, show periodicity, and typically exhibit superposition of arrest marks. On the other hand, faster plumes are longer and continuous, occur particularly in thinner layers, and show no superposition of arrest marks. There is a clear distinction between two en échelon segmentation end-members in the joint fringe, the ‘discontinuous breakdown type’ and the ‘continuous breakdown type’. There are also ‘transitional’ variations between the end-members. Only curved ‘discontinuous breakdown type’ boundaries of en échelon fringes can be equated with mirror boundaries.

The Application of the Battelle “Short Formula” to the Determination of Ductile Fracture Arrest Toughness in Gas Pipelines

Many models and formulae have been put forward, over the years, for the determination of the toughness necessary for the arrest of propagating ductile fracture in gas pipelines. One of the first, and most prominent, was that developed by Battelle Columbus Laboratories for the Pipeline Research Committee of the American Gas Association. As originally embodied, the model involved the comparison of curves expressing the variation of fracture velocity and of decompression wave velocity with pressure (the “two-curve model” — TCM). To aid in analysis, at a time long before a computer was available on every desk, a “short formula” (SF) was developed that provided a good fit to the results of the TCM for a substantial matrix of conditions. This SF has subsequently been adopted by several standards bodies and used widely in the analysis of the results of full-scale burst tests. Since the original description of the derivation of the SF is to be found only in a report to the PRC dating back to the Seventies, many in the pipeline industry today are left without a full appreciation of its range of validity. The present paper briefly discusses the original intent of the SF as a substitute for the TCM, and presents the results of extensive calculations comparing the results of the two. It can be concluded that the SF provides an excellent estimate of the results of the TCM over a very wide range of design and operating parameters, within the limitations inherent in the method.