Pipelines can be expected to play a significant role in the transportation infrastructure required for the successful implementation of carbon capture and storage (CCS). National Grid is undertaking a research and development programme to support the development of a safety justification for the transportation of carbon dioxide (CO2) by pipeline in the United Kingdom. The ‘typical’ CO2 pipeline is designed to operate at high pressure in the ‘dense’ phase. Shock tube tests were conducted in the early 1980s to investigate the decompression behaviour of pure CO2, but, until recently, there have been no tests with CO2-rich mixtures. National Grid have undertaken a programme of shock tube tests on CO2 and CO2-rich mixtures in order to understand the decompression behaviour in the gaseous phase and the liquid (or dense) phase. An understanding of the decompression behaviour is required in order to predict the toughness required to arrest a running ductile fracture. The test programme consisted of three (3) commissioning tests, three (3) test with natural gas, fourteen (14) tests with CO2 and CO2-rich mixtures in the gaseous phase, and fourteen (14) tests with CO2 and CO2-rich mixtures in the liquid (or dense) phase. The shock tube tests in the liquid (dense) phase are the subject under consideration here. Firstly, the design of the shock tube test rig is summarised. Then the test programme is described. Finally, the results of the dense phase tests are presented, and the observed decompression behaviour is compared with that predicted using a simple (isentropic) decompression model. Reference is also made to the more complicated (non-isentropic) decompression models. The differences between decompression through the gaseous and liquid phases are highlighted. It is shown that there is reasonable agreement between the observed and predicted decompression curves. The decompression behaviour of CO2 and CO2-rich mixtures in the liquid (dense) phase is very different to that of lean or rich natural gas, or CO2 in the gaseous phase. The plateau in the decompression curve is long. The following trends (which are the opposite of those observed in the gaseous phase) can be identified in experiment and theory: • Increasing the initial temperature will increase the arrest toughness. • Decreasing the initial pressure will increase the arrest toughness. • The addition of other components such as hydrogen, oxygen, nitrogen or methane will increase the arrest toughness.

To gain an understanding of the risks associated with a hydrogen pipeline failure, Air Products commissioned GL Industrial Services UK to perform two experiments where a buried 6″ diameter pipeline at an initial pressure of 60 bar was intentionally failed using an explosive charge to generate a full bore release of hydrogen gas from two open ends, simulating a pipeline rupture event in which a ground crater is formed naturally in the surrounding soil by the released gas. The first experiment was performed with the pipe buried 1m deep in a typical soil and the second experiment was performed with a 1m deep backfill of a mixture of sand and soil. The hydrogen released was ignited immediately following the pipeline failure. Following initiation of each experiment, the properties of the hydrogen gas release and resulting fire were measured. The two experiments were conducted under similar conditions, with the main differences being the nature of the soil used for the pipeline backfill and the wind speed (which was significantly higher in the first experiment). The initial pipeline pressure was very similar in the two experiments, with complete depressurisation of the gas pipeline and reservoir taking place over a period of approximately 80 seconds. Maximum flame lengths of up to approximately 100m were measured in each experiment. A number of previous experimental programmes have been carried out by GL in order to investigate the fire characteristics of natural gas releases from ruptured pipelines, conducted under nominally similar conditions. Recently, experiments of this type were also conducted to investigate releases of mixtures of hydrogen and natural gas. The paper will present a high level overview of the results including a discussion of the observed differences between the release and fire behaviour of the different gases.

Ruptures in gas and liquid pipelines are different. A rupture in a gas pipeline is typically long and wide. A rupture in a liquid pipeline is typically short and narrow, i.e. a slit or ‘fish-mouth’ opening. The decompression of liquid (or dense) phase carbon dioxide (CO2) immediately after a rupture is characterised by a rapid decompression through the liquid phase, and then a long plateau. At the same initial conditions (pressure and temperature), the initial speed of sound in dense phase CO2 is greater than that of natural gas and less than half that of water. Consequently, the initial decompression is more rapid than that of natural gas, but less rapid than that of water. A question then arises … Does a rupture in a liquid (or dense) phase CO2 pipeline behave like a rupture in a liquid pipeline or a gas pipeline? It may exhibit behaviour somewhere in-between the two. A ‘short’ defect that would rupture at the initial pressure might result in a short, narrow rupture (as in a liquid pipeline). A ‘long’ defect that would rupture at the (lower) saturation pressure might result in a long, wide rupture (as in a gas pipeline). This is important, because a rupture must be long and wide if it is to have the potential to transform into a running fracture. Three full-scale fracture propagation tests (albeit shorter tests than a typical full-scale test) published in the 1980s demonstrate that it is possible to initiate a running ductile fracture in a CO2 pipeline. However, these tests were on relatively small diameter, thin-wall line pipe with a (relatively) low toughness. The results are not applicable to large diameter, thick-wall line pipe with a high toughness. Therefore, in advance of its full-scale fracture propagation test using a dense phase CO2-rich mixture and 914×25.4 mm, Grade L450 line pipe, National Grid has conducted three ‘West Jefferson Tests’. The tests were designed to investigate if it was indeed possible to create a long, wide rupture in modern, high toughness line pipe steels using a dense phase CO2-rich mixture. Two tests were conducted with 100 mol.% CO2, and one with a CO2-rich binary mixture. Two of the ‘West Jefferson Tests’ resulted in short ruptures, similar to ruptures in liquid pipelines. One test resulted in a long, wide rupture, similar to a rupture in a gas pipeline. The three tests and the results are described. The reasons for the different behaviour observed in each test are explained. It is concluded that a long, wide rupture can be created in large diameter, thick-wall line pipe with a high toughness if the saturation pressure is high enough and the initial defect is long.

No abstract

Two full-scale fracture propagation tests have been conducted using dense phase carbon dioxide (CO2)-rich mixtures at the Spadeadam Test Site, United Kingdom (UK). The tests were conducted on behalf of National Grid Carbon, UK, as part of the COOLTRANS research programme. The semi-empirical Two Curve Model, developed by the Battelle Memorial Institute in the 1970s, is widely used to set the (pipe body) toughness requirements for pipelines transporting lean and rich natural gas. However, it has not been validated for applications involving dense phase CO2 or CO2-rich mixtures. One significant difference between the decompression behaviour of dense phase CO2 and a lean or rich gas is the very long plateau in the decompression curve. The objective of the two tests was to determine the level of ‘impurities’ that could be transported by National Grid Carbon in a 914.0 mm outside diameter, 25.4 mm wall thickness, Grade L450 pipeline, with arrest at an upper shelf Charpy V-notch impact energy (toughness) of 250 J. The level of impurities that can be transported is dependent on the saturation pressure of the mixture. Therefore, the first test was conducted at a predicted saturation pressure of 80.5 barg and the second test was conducted at a predicted saturation pressure of 73.4 barg. A running ductile fracture was successfully initiated in the initiation pipe and arrested in the test section in both of the full-scale tests. The main experimental data, including the layout of the test sections, and the decompression and timing wire data, are summarised and discussed. The results of the two full-scale fracture propagation tests demonstrate that the Two Curve Model is not (currently) applicable to liquid or dense phase CO2 or CO2-rich mixtures.