Carbon dioxide capture and storage (CCS) is one of the technologies that have been proposed to reduce emissions of carbon dioxide (CO2) to the atmosphere. CCS will require the transportation of the CO2 from the “capture” locations to the “storage” locations via large-scale pipeline projects. One of the key requirements for the design and operation of pipelines in all jurisdictions is fracture control. Supercritical CO2 is a particularly challenging fluid from this point of view, because its thermodynamic characteristics are such that a very high driving force for fracture can be sustained for a long time. Even though CO2 is not flammable, it is an asphyxiating gas that is denser than air, and can collect in low-lying areas. Additionally, it is well known that any pipeline rupture, regardless of the nature of the fluid it is transporting, has a damaging reputational, commercial, logistic, and end user impact. Therefore, it is as important to control fracture in a CO2 pipeline as in one transporting a flammable fluid. With materials specified appropriately for the prevention of brittle failure, the key element is the control of propagating ductile (or tearing) fracture. The determination of the required toughness for the arrest of ductile fracture requires knowledge of the decompression behavior of the contained fluid, which in turn requires accurate knowledge of its thermodynamic characteristics along the decompression isentrope. While thermodynamic models based on appropriate EOS (equations of state) are available that will, in principle, allow determination of the decompression wave speed, they, in general, have not been fully validated for very rapid transients following a rupture. This paper presents experimental results of the decompression wave speed obtained from shock tube tests conducted on pure CO2 from different initial conditions, and comparison with predictions by models based on GERG-2008, Peng-Robinson, and BWRS equations of state (EOS). These tests were conducted as a baseline before introducing various impurities.

Shock tube tests were conducted on a number of binary CO2 mixtures with N2, O2, CH4, H2, CO, and Ar impurities, from a range of initial pressures and temperatures. This paper provides examples of results from these tests. The resulting decompression wave speeds are compared with predictions made utilizing different equations of state (EOS). It was found that, for the most part (except for binaries with H2), the GERG-2008 EOS shows much better performance than the Peng–Robinson (PR) EOS. All binaries showed a very long plateau in the decompression wave speed curves. It was also shown that tangency of the fracture propagation speed curve would normally occur on the pressure plateau, and hence, the accuracy of the calculated arrest toughness for pipelines transporting these binary mixtures is highly dependent on the accuracy of the predicted plateau pressure. Again, for the most part, GERG-2008 predictions of the plateau are in good agreement with the measurements in binary mixtures with N2, O2, and CH4. An example of the determination of pipeline material toughness required to arrest ductile fracture is presented, which shows that prediction by GERG-2008 is generally more conservative and is therefore recommended. However, both GERG-2008 and PR EOS show much worse performance for the other three binaries: CO2 + H2, CO2 + CO, and CO2 + Ar, with CO2 + H2 being the worst. This is likely due to the lack of experimental data for these three binary mixtures that were used in the development of these EOS.

No abstract

No abstract

The control of propagating ductile (or tearing) fracture is a fundamental requirement in the fracture control design of pipelines. The Battelle two-curve method developed in the early 1970s still forms the basis of the analytical framework used throughout the industry. GASDECOM is typically used for calculating decompression speed, and idealizes the decompression process as isentropic and one-dimensional, taking no account of frictional effects. While this approximation appears not to have been a major issue for large-diameter pipes and for moderate pressures (up to 12 MPa), there have been several recent full-scale burst tests at higher pressures and smaller diameters for which the measured decompression velocity has deviated progressively from the predicted values, in general towards lower velocities. The present research was focused on determining whether pipe diameter was a major factor that could limit the applicability of frictionless models such as GASDECOM. Since potential diameter effects are primarily related to wall friction, which in turn is related to the ratio of surface roughness to diameter, an experimental approach was developed based on keeping the diameter constant, at a sufficiently small value to allow for an economical experimental arrangement, and varying the internal roughness. A series of tests covering a range of nominal initial pressures from 10 to 21 MPa, and involving a very lean gas and three progressively richer compositions, were conducted using two specialized high pressure shock tubes (42 m long, I.D. = 38.1 mm). The first is honed to an extremely smooth surface finish, in order to minimize frictional effects and better simulate the behaviour of larger-diameter pipelines, while the second has a higher internal surface roughness. The results show that decompression wave speeds in the rough tube are consistently slower than those in the smooth tube under the same conditions of mixture composition and initial pressure & temperature. Preliminary analysis based on perturbation theory and the fundamental momentum equation indicates that the primary reason for the slower decompression wave speed in the rough tube is the higher spatial gradient of pressure pertaining to the decompression wave dynamics, particularly at lower pressure ratios and higher gas velocities. The magnitude of the effect of the slower decompression speed on arrest toughness was then evaluated by a comparison involving several hypothetical pipeline designs, and was found to be potentially significant for pipe sizes DN450 and smaller.