Carbon dioxide (CO₂) as a major greenhouse gas (GHG) has increased significantly impacts on the earth owing to human activities such as burning of oil and gas and the discharge of exhaust gases. The Intergovernmental Panel on Climate Change (IPCC) reported that Carbon Dioxide Capture and Storage (CCS) can eliminate 20–40% of global carbon emissions (Metz et al., 2005). Carbon dioxide usually would be captured at a large point emission source (e.g., power plants) and be transported via long pipelines to another spot for use [e.g., oil field for enhanced oil recovery (EOR)] (Ziabakhsh-Ganji and Kooi, 2014). Currently more than 50 million tons of CO₂ is transported by over 6,400 km of pipelines in the United States (Metz et al., 2005). And the most pipelines are under supercritical/dense state which is considered as the most efficient way. Under stricter environmental policies, up to 200,000–360,000 km by 2050 could be built and operated in the United States, China, and Europe (John and John, 2004). This would require more attention to CO₂ transportation safety.

Potential leakage can happen with the development of pipeline corrosion and other outside forces, such as construction defects, solid movement, etc. The discharge and dispersion of high-pressure CO₂ pipeline different from the hydrocarbons pipeline involve complex physics including cool temperature, phase transition, sonic multiphase flow, and heavy gas dispersion. As an asphyxiant at high concentrations and heavier than air, the leaked CO₂ would accumulate in low-lying land and harm safety of living creature nearby (Wang et al., 2020). For safety issues related to CO₂ transportation, it is necessary to determine how CO₂ is released in the case of failure. More importantly, there exists little quantitative information on the source terms including near-field characteristics, which are useful for establishing appropriate models in release and dispersion.

In recent years, many researchers have made a lot of achievements in the numerical simulation of CO2 release and diffusion. (Webber, 2011) developed the two-phase flow model for flashing jet of CO₂. It revealed that two-phase homogeneous equilibrium flow models may be generalized to cover such a release. (Liu et al., 2016) simulated the CO₂ dispersion over two hypothetical topographies. This study provides a viable method for assessment of risks associated with CCS. (Wareing et al., 2013a) present a composite equation of state accounting for the three-phase CO₂ in the modeling of liquid CO₂ release. The paper predicted the near-field structure of the jet and the fraction of solid CO₂. (Liu et al., 2014) simulated the highly under-expanded single-phase CO₂ jets using CFD software Fluent implanted with Peng−Robinson (PR) equation of state (EoS) for accounting for real gas behavior. The two-stages simulation approach was used and resulted in heavy computational workloads. A consequence model with a pseudo source is employed to predict the dispersion of supercritical CO₂ from a high-pressure pipeline (Joshi et al., 2016). They assumed that the pseudo source plane is in the plane that is approximately 3.5 times the diameter of the orifice away from the exit plane. And in this plane, it is homogenous flow. (Woolley et al., 2014) designed an effective multiphase jet expansion model to predict the leakage of CO₂ after accidental damage to the high-pressure casing. The evolution of dry ice has been considered in some research. The behavior of CO₂ particles during the release of high-pressure liquids has been studied using a CFD, combined with a Reynolds stress turbulence model, Lagrangian particle tracker, particle distribution function, and turbulent shear agglomeration model for the particle evolution (Wareing et al., 2013b). The heavy gas dispersion models were developed based on the study of discharge models. In order to predict the dispersion consequences more accurately, the complex physics in near-field such as the structure and shock waves should be figured out. Unfortunately, due to a lack of experimental data in supercritical CO₂ releases focusing on near-field characteristics, currently the development of more complex models is limited.

Some experiments related to the supercritical CO₂ release almost focused on the macroscopic parameters, such as pressure, temperature, concentration, and velocity. However, few studies focused on microscopic parameters, such as the structure of shock waves, the evolution of solid CO₂, and the expanded angle. (Ahmad et al., 2013) carried out a controlled CO₂ release experiment from various circular orifices to obtain the thermo-hydraulic data of CO₂. A superheated jet was founded during the releases. In our previous work (Wang et al., 2019), effects of impurity concentration, initial inner pressure, and temperature on dispersion behavior were studied. (Guo et al., 2016) designed and built a large-scale supercritical experimental pipeline with a total length of 258 m and an inner diameter of 233 mm. The under-expanded jet flow structure and phase transitions in the near-field were studied for supercritical CO₂ released though different orifice diameters. Relating to the small-scale experiments focused on the near-field jet of CO₂. (Wareing et al., 2014) measured dry ice particles distribution along the jets in liquid CO₂ release and found that the sizes of particles are around 0.1–100 μm. And the study presented Mach disc in these releases is at a distance of around seven nozzle diameters along the centerline from the nozzle and the particles are likely to be close to equilibrium after Mach disc. The effect of superheat on flashing atomization characteristics and on the snow formation of liquid CO₂ has been investigated (Lin et al., 2013). Results show that the spray pattern transfers from jet spray to cone spray, and then to a bowl spray configuration with the increase of superheat. As mentioned above, a lack of near-field experiments data limits the development of models of discharge and dispersion of CO₂. In addition, CO₂ was released from a circular orifice or nozzle in almost current release experiments of CO₂ pipeline. However, the cracks in the damaged pipeline usually are not circular. The difference in orifice pattern may results in different consequence in an accident release.

In this paper, an experimental setup with a high-speed camera system was designed and constructed to study the near-field characteristics of flashing-spray jet of supercritical CO₂ from various orifice geometries. The work focused on (1) the near-field structures during CO₂ released from different orifice geometries; (2) the evolution of jet angle which can affect the dispersion region; (3) the shock waves system in the release from orifices with different geometries; (4) finally models of the flashing-spray jet of supercritical CO₂.

The jet from a high-pressure CO₂ vessel is different from general gas jet due to the phase transition and multiphase flow and must be considered. In view of general gas jet, the jet zone is divided into three sub-zones: flow establishment zone, tansition zone, and established flow zone, as shown in Figure 1. The fluid jet from the orifice to the ambient, which causes discontinuous velocity, further causes turbulence. As shown in Figure 1, extending the upper and lower boundaries of the jet to intersect at one point (Point O), and O is the virtual origin of the jet. And then θ is the jet angle.

A schematic of the highly under-expanded jet is shown in Figure 2. An expansion fan generates at the nozzle lip as the flow expands into the atmosphere. The pressure ratio P ₀/P _∞ is an important parameter to describe the expansion level, where P ₀ is a stagnation pressure in the vessel and P _∞ is the ambient pressure. When P ₀/P _∞ > 15, the complicated shock waves system will forms. Except for the intercepting shock in the interior of the jet, the Mack disc normal to the flow is unique for under-expanded jet. The flow front the Mach disc is supersonic, whereas the flow behind the Mach disc is obviously subsonic. The temperature will rise sharply near the Mach disc. However, the shock-wave structure in these jets also depends on a geometry of the nozzle and the property of gas (Velikorodny and Kudriakov, 2012). Some researchers also reported that the Mack disc cannot be observed when the flow jets from a elliptic nozzle (Menon and Skews, 2010).

For non-condensing gas jet, a theoretical analysis has been developed to predict the Mach disc location, X m d e = 0.5 γ P e P ∞ × ( γ + 1 γ − 1 ) 0.25 (1) where X _m is location of the Mach disc, P _e is the static pressure at the exit section. It should be noted that in the fact the Mach disc location weakly depends on γ, and it can be approximated by a commonly used experimental correlation of (Ashkenas and Sherman, 1966): X m d e = 0.67 × P 0 P ∞ (2)

In order to study the near-field structure of supercritical CO₂ released from the pressure pipe, a new experimental device was designed and built, as shown in Figure 3. The experimental apparatus consists of high-pressure vessel, gas source, CO₂ pump, refrigerating unit, thermostatic water bath, nozzle, and high-speed camera system. The rated pressure of the vessel with a volume 6 L is 15.0 MPa, and the material is 316L stainless steel. The container is filled with liquid CO₂ cooled by a refrigeration unit and controlled by a constant temperature water bath. To study the influence of different orifice diameters and orifice patterns (circular and rectangular) on near-field structure and dispersion of supercritical CO₂, six different orifices were used, as shown in Figure 3. The supercritical CO₂ near-field jet structure was observed with a single-lens reflector camera with a maximum frame rate of 200,000 FPS (frames per second). In this experiment, the frame rate of the high-speed camera is set at 3000 FPS.

It is very necessary to carry out the experiment under the premise of ensuring the safety of the experiment. In the process of supercritical CO₂ injection, steel frame is adopted to prevent the generation of reaction force, and the noise level is controlled in an acceptable range. To ensure the stable experimental conditions, the experiments were carried out indoor to avoid the impact of atmospheric turbulence. According to the actual transportation conditions of CO₂ pipeline, most of the initial conditions in the experiment are in the supercritical region. The main steps list as follows: (1) Before work, check whether the connection of the experimental device is loose and whether the container is damaged to ensure the normal operation of the equipment; (2) Open the cleaning mode to empty the air in the container to remove impurities; (3) The liquid carbon dioxide cooled by the refrigerator is fed into the container by means of a CO₂ pump; (4) When the appropriate amount of CO₂ is injected into the container, all valves are closed and a water bath heating sleeve is used to control the temperature in the container; (5) When the experimental conditions in the container reach the design conditions, the pneumatic valve in the pipeline will be opened quickly; (6) Record the experimental process with high speed camera.

This paper presents the experimental results of near-field structure when supercritical CO₂ releases from the orifices with different sizes and patterns. The main conclusions are summarized as follows:

1) The shape of supercritical CO₂ jet from a circular orifice is a near-cylinder structure and the process is a highly under-expanded jet marked with the Mach disc. However, when supercritical CO₂ is released from the rectangular orifice, the shock wave system has a fan-shaped structure, and the dispersion region perpendicular to the jet is wide.