Freezing and Freeze-Out of Carbon Dioxide In CCUS-Relevant Fluids

The accidental formation of solid carbon dioxide (CO2), commonly known as dry ice, in operational equipment and process facilities can lead to major technical issues and safety risks. For pure CO2 or any fluid with high concentration of CO2, accidental release or rapid decompression of the fluid, can result in the formation of dry ice. Consequently, this may cause solid accumulation, leading to partial or even full blockage of a pipeline, reactor, or any process unit. Even if this does not halt the entire process, it can significantly reduce process efficiency, and increase the risk of incidents within the operational unit. Therefore, a precise understanding of the P-T conditions for dry ice formation is essential to prevent, or at least minimise, the likelihood of operational disruption and, in the worst-case scenario, any serious incident in the operational unit. Additionally, in the event of dry ice formation, knowledge of the freeze-out conditions of the system is another critical piece of information for operators to ensure fast and efficient procedures to address the challenge effectively.

Although both freeze-out and freezing are phase changes governed by thermodynamic principles, the freezing process, due to its dependence on nucleation, exhibits a stochastic nature. In other words, the solid form of a compound or mixture (with a fixed composition) at a given pressure has a unique freeze-out temperature. However, for the same compound or mixture (in the fluid phase), various freezing temperatures can be recorded at a fixed pressure when the fluid temperature is reduced below the freeze-out point.
Despite a few modelling studies on dry ice freeze-out [1,2], a limited number of experimental studies on CO2 and mixtures containing CO2 (e.g., [3–6]) can be found in the open literature. Most of the generated experimental data are derived from systems with low CO2 concentrations, which are not realistic fluids for CCUS processes. Regarding the freezing data, this scarcity is even more pronounced for CO2-rich mixtures. Therefore, an extensive experimental study on freeze-out and freezing of pure CO2 and rich CO2 mixtures relevant to CCUS processes is essential to provide sufficient data for the validation of thermodynamic models and enable the development of models and correlations to estimate the freezing conditions of this type of fluids.
In this study, an experimental setup equipped with ten equilibrium mixing cells was utilised to measure the freeze-out and freezing P-T conditions for pure CO2, and a multicomponent CO2 rich mixture. The equilibrium cells were housed inside a thermal jacket to control the system temperature by circulating ethanol through the jacket using a high-performance cooling bath. During the measurements, the temperatures of the samples were determined using the calibrated average temperature of nine RTDs positioned close to the cells. The pressure of the fluid in each cell was measured using individual pressure transducers connected to each cell. Finally, all the measured data were recorded using a Keysight DAQ970A data acquisition system.
For each set of experimental measurements, the sample temperature was first reduced at a given cooling rate to 15-20 ‌̊ C below the expected CO2 freeze-out temperature to ensure formation of dry ice. After allowing the system sufficient time to reach equilibrium, the temperature was increased to -50 ‌̊ C to ensure complete freeze-out of the dry ice. For each system, starting from a given initial P-T condition, this cooling and heating cycle (with the given rates) was repeated several times to generate sufficient experimental data. Subsequently, the initial P-T condition, and if necessary, the heating and cooling rates were adjusted, and new sets of measurements were conducted. An example of the data obtained for pure CO2 and a multicomponent mixture with ~98 mol% of CO2 is shown in Figure 1. As illustrated in the figure, although the measured freeze-out data for both the pure and multicomponent mixture show relatively good agreement with the predictions of the Span and Wagner equation of state (EoS) for pure CO2 [7], all the measured freezing points for both systems, are consistently lower than the predicted freeze-out points by least ~ 1.5 ‌̊ C difference, and this difference in some cases even increased to ~ 10 ‌̊ C for data at higher pressures