(32p) Understanding Depressurization of CO2 Vessels Using Computational Modeling and Data-Driven Techniques for Safety in Carbon Storage and Transport

Rapid depressurization or blowdown of vessels is an important safety protocol in industries during an emergency (fire, etc.). By evacuating the contents of a vessel and discharging through the flare system, overpressure and stress rupture of the vessel can be avoided. However, the rapid depressurization process comes with certain challenges. Firstly, an important criterion is to determine an appropriate orifice size such that the depressurization completes within a specific time. Similarly, the temperature of the fluid in the vessel can drop significantly during depressurization leading to cold wall temperature which can fracture the vessel.

In the context of carbon storage and transport applications, there is significant interest in the depressurization of CO₂ rich vessels. At certain conditions, the depressurization of a CO₂ system could exhibit a couple of unique phenomena in addition to the above-mentioned design considerations, namely, (a) formation of large amounts of boiling liquid below the CO₂ critical pressure and (b) formation of solid CO₂ at colder temperatures. The former could have a significant effect on the metal wall temperatures due to the high heat transfer coefficient of the boiling liquid while the latter can cause blockages in the vessels and pipelines. An understanding of the conditions that could lead to the occurrence of both phenomenon is helpful to design the depressurization of CO₂ systems. Although there have been experimental^1,2 and simulation³ studies in literature demonstrating the qualitative influence of certain design parameters on the depressurization of CO₂ systems, there exists a lack of understanding about the quantitative or relative influence of each individual design parameter on both the formation of large amounts of boiling liquid and the potential of solid CO₂ formation.

In this study, we utilize computational modeling and data-driven approaches to establish a quantitative linkage between the design parameters and the occurrence of both phenomena associated with depressurization of CO₂ systems. The design parameters considered for this study are initial pressure, initial temperature, and orifice size. Based on a labeled dataset generated for this study, the quantitative influence of each design parameter can be deduced and the relative impact of tweaking a design parameter can be understood. The transient depressurization of CO₂ vessels is modeled using the BLOWDOWN Technology in Aspen HYSYS. We first validate the model by comparing the BLOWDOWN results for CO₂ system with the results reported for an experimental study in literature. Specifically, we compare the profiles for both pressure and temperature as a function of time. A labeled dataset for over 200 BLOWDOWN cases is then generated with a wide range of design parameters. Using machine-learning based techniques such as logistic regression and Random Forest classifier, we quantify the importance of each design parameter and establish a quantitative linkage between all the design parameters and the formation of both large amounts of boiling liquid and solid CO₂.

Our results reveal a good agreement between the BLOWDOWN and the experimental results. The fitted curves and the feature importance values indicate that the initial temperature and initial pressure strongly correlate with the formation of large amounts of boiling liquid, whereas the solid CO₂ formation is primarily dictated by the orifice diameter with a slight dependence on the initial temperature. The initial pressure is seen to have a negligible impact on the solid CO₂ formation. Broadly, this study provides insight into the design considerations for depressurization of CO₂ systems which is critical for carbon storage and transport applications.

References:

(1) B. Gebbeken et al. J. Loss Prev. Process Ind, 9 (4), 285-293, 1996

(2) Munkejord et al. Int J Greenhouse Gas Control; 109: 103361, Jul. 2021

(3) Umar Shafiq et al. IOP Conf. Ser.: Mater. Sci. Eng. 458 012077, 2018