1. Introduction
With the adoption of the Paris Agreement in 2015, which aimed to limit the increase in average global temperature to below 2 °C compared to pre-industrial levels and restrict it to no more than 1.5 °C, reducing greenhouse gas emissions has become a significant global issue in response to rapid climate change caused by global warming.
IRENA (2022) mentions carbon capture and storage (CCS) as one of the key technologies for achieving greenhouse gas emission reduction targets and expects it to account for approximately 20% of greenhouse gas reductions by 2050. Therefore, the need for research to effectively capture, transport, and store carbon is increasing.
Pipelines and liquefied CO
2 carriers can be considered for transporting captured CO
2 to storage sites efficiently. In particular, shipping was reported to be more economical for long-distance transport. In the case of shipping, liquefaction reduces the volume of CO
2, allowing more mass of CO
2 to be transported per unit volume.
Decarre et al. (2010) reported that transport by ship is economically more advantageous than pipeline transport when the distance to the storage site exceeds 1,000 km.
The problem of boil-off gas (BOG) can occur while transporting liquefied CO2 by ship because liquefied CO2 must be stored at temperatures lower than room temperature. Although onboard storage tanks are insulated, they cannot block heat ingress completely. Therefore, liquefied CO2 evaporates due to external heat influx, and the resulting BOG increases the pressure in the storage tank, potentially causing safety accidents. If the generated BOG is vented to the atmosphere to prevent overpressure, it results in cargo loss and greenhouse gas emissions. A BOG reliquefaction process to prevent this can be crucial in minimizing greenhouse gas emissions and ensuring the safety of the CCS supply chain.
Existing studies have mainly focused on the liquefaction process of captured CO
2.
Alabdulkarem et al. (2012) proposed a new liquefaction system applying various refrigerants to improve the efficiency of the CO
2 liquefaction process for CCS.
Lee et al. (2015) modeled the process for liquefying captured CO
2 and studied changes in process efficiency according to seawater temperature variations
Seo et al. (2015) used ammonia, propane, and R-134a as refrigerants in the CO
2 liquefaction process. They reported that ammonia showed the best efficiency.
Seo et al. (2016) modeled the Linde–Hampson process and a process using external refrigerants to liquefy captured CO
2 within the CCS value chain. They conducted a cost analysis according to liquefaction pressure.
Deng et al. (2019) modeled a liquefaction process using ammonia refrigerant and performed economic evaluations of captured CO
2 based on different scenarios.
Aliyon et al. (2020) modeled and analyzed energy, exergy, and economic aspects of the Linde–Hampson process, a process using ammonia refrigerant and an absorption refrigeration process for the liquefaction of captured CO
2. Nevertheless, these studies focused only on the CO
2 liquefaction process and did not consider the reliquefaction process of CO
2 BOG occurring during ship transport.
Some studies on CO
2 BOG reliquefaction systems have been conducted but have limitations.
Yoo (2017) studied a CO
2 BOG reliquefaction system using LNG cold energy on an LNG-fueled CO
2 carrier.
Lee et al. (2017a) proposed a CO
2 BOG reliquefaction process and presented optimal designs considering the seawater temperature and compressor discharge temperature.
Lee et al. (2024) modeled a CO
2 BOG reliquefaction system using the Linde–Hampson process, a process using ammonia refrigerant, and a subcooling process, comparing them based on the liquefaction pressure. On the other hand, these studies have the limitation of assuming CO
2 BOG as pure CO2. In the CO
2 capture process, various impurities can be mixed into the liquid CO
2. According to
Jeon et al. (2015), substances with low boiling points as impurities, such as nitrogen, rapidly vaporize within the liquid CO
2 storage tank, making CO
2 BOG a mixture containing more impurities than the liquid form. CO
2 BOG mixed with nitrogen requires lower temperatures for liquefaction; therefore, an effective reliquefaction process capable of cooling to lower temperatures than pure CO
2 is needed. This leads to increased energy consumption and costs, making research on efficient reliquefaction methods for impurity-containing CO
2 BOG an important topic in liquefied CO
2 transport.
Some studies examined reliquefying impurity-containing CO
2 BOG, but they had limitations.
Chu et al. (2012) reported that BOG generated during the transport of captured CO
2 from thermal power plants contains impurities, such as nitrogen and oxygen, and proposed an ethylene–propane cascade cycle as a reliquefaction process.
Lee et al. (2017b) proposed and compared four processes using CO
2 as a refrigerant to reliquefy CO
2 BOG containing 36 mol% nitrogen. Using the cascade cycle increased the process complexity because separate compression trains were required for each refrigerant. Cooling below −65.6 °C was necessary to achieve a reliquefaction rate of over 98% for impurity-containing BOG, which poses a risk of dry ice formation. In addition, using CO
2 as a refrigerant required compressing CO
2 to high pressures exceeding 100 bar compared to external refrigerants. To achieve a CO
2 reliquefaction rate of 90%, they proposed a process applying a distillation column to separate nitrogen and CO
2, but this increased energy consumption by approximately 1.5 fold compared to other processes.
The use of mixed refrigerants in the liquefaction of mixtures has also been studied widely. Existing low-temperature liquefaction processes, such as natural gas liquefaction and LNG BOG reliquefaction, use liquefaction processes using mixed refrigerants, reporting higher efficiency compared to single-refrigerant liquefaction processes. Using mixed refrigerants in natural gas liquefaction can increase efficiency compared to nitrogen single-refrigerant cycles or cascade cycles by reducing the temperature difference between hot and cold streams in the heat exchanger (
Moein et al., 2015;
Zhang et al., 2020). According to
Lee et al. (2023), using mixed refrigerants in LNG BOG reliquefaction reduced power consumption by 30% compared to using single nitrogen refrigerants.
Xu et al. (2013) reported that the composition of mixed refrigerants plays a significant role in the heat exchange efficiency of the process of natural gas liquefaction. On the other hand, such processes using mixed refrigerants have mainly been applied to natural gas liquefaction, and studies optimizing the composition of mixed refrigerants for impurity-containing CO
2 BOG mixtures are difficult to find.
This paper proposes an optimal process design that enables efficient BOG reliquefaction even under temperature constraints to avoid dry ice formation by incorporating a mixed refrigerant liquefaction process into the reliquefaction of impurity-containing CO
2 BOG mixtures. The CO
2 BOG reliquefaction process using ammonia single refrigerant proposed by
Seo et al. (2015) was compared. In the reliquefaction process, mixed refrigerants were applied, and a process incorporating a vapor–liquid separator before expansion was proposed to improve the reliquefaction rate.
3. Process Simulation Conditions
3.1 Carbon Dioxide Storage Conditions
Among the storage pressures for CO
2, the most frequently mentioned are 6.5 bar and 15 bar. Considering the density of liquefied CO
2 in saturated conditions, the density was 1154.6 kg/m
3 and 1031.7 kg/m
3 at 6.5 bar and 15 bar, respectively. This means that at 6.5 bar, the amount of liquefied CO
2 that can be stored per unit storage volume increased by approximately 12% compared to 15 bar (
Jeon et al., 2015). For this reason, the storage pressure of the CO
2 tank was assumed to be 6.5 bar in this paper.
The unit capacity of the CO
2 storage tank applied in this study was set to 100,000 m
3 following the assumption of
Chu et al. (2012), and it was assumed that the tank is filled up to an initial liquid level of 95% and operated. The boil-off rate (BOR), which is the amount of BOG generated during one day, is typically evaluated at 0.12%–0.2% (
Chu et al., 2012;
Yoo, 2017). In this study, a BOR of 0.13% was assumed, resulting in a BOG generation amount of 6,000 kg/h. The pressure of the generated BOG was assumed to be the same as the storage pressure, 6.5 bar.
The saturation temperature of CO
2 at 6.5 bar was −50.3 °C. However, in the case of large-scale storage tanks for liquefied cryogenic fluids, heated gas accumulates at the top due to external heat ingress, existing as superheated gas. Therefore, it was assumed that the temperature of the BOG increased to −28.5 °C when entering the reliquefaction facility (
Chu et al., 2012).
3.2 Carbon Dioxide BOG Composition
The CO
2 loaded onto the ship exists as both boil-off gas and liquid CO
2 after loading, and the volumetric ratio of the gas space in the storage tank is called the ullage. The ullage can be estimated based on the volumetric ratio of liquid and gas using
Eq. (3).
where
Vvap and
Vliq represent the volumes occupied by the gas and liquid in the tank, respectively.
vvap and
vliq denote the molar volumes of the gas and liquid at that time, and
x is the vapor mole fraction. This study calculated the vapor mole fraction corresponding to the volumetric ratio of liquid and gas using the commercial software Aspen HYSYS V11, and estimated the gas composition at that time as the BOG composition. When volatile impurities are present in liquefied CO
2, the composition of the boil-off gas at equilibrium changes according to the ullage ratio (
Jeon et al., 2015). The total volume of the tank was fixed to 100,000 m
3, and a decrease in the value of ullage means that the space occupied by BOG in the tank decreases. As the space occupied by BOG decreases, the mole fraction of nitrogen, which vaporizes more readily than CO
2, increases compared to when the space is larger.
Fig. 3(a) shows the change in the mole fraction of nitrogen in CO
2 BOG according to ullage when assuming that liquid CO
2 has a composition of 0.1 mol% nitrogen and 99.9 mol% CO
2, reaching equilibrium at 6.5 bar. The mole fraction of nitrogen increases as the ullage space in the tank decreases, causing the nitrogen mole fraction in the CO
2 BOG to vary from 0.1% to 8.7 mol%. The nitrogen mole fraction in CO
2 BOG was estimated to range from 7.6% to 15.4 mol% when the nitrogen concentration in liquid CO
2 was changed from 0.1 mol% to 0.2 mol% at 6.5 bar and 10% ullage (
Fig. 3(b)). In other words, even if only a small amount of nitrogen was included in liquid CO
2, a relatively large amount of nitrogen impurities may exist in the CO
2 BOG, which becomes an important consideration in the reliquefaction cycle. This study compared the optimized performance results of each reliquefaction cycle when the nitrogen content in CO
2 BOG was varied from 5 mol% to 15 mol%.
3.3 Carbon Dioxide BOG Reliquefaction Process
Fig. 4 presents process flow diagrams of the CO
2 BOG reliquefaction processes considered in this study. In Cases 1, 2, and 3, the CO
2 BOG is first compressed to high pressure and then cooled to low temperatures through heat exchange with a refrigerant in a heat exchanger. Subsequently, it undergoes partial liquefaction by expanding to the tank pressure via a valve, and it is separated into gas and liquid phases in a vapor–liquid separator. The non-condensable gases are then vented to the atmosphere, and the liquefied CO
2 is returned to the tank.
For Cases 1 and 2, ammonia refrigerant applied widely in CO2 liquefaction process studies was used as the refrigerant. Case 3 used a mixed refrigerant as the refrigerant, selecting ethane (C2), propane (C3), normal butane (nC4), and isobutane (iC4) as the components. The boiling point of pure CO2 at 6.5 bar is −50.3 °C, and propane, with a boiling point of approximately −42.1 °C at atmospheric pressure, is sufficient to liquefy it. On the other hand, a lower cooling temperature may be necessary for liquefaction for mixtures containing substances with lower boiling points, such as nitrogen. Therefore, ethane with a boiling point of approximately −88.5 °C at atmospheric pressure was included. Furthermore, normal butane and isobutane, with boiling points of approximately 0.5 °C and −11.7 °C, respectively, were included to improve the heat exchange efficiency in the high-temperature section of the heat exchanger.
Cases 4, 5, and 6 are process diagrams where an additional separator was installed in Cases 1, 2, and 3, respectively, to enhance the separation efficiency. Installing the additional separator serves to pre-release nitrogen, which vaporizes more readily than CO2, sending high-purity CO2 to the valve.
The Peng–Robinson equation of state was used to calculate the thermodynamic properties according to the pressure of liquid CO
2 (LCO
2). Each process was simulated using the commercial software Aspen HYSYS V11 under the conditions specified in
Table 2 to ensure an equivalent comparison.
3.4 Optimization Method for the Carbon Dioxide BOG Reliquefaction Process
Optimization was performed using the Particle Swarm Optimization (PSO) algorithm to find the optimal operating conditions for each reliquefaction process and nitrogen content case. The Specific Power Consumption (SPC)―the energy consumption per unit production of liquefied CO2― was set as the objective function to be minimized to achieve the optimal performance of the CO2 reliquefaction process.
While satisfying the conditions in
Table 2, optimization constraints were established to ensure that the temperature within the process remained above −55 °C even after the Joule–Thomson expansion of CO
2 to prevent liquid ingress and dry ice formation. If the constraints are violated, a penalty value is assigned to prevent the solution from being selected as the optimal value. The minimum SPC of the reliquefaction process was determined by considering key process variables, such as temperature, pressure, and refrigerant composition, as optimization variables for the reliquefaction process. These are summarized in
Table 3.
4. Results
Among the optimization results of the reliquefaction systems,
Table A1, A2 and
Table 4 lists the cases where the nitrogen content in the CO
2 BOG was 5 mol% and 15 mol%. According to
Table 4, when the nitrogen content in the CO
2 BOG was varied from 5–15 mol%, Cases 1, 2, and 3 showed a decrease in the total power consumption of the compressors as the compression ratio of the second compressor and the refrigerant flow rate decreased. In addition, the amount of reliquefied CO
2 decreased from 4,030, 4,080, and 4,296 kg/h at 5 mol% nitrogen to 570 kg/h at 15 mol% nitrogen. This is because the amount of cooling in the heat exchanger decreases, and the purge amount increases after Joule–Thomson expansion.
Fig. 5 shows the vapor fraction values of the stream after the Joule–Thomson expansion in each reliquefaction process. The vapor fraction after expansion increased as the nitrogen content in CO
2 increased, reaching up to 0.91 in Cases 1, 2, and 3. This was attributed to the minimum temperature constraint in the CO
2 reliquefaction process. If more nitrogen with a lower saturation temperature than CO
2 (−175.7 °C at 6.5 bar) is included, the phase equilibrium of the mixture shifts, resulting in the need for a lower temperature to produce the same amount of liquid under the same pressure conditions. In the BOG reliquefaction process, however, the lower temperature limit must be constrained to prevent dry ice formation. In this study, the formation of liquid after expansion was limited because the lower limit temperature is restricted to −55 °C.
Therefore, in Cases 1, 2, and 3, as the nitrogen content increases, most of the CO
2 is purged due to the temperature constraint, resulting in decreases in the compression ratio of the second compressor, the refrigerant flow rate, and the amount of reliquefied CO
2. Consequently, in Cases 1, 2, and 3, when the nitrogen content varies from 5–15 mol%, the reliquefaction rate decreases from 69.4–74.0% to 10.6% (
Fig. 6). Accordingly, although the total power consumption of the compressors decreased, the SPC increased exponentially from 87.5, 84.3, and 80.9 kWh/t to 380.0, 372.0, and 330.17 kWh/t, respectively (
Fig. 6).
On the other hand, in Cases 4, 5, and 6, when the nitrogen content in the CO
2 BOG was varied from 5 mol% to 15 mol%, the SPC increased from 86.2, 82.9, and 79.8 kWh/t to 112.6, 109.3, and 96.2 kWh/t, respectively. On the other hand, the increase was relatively low, only approximately 20–30% (
Fig. 6). This is because, in Cases 4, 5, and 6, the application of two-stage vapor–liquid separation keeps the vapor fraction after expansion low, at approximately 0.11–0.22, even though the total power consumption of the compressors increases to meet the lower saturation temperatures required as the nitrogen content rises (
Fig. 5). This allows for an increased amount of reliquefied CO
2 compared to Cases 1, 2, and 3.
By applying two-stage vapor–liquid separation, lighter nitrogen gas can be separated from CO2 before expansion, allowing for Joule–Thomson expansion of CO2 with relatively higher purity. This makes it possible to increase the amount of reliquefied CO2 even at the limited temperature of −55 °C compared to Cases 1, 2, and 3. Consequently, a reliquefaction rate of over 63.0% was achieved in Cases 4 and 5, even when the nitrogen content was increased to 15 mol%. In Case 6, which uses mixed refrigerants, a reliquefaction rate of 74.2% can be achieved under the same conditions.
Reliquefaction is possible when using mixed refrigerants because of the more efficient heat exchange.
Fig. 7 shows the thermal composite curves of Cases 4 and 6 when the nitrogen content is 15 mol%. The temperature difference between the hot and cold composite curves is, on average, 18 °C with a maximum of 38.5 °C in Case 4, whereas in Case 6, it is, on average, 11 °C with a maximum of 32. 2 °C. This suggests that more efficient heat exchange is occurring in Case 6 using mixed refrigerants. This efficiency is achieved because mixed refrigerants composed of substances with different boiling points can utilize the latent heat of each refrigerant in the BOG reliquefaction process.
As a result, the proposed two-stage vapor–liquid separation process increased the amount of liquefied CO2 compared to the single vapor–liquid separation process. When the nitrogen content changed from 5 mol% to 15 mol%, the reliquefaction rate of Case 1, which uses ammonia refrigerant in a single vapor–liquid separation process, decreased sharply from 69.4% to 10.6%. In contrast, Case 6, which applies mixed refrigerants and a two-stage vapor–liquid separation process, achieved a reliquefaction rate of over 74.2% in all cases.
Maintaining the reliquefaction rate also led to significant variations in the energy consumption required for the liquefaction process. When the nitrogen content was 5 mol%, the SPC of Case 1 was 87.5 kWh/t, whereas the SPC was 79.8 kWh/t in Case 6, showing an 8.8% reduction in SPC. As the nitrogen content increased, the effect of the two-stage separation process became more pronounced. When the nitrogen content was 15 mol%, the SPC of Case 6 was 96.2 kWh/t, showing a 74.7% reduction compared to the SPC of 380.0 kWh/t in Case 1.
5. Conclusion
Optimization was performed to minimize the specific power consumption (SPC) for cases where the nitrogen content in CO2 BOG varies from 5.0 to 15.0 mol%, considering the use of ammonia as a refrigerant (Cases 1 and 2) and the use of mixed refrigerants (Case 3). In the reliquefaction processes applying single-stage vapor–liquid separation (Cases 1, 2, and 3), the amount of reliquefied CO2 decreased sharply as the impurity content increased, leading to a significant drop in efficiency. This is because higher amounts of nitrogen, which has a lower saturation temperature, shift the phase equilibrium, requiring lower temperatures for reliquefaction. On the other hand, a temperature limit of −55 °C was imposed to avoid the possibility of dry ice formation, restricting liquid formation.
Therefore, cases were also optimized and compared by applying a two-stage vapor–liquid separation process (Cases 4, 5, and 6) to address the issue of decreasing reliquefaction rates with increasing nitrogen content. In the reliquefaction processes with an additional vapor–liquid separator before expansion (Cases 4, 5, and 6), efficient reliquefaction was possible because lighter nitrogen was removed before the Joule–Thomson expansion compared to CO2. As a result, the processes with an added vapor–liquid separator (Cases 4, 5, and 6) could achieve an SPC reduction of 1.6–74.7% compared to Cases 1, 2, and 3.
Among these, Case 6 allowed for efficient heat exchange using mixed refrigerants, making it possible to expect an additional SPC reduction of 7.4–14.6% compared to Cases 4 and 5. Consequently, in the process of applying the vapor–liquid separator and mixed refrigerants (Case 6), an SPC reduction of 8.8–74.7% could be expected compared to the existing process design. Only Case 6 showed a reliquefaction rate of over 74.2% across all nitrogen content ranges from 5% to 15%.
When impurities, such as nitrogen, increased in the CO2 BOG, the reliquefaction rate decreased gradually, and the amount of non-condensable gas vented to the atmosphere increased. This leads to increased greenhouse gas emissions and causes economic losses due to cargo loss from the perspective of CO2 carriers. Therefore, future studies will be needed to design reliquefaction processes considering economic aspects during CO2 transportation.