2.1 System constraints
A preliminary analysis investigated the possible integration of the capture and storing system within the truck cabin, restricting the system weight and volume. The volume constraint is more stringent than the weight constraint because heavy-duty vehicles reach the volume limit before the weight limit [30], so we focus more on the volume.
As the upper boundary limit, we looked at the volume of the battery electric vehicles needed to cover the longest distance with one charge only. The battery unit’s volume depends on its energy size and rated power. The International Council of Clean Transportation correlated the required battery energy size to the kilometer-range capacity for one charge of the battery[30]. Their estimates are validated by Volvo’s last commercial electric heavy-duty truck, which covers 450 km with one charge and a battery energy capacity of 565 kWh (https://www.volvotrucks.us/trucks/vnr-electric/).
Argonne’s BatPac5.0 [31] predicts lithium-ion battery performance in 2030 and provides the overall battery system volume once the battery rated power and energy capacity are given. Their database predicts the need for seven battery units, each with 300 kW as rated power and 85 kWh of usable energy, to ensure Volvo’s features of 565 kWh for 450 km with one charge. Additionally, it estimates the volume of each battery in 0.35m3, and the overall battery units system in 2.5 m3. Then, we further lowered the upper limit of the volume of the onboard carbon capture to 1.5 m3, considering future developments of battery technology, and their decrease in size. The lower volume helps placing the mobile carbon capture and storage system behind the constrained truck cabin space.
Due to such volume limitations, CO2 cannot be stored at atmospheric pressure because of its low density. Instead, it may be compressed to a pressure that balances between storing volume and compression power. Indeed, storing CO2 in a cylinder at the gas phase saves energy but requires a larger volume, whereas storing it as a liquid in cylinders involves lower volume but excessive compression energy, increasing the extra power needed to sustain the process. We limited the extra power to lower than 10% of the engine power. A solution that meets space and energy limits is storing CO2 with an adsorbing material where adsorbed CO2 has a density similar to the liquefied CO2 but at lower pressure.
As discussed, we selected Al-soc-MOF-1 for CO2 storage, it stores six to eight times more CO2 than any cylinder in the gas phase at 30 to 45 bar, which is the proposed storing pressure range.[15] The adsorbed-phase density values for CO2 uptake and CO2 are listed in Table 1S in the Supplementary Information, and the MOF density is 340 kg/m3.[15]
2.2 Overall system description
Figure 2 illustrates the turbocharged diesel engine, and its post-treatment systems coupled with the two proposed mobile carbon capture and storage process designs. After combustion, the hot, exhausted gas flows through the post-treatment system and the heat exchanger, where it is cooled before the capture process. The heat exchange powers an ORC, which ensures an average of 8.1 kW [24] to sustain the capture and storage process. In the first design (Fig. 2a), we lay out the case in which 5% of the hot, exhausted gases provide heat to the sorbent for desorption without affecting the overall power generated by ORC. The second design (Fig. 2b) uses the hot gas from the compression stages as a heating medium to regenerate the material.
2.3 One-dimensional engine simulations
The CO2 capture process performance depends on the tailpipe stream properties, such as the mass flow rate and temperature that change according to the load or rotational speed variation, making the capture system highly influenced by the engine dynamics. Therefore, it is impossible to calculate capture performance without first pinpointing exhaust engine conditions.
During the engine operation, as more diesel is injected more air enters the cylinder due to higher air pressure to satisfy the higher power demand that higher loads require. Thus, the exhaust gas flow rate and temperature increase with the load (see Fig. 1Sb). However, the injected diesel and compressed air increases are not proportional to their stoichiometric ratio, and a higher load leads to more concentrated CO2 in the exhaust stream. Further, Fig. 1Sc illustrates this trend, where CO2, H2O, and O2 are functions of the engine load, whereas Fig. 1Sd displays engine emissions on a dry base. Rijpkema et al. [24] reported a similar mass flow exhaust rate within a 10% difference range. After calculating the exhaust molar composition, temperature, and flow rate, we can evaluate the capture process performance, considering the dry flow results to be input for the capture process simulation.
2.4 Isotherms and kinetic parameters regression modeling
Among all the available state-of-the-art adsorbents, KAUST-7 MOF is the chosen material for the CO2 capture process because of its outstanding properties [13, 14] and the fulfillment of most of the capture material characteristics. The remarkable CO2 capture property at high temperatures in presence of water vapor, coupled with low regeneration temperature and high stability makes it an ideal candidate for onboard CO2 capture. Figure 3a-b shows the adsorption isotherms for CO2 and N2 for KAUST-7, providing the first insight into the material thermodynamic adsorption properties. As evident from CO2 isotherm, KAUST-7 has steep CO2 uptake and saturates quickly at relatively low concentrations, indicating a high affinity for CO2 [32]. We used Langmuir–Freundlich model to fit the data. On the contrary, KAUST-7 only has little N2 adsorption around atmospheric pressure, indicating low N2 affinity [33]. We used the Langmuir isotherm model to easier manage the input data once we simulated the capture process. The parameter values were obtained from nonlinear multivariable regression results listed in Table 2S in the Supplementary Information.
As reported elsewhere for similar cases [34], the boundary layer between gas and sorbent does not add visible resistance to the mass transfer. We ensured that axial dispersion is not significant in the breakthrough experiments. In addition to following Mears’ criteria and guaranteeing a higher Peclet number than 100 for each breakthrough run, Fig. 3c demonstrates that, for the particle size in the range of 300 to 710 µm, the breakthrough shape does not change for different spatial velocities. Thus, axial dispersion does not dominate the other mass transport resistances. The slope of the CO2 breakthrough curve stays constant when the flow increases without affecting the mass transfer resistance. Moreover, Fig. 3d presents a sharper breakthrough response and a faster CO2 transport from gas to the sorbent as particle size decreases, indicating that the macropore diffusion controls the rate of adsorption.
Adil [14] and Bhatt [13] reported high CO2/N2 selectivity and no N2 adsorption in breakthrough experiments. Their finding was confirmed both experimentally and computationally in this study. In fact, the N2 overall lumped mass transfer coefficient is almost equal to zero in all the breakthrough experiments. Table 1 lists the CO2 overall mass transfer coefficients at different experimental conditions. We used the value at dp < 150 µm for the simulations of the capturing process because of its lowest mass transfer resistance and closeness to the KAUST-7 intrinsic transport phenomena under the studied conditions.
The material was also tested in a hot and wet CO2-containing stream to mimic a realistic post-combustion scenario because, in an exhaust gas stream, the CO2 and H2O content is almost equal (Fig. 1Sc). The performance of physisorbed materials generally drops significantly when shifting from a dry to a humid environment [35, 36]. We performed a humid (33% RH) breakthrough experiment at 75°C to evaluate the effect of water vapor using a ternary gas mixture of 13.1% CO2, 12.9% H2O, and balance N2. Remarkably, the comparison of humid and dry experiments suggests no significant decrease in CO2 uptake of KAUST-7 due to competition from such a high amount of water vapor (Fig. 2S).
Table 1
CO2 overall mass transfer coefficients at the different particle sizes and superficial velocity.
Experiment
|
Particle diameter
|
Flow rate [SSCM]
|
Mass [mg]
|
Mass transfer coefficient [s− 1]
|
#1
|
300 µm < dparticle<710 µm
|
4.2
|
390
|
0.009
|
#2
|
300 µm < dparticle<710 µm
|
6.2
|
570
|
0.009
|
#3
|
300 µm < dparticle<710 µm
|
7.8
|
720
|
0.009
|
#4
|
150 µm < dparticle<300 µm
|
5.4
|
500
|
0.012
|
#5
|
dparticle<150 µm
|
4.2
|
390
|
0.014
|
2.5 Vacuum swing adsorption modeling
Once the isotherms and kinetic parameters are determined, the next step is to model the full adsorption system. First, we evaluated conditions that can maximize CO2 recovery and purity and minimize the power requirement during the adsorption process at a constant temperature by tuning the desorption pressure, light-gas recycling, and adsorption/desorption time. We investigated a case study that considered two beds and four steps that operate with a simplified exhaust binary mixture of CO2 and N2 at atmospheric pressure with yCO2 = 0.10, exhaust gas flow rate = 100 g/s, and 50 kg of KAUST-7 per bed column with aspect ratio = 5 and dp = 1 cm. Details on the system operation can be found in the method section and the supporting information. A flowsheet of the VSA system is depicted in Fig. 14S.
If the process is constrained to have a constant temperature, then only changes in pressure ensure process continuity and bed regeneration. KAUST-7 isotherms show that the pressure affects the CO2 working capacity only at temperatures higher than 348 K, therefore this was the minimum process temperature to ensure material regeneration. At this temperature, the sorbent has a working capacity when the pressure swings between 0.1 bar and 0.01 bar. Pressure levels lower than 0.01 bar are not considered to avoid excessive power requirements to create the vacuum. Adsorption and desorption steps are 100 s each.
The CO2 purity and recovery present an opposite trend to the light-gas recycle flow (see Fig. 3Sa‑b). The light-gas recycle flow (described in the Supplementary Material, Adsorption process description: VTSA) is depleted in CO2 because it derives from the adsorbing bed. Thus, the more the light-gas recycle flow enters the regenerating bed, the more diluted the recovered stream becomes, and thus CO2 purity drops. However, the more diluted the recovered stream, the lower the CO2 partial pressure, the higher the sorbent working capacity, and the higher the bed regeneration and CO2 recovery.
Desorption pressure affects CO2 purity and recovery equally. A higher vacuum results in higher recovery and purity at the same light-gas recycle flow due to a higher material regeneration and working capacity. As expected, achieving Pdes = 0.01 bar requires more power than Pdes = 0.1 bar; however, normalizing the absolute power requirements over the captured CO2 per cycle leads to similar or lower values for cases of Pdes = 0.01 bar over Pdes = 0.1 bar, as illustrated in Fig. 3Sc. The normalized power requirement similarity or higher performance happens because the Pdes = 0.01 bar case captures CO2 more efficiently. Indeed, it has double the CO2 recovery of Pdes = 0.1 bar and higher purity, making Pdes = 0.01 bar the chosen desorption pressure for the system due to its similar or lower energy performance and higher CO2 recovery and purity compared with the Pdes = 0.1 bar case.
At 348 K, the system exhibits slightly lower power intensity results. Thus, 348 K is the chosen temperature for the adsorption process and the exhaust gas inlet temperature in the packed bed.
Adsorption/desorption time optimization is achieved through a sensitivity analysis, where the objective function is CO2 recovery and purity. Table 3S in the Supplementary Information lists the sensitivity analysis results. The tads/des change slightly affects CO2 recovery and purity performance, we selected 30 s as tads/des because of the highest CO2 recovery.
The optimized isothermal VSA cycle uses 50 kg of KAUST-7 per packed bed, where Pdes = 0.01 bar and tads/des = 30 s. In addition, 348 K is the process temperature, and the light-gas recycle flow rate is 0.5%v of the feed. The optimized VSA cycle achieves a 90% CO2 purity and 53% recovery.
2.6 Vacuum Temperature Swing adsorption and Storage modeling
Aiming to improve the results of our VSA model, we next evaluated the Vacuum Temperature Swing Adsorption (VTSA) option, starting from the optimized parameters obtained from the previous isothermal VSA. The energy balance for the gas and sorbent is here included, as explained in Section 4.3.2.
Figure 4a displays how CO2 purity and recovery change with engine load for the base scenario where 50 kg of KAUST-7 was used for each bed, the desorption pressure is 0.01 bar, and 5% of the exhaust gas was used to heat the column during the desorption step. The engine load does not affect the CO2 purity, which remains almost constant. CO2 purity primarily depends on the ratio between the light-gas recycle and exhaust flow rate, which is constant for all engine conditions because it is a post-capture process parameter and not engine-related.
However, the higher the engine load, the lower the CO2 recovery. A higher load has a higher temperature and, thus, higher heat during desorption. This relationship increases the bed regeneration and sorbent working capacity, but a higher load has also a higher gas flow rate at the feed. Therefore, the adsorption column processes more flow, the gas-solid contact time is reduced, and the capture performance drops because the higher heat flow cannot outweigh the negative effect of the higher mass flow and lower gas-solid contact time.
The adsorption cycle requires different power inputs depending on engine load and storing pressure (see Fig. 4b). The power from the ORC is averaged for all loads and is taken as a constant at 8.1 kW [24]. The power for vacuum and compression increases according to the mass flow directed to storage. Despite the CO2 recovery decreases at a higher load, the overall stored CO2 flow rate increases (see Fig. 4S in the Supplementary Material). Therefore, a higher load results in higher power requirements. For the base-case scenario and its daily usage, the average CO2 purity is 96.5%, recovery is 59.3%, and the extra required power is 9.8% if CO2 is stored at 30 bar or 10.8% at 45 bar.
As stated in Section 2.1, Al-soc-MOF-1 is the chosen sorbent material, and the higher the storage pressure, the higher its CO2 capacity and the lower its mass and volume. The system captures 79.50 kg/h of CO2 for the base case, but the amount of Al-soc-MOF-1 depends not only on the captured CO2 but also on the storing time and pressure (see the matrix in Table 4S in the supplementary material).
The overall mass dependence on storage pressure is negligible in the capture and storing system. Indeed, the mass of the whole system is 2.7% of the truck payload at Pstorage = 30 bar and 2.6% at Pstorage = 45 bar, at 4 h of storing capacity. The storing time is the parameter that most affects the parasitic mass of the system: at 8 h, the mobile carbon capture and storage parasitic weight is 4.6% of the truck payload at Pstorage = 30 bar and 4.5% at Pstorage = 45 bar, twice the 4 h case.
Instead, system volume depends on the overall capture and storage pressure. The higher the storing pressure, the lower the volume, but the higher the power consumption. The optimization process is a trade-off between the needed storage volume and the power requirements. As already happens for the weight, moving from 4 to 8 h almost doubles the capture and storing volume requirement. Table 5S lists the volume requirements for the investigated base case study.
Fourteen additional cases were evaluated where boundaries were set on the maximum extra power requirements, the maximum volume of the mobile carbon capture and storage system, and, with CO2 recovery. Table 2 presents all cases.
Table 2
VTSA differs by desorption pressure, storage pressure, heat source type for desorption and adsorbent mass. The 14 cases are used to obtain the optimal conditions for the capture and storage system.
KAUST-7 mass [kg/bed]
|
Pdes [bar]
|
Pstorage [bar]
|
Heat source type for desorption
|
45
|
0.01
|
30
|
5% organic Rankine cycle
|
45
|
0.01
|
45
|
5% organic Rankine cycle
|
50
|
0.01
|
30
|
5% organic Rankine cycle
|
50
|
0.01
|
45
|
5% organic Rankine cycle
|
75
|
0.01
|
30
|
5% organic Rankine cycle
|
75
|
0.01
|
45
|
5% organic Rankine cycle
|
50
|
0.02
|
30
|
5% organic Rankine cycle
|
50
|
0.02
|
45
|
5% organic Rankine cycle
|
55
|
0.02
|
30
|
5% organic Rankine cycle
|
55
|
0.02
|
45
|
5% organic Rankine cycle
|
45
|
0.01
|
30
|
Hot gas from compression stages
|
45
|
0.01
|
45
|
Hot gas from compression stages
|
55
|
0.02
|
30
|
Hot gas from compression stages
|
55
|
0.02
|
45
|
Hot gas from compression stages
|
All 14 cases have a CO2 purity higher than 96%. In addition, Figs. 5S to 11S present a detailed report of the performance.
Figure 4a, CO 2 purity (black squares) and CO2 recovery (red circles) vs. engine load. b, Total power requirements vs. engine load at a fixed engine rotational speed of 1200 rpm; Pstorage = 30 bar in black squares and Pstorage = 45 bar in red circles. c, CO2 recovery vs. extra power. d, Stored CO2 volume after 8 h vs. extra power. Circles indicate Pdes = 0.01 bar, and squares indicate Pdes = 0.02 bar. Smaller circles or squares have 5% of tailpipe gas as a heat source for desorption; bigger squares or circles denote hot gas during desorption from compression stages. Color-filled symbols have Pstorage = 30 bar; empty circles indicate Pstorage = 45 bar. Blue values mark 45 kgKAUST-7 per bed, red marks 50 kgKAUST-7, green marks 55 kgKAUST-7, and gray marks 75 kgKAUST‑7. All points are calculated and averaged as explained in 4.1.1; see supplementary information for each case study.
Figure 4c indicates that a higher KAUST-7 mass per bed results in more CO2 recovered due to the higher contact time of the gas flow with the sorbent and has higher extra power because more gas is vacuumed and compressed (see 75 kgKAUST−7 case in gray). A higher Pdes lowers the sorbent regeneration and CO2 recovery but reduces the power requirement. Indeed, vacuum or compression power depends on the pressure ratio between the inlet and outlet pressure in addition to the mass flow (see Eq. 7).
Shifting Pdes from 0.01 to 0.02 bar decreases the extra power requirement by half and decreases the captured CO2 due to the lower sorbent regeneration. However, we outweigh the lower recovery by increasing the KAUST-7 mass per bed. For example, the base-case scenario (small red circles) has 59% CO2 recovery, whereas the same adsorbent material mass with a slightly higher Pdes = 0.02 bar (small red squares) has 46% CO2 recovery. The CO2 recovery rises to 50% by adding 5 kg per bed (small green squares), partially counterbalancing the effect of working at the highest desorption pressure. Therefore, we can obtain consistent energy savings at almost the same CO2 recovery by adding just a few kilograms of adsorbent material per bed.
Finally, the system using the hot gas from the compression stages has higher CO2 recovery than the system using 5% of tailpipe gas as the heat source at the same KAUST-7 mass during desorption (see big and small green squares and big and small blue circles in Fig. 4c). The reason is the higher heat derived from the compression stages; thus, this design results in deeper bed regeneration and higher CO2 recovery.
Storing the enriched CO2 gas at 30 or 45 bar does not affect CO2 recovery when 5% of the tailpipe gas is the heat source. The capture parameters remain constant regardless of the storing pressure (see small color-filled and empty circles or squares in Fig. 4c).
If the gas from the compression stages is the heat source during the bed regeneration step, the heat flow depends on the gas flow from the capturing bed, and its temperature on the final pressure after each compression stage (see Fig. 15S). Consequently, the higher the pressure ratio for each stage, the higher the outlet temperature after the compression stage, higher heat flow, and sorbent regeneration. Thus, more and more CO2 is released during desorption and captured, making the captured CO2 flow a function of the compression pressure ratio and vice versa.
Figure 4d indicates that CO2 recovery and stored CO2 volume follow the same trend because the higher the recovery, the more CO2 is stored. Compressing the enriched CO2 gas by up to 45 bar leads to an average volume reduction of 200–300 L compared to the 30 bar case with only a slight extra power increase (less than 1%). Thus, we selected 45 bar as the storing pressure. Among the 14 cases, only three were eligible for on-truck implementation considering the volume, extra power, and recovery limitations:
1. 45 kgKAUST-7 per bed at Pdes = 0.01 bar and Pstorage = 45 bar using 5% of the exhaust tailpipe gas as a heat source during bed desorption.
2. 55 kgKAUST-7 per bed at Pdes = 0.02 bar and Pstorage = 45 bar using 5% of the exhaust tailpipe gas as a heat source during bed desorption.
3. 55 kgKAUST-7 per bed at Pdes = 0.02 bar and Pstorage = 45 bar using the gas flow from the compression stages as the heat source during bed desorption.