https://doi.org/10.1016/j.seppur.2020.118193
“In a superficial analysis, the sensible heat duties with water-lean solvents should be lower than those with aqueous formulations simply because the heat capacity (CP) of organic diluents is usually smaller than that of water. For a given cyclic capacity Δq and a given difference in temperatures ΔT, this advantage is clear. What is obscure is if Δq and ΔT are the same in both types of solvents.
The cyclic capacity Δq is determined by the rich loading obtained in the absorber and the lean loading recovered in the desorber. As discussed in Section 4.1, because of the equilibrium shift, the rich loading in water-lean solvents will conventionally be lower than that of aqueous solvents. As for the lean loading, the previous Section 8.2 has shown that whether the lean loading is smaller or higher is somewhat dependent on the volatility of the diluent that is chosen, particularly for CCS applications. Therefore, it is unclear how different Δq can be in a water-lean solvent, making it a safe choice to suppose it is the same.
Meanwhile, ΔT shall be discussed with care. This temperature difference is calculated between the desorber feed and the reboiler, and between the absorber product and the desorber feed there is typically a cross-heat exchanger designed for warming up the rich solvent. Yuan and Rochelle [166] and Liu et al. [200] have pointed out that the amount of heat which can be recovered by an equipment will be directly correlated to the viscosity η, the heat capacity CP and the thermal conductivity λ of the solvent, and that this fact can be possibly detrimental to water-lean solvents.
Suppose we consider the solvents mentioned in the previous section. The physical properties of water, ethylene glycol and 2-methoxyethanol were obtained at 40 °C in Yaws [160], [201], Svoboda et al. [202] and Islam et al. [203]. These, together with the Reynolds, Prandtl and Nusselt numbers of each diluent calculated in a theoretical cross-heat exchanger, are shown on Table 8. For the calculation of the Reynolds number, the liquid velocity u = 0.40 m∙s−1 and characteristic length D = 0.004 m were adopted following the suggestion of Lin and Rochelle [204] for liquid flow in a plate-and-frame type equipment. For the calculation of the Nusselt number, the correlation suggested by Okada et al. [205] was employed as shown below.Nu=0.157·Re0.66·Pr0.4
Table 8. Physical properties of water, ethylene glycol and 2-methoxyethanol at 40 °C. Values obtained from Yaws [160], [201], Svoboda et al. [202] and Islam et al. [203]. Dimensionless numbers calculated under assumptions of u = 0.40 m∙s−1 and D = 0.004 m.
| Parameter | Water | MEG | 2ME |
|---|---|---|---|
| ρ/g∙cm−3 | 1.000 | 1.100 | 0.943 |
| η/mPa∙s | 0.646 | 9.906 | 0.942 |
| CP/kJ∙K−1∙kg−1 | 4.241 | 2.486 | 2.363 |
| λ/W∙m−1∙K−1 | 0.624 | 0.132 | 0.260 |
| Re (ρ∙D∙u∙η−1) | 2480 | 180 | 1600 |
| Pr (CP∙η∙λ−1) | 4.4 | 186 | 8.6 |
| Nu (h∙D∙λ−1) | 49.3 | 38.8 | 48.4 |
| h/W∙m−2∙K−1 | 7700 | 1280 | 3145 |
With the three dimensionless numbers Re, Pr and Nu, the convective heat transfer coefficient for each diluent can be calculated. These are also shown on Table 8.
Table 8 does not take into account the effect of CO2 loading in shifting solvent properties such as viscosity. As we have discussed in Section 6.2, the viscosities of water-lean mixtures are not only typically higher than those of aqueous ones, they also increase more steeply with CO2 loading. Guo et al. [80] have reported the viscosity of loaded 2ME with 30 %wt. MEA as being 10.97 mPa∙s (40 °C, α = 0.357 mol CO2∙mol MEA−1). For comparison, a water-lean solvent containing MEG has η ≈ 31 mPa∙s, while aqueous 30 %wt. MEA has η ≈ 2 mPa∙s at similar loading and temperature [59], [164]. This implies convective heat transfer coefficients far lower for water-lean solvents than for aqueous amines. And still, even if one ignores the effects of loading on viscosities, Table 8 already shows how this discrepancy arises.
The global heat transfer coefficient of a cross-heat exchanger is directly proportional to the average convective heat transfer coefficient of the solvent inside it [204]. Therefore, as seen on Table 8, thermal recovery with aqueous solvents will be about two times as effective as in nonaqueous ones, sevenfold so if one considers the particular case of ethylene glycol, and higher than that once CO2 loadings are taken into account. The reasons can be pinpointed by analyzing each row of Table 8. The high viscosity of organic diluents, their low thermal conductivity and even the low heat capacity will have an impact on the performance of the equipment, meaning that one of two options will apply:
- a.
-
The cross-heat exchanger in plants operating with water-lean solvents will be typically bigger, and therefore more expensive than the ones operating with aqueous solvents.
- b.
-
The amount of heat that can be recovered from the lean solvent leaving the reboiler will be less for water-lean solvents than for aqueous ones, meaning that the ΔT associated with sensible heat duties in the reboiler will be typically higher.
How much bigger the ΔT is will depend on other parameters of the cross-heat exchanger. What becomes clear, however, is that there are more factors at play than the diluent CP to determine if sensible heat duties are higher or lower in water-lean solvents. Still, this is typically not really that relevant since sensible heat duties have the least impact on overall reboiler heat duties [206].
“