Table of Contents:
Chimney stack exit penalty
When flue gas flows out of a chimney stack into the atmosphere it mixes with the surrounding air, disperses and blows away from the stack. This process—mixing is indeed a process—is hard to observe, except when motion of the flue gas plume is marked by white condensation of water vapour as it cools down. The mixing process, though almost invisible, nevertheless causes a large increase in entropy, and thus, a loss in Gibbs free energy, which corresponds to a loss of potential work. In theory, a thermodynamic process which makes use of the difference in CO, concentration between flue gas and atmosphere to produce useful work could be created. Allowing the flue gas to mix irreversibly with the atmosphere destroys this possibility. As a result, once the flue gas has left the chimney stack and mixed with the atmosphere, it takes much more work to capture the carbon dioxide than it would have done if it had been captured from the flue gas where it is more concentrated.
This chimney stack exit penalty is quantified in Figure 5, for two flue gas concentrations representing power plants firing natural gas (y = 0.04) and coal (y = 0.12). Capturing carbon dioxide from air takes about twice as much reversible work as capturing it from a natural gas power plant flue gas. Compared with FGC at a coal-fired power plant, DAC requires three times as much reversible work. As we show in the Appendix, this energy penalty translates into a significant cost penalty-DAC costs around 72-88 USS/'tCO, more than FGC.
Choice of solvent and cost of DAC
The IPCC’s 2005 report on CCS excluded Direct Air Capture from consideration because the CO, concentration in ambient ah was “...a factor of 100 or more lower than in flue gas. Capturing CO, from air by the growth of biomass and its use in industrial plants with CO, capture is more cost-effective based on foreseeable technologies...” (Metz et ah, 2005). However, as we have seen, the energy required for separation varies inversely with the logarithm of the concentration, so that the reversible work cost of air capture (Figure 5) is only 2-3 times that of FGC. This still gives a great advantage to FGC, which also needs only to treat a much smaller volume of gas to catch a tonne of carbon dioxide. Ah is a somewhat easier material to handle than flue gas, and the ability to site DAC plants almost anywhere, such as at locations where energy and/or storage are cheap, might outweigh some of the advantage that FGC would otherwise enjoy.
This chapter has considered the energy requirements of DAC in detail, as this process might become a huge industry later this century-—it is based on known technology that could be applied if other more novel techniques cannot be successfully developed. However, there has been some confusion and disagr eement about DAC in the literature, of which the very wide spread of cited costs is symptomatic. These vary from around 60 to 1000 USS per torme of CO,, though it does rather depend on how the necessary energy is supplied (Fuss et al., 2018). In then review of a range of NETs, Fuss et al. (2018) state “...a significant amount of thermal energy is often required for DAC due to the requirement of strong binding of the capmred material because of the extreme dilution of atmospheric CO,.” This is a misunderstanding. Strong binding of the sorbent with CO, would be required if it were necessaiy to reduce the concentration of CO, in the treated air to a low level—this would need a low equilibrium back pressure from the sorbent (low pK^). This is not required in DAC. Atmospheric carbon dioxide currently has a concentration of about 400 ppm, and this is not “extreme dilution” as far as gas treating is concerned. In processing natural gas with counter-current contact, amines of intermediate strength are able to remove H,S down to pipeline specifications (< 4 ppm). H,S is a weaker acid than CO„ but it reacts very rapidly in alkaline solutions.
Figure 5. Chimney stack exit penalty. Penalties shown are the additional reversible work needed for DAC rather than FGC,
for natural gas firing (yc= 0.04) and coal firing (yc= 0.12), calculated for a = 1.0,/?= 0 (100% recovery of pure CO,). T0 =
293 К and mole fraction in air is 0.0004.
It is true that the rate of absorption into solvents will be accelerated by reaction in the liquid phase, and CO, undergoes a second order reaction with the hydroxyl ion (Astarita et al., 1981), which is, thus, faster at high pH. This and other reactions in the liquid will enhance the absorption rate, enabling reduction in the size of the absorber, which is important in DAC. So, what is required is a fast reaction between CO, and sorbent, with a base of intermediate strength that can be regenerated with less energy than one with strong binding of CO,. “Fast reaction” means that the characteristic time of the reaction is small compared with the characteristic time of mass transfer in the liquid phase. Speed of reaction should not be confused with base strength. Piperazine, for example, reacts quickly with CO,, and is an effective promoter of absorption in potassium carbonate (section 2.2.1) and slower amines, such as methyldiethanolamine. It is a weak base. The development of DAC seems to have been unduly influenced by the consideration of strong (i.e., low pKb) bases, which are costly to regenerate.
Although aqueous alkanolamines as a class are receiving a great deal of attention as possible solvents for carbon dioxide capture, the sourcing of these chemicals, potentially at a very large scale, requires consideration. They are commonly made by reacting ethylene oxide with ammonia or amines. Current global production of ethylene oxide (EO) is around 29 Mt/y, and scaling this up quickly by the order of magnitude that might be needed would be a challenge. EO is a flammable, explosive and toxic gas (normal boiling point 10.7 °C) and its manufacture and handling require great care. It is important that sorbents considered for large-scale application in carbon capture har e a feasible supply chain. In some cases, this might require novel chemistry, and new manufacturing routes.
In the Appendix, we estimate the cost of capnuing carbon dioxide from the air with a solvent, extrapolating cost and technical data for current flue gas capture processes. The estimated cost in 2013 for this illustrative case is USS 155 per tonne of net carbon dioxide captured, including compression but excluding transport and storage. For every tonne of CO, captured fr om the atmosphere. 1.46 tonnes must be stored, assuming the energy required is supplied by burning natural gas; this ratio is similar to that for FGC at coal-fired power plants. These estimates are approximate and not based on any particular solvent, though they assume a process with characteristics like those using alkanolamine solvents, but with a very low pressure-drop contactor. For comparison, the levelised cost of carbon dioxide cap trued by the CE DAC process, when fired solely by natural gas as we have assumed, is reported to be in the range 168-232 US$/tCO„ projected to fall to 126-170 USS/tCO, as plant and process improvements result from operational experience (Keith et al., 2018, Scenarios A and B).
Comparing FGC and DAC, the cost of net carbon dioxide avoided (~ 67 USS/tCO,) in a coal-fired power plant is less than half the cost of DAC, and the cost in a gas-fired power plant (~ 83 USS/tCO,) is only slightly more than half the cost of DAC. Including 1-19 USS/tCO, for transport and storage (Rubin et al., 2015), a carbon price can be calculated that would motivate the deployment of DAC, amounting to 156-174 USS/tCO,, roughly double that required for Flue Gas Capture.
Costs discussed in this review should, as always, be treated with caution. Like all predictions, they incorporate many assumptions and uncertainties.
DAC: Meeting the energy demand
To capture the target amount of 810 Gt CO, by 2100 (Figure 1) would require some 387 x 10|: MJ of reversible work (at a rate 21 MJ/kmol CO,, see Figure 4) yielding the purity required for pumping to underground storage. As we have seen, capturing this CO, by solvent-based DAC alone would require an amount of heat approximately 14 times as great: 5418 x 1012 MJ. Spread over a timescale of say 60 years, this is a heat requirement of ~ 90 x Ю12 MJ/year. In comparison, the global consumption of primary energy of all types in 2017 was 566 x Ю12 MJ (BP, 2018). Clearly, the use of DAC on a large scale will consume a significant fraction of the world’s energy.
There is one potential mitigating factor, in the possibility that the direct air capture process could be driven by what is now classed as “waste heat” that is available from various industrial operations. For example, in this chapter, we have taken the average power plant thermal efficiency to be 40%. The remaining 60% of the energy input to electricity production is “waste heat” and is dispersed to the environment in either water- or air-cooling. In 2017, fossil fuel was used to produce 58.5 x 101: MJ(e) globally, suggesting that waste heat totalling around 88 x 10i: MJ(th) was also produced. Some of this heat will be hot enough to drive a sorbent regeneration at perhaps 90-100 °C, and new sorbents might be developed which are more easily regenerated at a lower temperature; this would benefit both DAC and FGC, and the cost reductions could be substantial.
The possibility of utilising waste heat is supported by calculations of Rattner and Garimella (2011) who analysed data for 2007 published by the U.S. Energy Information Agency. They concluded that waste heat from U.S. power plants amounted to 60% of input energy. Total waste heat available above 30 °C from this source in 2007 amounted to 24.2 x Ю12 MJ, with a waste heat weighted average temperature of 88 °C. This would provide usable heat for a significant quantity of DAC sorbent regeneration, but not for processes involving calcination, which needs much higher temperatures. In Rattner and Garimella’s analysis only ~ 8% of the waste heat was furnished by exhausts at temperatures gr eater than 450 °C.
Hauak et al. (2017) describe the Origen Power process, in which a solid oxide fuel cell converts fuel to electricity, and the high temperature waste heat is used to calcine a carbonate solid. The calcination produces high purity carbon dioxide which can be pumped to storage, and the calcined solid can be used to draw down more carbon dioxide. Such processes for power generation become viable when carbon capture is mandated or incentivised.