Per MMscfd Freed, CO2 Room Costs the Amine Reboiler 2.3 Times What Nitrogen Room Does

On the synthetic three-field case from our inlet-limit piece, a 10% looser CO2 limit frees 30.2 MMscfd of feed and adds 9.5 MW of amine reboiler duty, about 0.31 MW for each MMscfd. A 10% looser nitrogen limit frees 52.3 MMscfd for 7.1 MW, about 0.14 MW each. Per MMscfd freed, CO2 room costs the reboiler 2.3 times what nitrogen room costs. That piece priced each inlet limit in feed gas and found nitrogen room worth twice CO2 room. This one prices the same limits in the treating units' energy, and the two answers point the same way for a reason the gas price alone does not show: room on the CO2 limit is also permission for the gas already flowing to carry more CO2.

Duty is an area

The amine unit's reboiler regenerates the solvent, and the heat it needs scales with the CO2 the solvent picks up. We take the CO2 removed as the feed times its CO2 content, down to a treated-gas slip of 50 ppmv, and the duty as that tonnage times the solvent's heat of reaction times a reboiler multiple:

$$m_{CO_2}\,[\mathrm{t/d}] = q\,[\mathrm{MMscfd}] \times \frac{10^6 \times 0.0283168}{0.023645} \times \frac{c_{CO_2} - c_{slip}}{100} \times \frac{44.01}{10^6}, \qquad Q\,[\mathrm{MW}] = \frac{m_{CO_2} \times \Delta H_r \times k}{86.4}$$

FormulaLaTeX

Copy LaTeX

The first expression turns standard cubic feet into moles on a 15 C basis and then into tonnes of CO2. The second turns tonnes into megawatts. Because duty is proportional to feed times CO2 content, on a plane with feed across and CO2 up a schedule's duty is the area of its rectangle. The gas supply engineer reads the width and the utilities engineer reads the area. They are arguing about the same rectangle.

The heats of reaction are Polasek and Bullin's: 600 Btu/lb of CO2 for MDEA, 653 for DEA and 825 for MEA, or 1.40, 1.52 and 1.92 GJ per tonne [1]. The heat of reaction leaves out the sensible heat of the rich solvent and the steam that strips it, so the bench multiplies it by k = 2.0, which is our assumption. We found no open page that gives a full reboiler duty per tonne for a natural-gas MDEA unit. For a check on the order of magnitude, k = 2.0 puts MEA at 3.84 GJ/t, a little above the 3.6 to 3.8 GJ/t a pilot plant measured for 30 wt% MEA on steam-reformer flue gas at 90% capture [2]. That is a different solvent on a different gas, so it bounds the guess rather than calibrating it. With MDEA, k = 2.0 gives 2.79 GJ/t.

The plan needs six times the duty of the optimum

Schedule	Feed, MMscfd	CO2, mol%	CO2 removed, t/d	Duty, MW, MDEA at k = 2.0	Boiler fuel, % of feed
Plan	1,035	16.83	9,180	297	2.73
Plan without well A-03	955	15.13	7,612	246	2.46
Largest feasible feed	363	8.00	1,530	49	1.30
Same, CO2 limit at 6 mol%	280	6.00	885	29	0.97

The fuel column assumes a fired boiler at 85% efficiency burning gas at 1,010 Btu/scf, so it is an upper bound for a plant that raises reboiler heat from turbine exhaust. Switch the plan to MEA and the duty rises by the ratio of the heats of reaction, to 408 MW and 3.76% of feed.

The optimum takes 2.85 times less feed than the plan and needs 6.0 times less duty, because the linear programme that found it also chose the cleanest gas: it runs the sour sandstone wells and the carbonate's lowest-CO2 well and leaves the rest shut. At 1,035 MMscfd, each mol% of CO2 in the feed is 17.6 MW of reboiler duty. Shutting in A-03, the dirtiest carbonate well, saves 51 MW on the plan and nothing on the optimum, which had already left it at zero.

What more room on a limit costs

The inlet-limit piece found that at the optimum, CO2, nitrogen and mercury all bind, and that relaxing nitrogen returns more feed than relaxing CO2. Re-solve the same programme with each limit 10% looser and price the new optimum in duty:

10% more room on	Gas freed, MMscfd	CO2 removed added, t/d	Duty added, MW	MW per MMscfd freed	Strips like gas at, mol% CO2
CO2, 8 to 8.8 mol%	30.2	293	9.5	0.31	18.4
Nitrogen, 2 to 2.2 mol%	52.3	220	7.1	0.14	8.0

The last column is the CO2 content the freed gas would need to carry to add that much duty on its own. Nitrogen room keeps CO2 pinned at 8 mol%, so the gas it frees costs exactly 8 mol% gas. CO2 room frees less gas and also lets the 363 MMscfd already running rise from 8 to 8.8 mol%, and that second term is about half of the added duty. Each MMscfd it frees strips like 18.4 mol% gas, dirtier than the plan's own blend at 16.8. Per MMscfd, the gas a looser CO2 limit buys costs the reboiler 0.31 MW, more than the 0.29 MW the plan averages.

The ratio depends on the size of the step. It is 1.90 for a 1% relaxation, the limit of the shadow prices themselves, 2.30 at 10% and 2.64 at 50%, because each further step on CO2 lets in dirtier carbonate gas while nitrogen room keeps bringing in gas at 8 mol%. It does not depend on the solvent or the multiple: both rooms are priced by the same tonnes, so the energy per tonne cancels.

loosen the CO2 or N2 limit by 10%reboiler multiple 2.0 (assumed; heat of reaction x this)

draw the duty added by:CO2 roomN2 roomsolvent:MDEADEAMEAzoom to the optimum

the plan297 MW9,180 t/d of CO2; 2.73% of feed as boiler fuel

the optimum49.4 MW2.85x less feed, 6.0x less duty

multiple 1.5 to 3, the plan222 to 445 MWnow 2.0: 2.79 GJ per tonne of CO2

one mol% of CO2 at 1,035 MMscfd17.6 MWMDEA, multiple 2.0

Now. 10% more CO2 room adds 9.5 MW for 30.2 MMscfd, 0.31 MW each; 10% more N2 room adds 7.1 MW for 52.3, 0.14 MW each. Ratio 2.30.

The exhibit opens on the 10% case with the CO2 room drawn. The amber area is the duty that room adds, and about half of it is the thin strip along the top of the optimum's rectangle, the existing feed carrying 0.8 mol% more. Switch to nitrogen room and the amber area becomes a column of new gas only. The relief slider moves both corners along their paths. The multiple and solvent move every MW figure and the curves of equal duty, and leave the ratio in the corner exactly where it was.

The multiple outweighs the schedule

The multiple is the least certain number in the calculation. At 1.5 the plan needs 222 MW; at 3.0 it needs 445. That spread of 223 MW is four times what shutting in A-03 saves, and larger than any change to the plan short of the optimiser's schedule. The tonnage is well defined once the schedule is; the energy per tonne is not, and on a real plant it is the first figure to measure. An MDEA unit treating high-CO2 gas can also take much of its bulk CO2 off by flashing the rich solvent at low pressure with little heat added [1], which would put the effective multiple below what a reboiler-only model implies.

A published comparison gives the scale of what is at stake. In De Guido and Pellegrini's simulations of methane and CO2 feeds from 10 to 70 mol% CO2, the MDEA acid gas unit's energy rose from 2.7% to 55.8% of the processed methane, in net equivalent methane [3]. C3MR liquefaction of the MDEA-treated gas took 4.7% at every CO2 content, because the energy for liquefaction and the methane left to liquefy fall together, so the acid gas unit used less energy than liquefaction at low CO2 contents and more at high ones. Their acid gas figure includes CO2 compression, which ours does not.

Four checks before the cold box

The treating units after the amine unit carry their own limits, and three of them scale with the schedule.

Freeze-out. Shen and co-workers measured CO2 solubility in saturated liquid methane at nine temperatures from 112 to 170 K; at 112 K, about minus 161 C, it is 213 ppm [4]. A fit of ln x against 1/T through the nine points stays within 6% of each and gives 212 ppm at minus 161 C, so the common 50 ppmv treated-gas rule carries a margin of 4.2. The most CO2 that keeps a margin of 2 is 106 ppmv. Ten degrees warmer, at minus 150 C, the fit gives 769 ppm. Nitrogen and ethane in the liquid each raise the solubility slightly in the same data, so pure methane is the cautious case. In this model a 50 ppmv slip against 100 ppmv changes the duty by less than 0.1 MW, because duty follows the bulk tonnage; reaching a lower slip asks more of the absorber and of lean solvent quality, which the model does not see.

Mercury. The plan carries 27.6 µg/Nm3, 280 kg of mercury a year at 1,035 MMscfd, with normal cubic metres at 0 C. A sulphur-impregnated carbon can hold up to 30% of its weight in mercury and bring treated gas under 0.01 µg/Nm3, by its maker's bulletin [5]. Loaded to half that capacity, the plan uses 1.9 t of carbon a year. The optimum carries 178 kg a year, 1.2 t of carbon: its feed is 2.85 times smaller, but its mercury is only 1.6 times smaller, because it runs the sour wells that carry the most mercury and sits at the 50 µg/Nm3 limit.

Water. Saturated at 30 C and 6 MPa, by the Bukacek correlation [6] with a vapour pressure of 4.247 kPa [7], the feed carries 44.6 lb of water per MMscf, 940 ppmv and 33% more than the ideal-gas term alone. Drying the plan removes 21 t of water a day; drying the optimum removes 7.3.

Heating value. Nitrogen carries no heating value, so it lowers the Wobbe number of the gas a terminal sends out. In Great Britain the Gas Safety (Management) Regulations set the Wobbe number between 46.50 and 51.41 MJ/m3; the lower limit was cut from 47.20 from 6 April 2025 [8]. The draft of this piece computed the treated feed's Wobbe number on an assumed hydrocarbon split, and the answer moved with the split by more than the gap between the field cases, so we leave the number out. A cargo's Wobbe number is also set after end flash and heavy-ends removal, not at the inlet.

Key takeaways

1. 1Amine reboiler duty is proportional to feed times CO2 removed, so on a feed-by-CO2 plane a schedule's duty is the area of its rectangle.
2. 2On the synthetic case the 1,035 MMscfd plan needs 297 MW with MDEA at an assumed multiple of 2.0, and the 363 MMscfd optimum needs 49 MW: 6.0 times less duty for 2.85 times less feed.
3. 3A 10% looser CO2 limit frees 30.2 MMscfd for 9.5 MW; the gas already running may then carry 8.8 mol%, so each MMscfd freed strips like 18.4 mol% gas, dirtier than the plan's 16.8.
4. 4A 10% looser nitrogen limit frees 52.3 MMscfd at exactly 8 mol% for 7.1 MW, so per MMscfd freed CO2 room costs the reboiler 2.3 times what nitrogen room costs.
5. 5That ratio runs from 1.90 for a 1% step to 2.64 for 50%, and no solvent or multiple changes it, because both rooms are priced in the same tonnes.
6. 6The assumed multiple moves the plan from 222 to 445 MW, more than any schedule change short of the optimum, so duty per tonne is the first number to measure on a real plant.
7. 7CO2 solubility in liquid methane is 213 ppm at 112 K by measurement, a margin of 4.2 over a 50 ppmv treated-gas rule.

Limitations

The fields, wells, analyses and limits are the synthetic composition bench's, and every schedule here is a solve of that invented case. The ranking of the two rooms, and the 2.3, belong to it; a train with different wells will price its limits differently, which is the reason to compute them from its own analyses.

The duty model is linear in tonnes and steady. It has no rich or lean loading model, no circulation rate, no absorber, and a heat of reaction that Polasek and Bullin note varies with loading [1]. The multiple of 2.0 is assumed and is the largest single uncertainty: it is set by order of magnitude against a flue-gas MEA pilot, not calibrated on a natural-gas MDEA unit, and flash regeneration could put it lower. The fuel share assumes a fired boiler at 85% efficiency.

Nitrogen room is priced here only through the reboiler. Rejecting nitrogen has its own energy price, in end-flash gas burned as fuel or in a nitrogen rejection unit, and that cost would fall on the side of nitrogen room. We do not model it, so the 2.3 is a reboiler ratio, not a plant ratio.

The cold-box checks use pure-methane solubility, a sweet-gas water correlation at one assumed saturation state, and the vendor's upper capacity for mercury. Each is a screening figure for a process simulation to replace.

References

1. J. C. Polasek and J. A. Bullin. Selecting amines for sweetening units. Proceedings of the GPA Regional Meeting, Tulsa, September 1994, Bryan Research and Engineering: heats of reaction of 600, 653 and 825 Btu/lb CO2 for MDEA, DEA and MEA, their variation with loading, and flash regeneration of MDEA for bulk CO2 removal: bre.com.
2. M. Biermann, F. Normann, F. Johnsson, R. Hoballah and K. Onarheim. Capture of CO2 from steam reformer flue gases using monoethanolamine: pilot plant validation and process design for partial capture. Industrial and Engineering Chemistry Research 61, 2022, specific reboiler duty of 3.6 to 3.8 MJ/kg CO2 at 90% capture: research.chalmers.se.
3. G. De Guido and L. A. Pellegrini. Application of conventional and novel low-temperature CO2 removal processes to LNG production at different CO2 concentrations in natural gas. Chemical Engineering Transactions 74: 853 to 858, 2019: aidic.it.
4. T. Shen, T. Gao, W. Lin and A. Gu. Determination of CO2 solubility in saturated liquid CH4 + N2 and CH4 + C2H6 mixtures above atmospheric pressure. Journal of Chemical and Engineering Data 57(8): 2296 to 2303, 2012, nine binary points via the NIST ThermoML record: trc.nist.gov.
5. Calgon Carbon. HGR activated carbon for mercury removal, product bulletin, a vendor document: capacity up to 30% by weight and treated gas under 0.01 µg/Nm3: calgoncarbon.com.
6. M. Moshfeghian. Ideal water content correlation for sweet natural gas. Campbell Tip of the Month, January 2018, describing the Bukacek correlation and its range: jmcampbell.com.
7. NIST Chemistry WebBook. Thermophysical properties of water, saturation pressure at 30 C: webbook.nist.gov/chemistry/fluid.
8. The Gas Safety (Management) Regulations 1996, Schedule 3, Wobbe number at most 51.41 and at least 46.50 MJ/m3: legislation.gov.uk; and the Gas Safety (Management) (Amendment) Regulations 2023, SI 2023/284, substituting 46.50 for 47.20 from 6 April 2025: legislation.gov.uk.
9. EarthScan pretreatment and composition benches, synthetic throughout, the source of every figure from the three-field case: available on request.