Hydrogen sets 94 % of the cost of methanol made from captured CO₂ and PEM electrolysis

CheAI Research Inc. | Research article | September 2026

Abstract

1. Introduction

Methanol made from captured CO₂ and hydrogen from low-carbon electrolysis can have a lower footprint than methanol from natural gas [1]. The Joint Research Centre (JRC) re-simulated and costed a loop designed by Van-Dal and Bouallou [2]-[4]. The prices of hydrogen and captured CO₂ kept that plant from being financially viable [2], and a 5 kt/d study named electrolytic hydrogen as the main reason [5].

These studies treat the hydrogen price as an input. A cost and a footprint describe the same product only if they refer to the same hydrogen, at one price, pressure and carbon intensity. The reference loop also purges 1 % of the recycle gas to avoid inert accumulation [3]. With 99.95 mol % CO₂ from an ammonia plant [6] and inert-free electrolytic hydrogen, it was not clear what sets the purge. Hydrogen sets almost all of the cost and footprint, and recycle-side capital sets the purge.

2. Methods

The hydrogen comes from a published 50,000 t/yr PEM plant [7]. Its levelized cost of 11.34 USD/kg at 30 bar, plus the compression bridge of its footprint study [8], gives 11.361 USD/kg at 80 bar, so cost and carbon share one delivery pressure. The CO₂ is the compressed stripping vent of an ammonia plant at 19.0 USD/t in 2018 dollars [6].

A Python adiabatic plug-flow reactor with Vanden Bussche-Froment (VBF) kinetics [9] was coupled to a DWSIM 9.0.5 flowsheet [10] with Peng-Robinson thermodynamics, a 72 bar flash, a recycle compressor and a column. Because the reference kinetic re-fit [11] was not openly available, we fitted one activity multiplier to the reference outlet methanol and carried the original kinetics beside every result.

The purge fraction was the decision variable. The return-side pressure drop, 3.956 bar at 1 % purge, was inferred from the reference compressor power [4] and scaled with the square of the recycle flow. We minimized variable cost over purge sweeps and evaluated the levelized cost of methanol (LCOM) at 22 costed DWSIM nodes. Costs are AACE Class 5 (-30 % to +50 %) in USD 2024, with a 10 % discount rate.

The footprint is a cradle-to-gate screening inventory, not an ISO life cycle assessment, using the PEM study's hydrogen footprints [8]. The benchmark is an indicative natural-gas methanol value of 500 kg CO₂e/t [12]. The break-even hydrogen footprint is the value at which the methanol footprint equals it. Independent re-derivations and check scripts confirmed the calculations, not the physical models.

3. Results

3.1 The reference reactor needs a calibrated activity

With the original VBF parameters the reference reactor leaves at 229.5 °C against the reported 290 °C, with 1.46 vol % methanol against 4.7 vol %. An activity multiplier of 2.437 gives 288.5 °C and 4.70 vol %, and the loop's CO₂ and hydrogen use fall within 5 % of the references. The factor is a calibration to one reported number, not a prediction.

3.2 Hydrogen is 94.4 % of the levelized cost

At 0.9 % purge the plant makes 248,846 t/yr of methanol. It uses 0.2009 t of hydrogen per tonne, 6.5 % above stoichiometry, and the purge burns 5.3 % of the hydrogen fed. The LCOM is 2418.3 USD/t. Hydrogen contributes 2282.8 USD/t, capital 46.1, CO₂ 36.9, other operating costs 50.0 and electricity 2.5 USD/t.

Each 1 USD/kg change in the hydrogen price moves the LCOM by 200.9 USD/t (Figure 1). Over the P10-P90 range of the PEM study's cost, 9.77-15.16 USD/kg, the LCOM spans 2099.1-3182.1 USD/t. The CO₂ price is second, at 2381.4-2503.0 USD/t. The Class 5 capital range moves the LCOM only between 2401.5 and 2446.2 USD/t. With the original kinetics the LCOM is 2894.2 USD/t, 476.0 USD/t higher, because its larger recycle raises capital.

3.3 Recycle-side capital sets the lowest-cost purge

Hydrogen is 97.0-98.8 % of the variable cost at every purge. The variable cost therefore falls with the purge, from 3252.0 USD/t at 10 % to 2291.2 USD/t at 0.4 %, the lowest converged purge (Figure 2c). A CO₂ feed with 4 vol % N₂ leaves this minimum at 0.4 %.

The full LCOM has its lowest evaluated value at 0.9 % purge (Figure 2d). Below 1 % the exchanger and compressor grow with the recycle flow, and at 0.4 % purge the total capital is 172.6 MUSD and the LCOM 2462.9 USD/t. At 2 % the hydrogen loss raises the LCOM to 2514.9 USD/t. Near the lowest node the curve is flat, at 2419.3, 2418.3 and 2423.2 USD/t for 0.8, 0.9 and 1 % purge.

3.4 The hydrogen supply sets the footprint

On the US average grid the footprint is 4822.6 kg CO₂e/t, and hydrogen contributes 95.5 % of it (Figure 3a). With nuclear hydrogen it is 163.8 kg CO₂e/t, with wind 247.3 and with solar 553.6.

The break-even hydrogen footprint is 1.40 kg CO₂e/kg with the plant's other electricity on the US grid and 1.80 with it on the electrolyser's supply. These are two scenarios, not an uncertainty range. Nuclear (0.155) and wind (0.563) hydrogen fall below both. Solar hydrogen (2.059) exceeds both, but its methanol lies below the top of the gate range of 457.7-608.9 kg CO₂e/t derived from the IRENA tables [12]. Purchased steam and burned purge and vents add a floor of 132.1 kg CO₂e/t that cleaner hydrogen cannot remove (Figure 3b). With the original kinetics the break-even falls to 0.18-1.27 kg CO₂e/kg, and wind hydrogen passes only when the plant's other electricity is also wind.

4. Discussion

The capital range moves the LCOM by -16.8 to +27.9 USD/t, the P10-P90 hydrogen range by -319.1 to +763.8 USD/t. This agrees with earlier findings that hydrogen is the main barrier [2], [5]. Here the hydrogen price, pressure and footprint come from one plant, so cost and carbon refer to the same hydrogen.

The reference loop purges for inerts [3]. This feed holds almost no inert, so lowering the purge saves hydrogen, and recycle-equipment capital creates the interior lowest-cost node. The 1 % choice costs only 4.9 USD/t more.

For the methanol to fall below the indicative natural-gas value, the hydrogen footprint must be 13-16 times lower than US average hydrogen. The modeled wind and nuclear supplies meet this on the screening boundary.

5. Limitations

The kinetics is calibrated to one reported value, and the bed runs above the 280 °C and 51 bar range of the VBF data. The gap to the original kinetics is a structural sensitivity, not a bound. The reference sources also disagree on conversion per pass (22 % and 33 %), and no single activity matches both the JRC conversion and outlet methanol. Peng-Robinson fails four of eight held-out solubility checks. Only one sensitivity case (+0.54 %) tests their effect, so the full property uncertainty is not quantified.

The reactor has no pressure drop and an effectiveness factor of 1. Heating and cooling demands exceed the JRC values by 25.1 % and 120 %, for reasons not traced, and compression electricity is 6.7 % below the reference. Dissolved gas bypasses the column, which failed its CO₂ balance. Boiler efficiency and column diameter are assumed, and the column cost is 84 % above the JRC value, unexplained. The model forms no by-products, so the acetone and ethanol limits were not checked, and no sourced maximum catalyst temperature was found.

The 0.9 % purge holds for hardware sized for 1 % purge with a Q² pressure drop that no independent sizing backs. A loop resized with the purge could move the lowest cost. The LCOM was evaluated at discrete nodes, and no operating point was found at 0.32 % purge within a 40 bar search. On the cost side, pumps, storage and the boiler are not costed separately, and fixed capital has no salvage value. Cost and footprint are separate sensitivities, and the LCOM uses the US reference hydrogen price for every supply.

The footprint excludes electrolyser manufacture, capture and plant equipment, the ammonia-plant share, catalyst, wastewater treatment and minor utilities, so it understates a fuller inventory by an unknown amount. This matters most for low-carbon cases, and the wind margin is not shown to survive a harmonized boundary. The upstream-gas line uses AR6 potentials, the rest AR5. The IRENA value has no stated boundary, and the gate range depends on our removal of a combustion term from one row. No CO₂ removal credit was taken, and the product releases 1373.5 kg of fossil CO₂ per tonne outside the gate.

6. Conclusions

A CO₂-to-methanol loop sized on a specified PEM hydrogen plant makes methanol at a cost set almost entirely by the hydrogen price, with a screening footprint set by the electricity behind the hydrogen. The modeled wind and nuclear cases fall below an indicative natural-gas methanol value, and the US grid cases do not. With an almost inert-free feed the purge mainly loses reactants, and on hardware sized for 1 % purge recycle-side capital sets the lowest-cost purge. The calibrated kinetics, fixed-hardware hydraulics, property-model misses and screening boundary limit these results.

References

1. González-Garay A, Frei MS, Al-Qahtani A, Mondelli C, Guillén-Gosálbez G, Pérez-Ramírez J. Plant-to-planet analysis of CO₂-based methanol processes. Energy & Environmental Science 12(12), 3425-3436, 2019. DOI 10.1039/C9EE01673B
2. Pérez-Fortes M, Schöneberger JC, Boulamanti A, Tzimas E. Methanol synthesis using captured CO₂ as raw material. Techno-economic and environmental assessment. Applied Energy 161, 718-732, 2016. DOI 10.1016/j.apenergy.2015.07.067
3. Pérez-Fortes M, Tzimas E. Techno-economic and environmental evaluation of CO₂ utilisation for fuel production. Synthesis of methanol and formic acid. JRC Science for Policy Report EUR 27629 EN, European Commission, Joint Research Centre, Luxembourg, 2016.
4. Van-Dal ÉS, Bouallou C. CO₂ abatement through a methanol production process. Chemical Engineering Transactions 29, 463-468, 2012. DOI 10.3303/CET1229078
5. Nyári J, Magdeldin M, Larmi M, Järvinen M, Santasalo-Aarnio A. Techno-economic barriers of an industrial-scale methanol CCU-plant. Journal of CO2 Utilization 39, 101166, 2020. DOI 10.1016/j.jcou.2020.101166
6. Hughes S, Zoelle A. Cost of capturing CO₂ from industrial sources. National Energy Technology Laboratory, Pittsburgh, PA, 2022.
7. Ho TD. Green hydrogen by PEM electrolysis, what it costs to make. Techno-economic analysis of a 50,000 t/yr plant (AACE Class 5). CheAI Research Inc., 2024. cheairesearch.com/reports/green-hydrogen-tea.html
8. Ho TD. Grid carbon intensity, not electrolyser efficiency, determines the well-to-gate footprint of PEM hydrogen. CheAI Research Inc., 2026. cheairesearch.com/reports/pem-hydrogen-lca.html
9. Vanden Bussche KM, Froment GF. A steady-state kinetic model for methanol synthesis and the water gas shift reaction on a commercial Cu/ZnO/Al₂O₃ catalyst. Journal of Catalysis 161(1), 1-10, 1996. DOI 10.1006/jcat.1996.0156
10. DWSIM, open-source chemical process simulator, version 9.0.5. 2026. https://dwsim.org
11. Mignard D, Pritchard C. On the use of electrolytic hydrogen from variable renewable energies for the enhanced conversion of biomass to fuels. Chemical Engineering Research and Design 86(5), 473-487, 2008. DOI 10.1016/j.cherd.2007.12.008
12. IRENA and Methanol Institute. Innovation outlook. Renewable methanol. International Renewable Energy Agency, Abu Dhabi, 2021.

Related studies

• Green hydrogen PEM plant, techno-economic analysis, the hydrogen cost basis used here
• PEM hydrogen well-to-gate carbon footprint, the hydrogen footprint basis used here
• Syngas CO₂ capture on a steam methane reformer, break-even carbon price, which compares blue hydrogen with the same PEM design