Bashing blue hydrogen: questionable numbers, flawed models

“Quite high” is politely vague. In the body of the paper, H & J endeavor to show that fugitive methane emissions associated with its production would actually give blue hydrogen a higher global warming potential (GWP) than coal or diesel oil. That’s nonsense, but the authors deploy questionable numbers and flawed modeling to lend their thesis a veneer of academic credibility.

Not surprisingly, there has been pushback. In this twitter thread, initiated by @TedNordhaus of the Breakthrough Institute, two principal criticisms stand out. One is that the study employs a 20-year window to assign a greenhouse gas equivalency to methane instead of  the 100-year window that the IPCC uses. The other is that the study assumes a capture efficiency of just 65% for production of blue hydrogen from natural gas. That’s unreasonably low.

Both criticisms are valid and significant. Let’s consider first the 20- vs. 100-year window issue.

Methane has an expected lifetime in the atmosphere of only around 12 years. While it’s present, it is a very potent greenhouse gas. Gram for gram, it’s over 100 times as potent as CO₂. But it goes away. Integrating its warming potential over a 20-year window gives it a CO₂ equivalency of 86 -- the figure that H & J use in their study. Over a 100-year window, however, it’s only 28. That is one third of the figure that H & J employ. It’s lower, because over most of that 100 years, the emitted methane will be long gone. Using the 20-year equivalency exaggerates the overall impact of methane emissions. It runs the risk of warping our priorities for remedial actions.

As to the carbon capture efficiency for the reforming process, that’s a more complicated matter. The large fraction of uncaptured CO₂ emissions that the study asserts obviously challenges blue hydrogen’s credentials as a clean alternative to green hydrogen -- as the authors’ intended. But they anticipated criticism of their capture efficiency assumption. They acknowledge that higher efficiency is possible, but assert that it would require more natural gas to power the capture process. They contend that the fugitive emissions associated with that added natural gas would more than cancel out any GWP benefit. 

Are they right? Well, if the fugitive emissions associated with any use of natural gas were really as high as they contend, if the 20-year window for the CO₂ warming equivalency of methane were appropriate, and if the naive models that H & J posit for SMR and carbon capture were actually representative of what gas companies would deploy, then they would be. As it happens, though, they’re very wrong.

A seriously flawed study

None of the above predicates required for the authors to be right actually holds. The study: 

1. greatly overstates the magnitude of fugitive emissions associated with use natural gas for blue hydrogen production;
2. uses an inappropriately inflated value for the global warming equivalency of methane (the 20- vs. 100-year window issue); and
3. uses a naive model for the SMR process and carbon capture that is not what would likely be deployed for production of blue hydrogen.

I think we’ve adequately covered point 2) about the window for calculating CO2 equivalency of methane. Let’s now consider the other two.

Point 1) is tricky, because it involves more than just an overall estimate for fugitive methane emissions associated in some way with oil and gas production, distribution, and use. Such an estimate may provide a useful indicator of the global warming impact of our ongoing dependence on fossil fuels, but it’s not what we need for analyzing policy choices. We need detailed knowledge of where the emissions come from, what they’re associated with, and what the most effective approaches would be for reducing them. H & J’s study does not provide that. What it purports to provide for the particular case of blue hydrogen production is grossly misleading.

For a life cycle analysis of particular products or practices, there has to be an actual causal connection between the thing analyzed and the costs imputed to it. “If we do X, then we will incur Y as a cost”. Conversely, “If we don’t do X, then we won’t incur Y as a cost”. The authors do not establish a causal connection between production of blue hydrogen and the specific amount of fugitive emissions that would result. They don't even seem to recognize that a causal connection is necessary. They simply distribute all estimated emissions proportionally across all applications that consume natural gas.

Estimate of fugitive methane emissions

H & J’s study arrives at a figure of 3.5% for the fugitive emissions from all gas production. It’s the adjusted sum of two parts. The first is an estimate of upstream emissions from oil and gas field operations. They put this at 2.6%. The second is an estimate for everything else. They put that at 0.8%. The sum, 3.4%, is then adjusted upward by 0.1% to account for a small difference between net production (excluding upstream leakage) and net consumption. They don’t specify the source of difference. Presumably it’s from gas consumed in gas processing and pumping operations.

One could get lost in debating whether the 3.5% estimate is too high, too low, or about right. It’s lower than some that one can find and higher than most others. The 2.6% part comes from an unpublished synthesis by Howarth of 20 studies of 10 non-conventional oil and gas basins. It’s based on satellite measurements of methane concentrations, combined with statistical models of winds and gas dispersion in the lower atmosphere around the times of the observations. Let’s assume it’s good, at least for those fields.

All of which is rather beside the point. Gas emissions from non-conventional fields -- shale plays where fracking is required to produce oil and gas -- are known to be substantially higher than from conventional fields. The reason? Flaring in non-conventional fields is much higher than in conventional fields. The preponderance of methane leakage will be from flaring. A recent study by the NRDC found that 11% of flares in the Permian Basin were partially or wholly unlit, emitting large amounts of methane. 

The salient point, however, is that virtually all flaring activity, in terms of volume flared, involves associated gas from oil wells! They are wells that, for one reason or another, are not connected to the system for collection and distribution of natural gas. Divorced from that system, they have nothing to do with end use applications that consume natural gas! But wells that do feed into the natural gas system? They very rarely flare gas. Aside from tiny amounts that are released in maintenance operations, they have no reason to flare. All the gas they produce feeds into the natural gas system.

What this means is that the 2.6% figure for upstream emissions of methane grossly overstates the emissions that are actually associated with use of natural gas. If we decline to use natural gas for hydrogen production, it will not reduce those emissions. If we do use it for hydrogen production, it will not increase them. It could even have the opposite effect. The added demand could provide oil well operators with incentive to connect their wells to the gas infrastructure and stop wasting so much associated gas in unreliable flares. According to this report, flaring accounts for fully 4% of global gas consumption. It’s “low hanging fruit” for increasing the supply of natural gas while reducing fugitive emissions.

That brings us to point 3), regarding the authors’ model for the SMR process. 

The SMR process and blue hydrogen

Technical aspects of blue hydrogen production, whether by steam methane reforming (SMR) or other methods, are the domain of chemical process engineers. Neither Howarth nor Jacobson are chemical process engineers. Howarth lists himself as a “biogeochemist”, while Jacobson is a professor of civil engineering. They are credentialed, and affiliated with respected universities. But they either have a limited understanding of reforming processes and efficient ways to capture CO₂, or they have deliberately chosen to employ naive models for their study. The models they use are highly suboptimal for production of blue hydrogen. 

For gray hydrogen, the authors cite an IEEE article from 2004 to say that it takes between 2.0 and 2.5 kWh to drive the production of one cubic meter of hydrogen. That range is, in fact, realistic for a lot of SMR plants built over the years. Most are associated with oil refinery operations or large fertilizer or petrochemical plants. In that environment, it’s expedient to burn additional feed gas with air in a furnace to supply heat to the reaction tubes in which the SMR reaction takes place. It’s not the most energy-efficient way to produce hydrogen, but it’s relatively cheap and does the job. If one is not concerned about carbon emissions, it’s an appropriate and well-proven design.

If one is concerned about carbon emissions, then burning additional feed gas with air is a bad design choice. It gives rise to the need to separate CO₂ from nitrogen in the flue gas. A better choice that stays close to the conventional SMR design is oxy-combustion. With that approach, the extra feed gas is burned in a mix of recycled CO₂ and oxygen. The “flue gas” is then a nearly pure CO₂ stream. No costly separation step is needed, and capture efficiency is 100%. Of course, a supply of nearly pure oxygen is needed.

That’s not particularly hard to arrange. If the reformer is part of an ammonia fertilizer plant, then there will already be a large cryogenic air separation unit (ASU) within the facility. It’s to provide nitrogen for making ammonia, but it also produces a “waste” stream of nearly pure oxygen. Another option is to marry blue and green hydrogen production systems. Electrolysis of water for one unit of green hydrogen provides enough “waste” oxygen for about three units of blue hydrogen using oxy-combustion.

SMR adapted for oxy-combustion and capture of nearly all CO₂ emissions is an existing technology. It’s been commercially implemented for supplying CO₂ to conventional oil fields in the Permian basin for enhanced oil recovery (EOR). But there are other candidate technologies that would be even better.

Better options

It’s important to note that when production of blue hydrogen begins to ramp seriously upward, we won’t be seeing CCS simply tacked on to existing SMR plants. The market will be large enough to justify new plant designs, specifically optimized for the production of blue hydrogen. They will be designed by chemical engineers who truly know what they’re doing, and they will leverage cheap renewable energy to minimize consumption of natural gas. Energy to drive the basic reforming reaction might be supplied by clean electricity or by concentrated solar thermal energy, so that the hydrogen output has a greater free energy for heat generation than the hydrocarbon feedstock from which the hydrogen was made. That would certainly confound suggestions that blue hydrogen would be dirtier than using natural gas directly!

I don’t know what the winning design for a blue hydrogen production plant will ultimately be. A very strong candidate, however, is described in a May 2019 report published in Science. It’s electrified methane reforming. In that process, the energy to drive the reforming reaction comes from ohmic heating from electrical currents in SMR reaction tubes. There are many potential advantages. One is that no feed gas is consumed to provide the heat. There is no flue gas, and no CO₂ needing to be captured from it.

If the electrical energy for electrified methane reforming comes from clean energy resources, the product can be considered a form of green hydrogen. But by my own calculations, the hydrogen yield per kWh of electricity input, would be about 6 times higher than for electrolysis of water. This could provide an interesting resolution to the ongoing blue vs. green hydrogen debates. Combine them, and enjoy the best of both worlds!

Electrified methane reforming is by no means the only advanced process that has been implemented or is under investigation. I can’t begin to list them all, but there are three in particular that readers interested in a deeper dive into options for blue hydrogen might find of interest:

1. An approach proposed by Norwegian technology firm GTI that employs fluidized calcium oxide to capture and remove CO₂ from the downstream water gas shift reactor of a conventional SMR plant. CO₂ removal shifts the reaction equilibrium toward more complete hydrogen conversion. It minimizes the concentration of CO in the product stream.
2. A chemical looping approach proposed by researchers at Norwegian University of Science and Technology and research firm SINTEF.  The approach is termed Gas Switching Reforming.
3. Economic assessment of membrane-assisted autothermal reforming for cost-effective hydrogen production with CO₂ capture. Membrane-assisted autothermal reforming is a different approach to reforming hydrocarbons. It has not been widely implemented, but appears to offers some advantages.

Going forward

To reiterate what I said above, I don’t know what the winning approach for blue hydrogen production will ultimately be. But I am sure that the inherent advantages that it offers will ensure that blue hydrogen will be produced, and produced at scale. It will play a key role in the energy transition. The attack by Howarth and Jacobson is entirely without merit and will soon be forgotten. 

We need clean hydrogen, for a variety of reasons. Eventually, we will have to pay the energy price to extract it from water. With or without CCS, we won’t be able to rely on cheap reforming of energy-rich hydrocarbons indefinitely. But until we actually reach a 100% clean energy economy, truly green hydrogen can only be produced by clean energy at times when there is no competing use. For now and for a decade or more to come, those times will not occur often enough to make green hydrogen economically competitive.

We need a practical way to clear a path for the hydrogen economy. To me, at least, that looks like blue hydrogen.