Prof. Lovins is incorrect in implying that no major technological breakthroughs are needed in fuel cells, other than those aimed at cost reduction. Major work is still required on reliability and durability (current warranties must be lengthened by up to an order of magnitude to be acceptable in a marketplace used to 50-100,000 mile vehicle warranties). Membrane life is a major unknown. Avoidance of membrane fouling requires ultra-clean air which, in turn, currently requires ultra-filtration and consequent parasitic losses. Catalyst loadings must be reduced and perhaps precious metals in the catalysts replaced with less costly alternatives while catalyst life is increased. If hydrogen is to succeed both technically and economically, significant cost-reducing breakthroughs are required in manufacturing (especially in the distributed, rather than centralized, manufacturing model preferred by Lovins), equipment for compression to pressures above 5 ksi, pumping, pipeline hardware, local distribution systems, delivery systems (for liquid and gaseous hydrogen) and numerous other areas. A massive effort will be required to reduce the energy consumed in manufacturing and transporting hydrogen. Unfortunately, the gas starts off with the disadvantage of having to be made from a primary energy source that could be used more efficiently if conversion could be avoided.

In the case of autos, Dr. Lovins must know of the many failed efforts that have been made to develop a cost-competitive ultralight, mass-manufacturable auto body (the writer managed such a program involving stampable composites in 1976-9). Notwithstanding the undoubtedly good design work at Ultracar, translating such designs into real manufacturable products that meet a wide range of engineering, cost and safety criteria and also attract the necessary large market has proven extremely difficult.

With respect to the future size of the required hydrogen industry, the current North American hydrogen industry produces about 10 million metric tons/year of hydrogen (not 15 million as Lovins estimates). To replace all of the current gasoline consumption (about 9 million barrels per day) would require about 130 million metric tons of hydrogen annually, a figure that depends on assumptions about use efficiency – a very substantial increase and not just “several fold bigger”, especially in view of the completely different manufacturing technologies that are likely to be needed.

2.   Comments on Specific “Myths”

The numbering and wording follows that of Lovins.

1.   A whole hydrogen industry would need to be developed from scratch.

1.    If we cut through Lovins’ obfuscating detail, he claims, based on DOE data, that the present world-wide hydrogen industry produces about 50 million metric tons, or about 500 billion cubic meters of hydrogen annually. A little under 50% of this comes from natural gas, 30% from oil and 18% from coal (but please note our earlier comments – even in the case of natural gas, at least 50% of the hydrogen is derived from water). All of these sources produce hydrogen by steam-hydrocarbon reforming, an alternative water-splitting process which in effect uses the carbon (instead of electricity) to split the water to produce the hydrogen. In this case the oxygen forms carbon oxides instead of being released as oxygen. Thus, water, rather the hydrocarbon, can be the most important source of the hydrogen. The lower the hydrogen content of the hydrocarbon (the most extreme case being coal), the more the water needed to produce the hydrogen (see also “myth” #14). For 4 mols of hydrogen product:

For methane, including the shift reaction:  CH₄ + 2H₂O à CO₂ + 4H₂

For carbon, representing coal:                    2C + 4H₂O à 2CO₂ + 4H₂

Note that carbon/coal produces about twice the amount of CO₂, which must then be sequestered if CO₂ atmospheric emissions are a concern (but see our earlier comments about global warming). At present, only about 2% of hydrogen production comes from the electrolysis of water. Current U.S production is about 9 million metric tons/year with Canada accounting for another 1 million tons. Lovins’ estimate of ~15 million tons is high, apparently because of double counting.

2.    Lovins is correct in saying that “the industrial infrastructure for hydrogen production already exists”. However, there are only some 460 miles of pipeline in the U.S., all fully dedicated to industrial users of hydrogen in Texas and Louisiana. A comprehensive infrastructure associated with centralized hydrogen production would require many thousand miles of new pipeline (natural gas pipelines are fully committed and are likely to remain that way; in any case, none were designed for hydrogen service – for a further discussion, see “myth” #5).

3.    Professor Lovins does not like overhead electrical transmission lines but, notwithstanding the recent blackout in Midwestern and Northeastern states, they have served us very well and will continue to do so. As he rightly points out, these lines experience transmission energy losses (these can be as much as 5% and are a function of line length and construction as well as the transmission voltage), but he has his numbers wrong in the comparison with hydrogen. There is no real-world experience with pumping high-pressure hydrogen (≥5,000 psi/350 bar) through long-distance pipelines but Eliasson and Bossel (see Lovins, Ref. #5) have shown convincingly that the energy losses will be substantial. Moving hydrogen at lower pressures requires a very large pipeline to move very large volumes because the energy content of hydrogen gas (or liquid) per unit volume is so low (as Lovins points out, the only time that the very low density of hydrogen may be an advantage is in space travel, and even then, as we know from experience with the space shuttle, the size of the liquid hydrogen tank presents significant design challenges).

4.    Lovins is probably correct in saying that distributed, rather than centralized production and (of course) use of hydrogen will have to characterize any future “hydrogen economy”, but this is precisely the problem – small scale production means that economies of scale are lost (however well the “reformers and electrolyzers work at small scale”) and that the probabilities, and associated dangers, of equipment failure are greatly increased. Furthermore, the energy source has to be connected to the distributed reformers or electrolyzers, although this should present no more of a challenge than distributing and delivering gasoline does today. A centralized or regionally distributed system (see the comments on Myth #9) offers much greater safety. A problem with any system involving large-scale hydrogen production, whether national or regional is the lack of really large-scale storage. Suitable underground storage such as proven gas-tight former natural gas wells or even salt caverns is not usually available where it is needed.

5.    Off-peak power may be less costly, but is likely to become much more costly as the U.S. and Canada invest in a much-needed renewal of their power generation and distribution systems. Electrical power has or will become far too costly for hydrogen production.

2.   Hydrogen is too dangerous, explosive or “volatile” for common use as a fuel.

1.    The hydrogen manufacturing industry may indeed have an “enviable safety record”. This writer has no statistics but recalls a few alarming incidents. Large-scale reformers are usually controlled remotely and automatically so that employee exposure is minimal. The hydrogen user industry (oil refining, ammonia production, etc.) is more prone to accidents, but many minor hydrogen-related incidents (usually compressor fires or explosions) go unreported.

2.     Hydrogen is not inherently safe because it “rapidly disperses up and away from its source”, particularly if this happens in a closed or poorly ventilated building. It easily leaks from equipment using it, especially at elevated pressure, but may not ignite at the point of egress. Any equipment using hydrogen must be equipped with hydrogen and fire detection sensors strategically located above the equipment, as must the building in which it is located.

3.    As anyone who has been involved in large-scale hydrogen fires or has used an oxy-hydrogen blowtorch will testify, the flame is intensely hot on contact (although, as Lovins says, it is not intensely radiant) and causes a lot of damage very quickly. Notwithstanding all of the theory about lower explosive limits, in practice hydrogen both ignites and explodes easily.  Hydrogen explosions, especially if the gas is at high pressure, are massively powerful (although in practice major hydrogen explosions often involve other energy sources such as gaseous hydrocarbons that are mixed with the hydrogen). The subsequent large-scale fires are often intense and very difficult to fight because the flame cannot easily be seen except in cases where hydrocarbon is present.

4.    We agree that the Hindenburg story is irrelevant. Both airships and hydrogen technologies have made considerable progress since 1937.

5.    As NBC Television learned the hard way, staged demonstrations of vehicle fires seldom relate to real-world experience. No one experiences much heating from an oxy-hydrogen blowtorch flame, even a big one, but hydrogen explosions at even 3,000 psi (200 bar) have been lethal and have done immense damage (e.g., that at the Esso (now Exxon) refinery in Linden, NJ in 1970).

3.   Making hydrogen uses more energy than it yields, so it is prohibitively inefficient.

This entire discussion is an excellent example of the smoke-and-mirrors (or perhaps apples-and-oranges) method of comparing energy sources. We will examine the claims point-by-point:

1.    Any conversion of energy from one form to another is, indeed, costly although it is not true that such conversions “always consume more useful energy than they yield”. In addition, most of our current energy resources require no conversion – just a little chemical modification and fractionation for oil and usually only moisture and sulfur removal for natural gas. This means modest well-to-tank energy consumption (10-30% of that in the original source) as Lovins correctly points out, but no conversion energy costs of the kind applicable to hydrogen.

2.    In the case of manufacturing hydrogen by electrolysis, for example, the equivalent of the “well-to-tank” investments of energy must be made just to get the fuel (coal, oil or natural gas) to the power station. Then the power must be generated, typically at a low efficiency of ~30% or less (more if in a combined-cycle facility), and only then do we have electrical energy available for conversion into hydrogen. So far, we have an overall efficiency of about 22.5%. Now we convert AC power to DC (5-10% loss), and electrolyze water to produce hydrogen (electrolyzer losses are more like 35% in commercial operation but are improving). So far, our overall conversion efficiency is just under 14% based on the energy in the original resource.

3.    Having made the hydrogen at ambient pressure or a little above (and we will generously assume that this happens at the point of use, thus avoiding transportation costs), we have to compress it for distribution (if needed) and delivery, a process that can easily convert the small overall positive amount of net energy available (14% of that in the energy source) into a net loss of energy. These conversion losses and costs are not tolerable. It makes no sense, in a world that will soon be resource-limited, to invest massive amounts of additional energy just to achieve a fuel that offers somewhat greater end-use efficiency.

4.   The energy balance is a little better for the conversion of natural gas into hydrogen because the really nonsensical double conversion step – fossil fuel into electrical energy and electrical energy into hydrogen – can be avoided. Thus we have only to be concerned about energy source production efficiency (taken above at 75%, probably a little better for low-sulfur natural gas) and the reformer thermodynamic efficiency (also about 75%, lower for small units but expected to improve) for an overall efficiency for the well à hydrogen step of ~55% (not 70% as Lovins assumes). In this case, hydrogen makes greater sense although we are much less certain about the efficiencies (and certainly the safety) of future small point-of-use reformers. We expect them to be significantly lower – perhaps 55-60% for an overall figure of 40-45%.  At this point, we still have to compress the hydrogen (15-20% of total energy), so the net energy available, although positive, is not very large – and we lose all of that in fuel cell inefficiencies.

5.    The most critical issue facing the use of natural gas reforming to make large quantities of hydrogen, at least in the U.S., is that it represents an unwise use of a rapidly-diminishing resource (see response # 12-4, below).

6.    There have been many “well-to-wheels” analyses of the efficiency of petroleum fuel use. When being compared with hydrogen, pessimistic assumptions are usually made about the efficiency of modern internal combustion engines. Analyses for hydrogen that are used in comparison typically ignore some of the steps involved, or make optimistic assumptions about their efficiencies, so the differences are always exaggerated.  The Toyota analysis referred to by Lovins was chosen to make their Prius gasoline-electric hybrid vehicle look good relative to Toyota’s conventional vehicles.

7.    In-vehicle fuel cell efficiencies, when they are stated at all, are generally overstated.  If all accessory demands such as air compression are taken into account they typically range from 30-40% at high load to 40-50% or so at low load for an average of about 35-40%, depending on usage. The drive train (more accurately a drive system) loses a small amount of additional energy, leading to an overall tank-to-wheels average efficiency of about 35%. Electrically-driven air conditioning, steering and other loads will slightly reduce overall efficiency. Thus, using the well-to-tank estimate for reformer hydrogen (55%, which may be optimistic if small point-of-use reformers are used), we obtain a well-to-wheels estimate of 19%, or about the same low figure as Lovins quotes for the gasoline engine (gasoline-electric hybrids will soon achieve better than 30%).  Diesel-engine hybrids using ultra-low sulfur fuels and direct injection provide even better overall efficiencies (up to 40%) since no energy conversion is required and this advanced diesel engine is much more efficient than, and offers almost the same performance as, the gasoline engine of equivalent performance. As we have made clear elsewhere, we see the diesel-electric hybrid as a far better choice for future transportation needs (once the U.S. has low sulfur fuel available in about 2006-2008) than either gasoline-electric hybrids or hydrogen fuel cells.

8.    We have no quarrel with Lovins’ conclusions regarding fuel cells in power generation, although his efficiency figures are, again, the most optimistic that we have seen. Fuel cells are already showing their worth in peak-shaving (although more often with non-hydrogen fuels and alternate, e.g., solid oxide electrolyte, cell designs).

9.    Regarding underground storage of hydrogen, we note here only that old gas fields that are gas-tight for natural gas are often (although not always) unsuitable for hydrogen storage – they may leak too much. The result is dependent on the local geology.

4.   Delivering hydrogen to users would consume most of the energy it contains.

1.    This issue was dealt with quite effectively by Eliasson and Bossel (Lovins, Ref. #5).  A reading of Lovins’ comments on this paper makes one wonder if he has read it, at least in its latest form. The Swiss authors, both of whom are highly experienced engineers, looked very thoroughly at all real-world aspects of the transportation, distribution and delivery of hydrogen, not just for a centralized production system (with long-distance pipelines) but also for distributed production (e.g., using small reformers or even electrolysis). We recently reviewed their analysis once more and found it correct in its conclusions. Lovins, on the other hand, focuses on a few peripheral issues on which the Swiss authors’ case is less convincing and on that basis, incorrectly dismisses the entire paper. A few specific points:

2.    Eliasson and Bossel discuss liquid hydrogen in part because that is the hydrogen fuel preferred by at least two major U.S. or European auto manufacturers.

3.    The Swiss authors quite correctly point out the greatly increased safety hazard that results from the decentralization of hydrogen production (by natural gas reforming). Distributed manufacture means distributed danger and also (as Lovins suggests) the added cost of distributed carbon sequestration.

4.    Aside from the fact that “hydrogen appliances” is a really bad and confusing term (all other appliances that we can think of use energy rather than deliver or make it), RMI’s “suggested hydrogen transition strategy” (see Lovins, Ref 52 and page 13) makes little sense. The very idea of small-scale, badly-maintained (as they inevitably will be) high-pressure reformers, electrolyzers and storage systems, let alone vehicles, in highly populated areas just boggles the mind. However, the concept that hydrogen should not be transported over long distances but, instead, made as it is needed, clearly has merit if a system can be developed for doing this safely.  We prefer a model which transports coal-derived methanol for either direct use in fuel cells or for reforming into hydrogen.

5.    We are less persuaded by the argument for a low-carbon energy supply than is Dr. Lovins. We suggest that he examines very closely the supposed connection between carbon dioxide and global warming; it does not bear too much scrutiny. However, there may be other reasons to reduce our production of CO₂ – the well-being of the oceans, which is apparently where most of it goes, for example.
 

5.   Hydrogen can’t be distributed in existing pipelines, requiring costly new ones.

1.    This “myth” is substantially true. Any implementation of a centralized-production hydrogen economy would require new pipelines for several reasons. One is that the existing system, assuming that natural gas is available in sufficient quantities to continue present uses, will be operating at full capacity - even more so in the unlikely event that enough natural gas can be found for it to be used in addition as a raw material for hydrogen production.

2.    While a number of old and small-bore oil and natural gas collection systems have had their lives extended by the use of pull-through liners, this is unlikely to be economically feasible for the conversion of say 48” natural gas lines designed for use at 1,000 psi to hydrogen service. Add to this the cost of hardware conversion and of new compressors.

3.    Lovins should know that so-called “Hythane” mixtures can also lose hydrogen by selective diffusion (as the former coal gas industry knew very well many years ago – but that was generally viewed as OK since the loss of hydrogen raised the heating value of the gas).
 

6.  We don’t have practical ways to run cars on gaseous hydrogen, so cars must continue to use liquid fuels.

1.    We have only a few minor quibbles with this section of Dr. Lovins’ paper. In general, we are in full agreement. Notwithstanding our negative view of the hydrogen vehicle in general, the major challenge is not the car or the fuel cell that it contains, it is the manufacture and delivery of the fuel. Once we can deliver the hydrogen to the car at the right price, the rest is relatively straightforward or already done with the exception of major cost reduction.  The challenge is to get the hydrogen to the vehicle safely at a cost that will be acceptable in a marketplace that, in the U.S., has yet to accept $2/gallon gasoline. Hydrogen, based on our current projections, will cost much more (see response #9, below). The high cost of hydrogen may be more acceptable to the European market, which is used to higher gasoline prices (although there, hydrogen will no doubt be taxed in a way that makes it less attractive!)

2.    Dr. Lovins is optimistic about the ability of the fuel cell or automotive industries to reduce the cost of hydrogen fuel cells. Our information is that the projected cost of automotive fuel cells in volume production is a factor of ten more than would typically be acceptable for this application, based on current auto price expectations, which suggests that there is a very long way to go. The auto industry has always been surprisingly adept at cost reduction, but this challenge, if correctly stated, may be more than they can handle or reasonably expect any supplier to commit to.

3.    We are pleased to note that Dr. Lovins agrees with us in rejecting the idea of on-board reforming of gasoline for fuel cell autos. We agree – it was never a good idea, given the complexity of the system required and the consequent probability of failure and safety problems. On the other hand, we disagree on the potential of methanol. We agree that converting natural gas to methanol, transporting it to the point of use and then converting it to hydrogen for use, is unacceptable on many counts, one of which is that there is insufficient natural gas to allow this barring the use of imported LNG (or of methanol made offshore from natural gas). On the other hand, if hydrogen makes sense as a fuel (to repeat, we do not believe that it does for transportation use) then coal à methanol (for transportation to the reformer) à hydrogen makes a lot of sense both economically and technically.

4.    Methanol does not have a higher lifecycle cost than hydrogen if all costs applicable to hydrogen are included. Furthermore, Dr. Lovins forgets that we have been using methanol, diluted with varying amounts of ethanol or water, for many years in the form of methylated spirits (a fuel and cleaner now typically containing only about 5% methanol, the remainder ethanol), windshield washer fluid (35-45% methanol), wood spirits (100%), racing motor fuel (100%) and so on. It is toxic, but so are gasoline and ethanol. The MTBE fiasco that Lovins refers to was caused by the Environmental Protection Agency’s requirement that MTBE be used to replace the former tetraethyl lead before there had been an adequate evaluation of its downside risks. The costs were borne not by the transportation industry but by the energy industry (including the methanol producers)….and certainly not by the EPA!
 

7.   We lack a safe and affordable way to store hydrogen in cars.

5.     We agree that the manufacturing methods and performance of high-pressure containment vessels has made very significant progress in recent years. However, they are very expensive in an industry that has historically, and for the right reasons, been concerned about very small cost increments. A carbon fiber tank that can be manufactured “for just a few hundred dollars” and that is ten times the size of an equivalent gas tank (Lovins, p.17) is a very different proposition than a polyethylene gasoline tank with a ~$20 installed cost – especially in a vehicle that offers (per Lovins’ own figures) only modest on-road performance and is already projected to command a premium price. Under-skin space is already at a premium in small vehicles (unless the use of a fuel cell can somehow create more but so far we have seen no evidence of that) so a fuel tank that is 10X the normal size can represent an impossible design challenge.

6.    The primary concern about composite tanks of any kind is their poor resistance to penetration by sharp objects in a collision. We have recently heard much about fires in police-owned Ford Crown Victoria automobiles that appear to have been caused by sharp objects carried in the trunk penetrating the thin back wall of the trunk and the then the equally thin gas tank behind it. We are concerned about the possible results of a penetration failure of composite (e.g., filament-wound aluminum) tanks carrying hydrogen at nearly 700 bar (~10 kpsi). Such failures will occur, of course. Auto accidents almost always result in at least one unexpected outcome and, of course, the auto industry has essentially zero experience with accidents involving cars that use stressed structural composites in either body structure or fuel tank, other than through computer simulations.

7.    Although Lovins dismisses liquid hydrogen rather quickly (a conclusion with which we agree because of the large amount of energy required to liquefy the gas, a very inefficient process), BMW and General Motors seems to think otherwise and have recently been emphasizing liquid hydrogen.
 

8.    Compressing hydrogen for automotive storage tanks takes too much energy.

1.    Compressing hydrogen requires a lot of energy for the reasons discussed by Eliasson and Bossel. It is not clear that the relatively small compressors required for the volumes that have to be transferred in automobile refueling will offer even the rather poor efficiencies associated with the high-volume compressors used in the petroleum industry (which often uses 3 kpsi hydrogen). Intercooling may be practicable but using the rejected heat may not be. It is also not clear that any of the energy in the compressed gas can be recovered cost-effectively using a “miniature turbo-expander” (as Lovins suggests) but it is certainly worth trying.

2.    We are not aware of any commercially available, low-cost, small hydrogen compressors that will even reach 5 kpsi (~350 bar), let alone 10 kpsi (~700 bar), or offer integrated intercoolers. If the demand develops – meaning if the market buys hydrogen-fueled cars in sufficient numbers – such compressors will be developed although the cost will be high (see also Lovins, Ref. 5).

3.    Canada’s National Research Council has recently claimed to have developed an electrolytic hydrogen generator that operates at 5 kpsi. In agreement with Lovins, this may be the way of the future, although however the compression is achieved the energy required has to be paid for somewhere – in this case in reduced cell efficiencies. Again, compression can ‘cost’ 15-20% of total energy.
 

9.     Hydrogen is too expensive to compete with gasoline.

1.    This section of Lovins’ “Twenty Myths” paper lacks sufficient detail, so we were unable to check his claims that hydrogen can be competitive with the wholesale price of gasoline. We were unable to confirm that with our own detailed analysis. As we have pointed out previously, Lovins reaches many of his conclusions by (a) assuming the existence of hydrogen-generating equipment that may or may not be developed in the future; (b) assigning an operating cost and capital cost to that equipment, typically based on the most optimistic possible assumptions of operating efficiency; (c) ignoring many other costs (such as controlled-composition water for electrolyzers, costs of power conversion or voltage-reduction, costs of handling oxygen by-product, costs of sequestering carbon dioxide (assuming that this is selected), catalyst costs and so on. As we have previously noted, Lovins et al also make optimistic assumptions about in-vehicle fuel cell efficiencies.

2.    We will register here once more the concern that we have over the reliability and hence safety of a distributed hydrogen manufacturing “network”. Industrial-scale reformers are generally reliable in part because of their very large scale (even so, minor explosions are common). There are at this point no commercially-available small-scale reformers and performance or reliability projections can not be based on significant operating experience. Small-scale chemical “factories” of this kind (which are reminiscent of Mao Tse-Tung’s revolutionary but highly unsuccessful “blast furnaces in every back yard” directives of the late 1950s) have historically been notoriously unreliable unless rigorously maintained (and people being people, and maintenance labor costs being high, they will not be). Electrolyzers might be more reliable but the hydrogen would be unacceptably costly.

3.    Finally, it is unlikely that the discovery or production of natural gas will be greatly stimulated by an uncertain $6/MMBTU price. The best stimulant for exploration is stable gas pricing (preferably, of course, at a high level). Gas exploration may increase somewhat at present pricing, mostly in deep gas areas in the northern Rocky Mountain foothills, and some otherwise uneconomic shut-in wells may come into production, but these are unlikely to do more than maintain reserves at current levels. The United States has become hooked on natural gas in a very major way (consumption is 22 trillion ft³/year and consumption growth is currently supply-limited. The DOE, without regard to possible hydrogen manufacture for transportation use, predicts consumption at 32 trillion BTU annually by 2020). Major new discoveries within the contiguous 48 U.S. states or Canada, with the possible exception of the relatively unexplored offshore far north, are now considered very unlikely. Current production is not keeping pace with demand. A transportation fuel system based on natural gas would result in extraordinarily high prices and a greatly increased dependence on imported gas, a situation that we consider unacceptable.