Good point Davemart.

@Herm:
The problem isn't the fast chargers for batteries.
It is needing at least one charger per car.
That adds up quicker than the ~1 hydrogen station for 2,000 cars needed.

"but it is obvious that we can manage it [infrastructure] perfectly well."

Sure, and from the Palo Verde Nuclear plant, at 3 gigawatts in 1988, it is obvious that . . .

And from the Concord SST, in 1969 it is obvious that . . .

It is obvious that the future is not guaranteed by one step, however big.

Indeed, it is being pumped about... in very large pipelines, among a limited number of sites, across rather short distances (180 miles is tiny compared to the span of the natural gas distribution system).

The large size and relatively short span reduces pumping losses.  The small number of branches/sites allows a high level of inspection and repair activity to catch leaks.  None of this scales up to "hydrogen economy" sizes.  The kicker is that this network is explicitly for an application (chemical processing) where there is no substitute for H2.  Vehicular propulsion is not one of those applications.

It's a 1 billion SCFD pipeline, and from the construction pics it's about half as big in diamater as a man is tall.  It's not small.

The network does not appear intended for continuous traffic; it's there to interconnect regions which appear to be mostly self-sufficient except for intermittent outages.  The value of the pipeline is to allow all other processes to continue to run even if there is a temporary outage in something which supplies or consumes H2.  This would allow the processes to continue to run rather than requiring a shutdown and restart just because something else paused.

The problem with long-distance traffic in H2 is that the per-unit losses in pumping are far higher than even natural gas.  I've lost my references, but if memory serves the losses due to pumping power in NG transport can be 20% or more end-to-end.  H2 could get close to triple that.  By comparison, HVDC is 3% per 1000 km.  Even coal is being used to generate electricity right at the mine mouth; moving H2 on a continental scale just doesn't make sense.

Ah! Found an article about piping hydrogen!
Full of hairy equations again:
http://www.hydropole.ch/index.php?go=hydrogen_storage

'Hydrogen can be transported in pipelines similar to natural gas. There are networks for hydrogen already operating today, a 1500 km network in Europe and a 720 km network in the USA. The oldest hydrogen pipe network is in the Ruhr area in Germany and has operated for more than 50 years. The tubes with a typical diameter of 25-30 cm are built using conventional pipe steel and operate at a pressure of 10 to 20 bar. The volumetric energy density of hydrogen gas is 36% of the volumetric energy density of natural gas at the same pressure. In order to transport the same amount of energy the hydrogen flux has to be 2.8 times larger than the flux of natural gas. However, the viscosity of hydrogen (8.92·10-6 Pa s) is significantly smaller than that of natural gas (11.2·10-6 Pa s).'

And:
'The transmission power per energy unit is therefore 2.2 times larger for hydrogen as compared to natural gas. The total energy loss during the transportation of hydrogen is about 4% of the energy content. Because of the great length, and therefore the great volume of piping systems, a slight change in the operating pressure of a pipeline system results in a large change of the amount of hydrogen gas contained within the piping network. Therefore, the pipeline can be used to handle fluctuations in supply and demand, avoiding the cost of onsite storage.'

Talking about the great length presumably indicates that the 4% loss is for the 1500km Ruhr pipe, although it is far from clear.

So at least on the limited data I have been able to dig up, it seems that perhaps the losses you thought were the case are far too high.

If you manage to dig up anything better, let me know!
In the meantime I will let you play with the equations.
I will simply contemplate them from afar!

The EIA page you cite refers to Table XVIII for the pumping energy requirements, but it neither includes nor links to the table.  Given that, I'm not sure what the figures are supposed to come from, or even what they are.

The EIA figures on NG consumption for transport includes consumption by NGVs, but does not include NG burned in electric powerplants to power electric compressors.  The 686 BCF figure is a floor, and a rather low one.

I have found a paper on natural gas in the CIS:
http://www.oxfordenergy.org/wpcms/wp-content/uploads/2011/08/NG-531.pdf

I have not fully studied it, but Table 2 looks interesting.
it shows domestic supply as 662 of whatever units they are using, I have not spotted what that is, distribution losses as 11 and pipeline transport as 49, with the last presumably pumping etc.

That is around 1.6% for losses, which correlates well with the EIA's figure, and around 7.4% for the pumping etc.
That would put the total at about 10% energy cost.

2.2 times that would come to around 22% if it were hydrogen being transported that distance, which is a pretty big figure but then again the CIS is pretty big, and I can't think of many reasons why one would want to pump hydrogen so far, at least if you are producing it from natural gas.
It would be less lossy to transport it as NG and convert nearer to where you want it.

If you were using electrolysis to produce hydrogen presumably you would just transport the electricity or generate it by solar or whatever near to where you needed the hydrogen.

Hmm, not sure I have adequately allowed for hydrogen losses.
Clearly the CIS natural gas piping system is huge. Page 30 of the CIS document shows the Gazprom controlled part of it as 611,800 km, a totally different matter to the Ruhr hydrogen network at 1,500 km which apparently looses 4%.

I can't see much reason why hydrogen would normally need to be pumped huge distances or have anything like as extensive a network as it is not going to every home, but tomorrow I will see if I can get hold of any better figures on losses.

'Hydrogen leakage was assumed to occur during electrolysis of water, hydrogen compression, hydrogen storage at the filling station, vehicle fueling, in-vehicle hydrogen storage, in-vehicle flow through the fueling system, and hydrogen usage in the fuel cell stack. Studies have suggested a future hydrogen leakage rate of 3%, since the rate of natural gas leakage today is about 1% and that since hydrogen is a smaller molecule and more permeable than methane [Schultz et al., 2003; Colella et al., 2005]. Here, the leakage rate was similarly assumed to be 3%, the upper limit considered by Zittel et al.[1996] and Shultz et al. [2003]. This leakage rate is lower than that used in Colella et al.[2005], since they used a 10% leakage rate to ensure a conservative result, recognizing that the real leakage rate is most likely 1-3%.'

http://www.stanford.edu/group/efmh/jacobson/Articles/I/HydGlobGRLSOM0808.pdf

They are not specifically talking about pipeline transport, but perhaps the tendency to leak is the same.

So the best guess I can put on things is that the leakage rate might be around 3 times that of NG, and the pumping losses 2.2 times.

Not good, but not disastrous either.
Clearly efforts would be made to minimise losses by pumping natural gas and then converting it nearer point of use, and the same for electrolysis.

EP:
I find that assessment a real stretch.
Lots of things aren't simple, but that doesn't mean they can't be done.

What we were assessing is piping hydrogen.
It is clear that it is technically possible at reasonable losses, and that for the past many years we have been accumulating expertise, considering there is no reason to build a comparable network to natural gas.

If you are using electrolysis there is even less reason to be pumping the hydrogen vast distances, as it could be made anywhere you fancy if you provide the electricity, as is actually being done in these portable hydrogen stations:
http://www.itm-power.com/products/

Solar can of course also be used.

The comment about some imagined ill effects of releasing hydrogen sounds frankly completely random, as the releases would be trivial compared to the natural water cycle.

At least once you start running the car, a fuel cell and an ICE emit comparable amounts of water, the only difference being that the emissions from the ICE contain lots of nasties as well as water, and those from fuel cells don't.