Peak coal?

All resources are finite, and we tend to go for the easily extractable stuff first. Therefore, as we start to exhaust our reserves, production will tend to taper off, with the law of diminishing returns demanding ever more physical, economic and technological effort to access what’s left. In this sense, the concept of the ‘peak’, most commonly used in discussions of ‘peak oil’, is a conceptual no-brainer. Where things become controversial is that the methods used to calculate the timing of the peak, the remaining extractable reserves, and the post-peak production trajectory, often rely solely on extrapolating from the history of past production, taking few cues from geological knowledge, uncertainties about future economics, and the prospect of future technological advances. Or, more accurately, the ‘peakist’ argument seems to be that these factors have little long-term effect, and that the mismatch with generally more optimistic geologically based reserve estimates is a feature, not a bug.
It’s unsurprising that most analyses of this sort have focussed on oil, but you can theoretically do the same for any other resource, including, as has just been reported in Science, coal. David Rutledge of Caltech is one of those who has crunched the numbers, and just as with oil, has found that this method points to there being rather less remaining than more conventional reserve estimates would have you believe:

Applied to 14 major coal-producing regions, Rutledge’s method gives a world ultimate production of 660 billion metric tons. That’s only one-quarter of geologic estimates of ultimate production, he says.

As to when coal will peak, Rutledge declines to say, citing the way peak timing varied widely among regions already well past their peak. He will say, however, that in his projection the world will have produced a whopping 90% of its coal by 2069. Physicist Mikael H????k of Uppsala University in Sweden and his colleagues are willing to point to a peak. They have taken a similar approach to Rutledge’s but with some reliance on estimated reserves. Still, they see world coal production topping out by 2020, entering a 30-year-long plateau, and then declining.

If true, or even close to true, these conclusions have some rather thought-provoking ramifications. The standard estimates give us somewhere in the region of 150-200 years’ worth of coal left in the ground. On the one hand, this gives our fossil fuel-addicted economies a fall-back as oil and gas reserves are depleted; on the other hand, the impact of actually using all that coal on atmospheric CO2 levels, and hence our climate, are rather scary (the as-yet-unproven potential of carbon capture notwithstanding). If we do start running out of coal sooner rather than later, then raw economics – a force seemingly much more powerful than either our consciences or our sense of self-preservation – might force us down a less environmentally damaging path, but our room for manoeuvre in weening our economies off carbon is more severely restricted than we had previously imagined.

Categories: environment, geology

Comments (7)

  1. Miguelito says:

    I’m not convinced.
    Hubbert’s method was sort of good for conventional oil but not nearly as good as it’s portrayed to be. Peak-oil theorists use US production charts to show how prescient Hubbert was with US production topping out in the early 1970s, but those production charts almost always neglect to include Alaska north-slope oil, which ended up helping to create a second peak of US oil production around 1985 almost as big as the original peak and something that Hubbert’s model did not predict. This is important because, back when Hubbert made the prediction, Alaska represented a frontier area and it was, as far as I know, part of the United States. Well, those frontier areas still exist in the world today, not nearly as numerous, but there are places where discoveries can be made that weren’t really being considered back then because of significant technological advances (eg. sub-salt plays, deepwater plays on the continental slope, the Arctic, etc…).
    Hubbert’s method hasn’t considered unconventional oil either (like the oil sands of Canada, Venezuela, and I think there are some in Russia; and let’s not forget Bakken Formation oil and others). And then there’s enhanced recovery techniques, which are doing very well at taking geriatric oil (like the Ghawar field of Saudi Arabia) and squeezing every drop of oil out of it. Hubbert’s methodology does not include this because it models reserves data current to the year of the study and project future discoveries, not future increased recoveries or reserve additions through pool-size delineation.
    Hubbert’s methodology has been terrible at assessing natural gas thanks to additions of unconventional gas. First there was tight gas, now there’s shale gas, which has caused huge growth in US gas production (7.5% in 2008 alone and the market is now glutted) and just last summer there was the first, sustained, steady production test from gas hydrates in the Canadian Arctic (the second production test ever tried and they got steady rates and pressures; it’s pretty impressive). All of this is thanks to technological leaps.
    For coal, it’s early, but there’s still the possibility of alternative methods of extraction of deep coal seams (like gasification), which is being done around the world. It’s not mining per se, but it’s turning coal resources into burnable energy. And, like much of the above, it’s technology taking something unthought of and making energy.
    So, I conclude by saying that Hubbert’s methodology hasn’t worked all that well for other resources in determining production trends. Therefore, I’m not sure how much we can trust it to model coal production accurately. I’m sure my saying that is going to cause a flamewar (because Hubbert is very much an icon to many people and he made a very neat, superficially correct, but contextually wrong prediction), but it’s the truth. And I’ll also conclude by saying that there are plenty of fossil fuels in the world we can use to mess our climate up.

  2. Miguelit says:

    Small mistake in my post: Hubbert’s methodology doesn’t look at reserves, it looks at production. But it’s still a problem because a field in decline can be turned around with enhanced oil production. Things that could never produce before are producing thanks to horizontal drilling and multi-stage fracturing.

  3. Chris Rowan says:

    those production charts almost always neglect to include Alaska north-slope oil, which ended up helping to create a second peak of US oil production around 1985 almost as big as the original peak and something that Hubbert’s model did not predict.
    That’s not entirely accurate. If you look at US oil production, Alaskan production effectively slowed the decline from the 1970s peak, producing a ‘plateau’ through the early 1980s. This suggests that, in the case of oil at least, the impact of new discoveries/technology can be exaggerated – it seems that they can help to maintain high production for a period (and presumably increase the total amount extracted), but not really yield a net increase.
    More important is the point that, whatever the relative weaknesses of the peak calculations, many geological reserve estimates may also be less than robust (or, at best, tend towards the unrealistically optimistic). If this debate only serves to demand more rigour in more ‘realistic’ abundance estimates, then it’s useful.

  4. Lab Lemming says:

    Is peak coal like mountaintop mining?

  5. Kieran Sheehan says:

    Oil production is actually the key here, not the amount left in the ground. If we extract an energy resource then, to be viable, it must yield more energy than that used to extract it. The law of diminishing marginal returns works here too: it is getting increasingly energy-expensive to extract oil (and coal), therefore, once the energy cost of a barrel of oil (tonne of coal) is the same as the energy used to extract it, then the resource is useless as source of energy, no matter how much is left in the ground.
    If someone could impove the Energy Return on Energy Invested (EROEI) for oil and coal then that would solve the energy crisis in the short term – in the longer term, however, do not forget that law of diminishing marginal returns….

  6. Nick says:

    Hubbert based his analysis on production from a single field, I believe, but explicitly stated that his analysis was an analog of sorts for global production.
    My understanding is that it was deep water Brazil in the 80’s that “shifted” Hubbert’s peak forward. Presumably the arctic will do so in the ’10’s.
    However, the issue is really about costs (as Kieren pointed out). I read somewhere that Exxon’s New Hebrides platform operates at a 98$/barrel break-even point. So, that’s the low end of mega-deep water operations which presumably will need to operate to meet demand in the near future. Similarly, Iraq – with 110 bb sitting where we can cheaply get it – isn’t a cheap operation …

  7. Eric the Leaf says:

    Anyone can “do a Hubbert” using the statistical techniques outlined by Kenneth Deffeyes. I have done a “Hubbert Linearization” for all US oil production up to 2005 that yielded a peak of 1976 (actual peak was 1971). A similar analysis for world oil production yields a peak of 2008, which is so far correct. Examining the equations, one sees that oil production is linearly related to the quantity of oil yet to be discovered. As for coal, look for Richard Heinberg’s forthcoming (July 2009) “Blackout: Coal, Climate and the Last Energy Crisis.”