Skip to main content

Coal at its peak?

Hydrocarbon Engineering,

In the most facile sense of the word, the answer to the peak coal question is a resounding “yes.” The planet contains a finite amount of coal; the stuff is laid down over geological time, and there is not even a slightly plausible model of real-time resource replenishment to serve as a counterpart to the dubious theory of “abiotic” petroleum.

However, the peak resource question is more properly posed in the context of when economically exploitable reserves will hit half their initial levels, or when production rates will start to fall following a historical maximum. Under the model similar to that first applied by Hubbert to petroleum production, these two are essentially the same thing. For coal markets, such questions have been asked and answers have been given using various methodologies and with various results since at least the mid-19th century.1

It needs to be noted that context is crucial for such an assessment. For example, at a carbon credit price or tax of US$ 100/t of CO2e, coal use may effectively be phased out; peak production will have already occurred, and the remaining resource could stay in the ground indefinitely. At US$ 200/t or more, however, abatement technologies such as carbon capture and storage (CCS) become more economically viable and clean coal may fall into favour over alternatives; therefore production might surge and the peak in production could be moved to the relatively near future, with reserves becoming greatly diminished by the end of the century. In most specific cases, the path of production over time cannot be calculated from first principles: rather, economic, social and political circumstances intervened at various points to sustain, revive or curb production.

Peak concerns

For several decades leading up to the early 2000s, international coal trade had, relatively speaking, hit a plateau. Consequently, assessment of whether peak coal was imminent was not a front-burner issue. However, with a surge of trade since the mid-part of the last decade, driven particularly by large and quickly growing developing markets in Asia and elsewhere, the commonly agreed-upon idea of a coal production horizon stretching into the 22nd century has begun to undergo closer scrutiny. This has been disconcerting in many respects, especially since several major outlooks and indicators – not only those of the International Energy Agency (IEA) and the Energy Information Administration (EIA), but also the UN’s Intergovernmental Panel on Climate Change (IPCC) models on expected greenhouse gas (GHG) concentrations – have been driven in part by the assumption that coal use will increase through at least 2030.

In addition to the consumption boom, some highly-publicised cases of reserve overstating have also helped the idea of near or imminent peak global coal production gain traction. Questions as to the robustness of global reserves estimates have increased since at least 2004 when, over one year’s reporting time, the Government of Germany reduced the nation’s reserves levels by more than 99%.2 It has also been noted that reserve levels reported for certain countries, such as China, have remained static since the turn of the century or longer, even though significant coal production has occurred since the last revision.

Revisions and revisiting of the global coal reserve issue have been numerous. In 2007, the World Energy Council estimated that global peak coal could occur as soon as 2020.2 In 2010, Tadeusz Patzek and Gregory Croft conducted a Hubbert analysis of historical production in major coal producing nations, along with an estimate of likely reserves in major undeveloped regions, such as Alaska’s North Slope and Mongolia’s coalfields, and predicted that worldwide peak coal production would occur – around 2011!3 Also in 2010, David Rutledge published an estimate of future levels of coal production based on a Hubbert-type regression of production histories in various world regions.4 Although Rutledge did not derive a peak year, he estimated that 90% of total economically recoverable resources worldwide would be extracted by 2070.

There is reason, however, to believe that, while standard measurements of economic reserves are consistently overstated, use of predictive tools originally applied to petroleum and gas production under well-understood economic and political conditions may severely underestimate reserves. Certainly, results such as those from by Patzek and Croft (who gave the 2011 date as an estimate, not as a guarantee) seem counterintuitive and not borne out well by news. The increase in thermal coal trade volume in 2012 relative to the prior year, together with the persistence of oversupply in certain markets and falling index prices, are not altogether indicative of global post-peak conditions.5 In any event, if an actual peak in production did occur then, it would likely not become apparent until several years down the road.

Hubbert methodology

The primary mathematical assumption behind the “Hubbert peak” is the central limit theorem, which can be used to show that the overall output over time obtained by adding a large enough number of essentially random production curves (representing the output over time of oilfields, coal mines or any other sources of finite resources) will tend to occur as a “bell-shaped” standard distribution (Figure 1). This formulation was first applied in the field of fossil fuel production by M. King Hubbert, a US geophysicist who in 1956 predicted that oil production in the mainland US would peak between the late 1960s and the early 1970s (as it turns out, production in the lower 48 states seems to have peaked in 1970).

Figure 1. Hubbert curve drawn by M. King Hubbert, circa 1956, to illustrate projected world oil production. Source:

While Hubbert-type peak analysis has since been applied as a predictive method for determining peak and ultimately available reserves worldwide for petroleum, gas and coal, it has been pointed out that his original methodology was underpinned by a number of implicit assumptions:6

  • The overall producing population consists of a large number of small wells or fields. Where this is not the case, aggregate production will begin to look much less like a symmetric bell curve, and mathematical prediction of specific peak dates becomes untenable.
  • Exploration follows a smooth pattern, “unimpeded by political events or by significant economic factors.”6 As discussed above, it is often the case that future production will be interrupted, slowed or jump-started by inherently unpredictable factors.
  • Finally, it should be understood that basic Hubbert analysis does not work well in cases where there are newly discovered reserves or reserves that are not yet economically proven. Patzek and Croft, for instance, were not able to fit their model global coal production with a simple bell curve, but were forced to accommodate major undeveloped coalfields by adding a secondary, subsidiary production peak somewhere toward the end of the current century (Figure 2).

Figure 2. Growth of CO2e emissions under IPCC scenarios (coloured lines) and under Hubbert curve fit by Patzek and Croft (black line).  The secondary peak around 2100 is from production and use of resources that are currently not proven. Source: Patzek and Croft (2010).

Reserve estimate methodology

It has been seen that estimated economic coal reserve figures have proven to be notoriously volatile. In such cases, it is generally agreed that the adjustments tend to involve downward revision. Because proven reserve figures are assembled in a variety of ways and assessed by differing criteria, there will necessarily be large inherent sensitivities to the global reserves figure. Coal industry restructuring within a country, or adoption of a new system of reserves assessment, can also lead to sudden discontinuities or downgrades.

On the other hand, there exist compelling structural reasons to believe that the coal industry as a whole has not definitively estimated its reserves and resources and, indeed, that many private mine reserves – which still constitute the majority of reserves worldwide – are underestimated. Privately-owned companies (which do not have to publish overall asset balances) will not immediately prove all of their resources; instead, they tend only to define and publish reserve estimates pertinent to the specific business plans, which often only cover the lifetimes of specific mines. To a certain extent, this situation will change with the continued rise of large and publicly-traded international coal companies but, at present, smaller and more locally-oriented operations still own the majority of reserves worldwide.

Mining operations generally strand a high level of resource in place owing to factors such as depth, low calorific value (CV) or other suboptimal physical characteristics, and seam thinness. As has been demonstrated repeatedly in the oil and gas industry, it is likely that, as traditional reserves deplete, the ratcheting effect of higher prices and technological innovation will allow. for recovery of previously unmineable resources. On the demand side, boilers and processing systems are now being constructed to use off-spec, low CV or high ash coals, and higher efficiency power plants will be able to stretch the electricity produced from lower quality fuel consumption. New exploration technologies, such as enhanced mapping, are granting access to more fractured and deep seam systems. Finally, the emergence of new paradigms for power generation including underground coal gasification (UCG) could provide the capability to multiply reserve levels in many regions by allowing access to deep reserves or even reserves beneath the sea bed.7

There is little doubt that there still remain large overall global resources of coal. These resources remain finite, however, and it is not unreasonable to attempt to apply the best available tools to determine how much longer, and at what price, they might be expected to be viable. Nevertheless, based on industry structural differences, there are convincing reasons to believe that methods used to assess coal and gas or petroleum resources are incompatible. Although some of these uncertainties and differences in approach may indicate that reserves are overstated and peak coal will occur sooner than expected, other indications are that coal reserve levels stated by mines may actually understate accessible resource, and that, given the right economic and technological scenarios, resources that have so far been uneconomic because of factors such as depth or seam thinness will be utilised.


  1. See, for instance, JEVONS, W., The Coal Question (London: Macmillan & Co., 1866); and HULL, E., The Coal-Fields of Great Britain, (London: Edward Stanton, 1861).
  2. Energy Watch Group, Coal: Resources and Future Production (2007).
  3. PATZEK, T. and CROFT, G. “A global coal production forecast with multi-Hubbert cycle analysis”, Energy (15 May 2010).
  4. RUTLEDGE, D., “Estimating long-term world coal production with logit and probit transforms”, International Journal of Coal Geology (4 November 2010).
  5. See, for example, IHS McCloskey Coal Report 286, “Oversupply plagues thermal coal market” (1 June 2012).
  6. See, for example, LAHERRÈRE, J.H., The Hubbert Curve: Its Strengths and Weaknesses available online at:
  7. For further discussion of structural and technological drivers of coal reserve levels, see Energy Edge Ltd, Coal of the Future (Supply Prospects for Thermal Coal by 2030-2050), prepared for the European Commission – DG JRC Institute for Energy (2007) available at


Linus Adler is an analyst for Energy Edge Ltd, a cross-commodity strategic energy and power markets consultancy based in the UK.

Read the article online at:


Embed article link: (copy the HTML code below):