Supplementary Material to
“Fossil resource and energy security dynamics
inconventional and carbon-constrained worlds”
inClimatic Change
David L. McCollum1*, Nico Bauer2, Katherine Calvin3,Alban Kitous4^, Keywan Riahi1,5
1International Institute for Applied Systems Analysis, Laxenburg2361, Austria.
2Potsdam Institute for Climate Impact Research, Potsdam 14412, Germany.
3Joint Global Change Research Institute, College Park, MD 20740, USA.
4Institute for Prospective Technological Studies, European Commission Joint Research Centre, Sevilla 41092, Spain.
5Graz University of Technology, Graz 8010, Austria.
*e-mail:
^The views expressed are purely those of the author and may not in any circumstances be regarded as stating an official position of the European Commission.
Table of Contents
1.Additional information on the fossil fuel supply curves used by a subset of the models in this study
2.Additional information for the fossil resource consumption and price dynamics analysis
3.Additional information for the CO2 storage requirements analysis
4.Additional explanations and clarifications for the energy security analysis
References
1.Additional information on the fossil fuel supply curves used by a subset of the models in this study
As part of the Energy Modeling Forum 27 (EMF27) exercise, we collected information on producers’ cost curves for many of the models participating in this study. These curves provide information on fossil fuel availability and the cost of extraction as used by the models. Supplementary Figure 1shows these curves for coal, natural gas, and crude oil. For presentation purposes, we have truncated these curves around the levels of energy consumed over the course of the twenty-first century; to be sure, many models include significantly more fuels at higher costs. The curves shown are base-year costs in 2005$ per GJ. Some of the models, however, assume that costs change over time. GCAM and IMAGE, for instance,account for declining costs over time as a result of technological progress in mining/extraction; in GCAM this decline is exogenous, while in IMAGE it comes in the form of a learning curve. MESSAGE, on the other hand, assumes that costs will change in the future due to changes in both the cost of labor (which could result in cost increases) and technological progress. MERGE assumes implicit technical change in terms of the size of the ultimately recoverable resource (which scales the long-run cost curve), as it includesexpanded categories from the U.S. Geological Survey(USGS) assessment. It is important to note that model-realized prices of fossil fuel resources may differ from producer cost, due to the inclusion of other factors (e.g., transportation cost, taxes, subsidies, scarcity rent, monopoly rent, etc.). The difference between producer cost and price varies across models and can be significant.
(A) COAL/ (B) NATURAL GAS
(C) CRUDE OIL
Supplementary Figure 1. Global fossil fuel supply curves (coal, natural gas, crude oil) for selected EMF27 models.
2.Additional information for thefossil resource consumption and price dynamics analysis
The challenge of climate change mitigation crucially depends on the future consumption of fossil fuels, namely coal, oil, and natural gas. First, the use of these resources over the course of the twenty-first century absent any climate policy is important because it determines the CO2 emissions baseline from which reductions must be achieved. A second important factor relates to the flexibility of the energy system to reduce the consumption of freely-emitting fossil fuels, as this influences the effort required to deviate from the baseline. Thirdly, the future price levels of fossil fuels are a crucial factor because they will influence the opportunity costs of not using fossil fuels, as well as the deployment of non-fossil technologies, if there are concerted efforts to abate CO2 emissions. Finally, climate policies will lead to a restructuring of fossil fuel markets, in terms of expenditures on these resources as well as on carbon.
Supplementary Figure 2shows the cumulative consumption of coal, natural gas and oil over the 21st century for various scenarios and models of the EMF27 exercise. The figure supports the resultsdiscussed in the main text regarding the consumption of all fossil energy carriers.
The detailed results show that the reductionsin coal use (from the Base FullTechreference scenario) are the most necessary, in order to achieve the two different climate change stabilization targets. This effect is already significant for the 550 ppm CO2-eq target (550 FullTech). The additional reductions (from 550 ppm) to achieve the 450 ppm target (450 FullTech) are not quite as great. Moreover, the use of coal in combination with CCS is reduced for all models.
In addition, oil and gas consumption must be reduced to achieve the climate targets. The effect on total gas use is very mixed, but a majority of models show more significant use of gas with CCSin the 550case than in the 450 case. Similarly, oil use across the different scenarios is very mixed. Reductions are clearly larger in the case of high baseline consumption. The more optimistic assumptions regarding energy intensity development (Base LowEI scenario) are generally not sufficient to reduce the use of coal, oil, and gas enough so that the climate targets are met. For some models, cumulative oil and gas consumption in the 550 scenario is even higher than in the no-policy scenario of some other models. This result does not hold for coal, which shows a much more robust reduction. It is also worth noting that all models reduce the use of fossil fuels in the lower energy intensity scenario – i.e., there is no excessive rebound effect for a single fossil fuel in any of the models.
Supplementary Figure 2. Cumulative consumption of fossil energy carriers (2010-2100) across different scenarios and models. For gas and coal, the quantity of consumption in combination with CCS is also shown. For oil, the quantity with CCS is negligible in all models and therefore not shown.
Regarding global results (see Figure 1 of main text), in the 550 ppm CO2-eq climate stabilization scenario, the highest cumulative fossil resource use is reported by GRAPE (55 ZJ), which consumes a large share of its fossils in combination with CCS. GCAM and IMACLIM similarly consume large quantities of fossil fuels (45 ZJ) in this climate policy scenario, whereas MERGE (34 ZJ) and EC-IAM (22 ZJ) are on the low end. Hence, the range of cumulative fossil fuel consumption across models is less in the 550 FullTech scenario (34-45 ZJ) than in the Base FullTech (55-80 ZJ), when excluding the lowest and highest values in the former set; otherwise, the variations are similar to the baseline case. The cumulative amount of fossil fuel with CCS ranges between 2 ZJ (EC-IAM) and 28 ZJ (GRAPE) in the 550 FullTechscenario. When the stabilization target is tightened to the 450 CO2-eq level (450 FullTech[1]), the lowest total fossil fuel use is 20 ZJ (MERGE). In contrast, the highest value is roughly twice that level (GCAM with 39 ZJ). For all models, the use of fossils with CCS is generally lower in the 450 scenario than in the 550 case. However, there is significant variation across models in the amount of fossil fuels that can be consumed while still achieving this more stringent climate stabilization target.
What is perhaps most interesting about the reductions in fossil resource use across models in the climate policy scenarios (450 FullTech and 550 FullTech) is that they are somewhat independent of fossil use in the baseline (Base FullTech). For instance, the four models with the highest fossil consumption in the baseline (BET, MERGE, ReMIND and WITCH) simultaneously show relatively lower fossil fuel consumption in the climate policy scenarios (see Figure 1 in the main text). In particular, these models make less use of fossil CCS. Hence, the models that follow a fossil fuel-intensive pathway in the absence of emission restrictions also exhibit high flexibility in reducing fossil consumption when the need arises. Alternatively, certain models that exhibit lower fossil use in the baseline (GCAM, POLES and IMAGE) show smaller reductions in the stabilization scenarios and, thus, relatively high fossil fuel use in the latter. These trends can be attributed to differences in the flexibility of models to reduce energy demand and to substitute other fuels in the energy supply, which itself has much to do with model structure . (GCAM, POLES, and IMAGE are simulation models, for example.) Some models rely more on energy demand reductions to achieve climate policy targets than others, while others are more flexible in transforming the energy supply mix. In all models and for all scenarios, fossil fuels will be consumed up to the point where other alternatives, like renewables, become cost competitive. However, models differ in their ability to anticipate future energy prices, something that heavily influences energy consumption in the near term. For example, MERGE, ReMIND, and WITCH are all examples of inter-temporally optimizing models; thus, they make decisions about energy use today in anticipation of future carbon and energy prices. GCAM, in contrast, has limited foresight; it therefore assumes actors make decisions without full knowledge of future energy and carbon prices. The difference between these two modeling paradigms is one reason that fossil energy consumption in GCAM (and models like it) is higher, particularly in the near term.
The changes in the global energy system also have impacts on fossil fuel prices. This is shown most clearly in Figure 2 of the main text. Prices are seen to vary quite widely across models, even for the same level of cumulative fossil consumption, and in general prices vary more than consumption. In the baseline, models with high initial prices tend to show high price increases, and the other way around. On the other hand, the price response due to climate policies is ambiguous across models. Some models show no effect, others estimate lower prices under stringent climate mitigation. Still others show higher prices. The latter behavior is particularly interesting, as for no single model is the cumulative consumption of oil or gas resources in the climate mitigation scenario higher than in the baseline case.
The marked differences in price developments across models depend to a large extent on model structure and assumptions. For example, some models, such as POLES[2], calibrate initial prices to historical data, while others, like ReMIND, determine it endogenously. Second, inter-temporal optimization models like ReMIND – compared to recursive-dynamic models like GCAM and IMAGE – inherently include a price mark-up over extraction costs through anticipation of future supply scarcity and/or increasing production costs. For the same level of cumulative resource consumption, this mark-up ends up being higher in the baseline than in the climate mitigation scenarios. Third, the inclusion of short-term supply constraints and energy trade costs has an effect on prices in some models. Finally, as mentioned above, supply curve assumptions vary widely across models, and this has an important effect on resource prices at any point in time.
Supplementary Table 1 summarizes differences in fossil fuel price impacts across models for the year 2050. The ratio of the relative change of annual fossil use and the global average price can be interpreted as a market elasticity of fossil fuel output with respect to the price due to climate policies. A ratio of between 0 and 1 means that the percentage reduction of prices exceeds the percentage reduction of the quantity. If the ratio is negative, the price and quantity reactions move in opposite direction. The table illustrates the diversity of model results. In fact, there is no clear tendency. For each fossil energy carrier, the results cover the whole range of possible values. Also, only a few models show results for all three fossil fuels that are in sound relation to each other. More analysis is required to shed light on the reasons for these outcomes, and more fundamental research is needed to improve the representation of fossil fuel markets in models. The latter will hopefully help to narrow the range of model outcomes and thus reduce the uncertainty communicated to policy makers.
Supplementary Table 1. Percentage changes of coal, oil and gas global annual use and global average prices in the year 2050 and the ratio of the changes. The changes refer to the differences between the 450FullTechclimate stabilization scenario and the Base FullTechscenario in the year 2050. Note that global average prices are used. Regional reallocation of coal and gas could imply increasing average prices because of regional price differences.
Oil / Gas / CoalQ / P / Q/P / Q / P / Q/P / Q / P / Q/P
GCAM / -2% / 2% / -1.15 / -7% / -3% / 2.50 / -40% / -14% / 2.85
IMACLIM / -27% / -80% / 0.34 / -11% / -61% / 0.19 / -46% / -84% / 0.55
IMAGE / -30% / -7% / 4.52 / -19% / -6% / 3.24 / -38% / -4% / 10.29
MERGE / -15% / -44% / 0.34 / 4% / 18% / 0.20 / -60% / 0% / NAN
MESSAGE / -25% / -20% / 1.28 / -8% / 11% / -0.71 / -53% / -11% / 4.96
POLES / -21% / -54% / 0.39 / -22% / -55% / 0.40 / -52% / -10% / 5.32
REMIND / -7% / -28% / 0.25 / -23% / -31% / 0.75 / -80% / -69% / 1.17
TIAM-WORLD / -16% / -3% / 5.12 / -16% / 28% / -0.57 / -62% / 8% / -7.76
WITCH / -24% / -31% / 0.79 / -33% / -4% / 8.32 / -55% / -10% / 5.57
3.Additional information for theCO2 storage requirements analysis
Carbon dioxide is a waste product of fossil fuel consumption; and in the context of carbon dioxide capture and storage (CCS), the potential for disposing of this waste is of critical importance. One can even think of the world’s geological storage capacity for CO2 as a kind of “resource” whose utilization depends on the amount of fossil fuel consumed over the next decades, or perhaps centuries. The availability and geographic distribution of such reservoirs will become increasingly important in a carbon-constrained world, and concerns could eventually surface regarding the potential of this resource to permanently sequester the huge flows of CO2 that may be required going forward. Supplementary Figure 3 compares the cumulative(2010-2100) volumes of CO2 stored (regionally and globally) by the EMF27 models in the 450 FullTech climate stabilization scenariotoseveral estimates of carbon storage capacity from the literature. These estimates indicate a large uncertainty in capacity. For instance, the International Energy Agency (2010) has calculated that, theoretically, global storage capacity could be as high as 16,000 GtCO2; however, when restricting this capacity to what is actually deemed viable, the estimates are found to be much lower: 3,360 GtCO2 in the “20% Viable” case and 1,680 GtCO2 in the “10% Viable” case). Meanwhile, a more recent analysis conducted within the framework of the Global Energy Assessment (Benson et al. 2012) estimates that global capacity may actually be far higher: between 5,000 GtCO2 and 24,000 GtCO2. (Note that Supplementary Figure 3 only shows the lower estimates from the IEA and GEA.)
By the end of the century, EMF27 model results point to global CO2 storage requirements of around 1,300 GtCO2 (median across models), with the highest model requiring approximately 2,400 GtCO2. Thus, under most estimates, enough storage capacity exists to meet demand in the twenty-first century. However, should carbon capture and storage be needed post-2100 – and indeed this seems to be the trend in essentially all models, if one further extrapolates the scenario results into the twenty-second century – then CO2 reservoir constraints could potentially become scarce, depending on the resource potential considered.[3] The situation is less clear-cut at the regional level: whereas the OECD90 appears to have plenty of storage capacity, the other regions (especially ASIA) could approach, or even exceed, their capacity limits. To some extent, the global picture is distorted by the huge surplus of capacity in the OECD90, relative to the projected future size of this region’s energy system. It is not clear, however, that this surplus could be shared with the other more capacity-limited regions, as this would depend on the transport of CO2 over long distances via pipelines or ocean-going vessels. In sum, this comparison shows that concerns over the future scarcity of CO2 storage reservoirs are not unwarranted if one takes a fairly pessimistic view of the capacity estimates currently found in the literature. Yet, even with a bit of optimism, these concerns are alleviated, at least over the twenty-first century.
Supplementary Figure 3. Cumulative CO2 storage requirements (2010-2100) across modelsin the 450 FullTechscenario compared with storage capacity estimates from the literature. Minimum, maximum, and median values across all models are shown. (Note: we have excluded the GEA estimates from the regional charts since different methodologies were used to develop the different regional estimates. As a result, the estimates are not necessarily comparable across regions.)
4.Additional explanations and clarifications for the energy securityanalysis
Shannon-Wiener diversity index
The Shannon-Wiener diversity index (SWDI), referred to in the main text, has been used in previous studies to measure different aspects of energy system diversity(Jansen et al. 2004; Kim et al. 2009; Kruyt et al. 2009; Riahi et al. 2012; Stirling 1994). In this study, we apply the SWDI to (1) the diversity of the primary energy resource supply mix in individual regions, and (2) the diversity of the global markets for oil and gas exports. (Note that in the former the direct-equivalent accounting method is used for nuclear energy and non-biomass renewables.)A number of energy security indicators can be found in the literature (Kruyt et al. 2009), each with their own strengths and weaknesses. The reasons we chose to work withthe SWDI are multifold. First, it is analytically straightforward and can be reproduced by others fairly easily. Second, it offers a way to measure energy system resilience, a key component of security. Third, unlike compound indicators that consider multiple policy goals, the SWDI does not obscure policy trade-offs. And lastly, it does not require implicit assumptions about the stability of future political regimes or the implicit weighting of future risks.