Existing CCS Modeling Structure and Parameters in GCAM
For this study we used a version of the Global Change Analysis Model (GCAM) version 7.0, which is a technology-rich representation of climate and global energy, land, and water systems coupled to a physical Earth system model of atmosphere, oceans and terrestrial systems.56,67 GCAM features representation of carbon capture and storage (CCS) technologies for electricity generation, refining, hydrogen production, industry, and direct air capture with carbon storage (DACCS).56,68 In GCAM’s extant configuration, cumulative, graded resource supply curves for onshore CCS are based on Dooley and Friedman (2005) estimates for available CO2 storage volume in coal and gas basins, depleted oil plays, and deep saline aquifers.40 Onshore carbon storage supply curves from GCAM 3.0’s 14 regions (Table 1) are downscaled to the 32 regions in GCAM 6.0 on the basis of their relative land area. The CCS resource is split into 4 distinct grades, with the lowest grade costing ($0.10 per tCO2 encompassing 0.5% of the resource in each region, and increasing from there, with 60 percent of onshore storage available at costs below $10 per tCO2.41,69 Offshore storage is assumed be an unlimited resource where cost is a larger barrier to deployment than physical limits on storage availability. The offshore storage cost estimate of $96/tCO2 is not intended to serve as an exact point estimate but rather to represent a backstop reservoir for CCS when regions exhaust their land-based storage. Therefore, a conservative estimate is used (several times the $32/tCO2 estimate from Decarre et. al., 2010)70 owing to the large uncertainty of both offshore and onshore carbon storage costs and availability. Figure 7 summarizes GCAM’s existing volumetric carbon storage supply curves by aggregated groups of its major regions.71 For sensitivity analysis purposes, GCAM’s data system also produces optional input files where the assumed cost of a given quantity of CO2 storage can be scaled up by factors of 3 or 10, for more conservative estimates, or down by 20 percent for more optimistic estimates.
For this study, we extended GCAM’s “resource / reserve” modeling structure for depletable resources (oil, natural gas, coal, and uranium) to geologic carbon storage. Under this approach, as the market price of the resource increases, the model looks up the supply curve to determine the additional quantity available and moves that quantity of “resource” into a reserve” and assumes that reserve is produced over the lifetime of the well or mine. We assume a CO2 injection well lifetime equivalent to that of natural gas wells (30 years).5757 Initial analysis revealed this additional modeling capability slightly increased CCS cost and lowered deployment, but did not substantially affect the top-line results with respect to CCS under deep decarbonization scenarios.
Estimation of Injection Rate Constraints
For the development of dynamic CCS supply curves, we used estimates for cost and injection rates of over 680 formations from the U.S. National Energy Technology Laboratory’s (NETL) Saline CO2 storage cost model.58 This is a spreadsheet-based tool that estimates formation injectivity using simplified geologic engineering equations, then calculates first-year breakeven CO2 storage price for U.S. Environmental Protection Agency (E.P.A) Class VI injection wells over the project lifetime.7272 Model outputs were rank-ordered by cost to derive an upward-sloping supply curve of annual injection rates for the United States (Fig. 7). Like GCAM’s existing cumulative storage parametrizations, most of the storage is available at low cost, but annual injection rates are limited to a maximum to approximately 2.5 GtCO2-yr− 1, at which point additional injection becomes highly costly. We selected 7 points this curve to avoid excessively large input file size. These U.S. supply curve quantities were then linearly scaled to GCAM’s remaining 31 model regions based on their maximum volumetric production rate of oil and gas since 1971 at subsurface conditions (ρCO2 = 700 kg-m− 3, ρoil = 800 kg-m− 3, ρgas = 150 kg-m− 3), following an approach suggested by Lane et. al (2021)15,73 We translated the resulting rate-based supply curves downward by approximately $14 such that their lowest point is equal to zero to harmonize with GCAM’s existing cumulative supply curves and avoid double-counting of costs. This newly created, regionally explicit dynamic storage supply resource is consumed by both onshore and offshore carbon storage technologies and serves to restrict the maximum rate at which volumes (and therefore mass) of CO2 can be injected in any given GCAM region. To test the implications of CCS limits that do not depend as heavily on historical oil and gas production experience, we also developed a “CCS breakthrough” scenario that features rapid growth and a tripling of estimated maximum CCS rates for regions with oil and gas extraction volumes less than 1/3 that of the United States. (Methods Table 2).
We emphasize that our method of using oil and gas production data to estimate relative CCS capacity between countries and regions is only done in the absence of first principles estimates of practicable injection rate capacity. We further note that we excluded ‘unconventional oil’ from each region’s production estimates in estimating relative CCS capacity as suggested by Lane et al.15 However, due to challenges with data availability, unconventional gas production was included in the production totals from each region. This approach contrasts with that of Grant et. al., 2022, which included both conventional and unconventional oil and gas in their estimate of investible CCS potential in each region.43 Given the high levels of unconventional oil and gas production in the U.S. relative to other regions of the world, including unconventional oil in these ratios would further diminish the estimated CCS capacity of most regions outside of the U.S. relative to our estimate. While Lane et. al points out that neither approach with respect to unconventional oil and gas production is clearly superior for estimating practically achievable CCS rates,15 our method attempts to make use of limited available data while being conservative with the degree to which we further tilt the portion of estimated global CCS capacity towards the U.S.
Table 2
Estimated Maximum CCS Rates in GCAM
Region
|
Estimated Maximum CCS Rate
|
Estimated Maximum CCS Rate (Breakthrough)
|
Middle East
|
2739
|
2739
|
USA (From NETL data)
|
2558
|
2558
|
Russia
|
2269
|
2269
|
Europe
|
1884
|
5652
|
Asia (excl. China and India)
|
1673
|
5019
|
Africa
|
1047
|
3141
|
North America (excl. USA)
|
883
|
2649
|
China
|
542
|
1626
|
South America (excl. Brazil)
|
615
|
1845
|
Australia New Zealand
|
219
|
657
|
India
|
181
|
543
|
Brazil
|
170
|
510
|
Central America and Caribbean
|
133
|
399
|
Total
|
14913
|
29607
|
Estimation and Implementation of Growth Rate Constraints
Consistent with the conventions of transition theory, deployment constraints follow a logistic function dependent on the level of installed capacity. This results in a ‘S-shaped’ growth curve, that assumes accelerating uptake of a new technology in its early stages, followed by decelerating growth as that technology nears its saturation limit. While Lane et al (2021) point out that the practicalities of geological storage development might limit the potential to realize strong learning curve benefits, we suggest it not unreasonable to allow for growing deployment rates in the early stages of a region’s pursuit of CCS. High pace deployment first requires the region to configure the necessary research knowledge, commercial and regulatory capabilities, and policy settings to stimulate investment. Recent years’ experience would suggest that some regions are now witnessing that uptick in action.76
The GCAM implementation of this growth constraint applies a time-varying “efficiency” parameter on CCS for each model period, which allows the scale of CCS relative to a region’s estimated maximum CO2 injection capacity to be varied over time. Through to 2030, this limit follows IEA projections for CCS deployment (including operational, under construction, and planned projects),6363, and after that it follows a hypothetical logistic curve based on the historical growth of different possible analogues. The closed-form logistic function is fitted to observed technology capacity data, extracting the fit parameters that together predict capacity C(t) over time: the growth rate k, the inflection year t0, and the saturation level L:
$$\:C\left(t\right)=\frac{L}{1+{e}^{-k(t-{t}_{0})}}$$
That function is normalized so that it equals 1 when installed capacity is at the maximum allowed injectivity for each region. Note that, in the modelled GCAM scenario results, simulated CCS in any given year may be lower than the maximum limit if other abatement technologies prove to be a more cost-effective means of responding to the CO2 emissions policy.
The three growth scenarios are summarized in Table 3, with their basis described in the following section.
Table 3
Historical analogues for CCS and data sources (km = kilometers, GWe = gigawatts-equivalent, MW = megawatts).
Scenario
|
Technology basis
|
Units
|
Growth rate (% per year)
|
Data source
|
Slow
|
Natural gas pipelines
|
Km
|
3.2%
|
77
|
Breakthrough
|
Shale Gas (U.S. only)
|
Billion ft3
|
24%
|
78
|
Developing the growth constraint scenarios
To our knowledge, no meaningful CCS growth estimates have been modelled that reflect the complexities of developing an integrated CO2 capture, transport, and storage process train. Expectations are that overcoming those complexities, in the face of uncertain storage prospects, will present a major challenge in most if not all jurisdictions.15
In the absence of that, we use two abstracted scenarios (Slow Growth/Breakthrough) for testing the implications of different limits to the pace of CCS deployment. Each is based on logistic growth curves fitted to empirical data for a single infrastructure or technology type. While two of those might be considered to provide a potential analogue for components of the CCS process chain, all potentially lack the system complexity and potential barriers that may influence the evolution of CCS growth. Our scenarios are therefore chosen to explore a large spread of possible growth rate constraints and should not be taken as being an estimation of what might actually be possible. Table 1 in the Main manuscript summarizes our scenario design.
To test the effect of future CCS scaleups that differ from the present expectations of decisionmakers, we ran two additional below 2 ⁰C scenario variants. The first uses the endogenously solved CO2 price path resulting from the below 2 ⁰C scenario with no injectivity or growth rate constraints (#1) to represent relatively lower levels of mitigation effort consistent with the expected ability to rapidly scale CCS in both the near and long-term. However, we also applied the injection and slow growth rate limits from (#2a), which forces higher deployments of mitigation technologies that do not entail CO2 capture, but ultimately allows for less emissions reduction at the same CO2 price. The second scenario uses the solved CO2 price path from (#2a) to represent higher levels of mitigation effort consistent with the expectation of a far more limited role of CCS. However, we relaxed the CCS growth rate constraint to equal that of scenario (#2b) allow more rapid CCS scaling, which enables additional emissions reduction to take place at the same CO2 prices.
Unlimited CCS Rates Scenario (1)
This scenario uses GCAM’s existing (cumulative) carbon storage supply curves (see Methods Fig. 7). Per the GCAM implementation of SSP1 which as described below seeks explicitly to explore a more limited role for CCS, CO2 transport, and storage costs are increased by a factor of 10, and offshore CCS is disabled. With this up-scaling in costs, approximately 4000 GtCO2 of storage capacity, cumulatively, is available for below $200 per tCO2.
Slow Growth Scenario (2a)
The logistic curve is fitted to data for the global growth in natural gas pipelines over the years 1904–2021, compiled by the International Gas Union.77 Pipelines are efficient and low impact transportation modes for liquids and gases, and will likely be required in most locations to transport CO2 from the point of capture to geological storage locations. Like natural gas pipelines, CO2 pipelines and injection sites would require administrative, legal, and regulatory frameworks for site selection, land acquisitions, and rights-of-way and may also see delays due to public protest and opposition.61 In this, as well as the CCS breakthrough scenario described below, offshore CO2 storage is allowed, subject to the annual rate and growth limits for each region.
CCS Breakthrough scenario (2b)
The growth limits for this scenario are informed by the recent U.S. shale gas boom, which provides one of the most spectacular examples of learning curve effects seen in the energy sector. Data is taken from the U.S. Energy Information Administration’s time series covering the years 2007–2021, from which we calculate an average annual growth rate of 24%.78 Like geologic CO2 storage, shale gas extraction is a subsurface process that entails extraction and injection of large volumes of fluids from deep in the geosphere. Additionally, the potential need to fracture formations to allow higher rates and /or cumulative volumes of CO2 injection could bear many similarities to shale gas production. Hydrocarbon-bearing shale formations may also be well-suited to CO2 storage, with the possibility of waste CO2 itself being used as fracture fluid.79–81 Per Methods Table 2 (above) we also triple estimated maximum CCS rates for regions with oil and gas extraction volumes less than 1/3 that of the United States to reflect the potential for technology transfer and new storage space discovery that could allow CCS to be rapidly deployed in regions with relatively less experience with oil and gas production. However, the rapid expansion of shale gas production in the U.S. has benefited from public policy incentives including the relaxation of some environmental rules; the results of which, thus far, have not been replicated elsewhere.82,83 Additionally, natural gas extracted from shales and elsewhere has market value (energy supply) rather than being a waste treatment cost on the system as would be the case with CCS.
Socioeconomic and Policy Assumptions
All scenarios shown here use GCAM’s SSP1 (Shared Socioeconomic Pathway) 'sustainable development' assumptions marked by improved land use and other resource efficiency, a preference for renewable energy and other sustainable production methods, and investment in human development that together result in low challenges to both mitigation and adaptation.84–86 This choice of socioeconomic and technology assumptions is consistent with those of low emissions trajectories limiting end-of-century warming to below and well-below 2°C from the Working Group I contribution the IPCC's Sixth Assessment Report.87 Following GCAM’s standard SSP1 assumptions, strong policies are assumed to be put into place for pricing carbon emissions from land-use change. To represent transaction costs and long-term improvements in institutions for implementing land use policy, land use change emissions pricing is represented in GCAM as an increasing proportion of the fossil carbon price beginning after 2020, reaching 50% of the fossil carbon price by 2050 and then remaining constant through 2100.68,84
Projections of near-term CCS deployment for energy-economy model scenarios using the SSP1 assumptions, which were designed in part to explore more limited roles for CCS technology, already vastly exceed real-world deployments for the coming decade.54,63 Alternative sets of assumptions from the Shared Socioeconomic Pathway (SSP) scenario matrix have been shown to rely even more heavily on CCS under deep mitigation.68 The sharp limits on CCS deployment being explored here can therefore be expected to have even more drastic effects if combined with these alternative assumptions, including infeasibility of meeting the well-below 2°C or below 1.5 ⁰C temperature goals for many additional potential scenario permutations. These impacts are shown in Supplementary Fig. 18, which reports CCS by carbon source for below 2 ⁰C scenarios using the SSP2 “middle of the road” socioeconomic background assumptions88 and varying rate and growth limits (or lack thereof) on CCS.
Two constraints were imposed on end-of-century radiative forcing increases from pre-industrial levels: +2.6 W m− 2, consistent with limiting warming in 2100 to below + 2°C, and + 1.9 W m− 2 (below 1.5°C in 2100).89–91 These two end-of-century forcing targets were permuted across three potential limits (or lack thereof) on CCS injection and growth rates for each of GCAM’s 32 regions (unconstrained, slow, and breakthrough), for a total of 6 scenarios as described above. For each of these scenarios, GCAM solved for the lowest-cost, exponentially increasing CO2 price-path (beginning from 2025) to limit or return to each end-of-century radiative forcing limit. The atmospheric carbon budget consistent with a given level of radiative forcing increase is thus treated as an exhaustible resource to be depleted.92–94 The Hotelling rate (i.e. the annual rate of CO2 price increase after policy initiation, equivalent to the discount rate) is now set to 3% by default in the GCAM release. Discount rates are higher for developing countries that have higher rates of economic growth.95 Such higher rates, if applied globally, would tend to increase temperature overshoot under end-of-century warming targets by reducing near-term mitigation and increasing future carbon removal.96 This is especially the case if high CDR rates mostly underpinned by CCS are assumed as has been done in most energy-economy modeling frameworks to date, as well as in our scenarios in which storage and growth rates are unrestricted.