Selection of storage regions
For our study, we only consider those regions that have existing, or planned CCS projects with announced capture capacities published in the 2022 Global Status report by GCCSI that will reach an operational date prior to or during 203056. This narrows down our consideration to ten regions, encompassing countries including Australia, Canada, China, Indonesia, South Korea, Thailand, the UK, and the USA, along with the EU and the Middle East. For the consideration of storage resource availability in the EU, we constrained the resource base to the estimated offshore asset in the Norwegian North Sea. This is a simplified consideration given that the potential to develop onshore storage resources across continental Europe faces significant opposition and uncertainty57,58,59,60,61. For our consideration of countries that could contribute towards CCS within the EU region by 2030, we consolidated capacity from capture plants where announced, i.e., France, Belgium, Norway, Finland, Sweden, Hungry, the Netherlands, and Denmark (8 out of the 27 EU countries). While there are more EU countries that have announced projects, i.e., Italy and Ireland56, their annual capture capacity is yet to be determined and thus excluded. For the Middle East, we aggregated Saudi Arabia, Qatar, and United Arab Emirates, which have operational commercial-scale CCS plants and/or confirmed plans for future developments.
Finally, we differentiated our ten selected regions under three levels of readiness: high, medium, and low. For this, we make use of the 2018 Global CCS Readiness Index Report (GCCSI-RI)62 which scores each nation using a series of criteria categorised under four indicators: inherent interest of CCS which represents a nation’s dependence on hydrocarbon products, legal regime, policy measures that support CCS, and the level of maturity of storage resource assessment. A high level of CCS readiness in our study corresponds to regions scoring between 60–80 in the GCCSI-RI. A medium level indicates those regions where there is generally limited policy and regulatory support, but ongoing projects exists (scores 40–59 in the GCCSI-RI). Lastly, a low level of readiness is associated with regions that have limited demonstration of CO2 storage, a lack of adequate policy framework in support, and the volumetric estimate of storage resource base across oil and gas, and saline aquifers largely remain unknown (scores 20–39 in the GCCSI-RI).
Carbon dioxide storage projections in Integrated Assessment Models
Integrated Assessment Models (IAMs) are analytical frameworks capturing key interactions of the human-earth system to understand implications across disciplines of energy, economy, and the environment simultaneously63. For climate change mitigation, IAMs are employed to project greenhouse gas emission reductions that achieve long-term climate objectives (for 2100) or policy goals (generally for 2050) whilst exploring different cost-effective strategies to decarbonise the energy supply64.
The AR6 Scenario Explorer is an online platform hosted by the International Institute for Applied Systems Analysis (IIASA) that provides a comprehensive compilation of decarbonisation pathways generated by various IAMs underpinning the Sixth Assessment Report (AR6) of Working Group III by the IPCC5,8. We make use of this database to select scenarios of CO2 storage deployment at the global level that are compatible with two global warming levels, 1) limiting warming to 1.5 oC in 2100 (50% likelihood) with no or limited overshoot, and 2) limiting warming to 2 oC in 2100 (67% and 50% likelihood). We examine the key variable of total CO2 sequestration through CCS in Mt CO2 yr− 1 for 2050, including CO2 emissions captured from bioenergy use, fossil fuel use, and industrial processes. We refer to storage deployment modelled by IAMs as projected scenarios.
For our analysis, we assume all CO2 storage is within geological reservoirs including depleted hydrocarbon fields and saline aquifers. We do not differentiate modelled storage deployment by CO2 capture source.
Logistic growth modelling framework
We make use of the logistic growth model which is an empirical mathematical framework that has been traditionally applied in the hydrocarbon industry to forecast production outlooks and consumption trajectories39,43,38,65,41,66. Over time, the model’s application has been expanding beyond fossil fuel resources with several studies demonstrating the suitability of logistic curves to model energy production from renewable sources, nuclear, and the rates of technological substitution67,68,69,70,71,72.
The use of logistic growth models in the context of CO2 storage has been demonstrated at both global and regional scales44,45,51. In this context, the logistic growth model embeds an impact of the depletable nature of CO2 storage resources, capturing the interactions between market dynamics and the physical use of the resource base. Importantly, it avoids the explicit definition of highly uncertain socio-economic factors. This approach allows for an assessment of the feasibility of project storage demands modelled within IAMs and a quantification of the potential storage resource needs44,45,51.
A description of the cumulative storage, \(P\left(t\right)\) [GtCO2], and storage rate, \(Q\left(t\right)\) [GtCO2 yr− 1], of CO2 at time, \(t\) [yr], is outlined in Equations 1 and 2, respectively. The logistic growth rate, \(r\) [yr− 1], which hereinafter referred to as the growth rate, characterises the early part of the trajectory. This phase signifies near-exponential growth driven by expansive adoption of commercial practices, with minimal impingement from geological constraints. Subsequent to peak year, \({t}_{p}\) [yr], growth declines until the exhaustion of the storage resource base, \(C\) [Gt], influenced by both geological constraints and engineering capabilities.
We define the beginning of deceleration on the rate curve by the first inflection point, occurring in year \({t}_{n}\) when approximately 20% of the storage resource has been consumed, given by Eq. 3.
Constraints on the logistic growth model
The “Global Status of CCS 2022” report by the GCCSI provides the latest update on CCS project pipeline across the world56 (GCCSI, 2022). We use the capture capacity reported for both active and planned CO2 storage projects to estimate the cumulative storage of CO2 in 2030 for our ten selected regions (Fig. 5; Table 2). This serves as an external input, ensuring the consistency of modelled storage rate for 2050 with real world CCS development up to 2030, denoted as \(P\left(2030\right)\) [GtCO2]) in the logistic model. The cumulative storage for 2030 marks a “take-off” point at which we begin modelling the near-exponential part of the trajectory until the inflection points (Fig. 5).
Assessments have been undertaken across various regions of the world to estimate the available storage resource base for CO2 within geological reservoirs, including depleted hydrocarbon fields and saline aquifers73,74. Over the years, different concepts have been developed for storage resource assessment depending on the type of storage medium, trapping mechanism, and the geomechanical structure of the formation-seal system in saline aquifers, i.e., open, semi-open, or closed75,76,77. To define the parameter space for \(C\) [Gt] in Eqs. 1 and 2 which characterises the available storage resource base, we predominantly make use of the “Storage Resource Catalogue” from the Oil and Gas Climate Initiative, an energy company-led organisation that focuses on climate change mitigation23. This multi-year report offers an independent assessment that compiles and evaluates geologic CO2 storage estimates, focusing on saline aquifers and depleted hydrocarbon fields, consistent with the definitions of the Storage Resource Management System78,23. Their report describes that the prevailing approach used to make estimates within saline aquifers across our ten selected regions is derived from the volumetric model to estimate the proportion of total pore volume that can retain CO2. However, for the USA and China, the representation of storage resources we adopt is more conservative. These estimates are derived from more nuanced assessments that builds upon the volumetric estimates and considers technical restrictions that may impede access to certain storage resources 79,22. On the other hand, the approaches for storage resource estimated within hydrocarbon fields are more uniform across different assessments and straightforward. Commonly, the principle of voidage replacement is applied where the volume of CO2 stored is assumed to be the equivalent volume of oil/gas produced at reservoir conditions. Overall, we refer to these assessed storage resources as our central estimates (Table 2).
Saline aquifers possess the largest storage potential and dominates the resource inventory across our ten selected regions (85% -100% of central estimates)23. However, significant uncertainty between 1–2 orders of magnitude within storage resource estimates persists for saline aquifers44,22,80. This geological uncertainty is primarily attributed to the lack of project development within saline aquifers, resulting in sparse acquisitions of geological data. Such data are essential to describe detailed in-situ reservoir characteristics such as thickness, areal extent, permeability, porosity, and heterogeneity – critical properties governing the efficiency of the total pore volume to store CO281,75. In light of this, we follow the approach established by Zhang et al.51, to test sensitivity of the modelled storage rate. We analyse storage rates at two additional bound of storage resources: the hypothetical maximum (one order of magnitude greater than central estimates) and hypothetical minimum (one order of magnitude lower than central estimates) of available storage resource. We summarise the storage resource constraints in Table 2.
For growth rate considerations, we propose a parameter space for sustained annual growth of up to 20% for all regions except China. We consider this as technically feasible at the national/regional level, consistent with growth trajectories that are required to meet proposed CO2 storage targets outlined for Europe and the USA45,51. Given the rate of acceleration observed in large-scale energy infrastructure development in China82, we extend the feasibility of growth rate to up to 25%.
Finally, a key constraint is imposed on the peak year (\({t}_{p}\)) which is set to occur later than 2050 following Zhang et al.’s appraoch51. This is attributed to our main consideration for modelling scale-up trajectories of CO2 storage of using the early part of the trajectory in the logistic model. In such a case, we can understand the potential storage rate achievable when storage resource is sufficiently large that can sustain the long-term economic viability of CO2 storage but not in unlimiting quantities.
Table 2
Summary of modelling parameters of available storage resource base, cumulative storage in 2030 based on existing and planned capture capacity, and the level of CCS readiness for ten selected regions.
Regions
|
GCCSI CCS readiness
(Scored out of 100)
|
Cumulative Storage in 2030 [Mt]
|
Available storage resource base [Gt]
|
Hypothetical minimum
|
Central
|
Hypothetical maximum
|
Canada
|
71
|
High
|
121
|
40
|
404
|
4040
|
USA
|
70
|
1084
|
51
|
506
|
5060
|
EU + Norway
|
67
|
129
|
10
|
94
|
940
|
UK
|
65
|
258
|
8
|
78
|
780
|
Australia
|
62
|
110
|
50
|
502
|
5020
|
China
|
53
|
Medium
|
41
|
40
|
403
|
4030
|
South Korea
|
37
|
Low
|
6
|
20
|
203
|
2034
|
Middle East
|
36
|
71
|
2
|
18
|
177
|
Indonesia
|
31
|
26
|
2
|
16
|
159
|
Thailand
|
21
|
5
|
1
|
10
|
104
|
Modelling a storage trajectory database
To create a comprehensive database of geographically resolved CO2 storage scale-up models, we compute 1000 random iterations of prospective storage rates for each region. Solutions to Eqs. 1 and 2 are numerically calculated for each given storage rate by systematically exploring all combinations within our predefined parameter space for peak year, storage resource, and growth rate. We implement the solutions for the three parameters that exhibit the closest fit to our fixed input of established cumulative storage in 2030 by calculating a minimum squared difference. To generate a distribution of total aggregate storage rates, we take the sum of modelled storage rate across the ten selected regions in random combinations for 1000 iterations.
Scenario setting
We design a range of exploratory scenarios defined by a different set of conditions on growth rate, storage resource, and targeted limits on regional storage rate (Fig. 6). Comparisons of scenarios reflect the various drivers and limits to the modelled storage rate. Comparisons of modelled scenarios to the projections on storage rate from IAMs illustrate the range of technical feasibility. These scenarios are not predictive but provides an insight for the necessary market conditions, opportunities, and bottlenecks of what future long-term CO2 storage development has envisioned. The Reference scenario sets out a range of storage rate within the maximum bound of feasible growth rate (i.e., 20% or 25%) and reflects broad differences of modelled storage rate across geographies based on central estimates of available storage resource base for each region. The Minimum scenario reflects a conservative case on modelled storage rate, limiting growth rate to a maximum of 10% and a storage resource base of only 10% of central estimates (hypothetical minimum). The Growth10% scenario explores the extent of which modelled storage rate is limited by growth only compared to the Reference and Minimum scenarios. The Maximum scenario illustrates the opposite end of extreme where storage resource is essentially in unlimiting quantities, facilitating development of CO2 storage at a maximum growth of up to 20% or 25%. Finally, the US1Gt and EUUSChina scenarios illustrate the impact on modelled storage rate for other regions and the global aggregate storage rate when storage rate in USA, UK, EU, and China are limited to targeted amounts of 1 Gt yr− 1 ref.54, 0.175 Gt yr− 1 ref.55, 0.33 Gt yr− 1 ref.7, and 0.2 Gt yr− 1 ref.10, respectively.