Climate change threatens to increase average global temperature, raise sea level, and alter weather patterns in ways that could harm natural ecosystems and affect human civilization.1 The root cause for climate change is anthropogenic greenhouse gas (GHG) emissions, which are increasing the heat-trapping capacity of the atmosphere. A majority of scientists concur that to avoid many of the negative impacts anticipated to arise from global warming, GHG emissions will need to be significantly reduced or eliminated over the next century.1
One option for mitigating GHG emissions is carbon capture and storage (CCS).1 CO2 emissions from major industrial point sources, such as power plants, would be captured and transported to sites where the CO2 would then be pumped into large reservoirs for long-term storage. Candidate reservoirs on land include deep-saline aquifers, oil/gas reservoirs, and unmineable coal seams. The ocean is another possible reservoir, though the London Convention/Protocol suggests that any ocean storage will likely be limited to injection into sub-seafloor analogues of deep-saline aquifers (Fig. 1).1,2
Schrag3 argues that sequestration beneath the sea floor may offer the best option for large-scale storage of industrial CO2 emissions because it would avoid potential hazards from direct ocean injection, including harming ocean ecosystems, while also avoiding challenges/risks posed by onshore geologic sequestration. The latter include (i) the need for the onshore reservoirs to be isolated within sufficiently large and/or numerous geologic traps that they can safely secure significant quantities of CO2 underground; (ii) management of reservoir pressure; (iii) extensive monitoring of the reservoir for unacceptable levels of leakage; (iv) possible groundwater contamination by the CO2, contaminants leached by CO2 from reservoir rock, or saline pore waters displaced by CO2; (v) uncertain legal and regulatory regimes surrounding subsurface storage of CO2; and (vi) likely opposition from land or mineral-rights owners above the storage sites.3 Though sub-sea-floor storage of CO2 would also require a secure trap, it would not threaten groundwater, it would move the storage away human habitation, and where done in federal waters (i.e. generally >5 km from shore in the USA) would fall under federal regulation and be managed by government agencies that already oversee offshore oil and gas extraction.3 These advantages must be weighed against the additional cost of offshore storage relative to onshore options.
To date, there has been little examination of the economics of offshore storage in literature. Initial estimates reported in the Intergovernmental Panel on Climate Change (IPCC) Special Report on Carbon Capture and Storage were based on the techno-economic requirements for sub-sea-floor storage.4 The cost estimates ranged from $4.7–$12.0/ton as compared to $1.9–$6.2/ton for onshore storage.5
Subsequent analysis by industry also cites offshore storage as being more expensive than onshore but again without much consideration for the form of geological storage.6 This instead has been the focus of additional academic research, which has carried out analyses of geochemical trapping of CO2 in both deep-sea submarine sedimentary strata7,8 and within submarine basalt layers buried beneath impermeable marine muds.9–11 These latter studies, however, have not addressed the economics of submarine storage any further.
Expanding upon the initial work by House et al.,7 we recently completed a global assessment of deep-sea ‘self-sealing’ sedimentary strata as a potential resource for sequestering industrial CO2 emissions. These are strata that would not need a cap rock for storing CO2 because they lie below the Hydrate Formation Zone and the Neutral Buoyancy Zone for CO2, i.e. the pressure-temperature zones in marine strata where CO2 is trapped in hydrates and becomes denser than seawater, respectively (Fig. 1).12 Our analysis suggests there are thousands of gigatonnes of potential storage below ∼3000m water depth.
Relatedly, we have developed a geo-economic model for estimating how much CO2 can be stored in deep-saline aquifers (DSAs) on land and at what cost.13 In applying this model to DSAs distributed about continental USA, we have found that these reservoirs could store at least hundreds of gigatonnes of CO2, much of it at relatively low cost.13
In this paper, we expand upon our onshore economic model to create a similar model for the offshore and use the latter model to estimate the cost of sequestering CO2 in marine strata, including not only the deep-water self-sealing storage described previously but also offshore storage in non-self-sealing strata assuming the latter possess a caprock seal (Fig. 1). We also use a basic transport optimization algorithm to estimate the cost of conveying CO2 from point sources within the continental USA to onshore DSAs as well as the offshore reservoirs. The final result is a supply curve in the manner of Dooley et al.,14 Wildenborg et al.,15 Dahowski et al.,16 and others (e.g. Middleton and Bielicki17)that relates cumulative annual US emissions to the marginal cost of CO2 transport and storage, allowing us to estimate the price premium for offshore storage relative to onshore storage.
Assembling the integrated cost of transport and storage requires three steps: (i) estimating the cost of storing CO2 in the different onshore and offshore reservoirs, (ii) sub-dividing these reservoirs into geo-referenced reservoir blocks to which transport from the CO2 sources can be optimized, and (iii) determining the lowest-cost option for combined transport and storage between the point sources and reservoirs.
Our evaluation of the cost for onshore storage capacity in DSAs is essentially identical to that already published in Eccles et al.13 To estimate storage capacity, we use the same digital, geo-referenced grids of reservoir properties we created in Eccles et al.13 for 15 deep-saline sandstone DSAs from data published by the University of Texas at Austin's Bureau of Economic Geology.18 These grids, which include sand thickness, porosity and permeability, represent the geology of the DSAs underlying ∼750 000 km2 of the 2.1M km2 that makes up the total surface area of the DSAs.
We also estimate storage costs for the DSAs using the same economic model in Eccles et al.,13 with one minor yet important change. In our original publication, we assumed the spacing of injection wells, which affects storage costs, to be constant. We now account for well spacing that varies as a function of the reservoir properties using a formulation developed by Eccles et al.19 to yield a more accurate estimate of storage costs.
We use data from a previous publication13 on offshore storage capacity to assess its cost. These data include geo-referenced grids of sea-floor depth, and sub-sea-floor sediment thickness and temperature, as well as digitized Deep-Sea Drilling Project and Ocean Drilling Project core data. Eccles and Pratson12 describe in detail how these data are used to estimate the net-to-gross thickness of offshore sands in which CO2 might be stored, which we extend to all offshore storage in addition to self-sealing storage. They also explain how the data are used to map out the locations and thicknesses of the self-sealing marine strata. For these details, the interested reader is referred to Eccles and Pratson.12
In this paper, we limit our analysis to that portion of our offshore dataset from within the US Exclusive Economic Zone (EEZ). Furthermore, we increase the resolution of our grids from 5’ (roughly 15 km) to 1 km and convert the grids to equal-area projections in order to keep bulk volume calculations accurate. Finally, permeability measurements from sand deposits in ocean sediments are limited, so we assume a uniform permeability for the grid cells of a low 22 mD. This value is both the average permeability in the onshore Mt Simon DSA, and the log-average permeability of all the sandstone DSAs in the BEG data. Permeability data from offshore sediments at this depth (e.g. Wetzel20) generally does not shed light on likely sand permeabilities because it focuses on other sediment types.
To estimate the cost of offshore storage, we modify the economic model in Eccles et al.13,21 into a version that is appropriate for the ocean environment. In this model, net sand thickness b is critical for calculating both storage capacity and cost.13,21 Unfortunately, net sand thickness is not known for ocean sediments,12 so we use the net-to-gross ratios we arrive at in Eccles and Pratson12 to estimate net sand thickness from total sediment thickness as done in the USDOE's Carbon Sequestration Atlas.12,22 We have estimated the net-to-gross ratio for self-sealing regions to be roughly 2.0%, while that for offshore sediments in water depths <∼3000m to be a similar 1.8%.12 We believe the latter is an underestimate for some regions beneath the continental shelf, but note that at even such low ratios, the estimated values for b average several hundred meters.12
The major cost components for offshore storage are same in nature to those for onshore storage but can differ in scale. In our economic model for onshore storage13 the cost components are represented by the index variable k, for example, k = 1 for injection well costs, k = 2 for site evaluation costs, and so on. A detailed explanation of each component and the basis for its cost in onshore storage is explained in Eccles et al.13
For offshore storage, we scale each component k by an associated scaling factor Bk to account for any difference in cost between the two environments, i.e.
The total cost for offshore storage is thus
For most components, Bk ∼ 1; e.g. depending on land use/land cover, ocean seismic surveys may actually be cheaper than their onshore counterparts.23 Two components, however, are assigned a Bk > 1. These are injection well construction, and pipeline infrastructure. For injection wells, we calculate Bk using the following empirical formula derived from regressing offshore rig day-rate data24 against the water depths d in which the rigs can operate
See Fig. 2 for a plot of data vs. this equation. Where d > 1000 m, we set Bk = 20, a multiplier appropriate for the only class of rigs regularly deployed in these depths.24
The other cost component for which we set Bk > 1 is offshore pipeline infrastructure. In this case, we use the same multiplier as the IPCC (2005) i.e. Bk = 2.
A storage site is likely to include several separate injection zones because the site may contain more than one reservoir, and the quality of each reservoir (including its seal) will probably vary spatially (Fig. 3(a)). Consequently, infrastructure for a storage site will likely be configured to include a central receiving point for incoming CO2 that connects to a system of distribution pipelines which then carry the CO2 to individual injection wells positioned about the storage site.19 In this study, we define a reservoir block as a region where CO2 would be delivered to a site hub for distribution and injection within the region (S3 in Fig. 3(e)). The 1 km2 area of the cells in our grids of the DSAs and offshore storage reservoirs have too little capacity and thus are too small to be treated as reservoir blocks, because the latter need to be large enough to store many years of commercial-scale injection. We map out our hypothetical reservoir blocks by grouping economically similar grid cells until their aggregate storage volume reaches a minimum threshold for what we arbitrarily define as a viable ‘storage site’, with the capacity to hold a project lifetime's worth of emissions. We set the capacity threshold for onshore reservoir blocks at 200 MT, or 1.1 GW worth of average coal-fired power plant emissions over 20 years.25 For offshore reservoirs, we raise the threshold to a five-fold higher minimum capacity of 1 GT (5.5 GW). This is to compensate for the original resolution of the data, and the lack information in our grids on the true distribution, extent and thickness of offshore sands.
Figures 3(b)–(d) are a schematic of our algorithm for aggregating grid cells based on storage capacity and cost. The algorithm begins with the lowest cost cell and searches upward in cost, adding cells to the block until its cumulative capacity reaches the minimum threshold. As differences in costs between cells can be very small, even small step sizes like $0.10/ton CO2 can lead the algorithm to produce very large blocks that exceed the minimum capacity threshold by a significant amount. The end result, though, is a series of discrete storage sites that are similar in terms of capacity, but can differ in average storage cost and in distance from a CO2 source.
Transport and storage optimization
The final step in our analysis is to match CO2 sources with sinks so that a marginal abatement cost curve showing the combined cost of transport and storage can be used to evaluate the economics of onshore vs. offshore storage. We first compute the Euclidean distance from each CO2 source to all possible sinks. In this analysis, we limit the CO2 sources to coal-fired power plants in the EGRID database.25 We also subdivide the storage options into three sets: onshore DSAs, offshore non-self-sealing regions, and offshore self-sealing regions.
Minimum distances between sources (pn, Fig. 3(e)) and sinks (sn, Fig. 3(e)) are converted to a cost per tonne using a simplification in Chandel et al.26 that relates the cost per tonne of CO2 per kilometer transported to the total amount transported. We select a transport mass of 10 MT/yr, which is the emission rate of 1.1 GW of average coal-fired power.25 This corresponds to a scenario in which large plants could use their own transport networks while smaller plants would cooperate to join a larger-scale transport network to capture economies of scale (the gray ‘p cluster’ in Fig. 3(e)); note that we only assume clustering of small plants into feeder networks and do not explicitly model this in our analysis.
Again, offshore transport is increased to twice the cost of onshore transport (triple lines in Fig. 3(e)),4 but we do not optimize offshore transport routes for cost, so there may be some transport savings in an actual deployment of offshore CCS (t5b in Fig. 3(e)). Similarly, no cost surface (a planning tool for which areas are assigned weights according to their relative or absolute cost of transport) is used to route pipelines onshore around high-cost areas such as urban areas, wetlands, or rugged terrain, which could increase the cost of transport (t1b). The benefit of cost surface calculations is well-established in transport modeling,17 but for source-sink networks of far less complexity than the one we deal with here, i.e. dozens of sources and sinks in the case of the former vs the 628 sources and roughly 3000 sinks we assess here. The scale of our analysis produces a computationally intractable problem if cost distance is used instead of Euclidean distance.
After sources have been optimally matched to sinks, a marginal abatement curve is assembled by ordering the sources according to their integrated cost of transport and storage, and then plotting the combined costs against annual emissions from the sources (less capture costs, see Fig. 3(e) inset).
Geological storage cost and capacity
Our results for onshore storage capacity and cost are essentially the same as those reported in Eccles et al.,13 despite the minor revisions to our geo-economic model to account for variations in the spacing of injection wells. The gridded cost and capacity estimates are plotted in Fig. 4.
Our offshore storage cost and capacity estimates, however, are an entirely new contribution to literature. Within the US EEZ, there is an enormous quantity of storage available (Fig. 4). In offshore storage regions close to the coast, there is 1424 gigatonnes available, 1111 gigatonnes of which may cost <$10/tonne. This low-cost storage generally lies along the Eastern Seaboard and in the Gulf of Mexico (Fig. 4), relatively old-age passive continental margins where sediments eroded from the continental USA have been deposited for over 200 million years to form extensive accumulations of marine strata.27 The West Coast, on the other hand, has much poorer storage options. This is due to it being an active margin setting where sediments are being uplifted and accreted onto the continent or being subducted beneath it north of the Mendocino Triple Junction.28 The seismic activity along this margin further reduces its attraction for use in CO2 storage.28
The marine strata along the Eastern and Gulf of Mexico USA are thinnest near to shore and thicken seaward across the continental slope before thinning again beneath the continental rise and abyssal plain. Continental shelf strata at the appropriate depth range for storing CO2 in a supercritical state are in general a relatively expensive storage option at >$10/tonne. An exception appears to be the shelf offshore New Jersey and Delaware where we estimate storage could run <$5/ton within 70 km of the coastline. There also appear to be regions along the continental shelf in Gulf Coast where storage could run ∼$10/tonne.
In the strata further offshore beneath the continental slope and uppermost rise, storage costs drop to <$10/tonne. This includes the storage potential in the deep-water self-sealing regions, which we estimate to total 414 gigatonnes, 226 of which could be available for <$10/tonne (Fig. 5). Extensive self-sealing storage resources are available off the Eastern Seaboard and GOM, though the broad continental shelves in these regions (e.g. Texas, Maine) raise the cost of accessing these strata (Fig. 4). The closest and thus likely cheapest available self-sealing storage is off the coast of North Carolina. Self-sealing storage is even closer to the West Coast, but along this margin these strata are so thin that they make a poor storage option.
In general, the cost for offshore storage in self-sealing strata ranges from $4.28/t up to over $1000/t. Non-self-sealing offshore storage starts at $3.76/t and likewise goes over $1000/t. The average grid value for self-sealing storage is $142 (std. dev. $241), while that for non-self-sealing storage is $57.9 (std. dev. $158).
Relatedly, the range for storage capacity in self-sealing strata is 53-3.34e6 t/km2 with the average being 6.36e5 (std. dev. 6.16e5). Non-self-sealing strata have storage capacities ranging from 17-3.51e6t/km2, and a mean of 1.04e6 (std dev 9.72e5). Note that the distribution of values in our storage capacity and cost grids is highly skewed, so the mean values do not adequately reflect the fact that most grid cells have lower cost values and higher capacity estimates.
Transport and storage
Figure 5 is our integrated transport and storage cost curve in which the cumulative annual emissions of the coal-fired power plants are plotted against the minimum cost per tonne of CO2 each plant faces to transport and store its emissions. This supply curve is broken down into its transport and storage components, and shows the quantity of CO2 emissions that could be sequestered for a given carbon price minus the cost of capture.
The results of the transport optimization process indicate that offshore storage is more expensive than onshore storage, which is not surprising given that the latter is both farther from the CO2 sources and is assumed to incur double the costs of onshore transport seaward of the coastline. Without exception, when offered all options for storage, power plants in the USA should find it most economical to transport to onshore reservoirs, at least until these reservoirs are depleted or circumstances arise that raise the cost of onshore sequestration. Roughly 370 Mt/yr of CO2 can be stored in onshore reservoirs before the combined cost of transport and storage reaches the minimum cost for offshore storage of roughly $5/t (Fig. 6). Onshore, 1 Gt/y can be stored for below $5.66/t, with nearly all current US CO2 sources (which produce 2.1 Gt/y of emissions) having access to storage for <$10/t.
In the case of offshore storage, the minimum cost starts at >$5/t (off the coast of New Jersey), with up to 500 Mt/y of storage available for <$10/t or $5/t more than what it would cost onshore. Half of this difference is due to the higher cost offshore storage, and half is due to the higher cost of offshore transport. Additional offshore storage is located farther from the coast, rapidly raising the cost of transport even where storage costs decline so that the combined cost of transport and storage jumps to $13.5/t for 1 Gt/y of storage and $31.2/t for 2 Gt/y storage.
Finally, self-sealing storage of and by itself has a minimum cost of just over $10/t. Up to 500 Mt/y of such storage capacity is available for <$14.5/t, which is about $5/t more than other offshore storage, and about $10/t more than storage onshore. Almost all of this greater cost for self-sealing strata relative to the other options is due to the greater transport costs need to reach the strata. Similar to non-self-sealing offshore storage, the cost for self-sealing storage rises quickly to $18.1/t for 1 Gt/y of capacity and $35.4/t for 2 Gt/y of capacity.
Our results show that there is substantial capacity for sequestering industrial CO2 emissions beneath the seafloor within the US EEZ, but that doing so would be nearly twice as expensive (including transport costs) as storing CO2 onshore in DSAs. Interestingly, we arrive at essentially the same cost ratios for onshore versus offshore storage estimated by earlier studies summarized in the IPCC special report and by more recent private assessments,4,6 all of which relied upon less information than we used. We also find that while standard offshore storage is twice as expensive as onshore storage, onshore storage costs about a third of that for self-sealing offshore storage, which becomes even more expensive for cumulative capacities >500 Mt/y.
We stress that even with the extensive information that forms the basis for our analysis, our storage capacity and cost estimates involve considerable uncertainty, not only with respect to the geology of DSAs and even more so offshore reservoirs, but also in the implementation of our reservoir block and transport models (Fig. 3(e)). We find that our cost estimates for onshore and offshore storage encompass previously reported estimates (e.g. those in Hendriks et al. or the BCG report5,6), and that our combined cost for onshore transport and storage is similar to that of Middleton and Bielicki,17 for example, but this appears to be the limit to which we can verify our estimates.
Our offshore cost/capacity maps depend heavily on our translation of the sediment thickness dataset into net sand thickness via the net-to-gross ratio we calculated from DSDP/ODP cores as described in Eccles and Pratson.12 There is significant variation among cores in the net-to-gross ratio, however, which undoubtedly corresponds to geospatial variations in sand distribution resulting from the different routes (e.g. submarine canyons, channels and fans) and mechanisms (e.g. sea-floor failures to turbidity currents) by which sands are carried into the deep sea.12
In fact, our maps of storage capacity show a correspondence to the regional geology of the US EEZ inherent in the sediment thickness data. Areas we estimate to have good storage potential are located in thick sedimentary basins on the outer continental shelf and slope such as the Carolina and Baltimore Canyon Troughs. In fact, we probably underestimate the sand potential in these Troughs, where Lower Cretaceous strata include ‘massive to thin’ sandstones (∼1000m), and underlying Jurassic sediments consist of even more (several ∼100m formations) sands.29 The same holds true for the Texas Gulf Coast, where massive Tertiary sand deposits extend many kilometers seaward across the continental shelf, and have formed the reservoirs for the significant amount oil and natural gas that has been produced in the region.30,31 Consequently, our simplifying assumptions for this analysis likely make our gross storage capacity estimates for the US EEZ minimum estimates.
While we do offer estimates of storage cost, the focus of our analysis is the difference in cost between onshore and offshore storage, which we estimate to be roughly $5/t and $10/t for non-self-sealing and self-sealing offshore storage, respectively. If these differences are accurate, they represent an interesting window of opportunity for offshore storage. On the scale of the total cost of the CCS system, in which capture costs may range from $50/t to over $100/t, $5–$10/t is not particularly expensive.32 In fact the cost difference may be offset by a variety of factors that could increase the expense and/or complexity of onshore storage projects, making offshore projects more attractive.
First, the offshore regulatory environment may be greatly simplified. Much uncertainty regarding regulation of CO2 injection wells was put to rest with the EPA Class IV rules,33 but state and local regulations could still pose a significant hurdle to the approval and operation of CCS projects. Perhaps more importantly, buoyant CO2 gas bubbles and dissolved CO2 gas in groundwater at CCS projects could now run afoul of new EPA regulations concerning the contamination of US Drinking Water (USDW), preventing onshore storage at many sites previously thought viable for CCS.33 Seaward of State jurisdiction in ocean waters (generally 3 mi), the US Bureau of Ocean Energy Management, Regulation and Enforcement (BOEMRE) is the only agency that would regulate offshore CCS projects, potentially simplifying the project approval and oversight process.
In addition to this, monitoring, measurement, and verification may be less complex in offshore storage projects, particularly for self-sealing storage where the combination of the NBZ and HBZ mechanisms for trapping CO2 offers a safety redundancy against leakage.3 Although the costs for monitoring, measurement, and verification (MMV) are relatively small compared to the cost of offshore injection in our analysis – recall that injection costs are adjusted upward, but MMV costs are not – any reduction in the former would still decrease the gap in total costs between onshore and offshore storage.
Finally, offshore property rights are managed by BOEMRE whereas onshore property rights for CO2 storage will involve a large number of private as well as public owners, making leasing of these rights costly in terms of both time and money. Research by Gresham et al indicate that the costs for such rights may range between $1–$5/t, which by itself could make offshore CO2 storage an economically competitive option.34 There is also the problem of overcoming issues with public acceptance. Finally, both onshore and offshore storage would have to deal with transport rights of way, but we anticipate that the cost of offshore rights of ways will be less, a factor not included in our transport cost analysis.35,36
These factors, alluded to previously in our discussion of Schrag's persuasive piece,3 could end up making offshore storage economically competitive with onshore storage and thus reduce barriers to adoption of the former.
However, there are also significant reasons against storing CO2 offshore. One is cost. Based on this analysis, we estimate that the total cost of a 10 Mt/yr offshore storage project could run nearly half a billion dollars more than its onshore equivalent, and be nearly a billion dollars more expensive if the project is involves deep-water self-sealing storage. This is a significant amount of capital that may prevent private investors or storage system operators from utilizing offshore storage resources regardless of regulatory risk or other barriers to adoption.
Another important reason for not storing CO2 offshore is the possibility of environmental damage if undersea storage goes awry. The recent Deep Horizon disaster demonstrates the complexity of maritime contamination accidents and their long-term effects, and CO2 storage projects may be no different from oil and gas extraction or, for that matter, proposed geo-engineer projects also designed to mitigate carbon emissions.37 Sub-seabed storage must be carefully evaluated with further research into both its physical and economic impacts and deployed only as a cost-effective measure to mitigate climate change that does not adversely affect the environment it is meant to protect.
This research was supported by DOE award No. DE-FE0001934.
Jordan Eccles is a postdoctoral associate in Duke University's Nicholas School of the Environment in the Division of Earth and Ocean Sciences. His expertise lies in coupled physical and economic models applied to carbon capture and storage research.
Lincoln Pratson is a professor in Duke University's Nicholas School of the Environment in the Division of Earth and Ocean Sciences. He conducts research into energy and environmental issues, evaluating future demand for and supplies of energy resources.