By Joe McNease, Sephora Yameogo, Yuesu Jin, Yingcai Zheng
Department of Earth and Atmospheric Sciences, University of Houston
*Currently, superscript or subscript text is not supported in Wix. They are denoted by color in this article. superscript in red and subscript in green
Abstract
Geological carbon storage (GCS) involves CO2 capture at sources, transport through high-pressure pipelines, and injection of its supercritical form into geological formations for long-term storage. To make meaningful impact on the global climate, the goal of gigatonne GCS per year is proposed by governments and funding agencies. In this study, we evaluate the concept of the Texas Gulf Coast, stretching from Corpus Christi to Port Arthur, as a carbon capture and storage (CCS) hub and its role in achieving the gigatonne GCS goal. Because there are currently zero operating Class-VI wells and zero active or pending Class-VI well permission applications in the Texas Gulf Coast, we only evaluate the GCS potential and costs in the region. The numbers are just estimates and discussions could be speculative. In this study, we performed pipeline modeling from many stationary CO2 emission sources in the study region to two potential GCS sites in Bayou Bend and Free Port. With annual injection rates of 10 MtCO2 and 30 MtCO2, we let the two storage sites fill up to their individual capacities over 10 years. Our pipeline modeling shows that the optimal pipeline networks favor longer pipelines to large, low-cost capture sources rather than clusters of smaller, high-cost capture sources. The annual costs for the two scenarios including capture, transport, and injection/storage, are $715M and $2.65B, respectively. The capture cost and the injection cost are comparable, but they are much higher (~10x-30x) than the transportation cost. Therefore, any future significant reduction in GCS cost must involve reducing the price of capture or storage or both. There could be considerable seismic hazards. A survey of other CO2 injection sites worldwide shows that induced earthquakes do occur, and the maximum magnitude is about 1.6 at a low injection rate of ~ 1 MtCO2 per year. In the gigatonne-per-year scenario, another phenomenon, called dynamic earthquake triggering, can be important. A triggered or induced earthquake of magnitude ~3 in Texas state waters is of concern as it may rupture the seal of the subsurface storage unit to cause leakage. To achieve gigatonne GCS per year, aggressive actions are needed to increase the number of Class VI wells by 3 orders of magnitude to ~1000 and scale up the GCS system studied here by 33 times, which could result in a total cost of $87B per year. Texas Gulf Coast may have the storage potential, but it has long way to go to become an actual CCS hub given there are virtually no activities yet.
Introduction
The greenhouse effect was shown by Joseph Fourier in 1824. Two main greenhouse gases are methane (CH4) and carbon dioxide (CO2). Methane which reacts with the hydroxyl radical (OH), the main oxidant in the troposphere, has a lifetime of ~ 10 years in the atmosphere. But emitted CO2 can stay in the atmosphere for about 100 years, which means that the cumulative emission matters. From 1750 to 2020, the CO2 concentration in the atmosphere has increased from ~277 ppm to ~412 ppm (by mole). This increasing trend will continue and can lead to many problems. Many countries and states have set or pledged net-zero emission goals. However, achieving these goals is no easy task. The global CO2 emissions from using fossil energy in 2022, were projected to be 36.4GtCO2 (gigatonnes of CO2), rebounding back to the 2019 level (Friedlingstein, Jones, O'Sullivan et al., 2022). To feel the magnitude of this number, if we pool all the petroleum consumed in a year in the U.S., it would be a whopping ~ 1.17 Gt (assuming 20 million bbl/day), but it is still much less than the total CO2 emission per year (~ 5.7 GtCO2 of which ~78% is from hydrocarbons, according to EIA).
To make a meaningful impact in slowing down (unlikely to reduce soon) the increase of atmospheric CO2 concentration by the means of GCS, the injection amount should probably be on the order of one gigatonne per year. Deep saline formations have the largest storage potential compared to other reservoirs such as depleted oil reservoirs and residual oil zones. The gigatonne scale is consistent with the scenarios evaluated in the Princeton Net-zero America study (Greig and Pascale, 2021) as well as the Carbon Negative Shot (announced November 5, 2021; one of the Energy Earthshots Initiatives, see https://www.energy.gov/fecm/carbon-negative-shot, last accessed Feb 16, 2023) which aims to capture gigatonnes of CO2 from the air per year at a price < $100/tonne and store it in geological storage (GCS), biobased and ocean reservoirs, or turn it into value-added products, by 2050, along with aggressive decarbonization. CCS can include many storage types but it is meant to be GCS in this article.
Before large direct air capture (DAC) is economically feasible, CO2 capture from stationary emission sources is necessary in GCS. In our study, we focus on Texas Gulf Coast because it hosts the largest oil and gas market in the United States, producing significantly more carbon emissions per year than any other state (683 MtCO2 or million tonnes of CO2), almost double the CO2 emissions of the trailing state of California (358 MtCO2, Figure 1) for the year 2019. Could there be an opportunity for direct facility-to-reservoir CO2 pipelines in this region? Here, we explore the Texas Gulf Coasts viability as a hub for CO2 storage by performing CO2 pipeline modeling from sources to sinks and a cost analysis. We then consider potential risks.
Emission datasets and method
The 2021 EPA Greenhouse Gas Reporting Program (GHGRP) data (https://www.epa.gov/system/files/other-files/2022-10/2021_data_summary_spreadsheets.zip, last accessed Feb. 13, 2023) contains 828 stationary carbon emitters for Texas that are responsible for 372.7 MtCO2 for the year 2021. These sources comprise many different types, including electrical (coal/gas), petrochemical manufacturing, petroleum refineries, hydrogen production, and ethylene to name a few. We restrict our analysis to the stationary carbon emitters within the quadrangle spanned by Beaumont, Waco, San Antonio, and Corpus Christi (see Figure 2). More specifically, we analyze the top 23 CO2 emitters (Figure 2a) (>2.5 MtCO2 per year) in this area, which account for 48% or 130.7 MtCO2 per year of the total 271.6 MtCO2 per year (in the quadrangle). We then choose two potential carbon storage reservoirs along the Gulf Coast for analysis, namely the Bayou Bend CCS (Carbon Capture and Storage) and Freeport LNG CCS sites (Figure 2b). These two sites are used for the purpose of this study but by no means are the only available sites. Bayou Bend covers 40,000 acres of land and is expected to be able to accommodate 225-275 MtCO2 while Freeport LNG is 500 acres and is estimated to accommodate 25 MtCO2 (Talos Energy January 2023 Investor Presentation, https://s201.q4cdn.com/120347489/files/doc_presentations/2023/01/2023.01.05-Talos-Energy-January-2023-Discussion-Materials-vFINAL.pdf). You can find more information about these sites at Talos Energy’s website (https://www.talosenergy.com/operations/carbon-capture-and-sequestration/default.aspx).
We used the SimCCS2.0 Network Optimization (Middleton, Yaw, Hoover et al., 2020) to find the most cost-effective network of pipelines to transport a required amount of CO2 to the two named GCS sites within a given timeline. We choose to find a pipeline that will fill both GCS reservoirs (300 MtCO2) over the course of 10 years. To find the optimal pipeline network, we need a mathematical description of what a pipeline route costs to traverse. Since we used the SimCCS Gateway (Pamidighantam, Wang, Christie et al., 2020; Ellett, Wang, Christie et al., 2021) to solve the optimization problem, we will briefly describe the cost function and numerical solution process SimCCS uses. For a more complete description, please see Middleton, Yaw, Hoover et al. (2020). The objective function we wish to minimize consists of four main pieces: the capture cost, pipeline use cost, pipeline build cost, and storage cost. These costs along with standard constraints that require upholding engineering principles such as continuity of mass at pipeline junctions create a constrained optimization problem. More specifically, this is framed as a mixed integer programming problem. The approach in SimCCS2.0realizes a solution based on new linearly approximated CO2 transport costs as opposed to the previous discrete fixed and variable CO2 transport costs, making much larger optimization problems tractable. Now that we have a formulation to minimize the cost of a network, we need to know what networks are available to take. This collection of paths is called the candidate network and is calculated from a cost map. The cost map is a combination of many different raster data with weighted costs associated with moving from pixel to pixel. In our case, the cost map contains land use data, road data, topography data, etc. to create viable paths to consider between all the nodes in the graph of sources and sinks. Now that we have a computational graph of sources, sinks, and a candidate network of viable paths between them, we can compute solution networks using the SimCCS2.0 software.
Pipeline Modeling Results
We considered two GCS cases: cumulatively injecting 100 MtCO2 and 300 MtCO2 (estimated max capacity), respectively, over a 10-year period. Figure 3 shows the pipeline network solutions to the optimization problem previously described. In each case the algorithm considers all configurations of sources and sinks along the candidate network and choses the least cost network of pipelines based on the required total capture amount. The large red and blue filled circles represent the sources and sinks, respectively, which are labeled in the chart to the right of the map. The grey lines and small red outlined circles represent the candidate network, and the green lines show the solution network.
The solution for the 100 MtCO2 scenario (Figure 3a) is a pipeline that stretches 567 km. It extends from the Bayou Bend CCS site to the Air Products Port Arthur facility and W.A. Parish plants, then diverging to the Corpus Christi Liquefaction center and Freeport LNG CCS site. Additionally, the pipeline diameters range from 12 to 16 inches. The solution for the 300 MtCO2 scenario (Figure 3b) follows the same path as in the 100 MtCO2 scenario except that it also extends from the W.A. Parish plant to the Sam Seymour plant near La Grange, TX. This additional stretch of pipeline required to meet the 300 MtCO2 target makes the total length 698 km. The extra CO2 accommodation also requires that the pipelines that connect the Sam Seymour and W.A. Parish plants to the Bayou Bend CCS site have diameters up to 30 inches.
An interesting feature of these networks is that they do not simply connect the sinks to the closest sources. In fact, the cost of transport is significantly less important than the cost of capture and storage (see Table 1). In both scenarios, the pipeline near Bayou Bend GCS passes by the Motiva and Exxon Mobil refineries that combined produce 11.28 MtCO2 per year. The solution network instead connects only the Air Products Port Arthur facility, which produces 2.63 MtCO2 per year, before making a connection to W.A. Parish coal-fired power plant. This is because CO2 capture costs vary significantly by sector (Figure 4), depending on CO2 concentration in gas stream, plant location, energy supply, etc. (IEA, 2020), making cost optimal source-sink connections non-obvious. In this case, the Air Products Port Arthur facility has capture costs that are 3.6 times lower than both the Motiva and Exxon Mobil refineries. A similar example that is also present in both solutions is the connection made from the W.A. Parish plant to the Corpus Christi Liquefaction center. The cost optimal pipeline passes the Formosa Point Comfort (5 MtCO2 per year) and Coleto Creek (3.5 MtCO2 per year) plants to connect to the much further Corpus Christi Liquefaction center (3 MtCO2 per year). Again, this is because the Corpus Christi Liquefaction center has capture costs that are 4.7 and 3.3 times lower than the Formosa Point Comfort and Coleto Creek plants, respectively.
Costs
Our cost analysis (see Table 1) shows that to optimally capture, transport, and store 30 MtCO2 over the course of 10 years would require an estimated $1.22B/year in capture costs, $46.5M/year in transportation costs, and $1.4B/year in storage costs. This amounts to an estimated $26.5B spent over 10 years for a 698 km pipeline with diameter ranging from 12 to 30 inches to potentially offset just ~0.5% of the annual U.S. CO2 output or ~44% of annual Texas output, for the year 2019. Clearly, more ambitious goals and aggressive actions are needed to meet carbon net neutrality. The capture cost and the injection cost are comparable, but they are much higher (~10x-30x) than the transportation cost. Specific offshore conditions can also significantly increase the storage and transport costs. We have included the increased storage factor in our cost analysis and simulation by assuming that the Bayou Bend CCS storage cost is $50/tCO2 while the Freeport LNG site is $5/tCO2. Storage costs for offshore reservoirs can easily cost upwards of $100/tCO2 while most onshore storage is below $10/tCO2, low enough to be profitable through government tax credits. Additionally, offshore pipelines can have increased costs of 40-70% compared to onshore (IEA, 2020), making it important to follow the economy of scale of pipeline costs for future offshore storage. Our findings demonstrate the important role of stationary CO2 sources and injection sites in both pipeline modeling and cost. Significant reduction of cost must involve reduction of the capture costs, storage costs or both.
GCS Risks
Texas coasts have the Corsair fault systems (Treviño and Meckel, 2017), which can pose seismic hazards and cause CO2 leakage. Sudden emission of large amount of CO2 can threaten nearby lives. For instance, CO2 eruption in Lake Nyos in 1986 (Kling, Clark, Wagner et al., 1987) in Cameroon killed 1700 people. In this regard, offshore storage sites are clearly advantageous compared to onshore populated places. The offshore Texas subsurface consists of alternating layers of sandstones and shales (Treviño and Meckel, 2017). The layer thicknesses are about 200-300 m. The top of the overpressure zone is at a depth of 3000 m. CO2 can be injected at its supercritical state into the sandstones, deeper than 1000 m but above the overpressure depth. Using the earthquake source scaling law, we note that if an earthquake magnitude is > M3, we should be concerned because it may rupture a length longer than the layer thickness. However, an M3 earthquake will not be felt by humans. Therefore, a seismic monitoring system should be installed to monitor the subsurface for possible leakage.
Fluid injection induced seismicity has already been a considerable risk for wastewater disposal. Earthquake events were reported near the fluid-injection sites in U.S and Canada (Ellsworth, 2013). The relation between injection operation and occurrence of earthquakes is frequently speculated by the pore-pressure rise which decreases the effective stress. When the effective stress drops below the Coulomb failure threshold, the fault rupture and slip might occur to create an earthquake. Although the seismic monitoring around the injection sites is called micro-seismic monitoring, several significant earthquakes have already happened. Three M5.0+ earthquakes occurred near Fairview, Pawnee and Cushing, Oklahoma in February, September, and November 2016, respectively. A statistical study found that the cumulative seismic moment has a positive correlation with the fluid injection volume (McGarr and Barbour, 2017). Last year, an M5.4 earthquake occurred in the west Texas which was the largest earthquake in the past three decades. Another M5.0 earthquake occurred in Permian Basin in March 2020 possibly induced by wastewater disposal. A recent study compared earthquake catalog near Pecos, Texas with industrial operations around this city, the authors concluded that the majority of seismicity is likely induced by an increase of wastewater disposal operations (Skoumal and Trugman, 2021).
According to the past study of induced seismicity related to the fluid injection, it is reasonable to consider the risks of induced earthquake of GCS projects. We collected the micro-seismic monitoring results in other operating GCS sites worldwide (see Table 2). The typical injection rate is about 1 MtCO2 per year at each site. The largest induced earthquake magnitude is M1.6 occurred in In Salah, Algeria. This small earthquake magnitude might be because of the small amount of injection and the sparse injection site. However, this is not the excuse to ignore the earthquake risks of CO2 injection because the gigatonne GCS is 3 orders of magnitude bigger than what has been done. The CO2 in the rock not only can reduce the effective stress by increasing the pore pressure, it could also damage the caprock integrity by chemical erosion (Vialle and Vanorio, 2011; Pimienta, Esteban, Sarout et al., 2017). The effect of long-time contact between the rock interface and the acidic CO2 solution needs more study. Chemical reactions might also introduce different fault nucleation mechanism than current chemical inert fluid such as water.
Earthquake triggering in fluid saturated regions by passing seismic waves is another concern. Recently, we observed a dynamic pressure amplification in the fluid-filled fracture by a laboratory experiment (Jin, Dyaur and Zheng, 2021). This phenomenon says that the fluid pressure in the fracture could be amplified at least one to two orders of magnitude when a seismic wave passes by. We call this the transient pressure surge effect, which introduces new possibilities that could trigger earthquakes in fluid saturated regions. The seismic waves could dynamically change the pore pressure near the fluid-filled fracture, which might lower the effective stress on the fault plane in transient and make the fault slip. If the fracture is filled with supercritical CO2, the large transient pressure variation might lead to phase transition such as bubbling. High pressure CO2 injected in the formation involves processes of multiple time and space scales. We need to understand the host of issues using interdisciplinary knowledge including rock physics, geomechanics, geochemistry, seismology, and geophysical instrument engineering.
Discussion
How far are we from achieving the gigatonne GCS per year? We do simple scaling and find that we need ~33 such GCS systems studied in this article to achieve this grand goal. The annual cost would be ~$87B to store one gigatonne of CO2 in the subsurface. The Bayou Bend and Freeport LNG CCS projects can contribute marginally toward the gigatonne goal per annum. Although there are currently not many active CCS operations, the U.S. Gulf Coast is estimated to have 310 GtCO2 storage potential (USGS, 2013). With increases in the Inflation Reduction Acts 45Q tax credit (H.R.5376 - 117th Congress, 2021-2022), carbon storage may become more financially viable and even potentially economical.
CO2 injection needs to be done in Class VI wells, regulated by EPA or states with primacy status. As of January 25, 2023, the number of ‘active’ Class VI wells in the U.S. is two, both in the state of Illinois. There are 34 permission applications in ‘pending’ status but only one is in Texas in the Ector County. There are currently zero active/pending Class-VI well applications in Texas Gulf coast. If one well injects 1 MtCO2 per year (see Table 1) according to experience in other sites, we need 1,000 wells for the Gigatonne scenario. It means the number of permission applications must increase rapidly by 2-3 orders of magnitude and the approval process must also be accelerated.
Conclusions
We evaluated GCS potential in the offshore Texas state waters in the Gulf of Mexico, stretching from Corpus Christi to Port Arthur. Stationary CO2 emission sources and two potential GCS sites in Bayou Bend and Free Port are used in the study. Pipeline modeling shows the optimal network solutions seem to favor building longer pipelines to large, low-cost capture sources rather than clusters of smaller, high-cost capture sources. The capture cost and the injection cost are comparable, but they are much higher (~10x-30x) than the transportation cost. A survey of other CO2 injection sites worldwide shows the maximum magnitude of an induced earthquake is about 1.6 at low injection rate ~1 MtCO2 per year. In the gigatonne-per-year scenario, dynamic earthquake triggering should be considered in risk assessment. In the Texas state waters, a triggered or induced earthquake of magnitude ~3 is of concern as it may rupture the seal of the subsurface storage unit. To achieve gigatonne GCS per annum, aggressive actions are needed to increase the number of Class VI wells by 3 orders of magnitude, and we need 33 GCS systems studied in this work.
Acknowledgments
The authors would like to thank the support from the endowment of the Robert and Margaret Sheriff Professorship in Applied Geophysics, held by Y.Zheng. We also thank Marlon Pierce of Indiana University for giving us access to the SimCCS gateway. Any possible mistakes are ours.
References
Ellett, K., J. Wang, M. Christie, S. Pamidighantam, E. Abeysinghe, R. Kammer, R. Middleton,
B. Hoover, S. Yaw, N. Wei, and X. Li (2021), Development of a Science Gateway Software Platform for CCS Decision Support and Stakeholder Engagement (March 15, 2021). , Proceedings of the 15th Greenhouse Gas Control Technologies Conference 15-18 March 2021, Available at SSRN: https://ssrn.com/abstract=3821941 or http://dx.doi.org/3821910.3822139/ssrn.3821941,
Ellsworth, W. L. J. S. (2013), Injection-induced earthquakes, 341(6142), 1225942.
Friedlingstein, P., et al. (2022), Global Carbon Budget 2021, Earth Syst. Sci. Data, 14(4),
1917-2005.
Glubokovskikh, S., E. Saygin, S. Shapiro, B. Gurevich, R. Isaenkov, D. Lumley, R. Nakata,
J. Drew, and R. J. G. R. L. Pevzner (2022), A Small CO2 Leakage May Induce Seismicity on a Sub‐Seismic Fault in a Good‐Porosity Clastic Saline Aquifer, 49(12), e2022GL098062.
Greig, C., and A. Pascale (2021), Annex I: CO2 Transport and Storage Infrastructure
transition analysis in Princeton’s Net-Zero America study pp.44 pp.
H.R.5376 - 117th Congress (2021-2022): Inflation Reduction Act of 2022.
(2022, August 16). https://www.congress.gov/bill/117th-congress/house-bill/5376
IEA (2020), CCUS in Clean Energy Transitions, IEA,
Paris https://www.iea.org/reports/ccus-in-clean-energy-transitions, License: CC BY 4.0
Jin, Y., N. Dyaur, and Y. J. T. S. R. Zheng (2021), Laboratory evidence of transient pressure
surge in a fluid‐filled fracture as a potential driver of remote dynamic earthquake triggering, 1(2), 66-74.
Kaven, J. O., S. H. Hickman, A. F. McGarr, S. Walter, and W. L. J. E. P. Ellsworth (2014), Seismic
monitoring at the Decatur, IL, CO2 sequestration demonstration site, 63, 4264-4272.
Kling, G. W., M. A. Clark, G. N. Wagner, H. R. Compton, A. M. Humphrey, J. D. Devine, W. C.
Evans, J. P. Lockwood, M. L. Tuttle, and E. J. Koenigsberg (1987), The 1986 Lake Nyos Gas Disaster in Cameroon, West Africa, Science, 236(4798), 169-175.
McGarr, A., and A. J. J. G. R. L. Barbour (2017), Wastewater disposal and the earthquake
sequences during 2016 near Fairview, Pawnee, and Cushing, Oklahoma, 44(18), 9330-9336.
Middleton, R. S., S. P. Yaw, B. A. Hoover, and K. M. Ellett (2020), SimCCS: An open-source tool
for optimizing CO2 capture, transport, and storage infrastructure, Environ Modell Softw, 124, 104560.
O'Brien, S. (2017), The Quest CCS Project-MMV Technology Deployment Through Two
Years of Operation, AGU Fall Meeting Abstracts, 2017, S54A-01,
Pamidighantam, S., J. Wang, M. Christie, E. Abeysinghe, K. Ellett, R. Kammer, S. Yaw,
R. Middleton, and B. Hoover (2020), A Science Gateway for Simulating the Economics of Carbon Sequestration Technologies: SimCCS2.0, in Practice and Experience in Advanced Research Computing, edited,vol. pp. 262–269,PEARC '20,Series Vol. Series editor Association for Computing Machinery, New York, NY, USA.
Parker, N. (2004), Using Natural Gas Transmission Pipeline Costs to Estimate Hydrogen
Pipeline Costs, UC Davis: Institute of Transportation Studies. Retrieved from https://escholarship.org/uc/item/2gk0j8kq.
Pimienta, L., L. Esteban, J. Sarout, K. Liu, J. Dautriat, C. Delle Piane, and M. B. J. I. J. o. G. G. C.
Clennell (2017), Supercritical CO2 injection and residence time in fluid-saturated rocks: Evidence for calcite dissolution and effects on rock integrity, 67, 31-48.
Sawada, Y. (2019), Progress of Tomakomai CCS Demonstration Project, JCCS Report.
Skoumal, R. J., and D. T. J. J. o. G. R. S. E. Trugman (2021), The proliferation of induced
seismicity in the Permian Basin, Texas, 126(6), e2021JB021921.
Stork, A. L., J. P. Verdon, and J.-M. J. I. J. o. G. G. C. Kendall (2015), The microseismic response
at the In Salah Carbon Capture and Storage (CCS) site, 32, 159-171.
Treviño, R. H., and T. A. Meckel (2017), Geological CO2 Sequestration Atlas of Miocene
Strata, Offshore Texas State Waters, The University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 283, pp.74.
U.S. Geological Survey Geologic Carbon Dioxide Storage Resources Assessment Team, 2013,
National assessment of geologic carbon dioxide storage resources—Results (ver. 1.1, September 2013): U.S. Geological Survey Circular 1386, 41 p.,
https://pubs.usgs.gov/circ/1386/. (Supersedes ver. 1.0 released June 26, 2013.)
Verdon, J. P., J.-M. Kendall, A. L. Stork, R. A. Chadwick, D. J. White, and R. C. J. P. o. t. N. A. o. S.
Bissell (2013), Comparison of geomechanical deformation induced by megatonne-scale CO2 storage at Sleipner, Weyburn, and In Salah, 110(30), E2762-E2771.
Verdon, J. P. J. I. J. o. G. G. C. (2016), Using microseismic data recorded at the Weyburn
CCS-EOR site to assess the likelihood of induced seismic activity, 54, 421-428.
Vialle, S., and T. J. G. R. L. Vanorio (2011), Laboratory measurements of elastic properties
of carbonate rocks during injection of reactive CO2‐saturated water, 38(1).
The views, interpretations, and conclusions expressed within these articles represent those of the author (authors or other entities) and are not necessarily shared or representative of the GSH, GSHJ, or any other entity associated with the journal or society. The GSHJ does not claim to be or represent a peer-reviewed manuscript publication.