Earthquake triggering and large-scale geologic storage of carbon dioxide

Abstract

Despite its enormous cost, large-scale carbon capture and storage (CCS) is considered a viable strategy for significantly reducing CO₂ emissions associated with coal-based electrical power generation and other industrial sources of CO₂ [Intergovernmental Panel on Climate Change (2005) IPCC Special Report on Carbon Dioxide Capture and Storage. Prepared by Working Group III of the Intergovernmental Panel on Climate Change, eds Metz B, et al. (Cambridge Univ Press, Cambridge, UK); Szulczewski ML, et al. (2012) Proc Natl Acad Sci USA 109:5185–5189]. We argue here that there is a high probability that earthquakes will be triggered by injection of large volumes of CO₂ into the brittle rocks commonly found in continental interiors. Because even small- to moderate-sized earthquakes threaten the seal integrity of CO₂ repositories, in this context, large-scale CCS is a risky, and likely unsuccessful, strategy for significantly reducing greenhouse gas emissions.

The combustion of coal for electrical power generation in the United States generates approximately 2.1 billion metric tons of CO₂ per year, ∼36% of all US emissions. In 2011, China generated more than three times that much CO₂ by burning coal for electricity, which accounted for ∼80% of its total emissions. (According to the Energy Information Agency of the US Department of Energy, total CO₂ emissions in China were 8.38 billion metric tonnes in 2011, with 6.95 billion tons from coal burning, nearly all of which is used electrical power generation.) From a global perspective, if large-scale carbon capture and storage (CCS) is to significantly contribute to reducing the accumulation of greenhouse gases, it must operate at a massive scale, on the order of 3.5 billion tons (1) of CO₂ per year, a volume roughly equivalent (2) to the ∼27 billion barrels of oil currently produced annually around the world. (Under reservoir conditions, one billion tons of CO₂ occupies a volume of ∼1.3 billion cubic meters, equivalent to 8.18 billion barrels. Thus, 3.5 billion tons of carbon dioxide would correspond to a volume of approximately 28.6 billion barrels. There are currently ∼850,000 wells producing oil around the world.) Moreover, a leak rate from underground CO₂ storage reservoirs of less than 1% per thousand years is required for CCS to achieve the same climate benefits as renewable energy sources (3).

Before embarking on projects to inject enormous volumes of CO₂ at numerous sites around the world, it is important to note that over time periods of just a few decades, modern seismic networks have shown that earthquakes occur nearly everywhere in continental interiors. Fig. 1, Upper shows instrumentally recorded earthquakes in the central and eastern United States and southeastern Canada. Fig. 1, Lower shows instrumentally recorded intraplate earthquakes in south and east Asia (4). The seismicity catalogs are complete to magnitude (M) 3. The occurrence of these earthquakes means that nearly everywhere in continental interiors a subset of the preexisting faults in the crust is potentially active in the current stress field (5, 6). This is sometimes referred to as the critically stressed nature of the brittle crust (7). It should also be noted that despite the overall low rate of earthquake occurrence in continental interiors, some of the most devastating earthquakes in history occurred in these regions. In eastern China, the M 7.8, 1976 Tangshan earthquake, approximately 200 km east of Beijing, killed several hundred thousand people. In the central United States, three M 7+ earthquakes in 1811 and 1812 occurred in the New Madrid seismic zone in southeast Missouri.

Fig. 1.

Upper: Instrumentally recorded seismicity and damaging historical earthquakes in the central and eastern United States and southeastern Canada. Red dots indicate sites of reservoir-induced seismicity. Lower: Seismicity of south and east Asia and sites of reservoir-induced seismicity. Both data sets are available from the US Geological Survey (4).

Because of the critically stressed nature of the crust, fluid injection in deep wells can trigger earthquakes when the injection increases pore pressure in the vicinity of preexisting potentially active faults. The increased pore pressure reduces the frictional resistance to fault slip, allowing elastic energy already stored in the surrounding rocks to be released in earthquakes that would occur someday as the result of natural geologic processes (8). This effect was first documented in the 1960s in Denver, Colorado when injection into a 3-km-deep well at the nearby Rocky Mountain Arsenal triggered earthquakes (9). Soon thereafter it was shown experimentally (10) at the Rangely oil field in western Colorado that earthquakes could be turned on and off by varying the rate at which water was injected and thus modulating reservoir pressure. In 2011 alone, a number of small to moderate earthquakes in the United States seem to have been triggered by injection of wastewater (11). These include earthquakes near Guy, Arkansas that occurred in February and March, where the largest earthquake was M 4.7. In the Trinidad/Raton area near the border of Colorado and New Mexico, injection of produced water associated with coalbed methane production seems to have triggered a number of earthquakes, the largest being a M 5.3 event that occurred in August. Earthquakes seem to have been triggered by wastewater injection near Youngstown, Ohio on Christmas Eve and New Year’s Eve, the largest of which was M 4.0. Although the risks associated with wastewater injection are minimal and can be reduced even further with proper planning (11), the situation would be far more problematic if similar-sized earthquakes were triggered in formations intended to sequester CO₂ for hundreds to thousands of years.

Deep borehole stress measurements confirm the critically stressed nature of the crust in continental interiors (12), in some cases at sites directly relevant to the feasibility of large-scale CCS. For example, deep borehole stress measurements at the Mountaineer coal-burning power plant on the Ohio River in West Virginia indicate a severe limitation on the rate at which CO₂ could be injected without the resulting pressure build-up initiating slip on preexisting faults (13). Because of the low permeability of the formations at depth, pore pressure increases would be expected to trigger slip on preexisting faults if CO₂ injection rates exceed approximately 1% of the 7 million tons of CO₂ emitted by the Mountaineer plant each year. Similarly, stress measurements at Teapot Dome, Wyoming, the US government-owned oil field where pilot CO₂ injection projects have been considered, show that very small pressure buildups are capable of triggering slip on some preexisting faults (14).

Dam construction and water reservoir impoundment produce much smaller pore pressure changes at depth than are likely to occur with CO₂ sequestration, but many have triggered earthquakes at various sites around the world (15) (red dots in Fig. 1). Except for the much smaller pore pressure increases at depth, reservoir-triggered earthquakes are a good analog for the potential for seismicity to be triggered by CO₂ injection. Both activities cause pore pressure increases that act over large areas and are persistent for long periods. Three reservoir impoundments in eastern Canada (located in the ancient, stable core of the North American continent) triggered earthquakes as large as M 4.1 and M 5 at the two sites (Fig. 1), despite the fact that the pore pressure increases at depth were extremely small.

Triggered Earthquakes and Seal Integrity

Our principal concern is not that injection associated with CCS projects is likely to trigger large earthquakes; the problem is that even small to moderate earthquakes threaten the seal integrity of a CO₂ repository. In parts of the world with good construction practices, it is unusual for earthquakes smaller than approximately M 6 to cause significant human harm or property damage. Fig. 2 uses well-established seismological relationships to show how the magnitude of an earthquake is related to the size of the fault that slipped and the amount of fault slip that occurred (16). As shown, faults capable of producing M ∼6 earthquakes are at least tens of kilometers in extent. (The fault size indicated along the abscissa is a lower bound of fault size as it refers to the size of the fault segment that slips in a given earthquake. The fault on which an earthquake occurs is larger than the part of the fault that slips in an individual event.) In most cases, such faults should be easily identified during geophysical site characterization studies and thus should be avoided at any site chosen for a CO₂ repository. (Faults in crystalline basement rocks might be difficult to recognize in geophysical data. We assume, however, that any site chosen as a potential CO₂ repository would be carefully selected, avoiding the possibility of pressure changes in the CO₂ repository from affecting faults in crystalline basement.) The problem is that site characterization studies can easily miss the much smaller faults associated with small to moderate earthquakes.

Fig. 2.

Relationships among various scaling parameters for earthquakes. The larger the earthquake, the larger the fault and amount of slip, depending on the stress drop in a particular earthquake. Observational data indicate that earthquake stress drops range between 0.1 and 10 MPa.

Although the ground shaking from small- to moderate-sized earthquakes is inconsequential, their impact on a CO₂ repository would not be. Most of the geologic formations to be used for long-term storage of CO₂ are likely to be at depths of approximately 2 km—deep enough for there to be adequate sealing formations to isolate the CO₂ from the biosphere but not so deep as to encounter formations with very low permeability. Given large volumes of CO₂ injected into selected formations for many decades, if a small to moderate earthquake were to be triggered in a geologic formation at approximately 2 km depth, it could jeopardize the seal integrity of the storage formation. For example, if a M ∼4 earthquake were to be triggered by CO₂ sequestration (17)—an event that would be widely felt in a populated area but for which shaking would be unlikely to cause harm or damage—it would be associated with several centimeters of slip on a fault several kilometers in size. Because laboratory studies show that just a few millimeters of shear displacement are capable of enhancing fracture and joint permeability (18), several centimeters of slip would be capable of creating a permeable hydraulic pathway that could compromise the seal integrity of the CO₂ reservoir and potentially reach the near surface.

Safe Sequestration

It is important to emphasize that CCS can be valuable and useful for reducing greenhouse gas emissions in specific situations. A good example is the injection of CO₂ into the Utsira formation (19) at the Sleipner gas field in the North Sea, where a significant amount of CO₂ is coproduced with natural gas. After separating the CO₂ from the produced gas, approximately 1 million tons of CO₂ per year has been injected over the past 15 y without triggering seismicity. Assuming isolation from the near surface, injection into highly porous and permeable reservoirs that are laterally extensive would produce small increases in pressure in response to CO₂ injection. Moreover, weak, poorly cemented sandstones are expected to deform slowly in response to applied geologic forces. In such reservoirs, the stresses relax over time, and such formations are not prone to faulting (20). In this regard, the Utsira formation is ideal for CO₂ sequestration. It is isolated from vertical migration by impermeable shale formations, and it is highly porous, permeable, laterally extensive, and weakly cemented.

To contribute significantly to greenhouse gas emission reductions (2), roughly 3,500 sites similar to the Utsira formation would have to be found at convenient locations around the world, assuming comparable injection rates of approximately 1 million tons of CO₂ per year. In fact, it would take approximately 85 such sites coming on line each year to reach a goal of storing approximately 1 billion tons of CO₂ by midcentury. Clearly this is an extraordinarily difficult, if not impossible, task if only highly porous and permeable and weakly cemented formations are to be used.

Of course, rather than using potentially problematic geologic formations close to coal-burning power plants for sequestration (as illustrated by the Mountaineer case study cited above), relatively ideal formations for CO₂ storage could be sought on a regional basis to accommodate emissions from a number of plants. One example of this is the potential use of the Mt. Simon sandstone in the Illinois basin. The Mt. Simon is porous, permeable, and regionally extensive. However, models of injection of 100 million tons of CO₂ per year for 40 y predicts (21) increases in pore pressure of several megapascals over a region of ∼40,000 km². The approximate area of significantly increased pore pressure resulting from injection is shown as the blue-shaded area in Fig. 3, essentially adjacent to the Wabash fault zone, where a series of moderate natural earthquakes occurred in the spring of 2008, the largest being M 5.2. Paleoseismic data indicate the occurrence of much larger nearby earthquakes (some greater than M ∼7) in the recent geologic past (22). Importantly, the 100 million ton annual CO₂ injection rate used in the modeling only represents approximately one seventh of the CO₂ generated by the coal-burning power plants in the Ohio River Valley alone.

Fig. 3.

Instrumentally recorded seismicity in the New Madrid and Wabash Valley seismic zones (modified from ref. 23). Red circles indicate earthquakes that occurred from 1974 to 2002 with magnitudes larger than 2.5 located using modern instruments. Green circles denote earthquakes that occurred before 1974. Larger earthquakes are represented by larger circles. The area shown in blue corresponds to the area where a pressure increase of several megapascals would result from injecting 100 million tons per year of CO₂ into the Mt. Simon sandstone in the Illinois Basin for 40 y (19).

Because of the need to carefully monitor CO₂ repositories with observation wells, geophysical and geochemical monitoring systems, etc., it is likely that most sites will have to be located on land or very near shore. Otherwise, highly porous reservoirs located offshore, like those adjacent to salt domes along the US Gulf Coast, would be relatively ideal sites because salt formations are known to be excellent seals for hydrocarbons.

Depleted oil and gas reservoirs are potentially suitable for CO₂ storage for a variety of reasons—an infrastructure of wells and pipelines exist, and there is a great deal of geologic and subsurface property data available to characterize the subsurface from decades of study. In addition, from an earthquake-triggering perspective, depleted reservoirs are attractive because at the time injection of CO₂ might start, the pore pressure would be below the value that existed before petroleum production. Thus, there could be significant injection of CO₂ before pressures increase to preproduction values, thereby reducing the potential for triggering earthquakes.

There are a number of potential issues to consider before using depleted oil and gas reservoirs for CO₂ storage, the most important of which are capacity and geographic distribution. The reasons that there is such interest in using saline aquifers for CO₂ storage is that they are potentially well distributed with respect to likely sources of CO₂, and they could presumably accommodate the enormous volumes of CO₂ that need to be stored. If one were only to consider the United States, storing the 2.1 billion tons of CO₂ currently generated annually by coal-burning power plants in depleted oil and gas reservoirs would require injection of CO₂ at a rate of approximately 17 billion barrels per year; a rate equivalent to eight times current US annual oil production and more than four times US peak annual oil production that occurred in the early 1970s. In addition, it is important to make sure that production-related activities, such as water flooding during secondary recovery, did not compromise the seal capacity of the reservoirs. There also needs to be careful study of the wells in the depleted oil or gas field to make sure that poorly cemented well casings, especially in older wells, will not be pathways for release of stored CO₂ (23). Finally, there are likely to be complicated legal questions concerning ownership and liability that will need to be worked out on a case-by-case basis.

Although enhanced oil recovery (EOR) using CO₂ (in which CO₂ is injected to dissolve in oil and reduce its viscosity) would be a beneficial use of CO₂, it is important not to confuse this with CCS. In CCS the goal is to inject large quantities of CO₂ into available pore space and store it there for hundreds to thousands of years. When CO₂ is used for EOR, the CO₂ dissolved in the oil is separated and captured from produced oil and then reinjected. Thus, smaller volumes of CO₂ are used, and the long-term storage capacity of the reservoir is not an issue.

Many CCS research projects are currently underway around the world. Much of this work involves characterization and testing of potential storage formations and includes a number of small-scale pilot injection projects. Because the storage capacity/pressure build-up issue is critical to assess the potential for triggered seismicity, small-scale pilot injection projects do not reflect how pressures are likely to change (increase) once full-scale injection is implemented. Moreover, even though limitations on pressure build-up are among the many factors that are evaluated when potential formations are considered as sequestration sites, this is usually done in the context of not allowing pressures to exceed the pressure at which hydraulic fractures would be initiated in the storage formation or caprock. In the context of a critically stressed crust, slip on preexisting, unidentified faults could trigger small- to moderate-sized earthquakes at pressures far below that at which hydraulic fractures would form.

As mentioned above, sequences of small to moderate earthquakes were apparently induced by injection of waste water near Guy, Arkansas, Trinidad, Colorado, and Youngstown, Ohio in 2011 and on the Dallas-Ft. Worth airport, Texas. Although these earthquakes were widely felt, they caused no injury, and only the Trinidad earthquake resulted in any significant damage. However, had similar earthquakes been triggered at sites where CO₂ was being injected, the impacts would have raised pressing and important questions: Had the seal been breached? Was it still safe to leave previously injected CO₂ in place?

In summary, multiple lines of evidence indicate that preexisting faults found in brittle rocks almost everywhere in the earth’s crust are subject to failure, often in response to very small increases in pore pressure. In light of the risk posed to a CO₂ repository by even small- to moderate-sized earthquakes, formations suitable for large-scale injection of CO₂ must be carefully chosen. In addition to being well sealed by impermeable overlaying strata, they should also be weakly cemented (so as not to fail through brittle faulting) and porous, permeable, and laterally extensive to accommodate large volumes of CO₂ with minimal pressure increases. Thus, the issue is not whether CO₂ can be safely stored at a given site; the issue is whether the capacity exists for sufficient volumes of CO₂ to be stored geologically for it to have the desired beneficial effect on climate change. In this context, it must be recognized that large-scale CCS will be an extremely expensive and risky strategy for achieving significant reductions in greenhouse gas emissions.