At the end of 2003, the American Geophysical Union issued an unequivocal message: "Human activities are increasingly altering the Earth's climate. . . . A particular concern is that atmospheric levels of carbon dioxide may be rising faster than at any time in Earth's history, except possibly following rare events like impacts from large extraterrestrial objects."

Barring an asteroid strike, human production of CO₂ is not likely to slow down soon — and there's an urgent need to find somewhere besides the atmosphere to put it. Sequestration in the ocean and in soils and forests are possibilities, but another option, sequestration in geological formations, already has a head start.

"The likeliest places to store CO₂ are depleted oil and gas reservoirs," says Karsten Pruess of Berkeley Lab's Earth Sciences Division. He notes that porous reservoir rock can soak up gas, and caps of impermeable rock can seal that gas in place over geologic lengths of time. "The very existence of natural gas in a reservoir proves it is capable of long-term storage."

Karsten Pruess, Tianfu Xu, and John Apps applied the TOUGHREACT simulation codes to model carbon sequestration in Gulf Coast geological formations.

Energy companies have long practiced pumping CO₂ back into the ground to force more oil out of depleting reservoirs. Unfortunately, says Pruess, "depleted reservoirs don't have enough capacity to store the huge amounts of carbon dioxide we're producing, and often they're too far from stationary sources, like power plants and refineries, where it's practical to collect CO₂."

An attractive storage alternative to depleted reservoirs is saline aquifers. In these formations, beds of porous sandstones containing very salty water — with a salinity comparable to that of sea water, thus not a potential source of drinking water — alternate with impermeable shales.

The only place where carbon dioxide injection into a saline aquifer is already in use is the Sleipner natural-gas field in the North Sea, operated by Statoil of Norway. Some of the natural gas tapped by Statoil contains 9 percent CO₂, which has to be reduced to 2.5 percent before it can be sold. After removal at a treatment platform at sea, the excess CO₂ is reinjected into a massive sandstone formation a kilometer deep.

These are the ideal elements of geologic sequestration: concentrated CO₂ production close to deep saline aquifers. One region in the United States, the Gulf Coast of Texas, meets both these requirements in spectacular fashion.

A natural laboratory

Few places on Earth have a higher concentration of refineries, power plants, and other fixed CO₂ sources than the Texas Gulf Coast. At over 160 million metric tons (in 1999), Texas produces almost twice as much CO₂ as runner-up California among U.S. states, more than the entire United Kingdom.

	The Gulf Coast of Texas has a high concentration of refineries, power plants, and other fixed CO2 sources, conveniently located atop enormous beds of deep saline aquifers.

These numerous sources of recoverable carbon dioxide waste sit atop enormous beds of deep saline aquifers, which formed as the gradually-rising mountains to the west continually eroded into the warm sea that once occupied the middle of North America. Periodic inundations by the ancient seas capped the sands with beds of clay, which under increasing weight and pressure metamorphosed into impermeable layers of shale. Today these sediments lie thickly along the entire Gulf Coast of Texas, extending into Mexico, Lousiana, and Mississippi.

The Texas Bureau of Economic Geology, centered at the University of Texas at Austin, has recently begun a pilot program for carbon dioxide sequestration in the region. Pruess and other members of the Earth Sciences Division have partnered with the Bureau's Gulf Coast Carbon Center to develop computer models that can better simulate what's likely to happen when carbon dioxide and other byproducts are injected into the Gulf Coast's bedded sandstone-shale sequences.

Key to these simulations is Berkekey Lab's versatile TOUGH2 general-purpose numerical simulation program. The predecessors to TOUGH2 were created 20 years ago to model the flow of heat and fluids, in multiple phases, through porous and fractured media like sand and rocks. Originally designed for modeling fluid flow in geothermal reservoirs, the program has been applied to a wide range of studies including the movement of pollutants underground, such as nuclear waste. TOUGH2 is currently in use by approximately 300 organizations in more than 30 countries.

Realistic modeling of CO₂ injection is more complex than simply calculating its movement through sand and rock, however. A formation's storage capacity depends not only on its porosity and other structural properties but on its chemistry. And few industrial sources emit pure carbon dioxide; instead, the CO₂ is likely to be mixed with other acidic gases, including such unpleasant contaminants as hydrogen sulfide or sulfur dioxide.

Even pure carbon dioxide reacts differently with different kinds of aquifer rocks. In limestone (calcium carbonate), chemical reactions with CO₂ dissolved in brine can't increase the rock's capacity to absorb carbon by much. But in impure sandstone, the dissolved carbon dioxide may react to form solid carbonates that precipitate out of solution, fixing more carbon. And in rocks composed predominantly of magnesium iron silicates, reactions can precipitate so much carbonate it not only fixes a lot of carbon but may clog the rock's pores.

A version of TOUGH2 called TOUGHREACT was recently developed by Pruess and his colleagues Tianfu Xu, Eric Sonnenthal, and Nicolas Spycher to address processes like acid mine drainage, waste disposal, and groundwater quality, where chemistry plays a critical role. The program considers fluid dynamics (as Pruess describes it, "how to push a dense gas into the ground") in concert with chemical reactions and the effects of heat and pressure.

Geochemist John Apps helped apply TOUGHREACT to the problem of CO₂ sequestration in the sediments of the Gulf Coast's Frio and Jasper formations. The modelers set out to analyze the transfer of CO₂ and other compounds between shale and sandstone layers, the consequent immobilization of the gases through mineral precipitation, and the impact of cocontaminants like hydrogen sulfide and sulfur dioxide.

"The key to modeling these systems is to incorporate accurate geochemical data," says Apps, "and then to validate what the model predicts against actual experience in the field." Apps, Xu, and Pruess presented the results of their work at the 2003 meeting of the American Geophysical Union in San Francisco.

The chemistry of the rocks

Over the past several decades, tables of the chemical properties of thousands of rock-forming minerals have been painstakingly compiled from laboratory experiments, which seek to determine phase equilibria, heat capacities and heats of solution, and mineral solubility. These studies typically subject a few grams of a solid sample or a solution to varying pressures and temperatures using a raft of experimental devices, including furnaces and "rod bombs," together with other independent investigative techniques.

Laboratory experiments on mineral samples, like this sandstone from a deep saline aquifer on the Texas Gulf Coast, determine chemical and thermodynamic properties.

The tables are used to calculate the thermodynamic properties of sets of minerals in systems at different temperatures and pressures; one aim is to determine under what conditions minerals are stable, and when they may dissolve and exchange chemical constituents with liquid and gas phases.

"Huge studies have been made to calculate the thermodynamic properties of minerals," says Apps, citing as an example the pioneering work of his colleague Harold Helgeson, a UC Berkeley geochemist who with his co-workers published such a comprehensive review and tabulation in the late 1970s. But tables compiled by different researchers using different techniques and different sources of information may disagree significantly. "A tendency to uncritically combine data bases can be a real mistake," Apps warns.

Even with consistent and dependable measurements, problems arise in applying the data to physical processes like carbon sequestration. Apps notes that much of Helgeson's initial work "was done on minerals that form igneous and metamorphic rocks, which reach equilibrium at high temperatures and pressures. The carbon-sequestration processes we're interested in occur at relatively low temperatures and pressures. Calculation requires a large extrapolation, which increases uncertainty."

Apps characterizes his role in applying the TOUGHREACT simulation program to CO₂ sequestration in the Gulf Coast saline aquifers as one of "looking at the chemical systems, identifying the minerals involved, calculating their thermodynamic properties, and making sure there is internal consistency among all the minerals in the system." It's a task that often requires recalculating from scratch, using the original raw data of laboratory experiments or drawing on new experiments.

Although Apps has "strongly held views" about what processes still need to be incorporated and fine-tuned in the TOUGHREACT program — for example, an understanding of the rules by which various minerals precipitate out of solution in order of their relative degree of instability — he says, "Our recent simulations have been eerily similar to what we observe in the field."

What may be particularly significant, Apps says, is that by modeling "we can see the evolution of the system in all its complexity. Too often field geologists are like the blind men feeling the elephant: they find one property of the system and ignore the others. We can see the whole elephant — albeit through a fuzzy lens."

TOUGH-code pioneer Karsten Pruess is excited about the successes TOUGHREACT is already racking up. Not only has the program contributed to a better understanding of proposed carbon sequestration along the Texas Gulf Coast, TOUGHREACT is also being adapted for applications as far afield as the Caspian Sea, where sulfur-laden crude oil and "sour gas" pose particular problems of separation and sequestration.

Deep saline aquifers with potential for sequestering carbon are mapped in green. Yellow areas mark aquifers that are too shallow for storage; blue areas are too deep.

"The development of these codes has always been problem-driven," says Pruess. "Although much of our attention has been devoted to reservoir engineering and to studying the movement of wastes and pollutants, the recent work with the Gulf Coast saline aquifers has addressed the important question of whether we can use these codes to model natural systems. And indeed, we've shown that TOUGHREACT simulations handle these very well."

It would be hard to overstate the potential benefits. According to the Department of Energy's Office of Fossil Energy, deep saline formations in the United States may have the capacity to store up to 500 billion metric tons of carbon dioxide — enough to store all the CO₂ produced in the country, at present rates, for a century or more.

Additional information

• More about the TOUGH2 and TOUGHREACT simulation codes
• More about geological carbon sequestration at Berkeley Lab's Earth Sciences Division
• More about carbon sequestration research at DOE's Office of Fossil Energy