Billions of barrels of oil lie in the Lost Hills and Belridge oilfields in Kern County, right in the middle of California. The catch is, less than ten percent of it can be extracted from the rock with current technology.

	Sinking to the bottom of shallow lakes and seas, the glassy skeletons of unimaginable multitudes of microscopic plants formed the remarkable rock known as diatomite.

The rock in question is diatomite, made of countless silica skeletons of microscopic aquatic plants. Diatomite's tiny pores range from a few billionths to a few millionths of a meter wide, creating a soft and fragile mineral that is up to 70 percent empty space -- light enough to float on water but so nearly impermeable that no one is quite sure how the oil got into the rock in the first place.

"The only way to get the oil out is to fracture the rock," says Tad Patzek of Berkeley Lab's Earth Sciences Division (ESD), who is also a professor of civil and environmental engineering at UC Berkeley. Patzek notes there was virtually no production from Lost Hills until 30 years ago, when the hydrofracturing technique -- breaking the rock with pressurized water -- was first used. The ChevronTexaco Company currently extracts some 10,000 barrels of oil a day in the Lost Hills Field alone.

One result of oil extraction was that parts of the oil field began to collapse and subside. In 1992 ChevronTexaco introduced a program of balanced waterflooding in the Lost Hills Field in hopes of reducing subsidence and increasing oil recovery, but results have been mixed.

Patzek and his colleague Dmitry Silin of ESD are working with the ChevronTexaco team, led by Jim Brink and Pat Perry, to find better ways to tap the rich but miserly diatomite formation. They are pursuing a proposal by Grigory I. Barenblatt of the Mathematics Department in Berkeley Lab's Computing Sciences Directorate, a professor of mathematics at UC Berkeley and a world-renowned expert in petroleum engineering and rock mechanics.

Barenblatt recently advanced a new concept of damage propagation in diatomite, with consequent modification of rock properties; initial computer simulations based on the model suggest new technologies for oil recovery. Recent work has also benefited from startling new microscopic images of diatomite made by Liviu Tomutsa of ESD and Velimir Radmilovic of the National Center for Electron Microscopy (NCEM).

"To understand and predict what's happening in diatomite reservoirs, the big picture is not enough," says Patzek. "We're tackling the problem on the very small scale as well, trying to understand the 3-D structure of the rock itself."

Oil and water

When water is injected into an oil-bearing formation under pressure two things happen -- at least in theory. The fractured rock is supposed to release the oil, and the water is supposed to push it into flow-conducting fractures, further toward the production wells. Meanwhile the water presumably fills the voids left behind, maintaining the pressure and structural integrity of the rock formation.

Over a period of 17 years the surface of the Lost Hills oilfield sank about three meters, as indicated in blue and purple. Meanwhile fruit trees in a nearby orchard, sections 28 and 34, grew to heights of five meters, as indicated in red and yellow. (Each section is about a mile square.)

But even though waterflooding at Lost Hills has been carefully planned and monitored, it has produced less than textbook results. Water injection has damaged the rock in unexpected ways, causing the ground to subside over wide areas -- in some regions, 10 feet or more -- and shearing off well casings. Much of the water pumped down the injection wells comes right back up the production wells without increasing the amount of recovered oil.

"You can't produce from diatomite without injection, but we've got to learn how to control the process," says Patzek. "On the large scale we're looking at the overall structure of the formation."

One approach is to look at where and how the ground has moved and compare that data with records of the water injected and the oil produced. InSAR images, radar images from space satellites taken from different viewpoints and combined interferometrically, graphically reveal ground subsidence in the oil fields. Subsidence as oil is pumped out shows up immediately. Waterflooding has done less than expected to stop subsidence, no matter how much water is injected.

The reason lies in the geological complexity of the Lost Hills formation. The diatomite beds are divided into discrete segments by vertical faults and horizontal strata, with layers varying in thickness from less than a millimeter to tens of meters. Not only can excessive water-pressure crush diatomite, it can cause the weakly bound layers to break apart. Channels open that carry water directly from injection well to production well without moving any oil.

	A Berkeley Lab and ChevronTexaco team designed and installed automatic injection equipment to maintain the right water pressure in the Lost Hills diatomite. From left, Gennady Goloshubin of the West Siberian Research Institute of Geology and Geophysics, a guest of the Berkeley Lab researchers; Monika Valjak, ChevronTexaco field engineer; Jim Brink, ChevronTexaco field advisor; Tad Patzek and Dmitry Silin of ESD.

To prevent the creation of channels and to keep soft diatomite from collapsing, water injection pressures must be monitored and adjusted at frequent intervals. Working with colleagues at ChevronTexaco, Patzek and Silin have developed a software system that acquires pressure and flow-meter reading from Lost Hills wells over a computer network. In automatic mode, the data are analyzed and the results immediately sent back to the field.

Since February, 2002, the Supervisory Control and Data Acquisition (SCADA) system at ChevronTexaco has controlled 12 automated water injectors in real time. "The data is transmitted right here to my desktop PC," says Silin. Early results have been encouraging, and the company has budgeted support for SCADA for coming years.

From macro to micro

Seeking a theoretical understanding of the structure and mechanical properties of oil-bearing rocks, one based on fundamental physical principles, Silin and Patzek, with their colleague Guodong Jin of the University of California at Berkeley, have used computer modeling to study how sediments become rocks in the first place.

Model sandstone formation begins with sedimentation (left), as grains of different sizes settle under the action of gravity. Adding the pressure of overburden causes compaction (center). Cementation (right) is one of the final processes that turns sediment into rock.

They begin with a collection of grains, idealized as spheres of different sizes, randomly distributed as in natural sediments. These are then compacted, as natural deposits are when deeply buried. They then model the process of diagenesis by which the compacted grains become cemented together. Finally they derive the mechanical properties of the resulting model rock.

Their model is unusual because it incorporates the dynamic processes of sedimentation and compaction, including gravity, contact forces, friction between grains and with the surrounding fluids -- all the ways that grains of different size and composition may move and interact.

The program has produced astonishing model rocks. Its reconstruction of sandstone is virtually indistinguishable from images of real sandstone made by synchrotron X-ray tomography. Eventually computer reconstructions may make it possible to generate realistic images and understand the mechanical properties of any kind of sedimentary rock -- including diatomite -- at will.

Picturing the real thing

While X-ray tomography can image relatively coarse sandstone at a resolution of a few millionths of a meter (microns), the only way to characterize the very fine pore structure of diatomite and chalk is at the nanometer scale -- billionths of a meter.

ESD's Liviu Tomutsa has been working with NCEM's Velimir Radmilovic, of Berkeley Lab's Materials Sciences Division, to produce remarkable pictures of real reservoir diatomite. They have used focused ion beams to image its invisibly tiny pores and connecting throats, the empty spaces that constitute over half its volume.

A beam of gallium ions shaves off a sample of diatomite layer by layer, imaging each layer. A computer will combine the succeeding images to reconstruct the rock in three dimensions.

For some time NCEM has prepared samples for transmission electron microscopy (TEM) with ion-beam milling. TEM images columns of atoms in a crystal by sending a beam of electrons right through it, a kind of shadow picture. For best results the sections must be extraordinarily thin, typically a few microns or less. No mechanical slicing or polishing technique can operate on so fine a scale without tearing the sample to shreds.

A sweeping beam of electrically charged gallium atoms, however, mills a sample that is nearly perfectly flat. The ions do more: as they sweep across the sample, sputtering off patches of atoms a few dozen at a time, they free secondary ions that can be focused into a high-contrast image of the surface at a resolution as fine as 10 nanometers.

"As the beam shaves off layers of material only a few nanometers thick, it simultaneously makes a stack of 2-D images," says Tomutsa. "These can be stored in a computer and combined into a 3-D picture that reveals the constrictions and connectivity of the pores, the factors that affect trapping of liquids."

Tomutsa and Radmilovic are now working to merge their individual 2-D images into pictures with three-dimensional depth. Says Tomutsa, "These will be the first-ever images of diatomaceous rock in 3-D." What's more, he says, by using the ion beam to mill and image samples at very low temperatures, "we'll actually be able to see the oil inside the rock, undisturbed, frozen in the pores."

Computer modeling and real imaging may soon converge to understand the structures of rocks on the finest scale, both theoretically and experimentally. One result, says Tomutsa, could be a single number that expresses the relative permeability of a particular kind of rock under many different conditions.

It will be a dramatic step forward, giving energy companies a head start on choosing the best techniques for extracting oil from tough formations and relieving the expensive trial and error that characterizes oil recovery today.

Additional information

• More on new technology for managing the Lost Hills oilfield
• More on computer reconstruction of sedimentary rock