Looking for Black Gold -- LBL Scientists Help Oil Industry Recover More Oil from Domestic Fields

Fall/Winter 1992

LBL expertise in modeling subterranean fractures and flow paths can help industry recover more oil from domestic fields.

By Jeffery Kahn ([email protected])

America's oil fields range from sea to shining sea, from the Gulf of Mexico to the Gulf of Alaska. Oil has been discovered in terrain as varied as the Great Plains and the Rocky Mountains. These natural resources have provided a vital advantage to America. Without native energy sources, the industrious and independent American way of life might never have materialized. Today, however, domestic oil production is in decline.

Throughout the 20th century, pumps have relentlessly drained oil from these repositories. Today, the many slowed, stilled, and rusted pumps in U.S. oil fields testify to the declining state of this country's petroleum output. Production peaked in this country in 1970. In oil fields across the American landscape, production has dwindled, often to the point that it is no longer profitable to pump. As wells go dry, the country has come to rely on foreign supplies, becoming the largest oil importer among nations.

Though domestic production continues to decline, bountiful amounts of oil remain. According to the U.S. Department of Energy, of the 513 billion barrels of petroleum discovered in the U.S., 368 billion barrels remain in the ground.

Discovered oil is not necessarily recoverable oil. Observes Ernie Majer, an earth scientist with Lawrence Berkeley Laboratory, "The problem in the U.S. is that we have already found the big oil fields. We have already pumped out the oil that is the easiest to recover. Now, we have to recover the oil that remains in these fields. Somehow, we have to find a way to drive out more of the oil."

Oilmen are famous for their luck, with some having an almost mystical ability to find crude, but finding the residual oil remaining in partially depleted fields might be the toughest prospecting challenge of them all. Oil fields can stretch over hundreds of square miles with the oil itself residing in multiple reservoirs that might be isolated, many miles apart from one another. In other cases, the oil might be in a maze of pockets connected by permeable paths such as fractures. New wells cost millions of dollars and cannot be located based on blind luck. Somehow, oil companies must find the right place to sink additional wells to tap into these hidden deposits of energy.

The odds against recovering such oil almost make this a job for Superman. X-ray vision would be an ideal way to find oil. Eyes that could look beneath the ground could see which pockets of oil are connected by flow paths and which are not. These fractures and permeable zones are a vital part of the natural plumbing required to pump and recover oil. In fact, if engineers could map the underground geology, they would be able to place a network of wells that could tap the oil that could flow in from linked subterranean reservoirs and avoid the blank intervening areas.

While the oil companies have made great strides toward characterizing underground sites, the ability to describe and model the underground landscape remains in its infancy. In an effort to improve their ability to see underground, British Petroleum, Conoco, Amoco, and Phillips Petroleum have called on Lawrence Berkeley Laboratory and scientists in its Earth Sciences Division as well as the Department of Energy's Office of Fossil Energy for help.

Over the past decade, LBL has emerged as a leading research institution in subsurface processes. In particular, LBL has devised new systems for characterizing fractured rock and for modeling fluid movement through it. These capabilities could be of vital importance not only in driving out residual oil but also in guiding the development of new oil fields.

LBL scientists developed their know-how while studying geothermal energy sites and while participating in the Department of Energy's Nuclear Waste Storage and Disposal Program. As they have learned how to detect and map fractured rock, they have been able to create better site models that predict subterranean fluid flow. Much of what they have learned about fluid flow in geothermal fields and underground waste sites is applicable to oil fields. Whether investigating flows of hot water in a geothermal field, potential pathways of leakage from a prospective underground nuclear waste site, or the flow paths in an oil field, researchers must be able to detect and map the connective pathways that enable fluids to flow underground.

For years, the petroleum industry has relied on a number of separate technologies--hydrologic tests, seismic-imaging techniques, and logging--to provide data about the subsurface. However, the industry has had a difficult time interpreting and integrating this data as well as in transforming it into the "bottom line," a predictive model.

In addition to Majer, a seismologist, the LBL team includes Larry Myer, whose field is geomechanics, and Jane Long, who develops subsurface models. In a cooperative, cost-shared research effort with BP, Conoco, Amoco, Phillips, DOE's Morgantown Energy Technology Center, and DOE's Bartlesville Project Office, these scientists will continue to develop LBL's unique interdisciplinary approach. This approach combines geologic, geomechanical, geophysical, and hydrologic information into a unified model for predicting fluid flow.

During the initial stages of this four-year project, LBL scientists will focus on two sites in Oklahoma. Conoco's Newkirk site consists predominantly of fractured limestone and shale sedimentary rock, whereas the University of Oklahoma's Gypsy site is a more heterogeneous tract, a "meander belt" with the buried, intertwined sand, gravel, and clay channels of an ancient river system. Reservoirs of petroleum often are found in these two types of geological formations.

These particular sites were chosen not because they have significant oil deposits--they do not. Rather, the value of the sites is that an extensive amount of geologic, hydrologic, and seismic data already exists. Thus, these sites are ready-made laboratories for field-testing and refining LBL's new combination of tools and techniques.

The quest to arrive at accurate predictive models is not impossible, but it is difficult. Says Larry Myer: "We are trying to develop a system to determine what is going on not only thousands of feet underground but over expanses that stretch over scores of miles. Every piece of land, every gas and oil field will be different, and there will never be one single discipline or technique that can give us the whole picture.

"Consequently, our aim is to develop a versatile battery of techniques and to improve our ability to integrate this information into a model. The site itself ultimately will end up dictating how we design our study, what tools we use, and in what order."

The tool kit includes the sciences of geology, hydrology, geomechanics, geophysics, and geochemistry.

A field investigation often starts with an examination of the surface geology. By systematically looking at outcrops, scientists can chart a rough picture of the subsurface regional geology. However, inferring the underground geology from what is seen above ground is an inexact science.

To identify the dominant fluid paths, researchers require additional data, often starting with geophysical information. Geophysics has much in common with medical imaging. Both rely on a signal transmitted through a body. Both attempt to determine and image what features in that solid body cause the signal to be transformed. However, the bodies being analyzed by geophysical techniques are not foot-wide human beings which can be poked and probed from every direction but miles-wide and deep expanses of earth which can be accessed only from 10-inch-diameter holes.

Geophysical techniques have been the bread and butter of the oil industry. Ernie Majer specializes in an emerging type of geophysics called tomography, a process that transmits vibrations through the ground for recording by a sensor. The resultant data are interpreted and transformed into a geological diagram or map.

Majer says that the resolution of what can be seen through tomography is related to the frequency of wave generated. Higher frequencies generally yield higher resolution in terms of being able to discern the presence and location of faults and the thickness of underground structures. However, the ground soaks up energy, and high frequencies are absorbed more readily than low frequencies.

To tap the potential of geophysics more fully, Majer says that LBL scientists are developing and testing a variety of new geophysical techniques and devices.

For instance, in conventional seismic reflection, vibrator trucks at the surface generate signals which are recorded at the surface by a network of geophones. This process can detect large underground structures, but smaller structures, such as fractures, remain invisible.

At LBL, scientists are refining an existing technique called vertical seismic profiling and developing a new process called cross-borehole tomography. In vertical profiling, vibrations are generated at the surface as usual but are recorded directly by sensors placed down boreholes or wells. While more expensive, this technique provides more detailed information.

In cross-borehole tomography, both the source of the signals and the receivers operate underground from inside boreholes. This provides yet higher resolution maps. When receivers are placed in two or more holes, cross-borehole can produce three-dimensional views.

One way to improve geophysical imaging is to make higher powered, high-frequency transmitting sources which generate signals that are not so easily absorbed by the ground. Majer recently successfully field-tested such a source, a piezoelectric transducer. The technology that makes the higher voltages possible was developed decades ago at LBL in its accelerator research program but never before used in tomography.

"With a piezoelectric transducer," explains Majer, "we put voltage into the transducer, and it moves or expands. This generates elastic waves. How often it expands and contracts depends upon the frequency of the electrical signal, which ranges from 100 to 15,000 cycles per second. Our device is unique in that we can lower it 9,000 feet down a borehole and, through a cable, drive it with 30,000 volts."

The field conditions under which these sources must operate are brutal. Lowered thousands of feet down a borehole, they must be able to operate under water, under high pressure, and at temperatures of 200 degrees Centigrade.

Larry Myer is developing a new electro-mechanical source that is designed to withstand these rigors. To achieve better efficiency and reliability, the device relies on a pneumatic spring. Myer credits Neville Cook, a researcher in the Earth Sciences Division and professor at the University of California, Berkeley, with the original idea. Myer says the vibrator will be relatively low frequency, shaking 40 to 400 times per second. While at short ranges it will not provide as high-resolution data as the piezoelectric source, it could become an important new imaging tool at longer distances.

No matter the type or the location of the source and the receivers, geophysics ends up producing a record that consists of a transmitted signal that has been slowed down and reduced in energy. The challenge is to transform this record of seismic or electrical variabilities into a picture of the subsurface. Myer leads the effort to interpret these geophysical data.

"When we do seismic tomography," says Myer, "in every shot we obtain data which gives us amplitudes and velocities. Why is there a different velocity or amplitude in one part of a picture as opposed to another? In my geomechanical research, I try to understand what properties of rock and earth cause these changes. Do these variations indicate we are looking at two different types of rock? Is it because there are conduits permeated with water? Or are we seeing a reservoir of oil?"

Laboratory research into how different properties of a rock alter a tomographic signal is an important element in improving the interpretation of geophysical data. "This is an active area of research," says Myer. "We are still trying to sort this out. For example, whether or a not a rock is partially saturated, whether it is fractured, its mineral content and its porosity, can alter both the amplitude and velocity. I want to find out whether we can distinguish what is responsible for these changes in our recorded signal."

Over years of research, LBL scientists have developed several theoretical models that describe how specific rock properties result in particular tomographic anomalies. Laboratory experiments have tested these models, which subsequently have been validated in small-scale field experiments. Other theories about why fractures can be detected by one type of seismic wave but not another can also be tested in the field.

"Why do these fractures show up at times but not at others?" asks Myer. "We have theories, and to test them, we are working in the field at sites of known fractures. We can change the properties of the fracture, flex it by pumping water in or out, then monitor how its geophysical signature changes. This allows us to improve our techniques so we can distinguish these features from other effects."

In order for fluids to flow underground, the subsurface must be not merely porous but permeable. Porosity indicates the presence of pores which can fill with fluids. Permeability denotes that these pores are connected by spaces through which fluids can move. Geophysics tends to see porosity. Hydrology assesses permeability.

As the geological and geophysical evidence comes in, earth scientist Jane Long pulls this information together with an eye toward identifying zones of permeability. Where there are ambiguities or uncertainties, Long conducts hydrological field tests. Hydrologists typically drill holes and see if they are able to pump water into or out of the subsurface at defined depths. This provides a direct but not always accurate indication of permeability. Water may flow into the subsurface, indicating the presence of fracture zones. However, rather than being miles-long flow paths that connect separate oil deposits, these fractures may end up being dead-ends.

At a certain stage, the LBL team decides it has assembled sufficient field information and is ready to create a preliminary, conceptual model of the subsurface. Long leads this effort to integrate the data.

The conceptual model provides a geometric framework of the hydrologic and hydraulically conductive features of the oil field. That is, the fluids, the fractures or paths through which the fluids can pass, and the features that govern flows are tentatively mapped in a computer model. Flows can occur along horizontal fractures, up vertical channels, or through slow leaks into an underground plane of fractures.

"This phase of the project is called diagnosis," explains Long. "We are looking for the hydraulic behavior of the oil field. The fracture zone may control the behavior. Or it may be the regional stress of the surrounding rock, which is under stress due to the weight of the rock above and lateral tectonic stresses. Other times, the material that fills the fractures can be the dominant factor. The filling can be less permeable than the surrounding rock, acting like a seal and creating a very complex situation."

The conceptual model evolves into a predictive model. This latter model is designed to predict the consequences of alternative ways that an oil field might be developed. Since wells can be sunk at any number of different locations, the model can point toward an optimal pattern of development. Later, as production begins to diminish in an oil field, secondary and tertiary recovery techniques can be used, such as pumping water or steam into the ground in an effort to force oil toward wells. As fluids are pumped out of or into the ground, underground fluid pressure changes. Fluid removed from the ground can cause rocks to crack and fail, either closing or opening permeable flow paths.

Long says the predictive model, the end result of LBL's multidisciplinary effort, is designed to foresee the consequences of a range of possible means for extracting oil. Typically, she says, predictive models are created and then, based upon subsequent geophysical and hydrological field tests, gradually modified and refined.

"The earth is perverse, and there are always surprises. So," says Long, "we have one iteration of the model, and then as feedback comes in from the field, another. Actually, from start to finish, the entire process is iterative. Each field test or model not only provides more data but points the way towards what test or refinement in the model should be made next."

This kind of feedback ultimately has allowed Long to evolve a unique new approach to modeling fluid flow in fractured rock like that found in oil fields. She calls it an equivalent discontinuum model.

Earlier models of below-ground fluid flow treat the ground as though it consists of uniformly connected material. Rather than account for the tortuous flow paths, this type of model simply breaks an area into grids and assigns each grid element a value for permeability. Long says that in fractured terrain, this approach can be blindsided. Underground fractures often manifest as flat planes. Flow is determined by whether or not these planes are connected to other fracture zones. Whereas standard models can fail to account for these unconnected areas, a discontinuum model can incorporate connectivity determined through hydraulic tests in the field. Long says this allows the discontinuum model to recreate the actual behavior of a real system.

Independently of one another, each of the LBL project scientists singled out one element that all cited as the key to their success.

Long speaks for the team: "Imagine trying to figure out what is going on down there, thousands of feet beneath the ground. To accurately predict the behavior of a gas or oil field calls for a team of people representing a range of disciplines. And still, we must be able to integrate our different approaches and distinct types of data into a single model.

"How do we do this? We constantly talk to one another. We are developing cross-disciplinary understanding and visualization tools. With this kind of continuous feedback, we can create better models that tell oil companies where to place their wells and how much oil they will be able to pump out."