HSE & Sustainability

Filling in the Blanks To Limit Earthquake Risks

It is known that a well injecting a lot of water near a big fault can lead to earthquakes. The problem is, more often than not those faults are not known until after a tremor.

jpt-2017-09-inducedseis-art2hero.jpg
Maps plot the earthquake epicenters over maps showing various versions of gravity and magnetic studies.
Source: RPSEA 12122.

It is known that a well injecting a lot of water near a big fault can lead to earthquakes. The problem is, more often than not those faults are not known until after a tremor.

“Only 34% of these earthquakes occur within 2 km of any known fault,” said Jeremy Boak, director of the Oklahoma Geological Survey, which is working on multiple studies to describe and understand how water injection activates critically stressed faults to cause earthquakes.

“We are learning where many faults are,” Boak said at the American Association of Petroleum Geologists annual meeting this spring, where he delivered a paper (2017) along with researchers from Texas, Kansas, and Colorado talking about what they are doing to fill the knowledge gaps.

While the Oklahoma Geological Survey is not a regulator, its work has guided what the state has done to reduce the number of earthquakes by limiting injection in those areas with the most problems.

“The drop is due to decreases in injection,” related to a slowdown in activity since oil prices dropped and the orders to reduce injection in affected areas by the regulator, the Oklahoma Corporation Commission, Boak said.

Speaking at the ­Unconventional Resources Technology Conference in July, Boak predicted that the number of earthquakes strong enough to feel will total around 300 this year—one-third of the number of the peak that occurred in 2015. Boak said it is far from an acceptable rate of activity in a state where the annual average used to be 1.6 events a year.

He and others working on this problem are concerned about big gaps in the information needed to measure and avoid spots predisposed to seismic activity if too much water is injected. That is as big a problem in Texas and Kansas as it is in Oklahoma.

The future economics of oil exploration in these areas depend on identifying specific solutions—problem places or practices—rather than wholesale limits on saltwater disposal.

A study in Colorado identified a previously unmapped fault zone to avoid, which was responsible for a cluster of earthquakes within the Denver Basin.

Identifying and avoiding places and practices that trigger problems can answer the public pressure to ban salt­water injection, according to the paper by two graduate students at the Colorado School of Mines (Harty and Bauer 2017). They estimated that treating the water could cost six times more than disposal.

“We need to improve our knowledge of subsurface faults and fractures,” said Tandis Bidgoli, an assistant scientist for energy research at the Kansas Geological Survey. Kansas recently completed a new fault map of the state, which shares a border with Oklahoma, with support from the US Geological Survey.

Looking Deep

The faults that cause problems often go unnoticed because they are buried in the basement.

More accurately, the faults and fractures are found in the crystalline basement rock. These ancient layers, often Cambrian and Pre-Cambrian rock, are found under the sandstones or carbonates that hold hydrocarbon reservoirs.

“These are important because a lot of epicenters occur along the faults and fractures in the pre-Cambrian,” said Harvey Eastman, a consulting geologist, who did a study mapping faults in Johnson County, Texas, which is in the Barnett formation (Eastman and Murin 2017).

As water flowing down from higher levels reaches critically stressed faults it can cause changes, which when combined with natural stresses can cause the opposing rock faces to slip, causing an earthquake.

Researchers are studying old seismic surveys turned over by oil companies looking for faults and fractures in the zones above the basement that transport injected water down to it. These images have their limits. The features are relatively small compared to the resolution of the seismic, and the image quality is often poor because these surveys were generally not designed to image below the layers that hold potential oil and gas deposits.

Surveys available normally predate the shale boom. For example, in Kansas companies are required to allow the state to use seismic after 10 years, Bidgoli said.

For the Johnson County study, Eastman relied on 220 well logs from Texas regulators to define the underlying strata and combined that with a map of the surface contours to identify major faults, which he said were generally in line with past studies. In his case seismic was not available.

The Colorado study combined multiple sources—well logs, records of wells that penetrated the basement rock, gravity, and magnetic studies. These data were used to identify a basement structure near the source of the earthquakes and determine if the direction of natural stresses is aligned in a way that could lead to movement along a stressed fault if injected water was added.

“We used this to create a stress map to identify where there would be critical stress in the basement that could trigger faults,” said Bauer. He added that “these zones are also confirmed by the seismic history in the DJ basin.”

Magnetic and gravity studies were also used in a major Oklahoma study (RPSEA 12122). That project, backed by RPSEA, a federally backed research organization, concluded that a study of earthquake clusters in two counties “clearly illustrate the need to acquire new high-resolution gravity, magnetic, and geologic data along with improved geologic models of the sedimentary formations and structures above the crystalline basement.”

jpt-2017-09-inducedseis-art2fig1.jpg
Earthquakes in Oklahoma have risen and fallen with wastewater injection rates. Source: RPSEA 12122.

Managing Risk

The fix can be as simple as reducing the injection rate per well. “The first-order thing that creates pore pressure is the volume per well,” Bidgoli said, adding, “if you go from 16,000 B/D in one well and make it 8,000 B/D in two wells, the pressure reduction is substantial.”

In Kansas, the total amount of brine disposed in two counties was about the same—84 million bbl each in Ellis and Harper counties in 2015. Much of the earthquake activity was concentrated in Harper County, while there was none in Ellis County, which had twice as many injection wells, according to Bidgoli and Jackson (2017).

Reducing the depth of an injection well may also help by cutting off flows down to the basement rock. In Colorado, a 10,800-ft-deep injection well associated with a cluster of seismic events in Weld County was plugged back, so the deepest injection point was at 9,800 ft. That change reduced the rate of earthquake events, suggesting “reduced or removed basement communication,” according to a previous paper by Bauer and Harty (2016).

Kansas researchers used software from the Southwest Research Institute to analyze how the stress on a fault or fracture could be altered by injections.

The 3D Stress program was created to point to spots where subsurface stress change may cause a fault or fracture to shift, potentially causing an earthquake, said Alan Morris, a staff scientist at the Southwest Research Institute in San Antonio.

The program is designed to evaluate the risk of seismic activity using “data inputs that can range from an educated guess as to the stress state, to seismic activity and wellbore data and full 3D fault interpretations from seismic reflection studies,” he said. The program’s quick turnaround time allows a user to consider multiple possible outcomes when using differing inputs that reflect the level of uncertainty in the data.

The example Morris uses to explain how it works is an injection well outside of Youngstown, Ohio, that triggered a 4.0 magnitude earthquake in 2011. In the month before the earthquake there were small events below the level which could be felt, which had no impact on the injection rate. At the time, the well operator was seeking permission to increase injections.

“Careful analysis of these smaller events could have identified the fault before the magnitude 4.0 earthquake occurred,” Morris said, adding that the program “could then have highlighted the potential for a felt earthquake before it happened.”

But as is often the case, the warning signs and the underlying fault were only studied after the fact, which in that case led to the shutdown of the injection well.

Induced seismic activity in Texas dates back to the booming years of the first shale play, the Barnett around Fort Worth. Now the epicenter for shale drilling and fracturing has moved west to the Permian Basin.

The induced seismicity in that sparsely populated area has not been an issue, but produced water disposal has grown with the number of wells.

Bridget Scanlon, an economist for the Bureau of Economic Geology for the University of Texas at Austin, said she has heard that saltwater injection wells are going deeper, sending brine into the Ellenberger, a formation just above the basement rock like the Arbuckle in Oklahoma, which was the target for high-volume injection wells there.

“Injecting water into the Deeper Ellenberger could cause induced seismicity,” Scanlon said, during a talk on her study of alternative uses for the water in the Permian.

While no state has experienced the level of earthquakes that hit Oklahoma, that could change. “I am not sure there is more faulting in Oklahoma than in other states nearby,” Boak said.

For Further Reading

RPSEA Project 12122-91. 4D Integrated Study Using Geology, Geophysics, Reservoir Modeling & Rock Mechanics to Develop Assessment Models for Potential Induced Seismicity Risk by Jeremy Boak, Oklahoma Geological Survey et al.

AAPG. 2017. Patterns of Induced Seismicity in Central and Northwest Oklahoma by Jeremy Boak, Oklahoma Geological Survey.

AAPG. 2017. Induced Seismicity in the Denver Basin Prompts Updated Basement Stress and Fault Configuration Model by Michael J. Harty and Matthew Bauer, Colorado School of Mines.

AAPG. 2016. Induced Seismicity in the Denver Basin Prompts Updated Basement Fault Configuration Model by Matthew W. Bauer and Michael J. Harty, Colorado School of Mines.

AAPG. 2017. Operational Practices and Their Influence on Injection-Induced Earthquakes: Lessons Learned From a Statewide Survey of Brine Disposal in Kansas by Tandis S. Bidgoli and Christa Jackson, Kansas Geological Survey.

AAPG. 2017. Geologic Characterization of Johnson County, Texas by Harvey Eastman, consulting geologist, and Timothy Murin, AECOM.