Scientists have put together a simple model of shale gas wells' productivity that accurately describes wells fracked on the Barnett Shale.
Using the model, scientists from the University of Texas, Austin, studied about 8,000 wells on the Barnett and found they would produce at least 10 trillion cubic feet of natural gas. The Energy Information Administration estimates that the Barnett contains 43 trillion cubic feet of technically recoverable gas in total.
The work was published yesterday in the Proceedings of the National Academy of Sciences.
The specific estimates are not the most important result of the study; rather, it is the ability of the model itself to predict the behavior of unconventional reservoirs. The authors said the model also describes the Haynesville and Fayetteville shales accurately, and preliminary data indicate it could be a good fit for the Marcellus Shale.
"The model tries to give a picture of how gas evolves from these wells from the beginning until the very end," said Michael Marder, an associate professor at UT Austin and co-author of the study.
The model was the underpinning for a landmark study released in August that found the Barnett Shale is an energy behemoth, containing more than 40 trillion cubic feet of economically recoverable natural gas (EnergyWire, Sept. 25).
The ongoing shale gas boom in the United States has led to an oversupply of natural gas and depressed prices. But some experts have questioned the boom's viability since unconventional gas wells decline rapidly, sometimes within 12 months of drilling. The parameters that determine the rates of decline are not well-understood.
The UT model does not try to iron out the parameters. Instead, it identifies wells at the threshold of steep decline and assumes these wells will experience exponential decline. The scientists tested their model in the Barnett Shale using field results from 8,000 wells operating for more than 18 months that had not been re-completed. They found that the field data agreed well with their model predictions.
Models are mathematical approximations of the real world and can be correlated to events underground. During hydraulic fracturing, millions of gallons of water, sand and chemicals are pumped into well bores at pressures high enough to fracture shale. The fractures run parallel to each other and extend into the rock like channels.
The model imagines the shale landscape as a grid, with the sides of each grid cell bounded by adjacent fractures. The top and bottom of each grid box would be the top and bottom of the shale play. The box contains rock with uniform pores containing methane. After fracking, the methane rushes through the pores, moving from the high-pressured center of each grid cell into the low-pressure fractures. The gas then flows up the well bore.
As the grid cell empties, parallel fracks begin interfering with each other by attracting the gas from opposite sides. There is little net movement into the fractures. The scientists call this phenomenon "interference."
Once interference happens, the well production declines exponentially, said Tad Patzek, a researcher at the Department of Petroleum and Geosystems Engineering at UT Austin and co-author of the study.
Using this model, the authors found that 2,057 wells experienced interference after about five years. And 6,237 wells were too young and had not yet experienced interference. They estimated that 8,294 wells they studied would produce at least 10 trillion cubic feet of gas in the first 10 years and up to 20 trillion cubic feet of gas over 50 years.
"We have performed a similar analysis for the Haynesville and Fayetteville shale, where many of the wells started interfering earlier," Patzek said. "The same model holds with a remarkable consistency."
While detailed models have existed for conventional gas wells for decades, scientists and industry experts are only now trying to understand unconventional wells. The challenge is that not much is known about shale.
"It is a very, very complicated process with many things happening at the same time, and we are just at the beginning, really, of understanding what is happening and what we can use to describe it," said Karen Schmid, a hydromechanics modeler at the University of Stuttgart, who is unaffiliated with the University of Texas study.
Any model of gas production would have to be simple enough to be fairly easily run and, at the same time, complicated enough to give good predictions, Schmid said.
The University of Texas model satisfies these requirements because it ignores some of the complicated geological processes without sacrificing its predictive abilities, Schmid said. The model works well for the Barnett Shale.
However, Schmid questioned whether the model could be applied in other, younger plays. The Barnett is the oldest producing shale basin in North America, and, as such, its wells have already matured.
"How far [the model] will take us with other shale plays is not clear yet," she said.
Yu-Shu Wu, a professor of petroleum engineering at the Colorado School of Mines, said the model may not be accurate in the long term, perhaps after the first 10 years of production when the gas flow patterns in the reservoir drastically change.
Brent Johanson, a reservoir engineer at ConocoPhillips Co. who studied the Barnett for his master's thesis, expressed doubts about the geological basis underlying the University of Texas model. The model assumes that interference between adjacent fractures leads to declining production. Johanson said it is more likely that the decline is due to the beginning of the end of a particular fracture treatment's efficacy.
This distinction may seem minor, but "it has major industry implications for designing fracture stimulation treatments with regard to the spacing between fractures [and] the spacing between parallel well bores," Johanson said in an email.
He also suggested the model should be refined to provide a smaller range of production estimates. Currently, the model finds the estimated ultimate recovery (EUR) for the roughly 8,000 wells is 10 trillion to 20 trillion cubic feet.
"When considering the high cost to develop a new horizontal gas well and the relatively low commodity price of natural gas, industry operators would benefit from models that can narrow the range between the upper and lower EUR predictions," he said.