Sunday, February 27, 2011

Considerations from a Fracking Expert

If you want to talk fracking, then Cornell engineering professor Anthony Ingraffea’s your guy. After all, he got his PhD in rock fracturing and conducted hydro-fracturing research with Schlumberger. So EPA has invited Ingraffea to their technical workshop focusing on the engineering aspects of drilling, fracture design and stimulation, and mechanical integrity in gas wells. The March 10 workshop is one of four that EPA is holding as part of the Hydro-fracturing Study.

For the past year or more Ingraffea has been traveling around the Marcellus region of NY and PA, sharing his insights into rocks, fracturing and the engineering of gas wells. The first thing to keep in mind, he cautions, is that unconventional wells are nothing like gas wells drilled even a decade ago. Those are vertical wells, drilled one per pad, each using about 80,000 gallons of water.

Unconventional wells in Pennsylvania currently have 8 wells per pad, Ingraffea says, with each of those wells up to 10-frack-stages in length. That means that a single well pad will use about 44 million gallons of fracking fluid.

As far as drilling in NY’s Marcellus, Ingraffea points to PSU geologist Terry Engelder’s estimates of 36,000 to 78,000 wells. With up to 10 fracks per well, that’s 360,000 to 780,000 frack stages – and each of those frack stages uses close to 500,000 gallons of fluid.

“The first one thousand gas wells unconventionally developed in NY State will use more frack fluid, and produce more waste, than all the gas wells ever drilled in the state,” says Ingraffea. “This kind of development is on a scale at least two magnitudes larger than we have ever experienced, and right now we have no regulations guiding shale gas development.”

So Ingraffea has lots of questions for the EPA. Topping his list: cement. Cement failure has been a chronic problem in the industry, Ingraffea says. He’d like to see cement logs required for each job.

Then there’s the question about how cement holds up under multiple hydro-fracks. Re-fracking is a real concern, says Ingraffea, because “each time you re-pressurize the wellbore for a frack job, it puts the cement at risk.” And, he says, the industry already knows that cement that has been stressed frequently has a higher failure rate.

Ingraffea also wants EPA to get better data on the cumulative impact of intensive drilling on neighboring wells. Pennsylvania drillers estimate that they’ll be putting in 8 to 12 wells on a pad, with the vertical wellbores spaced about 20 feet apart. “What happens to the first well when the second is drilled?” asks Ingraffea. “Do the vibrations damage the cement?”

The problem with articles in professional journals is that they always focus on impacts to a single well. But drillers in British Columbia report that wells drilled as far as 350 feet from each other can send lateral fractures into neighboring wells.

As for migration or fracking fluids and gas, “EPA needs a realistic model that gauges cumulative impact,” says Ingraffea. “If they’re not looking at it that way, they they’re missing the point.”


  1. We are trying to get Professor Ingraffea to come to Penn State (Marcellus University I think some might call us) and give a talk in early May.

  2. From his presentations - he states that the models are not capable of predicting the partings and movement because the material is not homogenous or isotropic and their efforts to simulate subsurface conditions did not work.

    It would be better to require the industry to provide microseismic or acoustic logs of actual wells to document the extent of the parting extent or actual movement of materials (gases and fluids) - this would help model development.

    I disagree with this statement:
    "As for migration or fracking fluids and gas, “EPA needs a realistic model that gauges cumulative impact,” says Ingraffea. “If they’re not looking at it that way, they they’re missing the point.”

    We need actual monitoring NOW during the development - to better develop a predictive model.