Microseismic: Petro.ai uses this cloud of points to locate where the hydrofrac is growing.
Microseismic: Petro.ai predicts productivity accurately with a deep learning model that incorporates SHmin and the shear fracture drainage.
The tensile fracture, the hydraulic fracture, connects to the shear fractures, hundreds of natural fractures some half the size of a football field, some the width of a piece of paper. Together they form a super permeable, superhighway network that rushes the injected fluid back to the hydrofrac and into the well via the completion stage. Petro.ai is mapping that drainage so reservoir engineers have a model to use for both productivity investigation and prediction based on a combination of SHmin and microseismic.
Microseismic is a locator of the shear fractures that follow the silent tensile fracture. The tensile or hydraulic fracture is propagating very fast in the rock, moving at nearly the speed of sound. The fracture leaves the well bore and behind it is the fluid causing that wedge to continue to grow. The fluid is intersecting into this rich network of natural fractures that are now also moving. Those natural fractures are connected back to the hydrofracture and they start to shear as the fluid communicates the pressure and a measurable microseismic event occurs.
Kyle La Motta, VP of Analytics breaks it down, “Imagine a freeway where the speed limits are much higher but you’ve got all these smaller roads with less traffic, fewer number of lanes and lower speed limits. The highway is connecting everything together, but you’re feeding the highway from these other smaller roads. Those natural fractures are connected back to the well through that superhighway that you create with that prop fracture. You pump the stage that opens up this big fracture. You hold that fracture open with the proppant that’s propping it open. Fluid is leaking off around it, creating these smaller roadways to get back into that main conduit. So, you’ve got a big well bore, tiny prop fractures that are about the size of a grain of sand and then even smaller fractures coming off the side of that which is that natural fracture network.”
“Now how it connects back to SHmin and the vertical stress profile is that the vertical stress profile is guiding that hydrofrac. You’re only going to get microseismic events in a situation where you don’t have depletion or pad scale faults,” Dr. Troy Ruths, CEO of Petro.ai explains. “We’ve found that shear fractures are anywhere from 10% to 70% of the well’s productivity depending on the basin, the abundance of natural fractures and the connection to the well. If you see your wells communicating and we’ve seen this in the Montney, in Midland, in the Permian, wells will have pressure responses through the shear fracture network.”
Ruths cautions, “You can’t optimize shale without taking the microseismic into account. If you’re trying to build models and you have anywhere from 10 to 70% of your productivity coming from a source that you’re not modeling, your accuracy falls off proportionally.”
Zoback and Hakso assert that, “the role of stimulated slip on pre-existing fractures and faults on production is estimated by modeling the area of the fracture network implied by microseismicity associated with hydraulic fracturing operations.”
Petro.ai is building deep learning models based on the shear fracture networks and the frac geometries created from them when you stimulate wells at different locations in the vertical stress profile. “The fracture grows in the direction of least principal stress,” Ruths reminds us. “That’s the type of information that this deep learning model takes into account.”
Below, the vertical stress profile is represented on the left. We’ve selected two wells and the normalized microseismic data for the two selected wells for those stages is on the right. Note the asymmetry due to parent wells and the associated depletion effect.
Further calculations create the visualization below to understand what the drainage in that shear fracture network would be for wells that are completed in the specific location associated with the points from the vertical stress profile graph.
Using the deep learning model, we can drop a point and predict. The well can be moved anywhere throughout the vertical stress profile and it will calculate the associated total drainage.
And how was analysis created? The raw data is XYZ point data from Petro.ai Earth. In 3Dviz, microseismic data can be visualized as a cloud of points.
Simplifying the visualization further. We can see that the activity is growing up from the bottom well. But it doesn’t keep growing. It appears to stop just above the top well.
And now, changing to the actual scale, without the augmented z direction, we can see the real relationship of the wells and the microseismic. Remember that the wells are two miles in the horizontal direction and span about three stories in the y.
There’s not much microseismic on the bottom, it’s growing upwards with something that’s stopping it from propagating further.
