Well-to-Well Variation—What differentiates Petro.ai is that we model the variation in one particular well. From one well to the next well, we capture that specific well-to-well variation. That’s the same thing as saying we are more accurate. Our conceptual model takes in the activity at the well bore interface and also monitors the far-field activity. We are a 3D view of your part of the full basin that you need to know and understand.
Pouring in Proppant—To use the correct amount of this limited and costly resource, you need to understand “the fracture network around each of your wells,” Ian Palmer in Forbes reveals from a study undertaken in Fayetteville shale, “a geomechanics model was developed.” That was in 2014. Petro.ai has expanded that understanding to a full drainage calculation using an AI platform for accurate results.
Accounting for anywhere from 6 – 25 percent of a well’s total cost, the amount of proppant used per well is increasing with every play. Proppant’s short history in the oil and gas industry has seen fast evolutionary changes as the understanding of frac size and the underlying geomechanical concepts surrounding shale have grown both onsite and in research-based labs.The standard 20-40 proppant sand, the smaller 100-mesh or the more common 40-70 sand, the artificial ceramic or resin-coated proppants, and the new 200-mesh DEEPROP option become the secret sauce for a company’s basin-dependent hydraulic fracturing program.
“Proppant is an interesting one,” Dr. Troy Ruths, CEO of Petro.ai explains, “The role of proppant is to prop open the hydraulic fractures that open during a frac job. There’s lots of things that affect it. Some critical questions are, can we put in more proppant to get more production? And is that production coming early in the well’s life cycle or throughout the life of the well? Not only are we asking for more production, are we changing the clime of the well because we’re using proppant?
“When we think of the effect of proppant, we need to ask, do we get an initial impact that’s improved, or do we get a better decline afterwards? Petro.ai can answer those questions. The way we manage that is by looking at decline curve parameters. In the graph above, as you put more proppant in your decline decreases in severity. The conclusion here is that we’ve managed to keep surface area accessible to the well for longer and that’s because it’s propped open.
“You might also ask, if I put in more proppant my fracs are getting bigger and so should I space the wells out more? Now the problem here is you just covaried something.
“What we do is, we normalize by our estimate of drainage. We capture that by using a coefficient that we fit on the drainage to see if you are extracting from it efficiently.
“Are we extracting that volume of reservoir? You know you’re getting more volume area because you know your rock isn’t changing.
“Now, if you’re getting an improved increase and/or normalizing for drainage that means you’re getting better placement of proppant. This isn’t exactly a one-to-one correlation, and you could actually hurt the well by putting in too much proppant. More proppant isn’t always better.
“Then you can take into account the economics of putting more proppant in. There may be a benefit to the production but is it beyond the cost of putting the proppant in. That might change from well to well. You might have a situation where you have access to a lot more proppant. You want to see, should I put it in this well or the other well.”
“That’s interesting from a tactical point of view also,” Kyle LaMotta, VP of Analytics at Petro.ai adds, “If you’re completing a well next week and suddenly the price of proppant goes up, market conditions change overnight, you can quickly rerun it and see the effect of that increased price of proppant. It’s not always a given how that price point may change a job. Maybe before when proppant price was low it made sense to pump a bit more because you saw a little bit better production but if the price doubles, then you probably want to pump a smaller job because the incremental gain just isn’t there.”
“And what we do is we take those considerations and turn that into a 1D response curve,” Ruths continues, “It’s 1D because it’s one dimension. We’ve taken all these variables and put it into that total proppant per stage. All that complexity that we just talked about is baked in for that particular well. We don’t need to look at so many sensitivities. The challenge with multi-variate regression is that people try to make it work for everything. It doesn’t need to work for everything. It just needs to work for the design parameters that you want to put in it. By making it specific to a well you can handle those well-to-well sensitivities better.
“You’re not running over too many of those sensitivities. What if we landed a little bit higher or lower? What if we had a parent well?
“All of those ‘what-ifs’ are now ‘we-know.’
“We know exactly where it’s landed, exactly how many parent wells are around it, we know the cost of completions. You’ve changed the ‘what-ifs’ to the ‘we-knows’ and you only have a couple of ‘what-ifs’ left.
“That’s what we’re doing with the CSA, Completions Scenario Analysis, and the PSA, Pad Scenario Analysis, the customized tradeoff decision matrices which distill your data and your options to one fast, easily updated, interactive overview. Your accuracy improves dramatically as your well-to-well ensemble of technical models create your specific 3D understanding of your specific reservoir.”