FracPrint AI: FracPrint technology is the strongest predictor in a multivariate model. Everybody can make multivariate models but Petro.ai’s are going to be the most informative and most accurate because we have FracPrint which results in the drainage volume. FracPrint lets your rock speak.
Well-to-well Variation: Well-to-well variation is explaining why one well is different than the other. And the reason? It’s the drainage. It’s the drainage because it was offsetting a parent. Or it’s the drainage because it’s in a different interval. Drainage is your rock speaking volumes.
Oil and Gas Industry desktops are littered with the remains of software that promised what it couldn’t produce. Solving your issues, easy to use, integration with other tools, meeting your feature requirements and accuracy combined are a tough find even in the marketplace of many options. Petro.ai has developed something more than software, something that lets your rock speak, something that customizes in all the right places to provide a decision matrix for your strategic issues, to move your data fast, to integrate with your business processes and to be accurate at the highest level in the industry. And, is enterprise ready to be placed in the hands of everyone on the team.
“What differentiates Petro.ai is that we can capture the variation in one particular well,” Dr. Troy Ruths, CEO of Petro.ai emphasizes. “From well to well, we capture the well-to-well variation better and that’s the same thing as saying we are more accurate.
“And if you can’t capture well-to-well variation, you can’t hear what your rock needs to say now. Instead, you use it as a look back. But as your accuracy gets better you can use it well-to-well. The foundation piece in Petro.ai for that accuracy is FracPrint.”
Dr. Brendon Hall, VP of Geoscience, continues, “Our FracPrint technology is a unique differentiator Petro.ai has over any other offerings or technologies. It bridges the gap between ad hoc empirical models of frac size and geometry based on observations or measurements and the fully physics based hydrofrac simulators that are also available. Both are useful tools but they’re hard to put into a practical workflow.
“For a reservoir engineer to look at tradeoffs between engineering parameters, economic parameters and geologic parameters they need to be able to run these scenarios quickly. They need to take into account the effect that geomechanics has on the shape of the frac. The mechanical state in the rock is really what governs how these fractures grow and propagate. If there’s no stress barrier or if there’s a stress barrier that blocks that frac energy they might go laterally not just vertically.
“If you start using rules of thumb you have from observations, you’re not taking all the geomechanics into account. If you’re on a fully physics-based simulator, you can take that into account, but the simulators take a long time to run. Minutes to days depending on the resolution of the model and the kinds of physics that you’re trying to capture. That’s not very practical for analyzing a lot of different scenarios.
“What Petro.ai does, it builds a surrogate model that can learn from observations or learn from simulations and incorporate that into a practical model that can generate stimulated areas, the FracPrint, in a very short amount of time which honors geomechanics and allows us to calculate these drainage areas. FracPrint is a very practical, concise and easy to use workflow that enables reservoir engineers to tie geology and geomechanics all the way to economics.”
“And here’s where the high level of accuracy comes in,” Kyle LaMotta, VP of Analytics reveals. “We’ve found from the results of the FracPrint, the frac geometry, is that drainage is key. Petro.ai predicts a 3D drainage volume around the wellbore, for each well. That’s the contacted rock that’s available to be drained by a well.
“From that volume of rock, a small percentage of oil and gas is extractable. We also calculate the extraction percentage for each well – called the cubic extraction rate (CER). The drainage volume coupled with the CER is a very strong predictor of well performance.
“In many multivariate modeling workflows we see that lateral length is one of the best predictors for well performance, which makes sense because we know that production scales up with longer wells (to an extent) as more rock is being contacted. The problem, though, is lateral length is a 2D measurement - it’s not really considering any of the rock around it, so when you use the multivariate model to predict a well’s performance, there’s a lot of variance that can’t be explained. Maybe it’s 60% accurate, which means that you can’t explain 40% of what’s going on. What we’ve found is Petro.ai’s calculated drainage volume is a much better predictor of well performance and improves multivariate model accuracy.
“That’s one of the missing links in a standard multivariate model, the ability to accurately calculate the drainage for each well individually, while considering the technical complexity of each well’s actual location.”
“It’s a very good feature for training machine learning models that until now hasn’t really been used by anybody,” Hall emphasizes. "Getting the machine learning out of the drainage model is a competitive advantage for Petro.ai. A lot of people have been chasing a way to get at drainage. Connecting geomechanics to the prediction of drainage is a holy grail that Petro.ai has been able to figure out. We’ve allowed the rock to have a voice.”
