Frac Hit/Pad Designer: Two apps devoted to well spacing inside the Petro.ai Platform that, working together, provide a consistent over 90% accuracy in production prediction for a new pad.
AI: The domain-based machine learning that uses complex geomechanical calculations to determine the drainage model necessary to achieve that over 90% accuracy in production prediction for a new pad.
Type Curves: A legacy decline tool that provides 70-80% accuracy in production prediction for a new pad.
Simply stated, a type curve is a description of production behavior that has been observed for a well. We also know that one type curve is not nearly as useful as aggregating and analyzing the average of many type curves that have a similar spacing in an unconventional field. Type curves have been carrying the weight long enough, now it’s time to add geomechanics. And geomechanics is a foundation piece of Petro.ai.
We’re not talking about throwing out type curves, they have a history of value.
What we are suggesting is adding on an analysis of the AI derived drainage calculations to find the representative region for determining well spacing and predicting ROI for a new pad. Petro.ai has seen results of far over 90% accuracy in unconventionals with the use of this trifecta: Frac Hit/Pad Designer, AI and type curves.
With any analytical tool, error factors become an important part of trusting only one technique. Type curves are dependent upon operating conditions, well length, completion stages, the order of drilling, fracking spacing, fluid, proppant, and just basic geology. In a typical type curve analysis, you have to assume that all these variables are similar for the range of wells you’re comparing. But in reality, plotting a hundred type curves becomes a blur of jagged lines across a large swath of graph, where the average usually doesn’t reflect any particular well.
Troy Ruths, CEO of Petro.ai explains, “When we make type curves right now, we discretize based on spacing. They’ve proven to pull their weight but only so far. You lose tons of information because these wells are spaced closer or further apart than the grouping indicates. And in shale there’s so much more going on. When you bin on discretization, it’s not a reliable variable. You take a lot of information that you already have and you don’t use it in a type curve. That’s why adding drainage is so important. It takes what engineers are doing as a discrete variable and makes it a continuous variable.”
And that’s where the Frac Hit App/Pad Designer needs to join the lineup. Ruths continues, “It started off named Frac Hit, but it is a well spacing application. Where we’re using Pad Designer to think about well spacing in the future or for tradeoffs, Frac Hit calculates it for actual wells and the results can be used for discovery in similar prospective pads.
“The Frac Hit pulls together some of the most important input features into our AI. It’s handling landing in your lithology. In your stratigraphic column, it’s handling the changes in stress. It’s handling sequencing and the distances between wells. All of that vital information is being used to calculate the most important input parameter to your model, drainage. With Frac Hit, you can analyze what you know in current wells, then take that representative information into the Pad Designer to get a best estimate or prediction of performance in a new area. Type curves can build your comfortable safety net of understanding. For full accuracy you need the Frac Hit/Pad Designer information applied to future pads.”
Kyle LaMotta, VP of Analytics has created direct pad comparisons between using just a type curve analysis and then adding the Frac Hit/Pad Designer combination. "The problem with type curves is you’re guaranteed to be wrong because there’s such a wide range of values. Type curves are maybe 70% or 80% accurate. In some cases, it’s a lot worse. So, you’re going to take $8 million and bet that you’re right, but that’s what companies do.
“What we do with the Pad Designer is take it a step further and make the estimation from the type curves more accurate. By incorporating the geomechanics and calculating the drainage and the multivariate AI aspect of the productivity model, we’re taking it far beyond what you could ever predict with just type curves.
“So then, what’s really important, is being able to dynamically change the model and rather than using type curves and getting one average, you can change the spacing and ask, what if they’re 50 feet further apart? You can really start to do a lot of that important ‘what if’ analysis. We’ve seen well over 90% accuracy on the Pad Designer with pretty simple data inputs. That’s why it’s so compelling. Type curves won’t be thrown out any time soon, but with the accuracy of Frac Hit and Pad Designer, they’ll soon be the secondary test of the system.”