In double-blind tests, where neither the operator nor Petro.ai knew the possible outcomes of predictions for well groups, accuracy from the Petro.ai DSU Design Service (DSUDS) modeling ranged from 92% to 99%. These consistently high accuracy results mean that the engineering scenario tradeoffs and engineering tradeoffs developed by Petro.ai for a variety of DSUs in basins across the US are dependable, trustworthy, and reliable.
“It’s very powerful to be this accurate,” Richard Gaut, CFO of Petro.ai emphasizes.“ Our customers have become accustomed to a huge uncertainty when underwriting their CapEx budgets. As a result, it becomes impossible to accurately forecast returns on capital, particularly for infill wells. These double-blind tests prove that operators don’t need to accept a wide error band anymore, even for complicated infill patterns.”
Until now, approaching the unknown of the shale subsurface has meant accepting the uncertainty and low accuracy that comes with traditional concepts like type curve areas and infill degradation factors. These existing reservoir engineering workflows are good for history matching, but are chronically unreliable when used to forecast new wells. This is due to the large number of engineering and geologic variations that happen from well to well (lateral length, completion intensity, landing zone, reservoir quality, distance from offsets, etc.).
Another challenge is the time lag from accurate history matching of offset wells to the well permitting and DSU design process. Since reliable Arps parameters can’t be extracted until about 12 months of production, operators don’t really know what they have until a well has been producing for a year. “Meanwhile, all of the economics are in the first 12 months,” Gaut continues, “You’re recovering about 40% of a well’s EUR in those first 12 months. While at the same time there is no certainty about how that well is going to perform vs. offset wells. This means more than half of your NPV is drawn down while you are flying blind.”
The results of the double-blind tests are staggering. “It means that DSUDS transforms the way that you allocate capital. You can transform the way that you make investment decisions. You can get a step change in improvements in your returns. These high-nineties double-blind results are evidence of the accuracy of the sensitivities the Petro.ai DSUDS provides - without waiting 12 months for each design modification to manifest in a production history.”
The Double-Blind Results
For the double-blind tests, undeveloped DSUs were selected. "We used the blank spaces where wells didn’t exist when we ran the model 12 months ago,” Gaut reveals.“ Now these wells do exist.
“We fed the operator’s plan for each DSU into Petro.ai.This included the usual data sets that we use for all of our clients (gun barrel targeting, lateral length, completion intensity, frac fingerprints, etc.). Petro.ai DSUDS created a time series production prediction for each DSU before the wells were spudded.”
Then came the hard part. We waited 12+ months while the wells were drilled, completed, and put on production. Once there was enough production data to hang a production curve, we compared our DSUDS predicted results (dashed line) with the operator’s actual results (solid line).
“Over and over again, the results were conclusive. From one DSU to another, from one operator to another, from one basin to another, the accuracy remained consistently high. From 92% to 99%, with an average of 96%, each DSUDS prediction painted the line of the production history for each DSU. Again and again, Petro.ai nailed it.
“The effect of all of this is industry changing.We don’t have to live with that bandwidth of doubt. There is a way to know with tremendous certainty what the reservoir is going to deliver, even after it has been drained by parent wells. The double-blind results give great confidence in our process—the physics works and the predictive power is extremely strong. These double-blind back tests are proving that the predictions Petro.ai is making are trustable, highly accurate, and no matter which basin you’re in, Petro.ai DSUDS models will deliver your best options for high returns in your AOI.
“The sensitivities DSUDS dials in are real. Every DSUDS tradeoff scenario is a highly accurate prediction about how that DSU would have performed. How would 2,500 pounds per foot have performed? What if we moved 50 feet further away from the parent? What if we staggered spacing instead of stacked? The sensitivities the DSUDS delivers are reflective of what would have happened, allowing operators to develop optimized sections with confidence.”