Frac-Hit Responses Using Pattern Recognition
Rapid Analysis of Offset Well Pressure Response during Fracturing — A Synopsis of SPE paper URTEC 3219
URTeC 2020 was a few weeks ago and despite the current circumstances (and being completely digital), there were a lot of great takeaways. In this post, we wanted to highlight just one of the many papers that we found interesting — specifically around fracture driven interactions (frac-hits, frac-interference, frac communication, offset pressure responses, etc.).
Many folks across the industry have greatly contributed to providing technical content around frac-hits or fracture driven interactions (FDIs). FDIs are a real problem that greatly impact designs, decisions, and economics. After several years of learning/debating how to characterize, mitigate, contingency plan, make real-time completions decisions, etc, the question became: so what?….what is an easy, practical workflow that allows me to derive a better understanding without adding substantial cost? With the most recent downturn, the common phrase “do more with less” continues to ring in our ears. In order to survive in this depressed market, it is more important than ever to understand the secondary and tertiary effects of our engineering practices. We need to better understand our reservoirs as well as apply lessons learned and best practices to future projects….yesterday (with a sense of urgency).
The paper we are highlighting today is URTeC 3129: Rapid Analysis of Offset Well Pressure Response during Fracturing: Distinguishing between Poroelastic, Hydraulic, and Frac-Hit Responses in Field Data Using Pattern Recognition published by Seth Puneet, Brendan Elliot, and Mukul Sharma at the University of Texas at Austin.
The authors of the paper (referenced above) do a great job of presenting a potential solution to this dilemma using a case study from the Permian basin and a Python-based approach using pattern recognition. They first differentiate and quantify different frac-hit or FDI response types as well as provide a conceptual schematic for typical corresponding distances for each response type (reservoir dependent). They then explain their algorithm in a few steps:
(1) They parse their raw pressure data into individual stages that can be overlaid.
(2) They then use automated pattern recognition to determine the type of well response and magnitude.
(3) They then use an algorithm that takes directional surveys and stage measured depths as inputs in order to calculate “a) the lateral distance between interacting stages along the wellbore, b) the normal distance between the stages (i.e. well spacing), c) the vertical offset between the stages and d) the total distance between the stages in 3D.” (Seth et al. )
(4) Next, they use the distance and response data to classify the pressure responses into the following categories — direct hit, mixed, poroelastic, and negligible. These categories are based on the notion that frac hit/response pressures are inversely proportional to the length/distance from the monitoring stage(along the lateral).
(5) Finally, they use the previously calculated distance values to calculate/verifying the frac azimuths.
It is a very well written, concise, and documented (lots of good diagrams/pictures) paper. So, if you’re having trouble or wrestling with unlocking more value from your treatment and offset monitoring data, I would highly recommend giving this paper a read. Remember: You can’t improve what you don’t measure. Best of luck to everyone out there. Until next time…
— Kevin Satterfield & John Kalfayan
If you enjoyed this content, please take a second to give it a clap, comment, recommend it, and share it with others!
Sources & Citations
[1] Seth, P., Elliott, B., and Sharma, M.M. Rapid Analysis of Offset Well Pressure Response during Fracturing: Distinguishing between Poroelastic, Hydraulic and Frac-Hit Responses in Field Data Using Pattern Recognition. URTeC 2020. URTeC-3129. http://dx.doi.org/10.15530/urtec-2020-3129.
