As laterals become longer with more stages, large datasets make evaluation of the information complicated and time consuming. Variations in the obtained data adds more uncertainty to the main industry challenges and questions, such as optimal well spacing, optimal completion (e.g., type of completion, stage and cluster spacing), optimal stimulation treatment (e.g., proppant type, mass and concentration, fluid type, volume and composition).
This paper describes a novel automated framework for completion optimization and post-fracture analysis. For pre-fracture analysis, this framework utilizes an integrated dataset from drilling and wireline logs, and automatically places clusters based on heterogeneity along the lateral. This information is then coupled with post-fracture data to examine fracture treating information (breakdown pressure, instantaneous shut in pressure, etc.) and correlate it to rock properties from each fracture stage. The newly identified correlation is looped back into the framework as a guideline for future frac design.
The strong correlation between subsurface rock properties and hydraulic fracture treatment parameters suggests that the previous captured heterogeneity reveals the fracture design efficiency, and the multivariable evaluation process can be accelerated by the machine learning platform. A statistical model was built on geomechanical and petrophysical properties used to design fracture treatments. Meanwhile, the platform automatically evaluates fracture treating signatures (surface treating pressure, pumping rate, ISIP, etc.) and links them with subsurface information for each stage. When the model is trained with a sufficient amount of data, it can be used as a real-time advisor that suggests a fracture treatment schedule (pump rate, sand concentration, fluid volume). The fracture geometry as well as the treatment efficiency can be optimized with the enhanced design. The model can eliminate or reduce the need for expensive subsurface characterization logging tools and provide quick recommendations for changes to the treatment.
This paper introduces an integrated solution to optimize fracture treatment design assisted by data analytics that can be further improved to solve other multivariable problems in the oil and gas industry.