Successfully discriminating future lateral well locations with above average production from those with below average production is important for financial planning and field development with unconventional resource assets. To address this, a predictive analytics approach is utilized that will maximize an asset's value through infill drilling prioritization, as well as drilling and completion design by creating a production metric model.
As with most unconventional resource plays there is an abundance of production data from numerous lateral wells in various landing zones. A predictive analytic production metric model can be constructed for each landing zone using a supervised neural network-training 3D depth migrated seismic attributes with this statistically-rich production information. The production metric model indicates where the best rock (above average production) is, where strata with below average production is, and where completion or well paths might be optimized to improve production.
While most lateral wells drilled into unconventional shale reservoirs or benches in the Permian Basin will be economical, some wells will have higher yields or better performance than others due to variations in shale mineralogy (Dopkin, et al, 2017). With modern high-fidelity seismic data, we can systematically identify variations in silica content, effective porosity and total organic carbon content either directly or by proxy. However, determining the best combination of these interdependent rock properties in terms of higher yields can be problematic, since the variability of shale reservoirs can demonstrate significant mineralogy changes across the extent of the asset (Bashore and Grant, 2018). The complexity and high variability of unconventional reservoirs (Dopkin, et al, 2108; Singleton, 2018) require a more generalized solution to classify asset areas with higher production potential beyond simple cross-plotting of wells with one or two seismic attributes and or simultaneous prestack inversions (with subsequent geo-mechanical or litho-porosity transforms). To that end, a predictive analytic production model is constructed using multi-attribute seismic data and normalized production metrics from lateral wells.
