The objective of this new approach is to provide a scoping tool for better economic completion design optimizations in unconventional reservoirs. The method utilizes extensive data mining of actual completion and production data and combines these data with the actual physics of hydraulic fracture growth and reservoir flow dynamics.
Current statistical multi-variate models (MVA) use either multiple linear regression or neural network type models to find correlations between relevant completion/reservoir parameters and production. While these models have shown merit, they may have poor predictive capabilities, especially beyond the range of existing data. The new hybrid approach uses physics-based variable transformations in the MVA to provide more realistic predictions of how completion parameters affect production.
The MVA model is then used in an economic optimization routine to minimize Cost$/BO to help find an optimized completion design.