In this study, we construct and calibrate simulation models of four field-scale datasets: two from the Bakken, one from the Delaware Basin, and one from the Midland Basin. The datasets include diagnostics such as: offset pressure observation wells, cross-well and injection-well fiber, interference tests, post-frac perforation imaging, and production data. For each dataset, we build a list of ‘key observations’ and then vary model parameters to match each observation. The pressure observation wells and the interference tests provide strong constraint on far-field pressure depletion. The calibrated models provide more confidence in the spatial and temporal evolution of fracture conductivity over time. Key conclusions are: (a) fractures are finite conductivity, exhibiting large and persistent pressure gradient during long-term production, (b) fracture conductivity diminishes with distance from the well, (c) fracture conductivity decreases with pressure drawdown, and also, fracture conductivity decreases as a direct function of time, (d) proppant ‘immobilization’ (or, ‘trapping’) caused by fracture roughness is a key determinant of the propped length, (e) the magnitude of proppant trapping appears to a function of both formation properties and mesh size, (f) more speculatively, the magnitude of proppant trapping may also be related to fluid velocity/rate per cluster, and (g) it is advantageous to perform fracture injection testing to measure the pressure and stress profile above the pay zone, especially in formations where significant pore pressure variability is suspected above the target zone.
