Integrated Dynamic Reservoir Modeling for Multilayered Tight Gas Sand Development

Abstract

This paper summarizes the approach used for applying integrated reservoir modeling to the tight gas sands of the Pinedale Anticline in western Wyoming. The simulation of tight gas sands such as those at Pinedale has always been challenging because of the high degree of heterogeneity that needs to be retained to replicate reservoir performance, coupled with computing constraints. Added to this, simulating the Pinedale reservoir has its own unique challenges due to its characteristically thick gross sand interval composed of multiple, heterogeneous sand bodies produced commingled in a well.An intensive data-gathering program to investigate optimum well spacing accompanied the simulation effort. A significant part of this program was the installation of pressure monitor wells 1 to detect communication with surrounding producers at the hydraulic fracture stage level. This was coupled with multiple time-lapse production logs. The two data sets together allowed better definition of stage performance at producing wells.Static models were built with fine resolution to duplicate reservoir heterogeneity. However, upscaling was necessary due to computing constraints. The upscaling procedure of Li and Beckner 2 was utilized to maintain substantial geologic heterogeneity. The upscaled model was calibrated to mimic fine scale well performance prior to history matching.Several sector upscale models were history matched using a statistical approach without compromising key aspects such as reservoir connectivity and proper mass withdrawal from each geologic sub layer. Hydraulic fractures in each stage were characterized through history matching. Given the geostatistical nature, an exact match on every frac stage and every pressure gauge located away from the producer should not be expected. Rather, a more statistical definition of a history match should be adapted to a level that still gives confidence in forecasting the value of future infill wells. The history-matched parameters were then statistically distributed to forecast more realistic future development wells.The availability of data including pressures at observation wells and production logs was critical in narrowing the range of uncertainty in the history-matched scenarios and reduced the degree of non-uniqueness in the model thus resulting in increased confidence in model forecasts. This paper describes the methods used to overcome many of the problems encountered in modeling heterogeneous tight gas sands, such as at the Pinedale Anticline.