Fullfield Numerical Simulation Using Geostatistical Reservoir Model of Upper Main Body "B" Reservoir, Monterey Formation, Elk Hills Field, California

Abstract

Abstract This reservoir is located in the southern San Joaquin Valley of central California, about 20 miles west of Bakersfield. The Upper Main Body "B" (UMBB) is part of the Upper Miocene Monterey Formation. This turbidite reservoir was initially under crestal gas injection from 1976 to 1988. The reservoir has been under peripheral waterflood since 1983. An integrated study including reservoir characterization and full field simulation was started in late 1998 to evaluate near term enhancements to production. Geological mapping and correlation generated nine zones based on logs from 725 wells, 413 with modern log suites, of which 28 were cored. Geostatistical models (~25 million cells) were built with 413 wells for net pay, porosity, and permeability. These models were evaluated with sequential Gaussian simulation (sGs) and cloud transform algorithms on selected "sector" areas. The geostatistical models were upscaled for numerical simulation to allow for realistic process time. The final numerical simulation grid consisted of 240×140×20 cells with ~82,000 active. The simulation model was built by integrating: 1) PVT data from nine wells, 2) 40+ samples of special core analysis data (relative permeability tables and capillary pressure tables), 3) completion data for 332 wells completed in UMBB, 4) 25+ years of allocated production and injection history, 5) more than 7,000 bottom hole pressure surveys, and 6) repeat formation tests from 43 wells. Parallel processing with four to eight processors on an SGI ONYX machine significantly cut the run time. Integration and communication between simulation engineer, geostatistician, petrophysicist, and reservoir management team has led to considerable improvements in the simulation model. The combination of this process and intensive cross-disciplinary reservoir characterization resulted in significant time-savings in building the model and simulation history match time. The prediction cases are being utilized for future development planning for the UMBB.