Material Balance Applied to Dynamic Reservoir-Surveillance Patterns

Abstract

Summary Determining the remaining spatial oil-saturation distribution or current reservoir-pressure distribution for a mature (water, solvent, CO2) flood is a cornerstone of reservoir management for improving sweep and selecting infill-well locations. Decisions of these types are typically supported by reservoir flow simulation models that have been calibrated to historical injection/production data. In this paper, we present a novel-pattern material-balance (PatMB) approach to estimating remaining fluids in place as an alternative to flow simulation. First, we use the historical injection/production volumes to solve for streamlines and streamline-derived pattern metrics such as well-allocation factors and injector/producer-pair reservoir pore volumes (PVs). Then, we apply material balance on these volumes over time to estimate the remaining oil in place (ROIP) and pressures at the end of history. Resembling reservoir simulation, the method considers changing well patterns through time, requires a 3D static geological model, and yields 3D saturation distributions of oil, water, and gas. However, unlike reservoir simulation, because historical injected and produced voidage terms are honored, calibration is only possible through the 3D distribution of PV and fluids initially in place. We present results for the Berrymoor-pattern waterflood and show that the ROIP distribution is a strong function of the original-oil-in-place (OOIP) distribution, well locations, and historical oil, gas, and water production/injection volumes. For this case, the ROIP distribution is almost insensitive to interwell permeability distributions, suggesting that the primary focus when estimating ROIP with the PatMB approach is to ensure a good estimate of OOIP, major flow units, and the correct injection/production data. We also compare our method to reservoir flow simulation for a large water/hydrocarbon miscible flood (HCMF), and we observed that the ROIP maps compare well, with both methods highlighting similar areas for potential infill locations. However, the remaining-gas-in-place maps differed with PatMB, showing a more diffused distribution than flow simulation of the gas. We attribute the difference to the fact that PatMB does not account for transport effects such as separation of the phases caused by density differences.