Reservoir characterization and reservoir simulations: Fractured and nonfractured systems

Numerical simulation of oil reservoirs has become an important tool in petroleum engineering. The simulators solve partial differential equations for fluid flow that require determination and assignment of different reservoir rock and fluid properties over the spatial simulation domain.; A unique experimental system that did not require use of mercury as displacement fluid was designed and fabricated to measure fluid thermodynamic and rock-fluid properties. Experimental procedures were streamlined and pressure-volume-temperature (PVT) and rock-fluid properties were measured at reservoir conditions. Some of these properties were later used in the numerical models.; Effects of scale of reservoir representation and sand connectivity on production from a fluvial deltaic reservoir were studied. Reservoir images were generated using stochastic simulations. The variability in production was greater for the area that was less developed because of lack of data. It was observed that models with data generated at the resolution closer to the available data predicted flow better than at other scales. In order to study the effect of scale on flow performance, production from a unit at two different reservoir scales was compared. The production of the wells on the boundary was affected due to stochastically generated properties in the vicinity of the boundary. A sand connectivity index was defined and connectivity functions were calculated for different stochastic realizations. The effect of sand connectivity on production showed that both oil in place and sand connectivity affected production.; Numerical flow models were developed to study production performance of a complex, naturally fractured reservoir using a conventional dual-porosity, dual-permeability approach. Most of the wells in the field did not perform to their potential (as expected from oil in place calculations and general reservoir quality), and formation damage was suspected. Time dependent transmissibilities in the numerical models were used to quantify formation damage. In some wells the near well bore permeabilities were reduced by as much as 99% of their original values. Extremely low values of matrix rock permeability and fracture properties were needed for a history match, suggesting that either the fractures were not contributing to the flow or that there was extensive formation damage near the well bore. Sensitivity of the model predictions to variation in reservoir properties (particularly, fracture properties) was studied. The effect of reservoir scale on production from two specific wells was determined. The results showed that the production performance of a well was affected by the scale of study.; A numerical algorithm was developed in order to perform parallel flow simulations of fractured reservoirs. The performance of the simulator on a shared memory machine, and a distributed memory machine. The computational times reduced significantly with increase in the number of processors.; A novel stochastic approach was developed to generate fracture frequency distributions. The fracture frequency distributions considered not only the 'hard' fracture frequency data but also lithotype distributions as the 'soft' or 'fuzzy' data. This approach reproduced the dependence of fracture frequency distribution on the rock type fairly accurately and better than other approaches.