PARALLEL SIMULATIONS OF COMMERCIAL SCALE POLYMER FLOODS

Despite advances in numerical techniques and computer hardwares, numerical modeling of large scale polymerfloods remains a challenge. A low order finite difference scheme with coarse grids, used as standard practice in oil industry, tends to smear the front and mask the benefit of polymer solutions whereas the finer grids require excessive computer times. A polymer module is added to in‑house simulator to benefit from parallel capability and efficient algorithms for fine grid field‑scale simulations. Application The simulator can be used for reliable predictions of large scale polymerflood performances. The impact of elastic properties of commonly used hydrolyzed polyacrylamide (HPAM) on oil recovery improvement can also be evaluated. Results, Observations, and Conclusions Polymers are added to water to improve the sweep efficiency in heterogeneous formations by reducing the water mobility. Polymer properties modeled are dispersion, adsorption, aqueous phase permeability reduction, shear thinning and shear thickening, and inaccessible pore volume. Aqueous phase property modified is viscosity that depends on salinity, polymer concentration, and shear rate. Polymer solutions exhibit non‑Newtonian behavior where the viscosity decreases as the shear rate increases. However, recently applied large HPAM molecules on the order of 20 MM molecular weight exhibit shear thickening behavior where above a critical shear rate the viscosity increases. Field‑scale simulations of polymerflood are performed using very fine grid and the results are compared with the commonly used grid resolutions. Significant of Subject Matter Polymerflooding is a mature technology to economically recover additional oil from medium to heavy oil reservoirs. A comprehensive polymer model coupled with fine grid simulations is required to accurately evaluate the field‑scale polymer flood performances. The biggest contribution is the capability of making commercial scale simulations of polymer flood with ultra large number of gridblocks and parallel capability on a cluster of PCs.