The accuracy of reservoir simulation depends on the grid block sizes: the smaller the grid sizes, the higher the
modeling accuracy, and also the closer towards the converged result. Using large sized grid blocks is often associated with severe
numerical diffusion. Chemical flood is a complex Enhanced Oil Recovery (EOR) process. Due to its complex mechanisms (the
emulsion phase behavior as an example), chemical EOR often requires very small grid blocks to achieve accurate simulation
results. In this work, we explore the possibility of using higher order numerical schemes to improve the accuracy of chemical
EOR simulations. First, we conduct a comprehensive review of the three major methods for improving the reservoir simulation
accuracy, including the upscaling method, adaptive mesh refinement and the use of higher order schemes. Considering the
particular problem of chemical flooding, we choose the use of a higher order scheme as our approach to solve this accuracy
improvement problem. Furthermore, we tested the one-dimensional and two-dimensional Alkaline Surfactant Polymer (ASP)
flood simulation problems. We found the simulation accuracy highly dependent on the grid block sizes used. In order to obtain
close to convergent results, chemical EOR requires much smaller grid blocks than normal water flood simulations. In one-dimensional
chemical EOR modeling, the larger the grids, the more artificial averaging we observe for key physical properties such as
surfactant concentrations, which ultimately lead to less recovery. Finally, by using second and third order numerical schemes,
we have found great enhancement in modeling accuracy when simulating one-dimensional and two-dimensional ASP flooding.
Using a coarse grid and higher order schemes may get the similar level of accuracy as fine grid simulation. Higher order schemes
serve as powerful solution to reducing numerical viscosity, increasing accuracy, reducing the required number of grid blocks and computational time. This method shall also assist the implementation of fast and accurate field scale chemical EOR simulations,
history matching, and optimizations.