Correlating Recovery Factors with Measures of Reservoir Heterogeneity

Field scale reservoir simulations are computationally intensive, and therefore also time consuming. This thesis focuses on the development and evaluation of computationally easy measures of reservoir heterogeneity, by comparing their correlation with recovery factors after waterflooding and tertiary polymer injection. Expected recovery is related to heterogeneity, as heterogeneities increase the chance of bypassed oil during a waterflood. While field scale simulations include multiple fluid phases and many time steps, heterogeneity measures are often based on steady-state solutions of single-phase flow, which require much less computations. Heterogeneity measures can therefore be used for rapid screening of reservoir models, including positioning of wells and other potential recovery strategies. This allows for the evaluation of more options than what is feasible with only field scale simulations. After the initial screening, the best options can then be used for more thorough considerations. The work has been focused on new heterogeneity measures based on a model of permeability description by the use of streamlines, and heterogeneity measures from a parameter called the diffusive time of flight. The permeability description divides the averaged permeability over a reservoir model into hydraulic conductance, tortuosity, constriction and effective bulk volume. New measures have further been compared with existing dynamic measures of reservoir heterogeneity. In addition, the required code to calculate both new and existing heterogeneity measures has been implemented as an extension to the Matlab Reservoir Simulation Toolbox (MRST), which is an open-source tool for reservoir simulation and evaluation developed by SINTEF. The code will be shared with the MRST team at SINTEF, and can be included in coming releases of the software. The results show that several heterogeneity measures are closely correlated to expected recovery. Measures from diffusive time of flight were more promising than measures based on the permeability model, though especially tortuosity also shows correlation with recovery. However, the best correlations were found among some of the existing heterogeneity measures, especially sweep efficiency at 1 and 2 pore volumes injected and the dynamic Lorenz coefficient. The best existing measures gave a Pearson correlation coefficient above 0.9, which is very significant. The close correlations are interesting, as they indicate that the flow fields from single-phase steady-state solutions, which the heterogeneity measures are based on, has resemblance to the average flow fields during multiphase simulations