Cold heavy oil production with sand has become one of the main non-thermal schemes for developing heavy oil reservoirs in Canada. One challenge in modeling the fluid flow in the reservoir simulation studies is reservoir heterogeneity. Several seismic attributes were used to estimate the porosity (ranging from 19% to 35%) at the Plover Lake oil sands reservoir in Canada. First, the top and the base of the reservoir were mapped based on several seismic attribute volumes that include the density. From petrophysical analysis we learned that density is a key physical property in differentiating between sand and shale within the oil sands. Probabilistic neural network (PNN) analysis was used to derive the relationship between density log data and external attributes (PP and PS migrated stacks, AVO attributes and inversion results). Secondly, we used geostatistics to estimate a porosity map within the reservoir. The study is based on a set of porosity logs at well locations and several seismic attribute maps. Kriging, cokriging, kriging with external drift (KED) and multiattribute analysis for maps plus KED, were tested in order to improve the results. The KED with porosity from multiattribute analysis is the most realistic, honoring the wells and the seismic.
An iterative refinement method for determining a layered resistivity model from a Schlumberger or Wenner sounding curve is adapted to determine a layered resistivity model by using apparent resistivity and phase derived from the magnetotelluric impedance. Magnetotelluric observations presented as a function of period are first converted to an approximate resistivity–depth profile using Schmucker's transformation and this is used to construct an initial guess (starting) model. A two‐stage procedure is then invoked. Keeping resistivities constant, layer boundaries are first adjusted to give a minimum misfit between measured data and responses and this is followed by resistivity adjustments with fixed layer boundaries to reduce the misfit further. The method is illustrated by application to some synthetic data both exact and with added noise, to a real field data set and to some magnetotelluric profile data obtained in a survey over the Carnmenellis granites in south Cornwall. The method is validated by recovering conductivity models from the exact and noisy 1D synthetic data. For complicated three‐dimensional data at a single site and along a profile of stations, the method is shown to produce acceptable solutions which may be used as starting models in further two‐ or three‐dimensional studies.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.