Time-Lapse Electromagnetic History Matching in a Fractured Carbonate Reservoir

Summary We applied an image-oriented ensemble history matching workflow to a complex fractured carbonate model using production and time-lapse electromagnetic data. The developed workflow facilitates the conditioning of complex reservoir models to high-dimensional geophysical data, by combining a feature-based assimilation approach with a data subspace localization method. A hybrid parameterization approach is used to represent the multiscale fracture distribution in which the spatial distribution of small-scale fractures is modelled by a truncated Gaussian simulation method. Over one million uncertain model parameters including matrix and fracture properties are updated using an iterative ensemble smoother. Instead of directly integrating the high-dimensional noisy EM tomography data, the contour information extracted from the resistivity is assimilated through a distance parameterization approach. Experimental results demonstrate the efficiency of the developed history matching workflow for large-scale reservoir applications.