Ekofisk Chalk: Core Measurements, Stochastic Reconstruction, Network Modeling and Simulation

Abstract

This dissertation deals with (1) experimental measurements on petrophysical reservoir engineering and morphological properties of Ekofisk chalk, (2) numerical simulation of core flood experiments to analyze and improve relative permeability data, (3) stochastic reconstruction of chalk samples from limited morphological information, (4) extraction of pore space parameters from the reconstructed samples, development of network model using pore space information, and computation of petrophysical and reservoir engineering properties from network model, and (5) development of 2D and 3D idealized fractured reservoir models and verification of the applicability of several widely used conventional upscaling techniques in fractured reservoir simulation.

Experiments have been conducted on eight Ekofisk chalk samples and porosity, absolute permeability, formation factor, and oil-water relative permeability, capillary pressure and resistivity index are measured at laboratory conditions. Mercury porosimetry data and backscatter scanning electron microscope images have also been acquired for the samples.

A numerical simulation technique involving history matching of the production profits is employed to improve the relative permeability curves and to analyze hysteresis of the Ekofisk chalk sample. The technique was found to be a powerful tool to supplement the uncertainties in experimental measurements.

Porosity and correlation statistics obtained from backscatter scanning electron microscope image are used to reconstruct microstructures of chalk and particulate media. The reconstruction technique involves a simulated annealing algorithm, which can be constrained by an arbitrary number of morphological parameters. This flexibility of the algorithm is exploited to successfully reconstruct particulate media and chalk samples using more that one correlation function. A technique based on conditional simulated annealing has been introduced for exact reproduction of vuggy porosity in chalk in the form of foraminifer shells. A hybrid reconstruction technique that initialized the simulated annealing reconstruction with input generated using the Gaussian random field model has also been introduced. The technique was found to accelerate significantly the rate of convergence of the simulated annealing method. This finding is important because the main advantage of the simulated annealing method, namely its ability to impose a variety of reconstruction constraints, is usually compromised by its very slow rate of convergence.

Absolutely permeability, formation factor and mercury-air capillary pressure are computed from simple network models. The input parameters for the network models were extracted from a reconstructed chalk sample. The computed permeability, formation factor and mercury-air capillary pressure correspond well with the experimental data. The predictive power of a network model for chalk is further extended through incorporating important pore-level displacement phenomena and realistic description of pore space geometry and topology. Limited results show that the model may be used to compute absolute and relative permeabilities, capillary pressure, formation factor, resistivity index and saturation exponent. The above findings suggest that the network modeling technique may be used for prediction of petrophysical and reservoir engineering properties of chalk. Further works are necessary and an outline is given with considerable details.

Two 2D, one 3D and a dual-porosity fractured reservoir models have been developed and an imbibition process involving water displacing oil is simulated at various injection rates and with different oil-to-water viscosity ratios using four widely used conventional upscaling techniques. The upscaling techniques are the Kyte & Berry, Pore Volume Weighted, Weighed Relative Permeability, and Stone. The results suggest that the upscaling of fractured reservoirs may be possible using the conventional techniques. Kyte & Berry technique was found to be the most effective in all situations. However, further investigations are necessary using realistic description of fracture length, orientation, connectivity, aperture, spacing, etc.