TX 75083-3836, U.S.A., fax 01-972-952-9435.
AbstractFirst, this paper presents a novel method used to derive accurate black-oil properties from a detailed EOS model. This entails the selection of the fluid sample, the choice of the simulated experiment and the representation of the composition gradient. In the case of significant compositional grading with a critical or close to critical transition within the reservoir fluid column, the table should be a compromise between (a) the in situ properties versus depth, which allows an accurate fluid in place representation and (b) the property change versus pressure during depletion. We then describe three field cases using this approach:A miscible gas injection into an oil reservoir. 1D fine grid simulations show that the black-oil model gives the same property profiles as the compositional ones. At the field scale, we take advantage of the lower CPU time in black oil to better represent some heterogeneity as fairways which could be detrimental to gas injection.A reservoir with a critical transition developed either by gas or water injection. With black-oil simulations both initial fluid in places and flow behaviour is correctly represented. But, even if the results are more stable than in compositional ones, the CPU speed-up is lower than expected.A black-oil representation for history matching a reservoir with a critical transition allows us to have a finer description of the geological heterogeneities with a run time lower than the work day.Finally, we propose improvements of our methodology based on the lessons learned from the simulations of the previous cases.