ESTIMATION OF NEAR-WELLBORE RELATIVE PERMEABILITY FROM SIMULATION AND INVERSION OF TIME-LAPSE MULTIPHASE PRODUCTION LOGS

摘要

Laboratory measurements of saturation-dependentrelative permeability are commonly used toconstruct reservoir models. However, because ofsample size and paucity, their reliability to predictreal-time production behavior of heterogeneousformations remains questionable. This paperintroduces a new method to estimate near-wellborehydrocarbon-water relative permeability frommeasurements acquired with production loggingtools (PLT). Production logs (PL) are simulated intime-lapse mode by effectively coupling fluid flowregimes taking place in both wellbore andhydrocarbon-producing formations.To simulate production logs, the reservoir-boreholecoupled flow model assumes slightly compressibleoil bubbles flowing through the incompressiblephase of water in the wellbore. We apply a onedimensionalisothermal two-fluid formulation tocalculate wellbore fluid properties in the presenceof velocity differences between fluid phases.Subsequently, a nonlinear inversion algorithmestimates saturation-dependent relativepermeability by minimizing quadratic differencesbetween borehole measurements of velocity,pressure, and holdup of the two phases, and theirnumerical simulations.Estimations of relative permeabilities using thenew method are within 5% of actual values invarious synthetic cases. Estimation accuracydecreases to 30% in the presence of 5% additiveGaussian noise for intervals where phase saturationis close to the end-point value of relativepermeability. Additionally, ignoring low-pressurecompartments in multilayer reservoirs withdifferential depletion causes underestimation ofrelative permeability. We also examine time-lapseproduction logs acquired in multilayer reservoirssupported by an active aquifer. Incremental waterproduction over time enables the inference oflayer-by-layer relative permeability. Feasibilitystudies with synthetic measurements indicate thatrelative permeability can be estimated with errorslower than 10%. However, the reliability of theestimation method is limited by the effective watersaturation window monitored in the wellboreduring the life of producing reservoirs.