NONLINEAR REGRESSION MODEL FOR PERMEABILITY ESTIMATION BASED ON ACOUSTIC WELL-LOGGING MEASUREMENTS

摘要

Full waveform sonic logging is routinely applied in the oil industry practice. The primary and secondary waves inform about the porosity, elastic parameters and the orientation of in-situ stresses acting around the borehole. The separation of Stoneley waves propagating in the drillhole enables to determine the permeability of hydrocarbon reservoirs. Permeability is generally estimated from the inversion processing of Stoneley interval transit-time data. Statistical methods can also be found in the literature, which offer quicker and simpler solution with revealing the empirical relationships between the characteristic data of Stoneley waves and permeability. These methods do not require the prior knowledge of porosity. Case studies assume a linear connection between the relative decrease in Stoneley-wave’s velocity of porous formations and the natural logarithm of permeability, which usually gives only a rough estimate. In this study, the regression model is improved by the determination of a more accurate nonlinear relationship between Stoneley wave slowness and permeability. The new algorithm is tested over a wide domains of petrophysical parameters using an exactly known permeability model and synthetic well logs. The exponential model is also applied to the statistical processing of real well-logging data, where the estimated permeability log is compared to laboratory data measured on core samples. The quality checks of synthetic and field results show that the application of the nonlinear model is highly recommended to get a more accurate and reliable estimation of hydrocarbon reserves.