Impact of data distribution on fluid sensitivity analysis: A quantitative investigation

Abstract Pore fluid is an important factor affecting the acoustical properties of reservoirs. Generally, variations in the physical parameters of rocks reflect their hydrocarbon saturation status. The intrinsic fluid sensitivity of these parameters can be quantified by the fluid sensitivity indicator (FSI). However, in practical application, the parameters selected by the FSI may not clearly separate fluid saturation statuses because of significant data perturbation (usually by lithology/structure variations in heterogeneous formations). Consequently, we introduced two dimensionless statistical measures: coefficient of data dispersion (CD), which represents the level of data perturbation, and overlap ratio (OR), which quantifies the discriminatory ability of two fluid saturation statuses in 1D/2D plots. The CD and OR provide tools for prescreening data and evaluating different rock physics parameters. They were successfully implemented on one set of synthetic data and two real datasets and yielded better results than FSI-derived parameters. For data from the sandstones of Daqing Oilfield (northeast China), the parameters obtained from OR reduced the overlap in the 1D and 2D parameter displays by up to 22.4% and 23.2%, respectively, relative to the parameters derived from the FSI. In the field example of the Sichuan Basin, China, the gas and brine layers are predicted based on crossplot analysis.