Abstract

This study describes a new technique utilizing multiple geological models for deriving porosity and lithology from wireline logs. For any given model, only those mineral combinations which are geologically probable are used. Each combination consists of matrix and fluid parameters that are predicted from sample data and log crossplots. Each model is expressed as a set of linear response equations and then is solved by either matrix inversion or least squares as appropriate. Diagnostic terms and reconstructed logs are computed to indicate inconsistencies that may result from poor log readings or the presence of unaccounted models. Recognizing these deviations, values of porosity and lithology can be computed accurately and with confidence.

Examples are presented which demonstrate this methods utility when the variability of the reservoir is not condusive to constant log parameters. A Kalimantan and North Sumatra clastic example demonstrate how this technique can be used to efficiently derive clay parameters that yield consistent results. Treating gas as a separate fluid component is shown to enhance porosity computations in North Sumatra carbonates.