Nonparametric Statistical Determination of Facies and Petrophysical Parameters from Well Data

E. Rodriguez, D. M. Tetzlaff, R. L. Anderson

A novel statistical method for well data analysis has been developed and implemented in a set of computer programs (HORIZON) for use on computers ranging from PCs to mainframes. Most well information can be handled by the package, including quantitative data such as well logs and lab core measurements and qualitative data such as lithofacies descriptions and formation fluid types. Quantitative applications include estimation of log curves for editing and quality control, and estimation of petrophysical parameters such as porosity and permeability. Qualitative applications include facies identification, fracture detection, and determination of type of fluid to be produced. Utility programs permit the squaring of well logs and the generation of additional data from the nume ical characterization of curve shapes.

A data base is built using reliable information from selected depth intervals in key wells. A nonparametric statistical technique is then used for estimation in other wells or other depth intervals within the key wells where data are incomplete. No predetermined relationships between variables are assumed. HORIZON also provides a measure of the reliability of the estimations. In addition, existing statistical procedures can be used to determine the applicability of the method.

Two application examples are presented. In the first example, the package is used to edit a log affected by adverse borehole conditions. The second example is an application of lithofacies determination. The examples show that this technique provides a powerful way to extend knowledge from key wells to nearby areas without the use of explicit deterministic relationships.

AAPG Search and Discovery Article #91022©1989 AAPG Annual Convention, April 23-26, 1989, San Antonio, Texas.