Improved Saturation Determination in Thin-Bed Environments Using 2D Parametric Inversion

Abstract

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThe analysis of deep-reading electromagnetic measurements is critical to the evaluation of hydrocarbon reserves. However, in thin bed formations, poor tool vertical resolution and corresponding low sensitivity to hydrocarbon presence make interpretation in the virgin zone difficult. A priori knowledge such as the formation geometry or auxiliary petrophysical information is necessary to overcome these difficulties. This paper presents a prototype code developed by Schlumberger S-RPC in collaboration with AGIP. Using this code, wireline or LWD, laterolog and induction measurements can be more correctly analyzed in thinly bedded environments (2-D geometry, fluid invaded layers perpendicular to the borehole).This code has been implemented in a software framework that provides a common environment specifically designed for electrical tool interpretation. Processing modules have a common interface and share common functionalities. Their organization reflects an implicit processing methodology, with progressive refinements that provides the interpreter with a robust and simple to use product, to better quantify reserves.A preliminary step is to determine the formation geometry, which is carried out by detecting bed boundaries and representing the formation as a vertical sequence of layers. Petrophysical analysis can be invoked to characterize certain formation properties such as shale volume and porosity. These steps are performed prior to resistivity log measurement analysis and serve as a form of a priori knowledge.Once the formation is described as a sequence of layers, wireline l ogging or logging while drilling (LWD) tool response can be computed using fast 2D simulators. The estimation of resistivity and the subsequent estimation of saturation will correspond to the minimization of a cost function, defined as the weighted squared difference between the measurement and the simulated response. Confidence outputs can be related to the local shape of the cost function at the end of the processing.Two important advantages of the new code must be emphasized: (1) the possibility to choose among several petrophysical models to better describe the environment and determine directly parameters such as hydrocarbon saturation, and (2) the possibility to group together beds which are too thin or too close to each other to be analyzed independently, into a so called single optimization interval described by a reduced set of parameters.This paper presents results obtained on selected benchmarks extracted from real data and compares them with those obtained through more traditional approaches.