Taizhong Duan¹, Timothy A. Cross², Margaret A. Lessenger¹

(1) Platte River Associates, Boulder, CO
(2) Colorado School of Mines, Golden, CO

ABSTRACT: Reservoir- and Exploration-Scale Stratigraphic Prediction Using a 3-D Inverse Carbonate Model

Accurate prediction of facies geometries, dimensions, and petrophysical properties from sparse control is a major goal of stratigraphic analysis. Another is to measure the accuracy and uncertainty of the predictions. Until recently, achieving these goals quantitatively was not possible. However, with the advent of stratigraphic inverse modeling, a new technology is available for predicting stratigraphic and sedimentologic attributes from limited well control.

A stratigraphic inverse model contains: (1) a forward model that simulates stratigraphy; (2) a set of observations comparable to forward model predictions; (3) an algorithm that calculates differences between predictions and observations; and (4) an inverse algorithm that simultaneously adjusts values of forward model parameters to improve the match between predictions and observations. The inverse model iteratively reduces differences between predictions and observations until a best match is achieved. The model calculates the degree of accuracy and uncertainty of the predictions.

Experiments on synthetic data show that the inverse model recovers the correct values of model parameters within narrow tolerances. Compared with our previous 2-D stratigraphic inverse model the error surfaces of our 3-D carbonate model are smoother and convergence of predictions to observations is quicker. The method is being applied to the caprinid/rudist-rich Albian-Cenomanian carbonate platform of southwest Texas, and the northwestern margin of the Plio/Pleistocene Great Bahama Bank.

Because the carbonate inverse model provides an accurate map of the 3-D distribution of lithofacies, model simulations may be used to calibrate and populate petrophysical properties for reservoir simulation. Stratigraphic and sedimentologic predictions of the model also may be used to locate optimal positions for exploration or step-out drilling.

AAPG Search and Discovery Article #90906©2001 AAPG Annual Convention, Denver, Colorado