Abstract: Prediction of Mesaverde Estimated Ultimate Recovery Using Structural Curvature and Neural Network Analysis, San Juan Basin, New Mexico
BASINSKI, PAUL M., ABDEL M. ZELLOU, and AHMED OUENES
Mesaverde estimated ultimate gas recoveries were qualitatively predicted through the integrated application of structural curvature and neural network analysis for a four township area in the northern San Juan Basin, northwest New Mexico. A relationship was established between geologic structure and Mesaverde fractured reservoir permeability enhancement.
The project location was selected after Mesaverde geologic and reservoir characteristics were extensively quantified by Burlington Resources. Performance is strongly dependent on the degree of natural fracture permeability which varies by an order of magnitude across the area, based on reservoir simulation and type curve matching of forty years of Mesaverde production history.
The project employed: 1) a Laplacian surface mapping technique to derive a structural surface from well control which averaged three wells per 2.6 square kilometers; 2) a generalization of the curvature method to describe multi-azimuthal structural derivative properties; and 3) neural network analysis to delineate the underlying complex, non-linear relationships between reservoir structural derivatives and established Mesaverde production.
The input for this analysis was the chronostratigraphic Middle Lewis Marker well log top located approximately 500 feet above productive Mesaverde sandstones. The output was estimated ultimate recovery from the original Blanco Field Mesaverde 'parent' wells. This technique represents an innovative prospecting method to potentially high-grade areas with significant natural fracture permeability development using minimal input data.
AAPG Search and Discovery Article #90946©1997 AAPG Rocky Mountain Section Meeting, Denver, Colorado