High Resolution 3D Mechanical Earth Model Using Seismic Neural Netmodeling - Integrating Geological, Petrophysical and Ge

A 3D Mechanical Earth Model (3D MEM) was built integrating geological, petrophysical and geophysical data. The 3D MEM represented the close-to-initial rock property conditions, and is an ideal basis for e.g. wellbore stability and formation integrity forecasting during initial and depleted field conditions. Using neural networking, dynamic rock properties at the wellbore level and derived from seismic inversion were integrated to obtain a "true" 3D property model. The lateral resolution of the model was driven by seismic trace density; the vertical resolution was driven by upscaled well log information. The workflow was developed for constraining regional property trends, that honor structural elements in the field, and allow to derive realistic rock property values and distributions even in areas where only limited well log information exists. The 3D MEM demonstrated that (1) Neural networking driven by seismic inversion produces rock property values that are nearly identical with the upscaled logs, which is a remarkable improvement; (2) The distribution of the rock properties is controlled by main structural elements such as faults, erosional surfaces, layering and topography; and (3) The distribution of the rock properties has a high lateral and vertical resolution even in areas where wellbore information was sparse.