Rock physics analysis of well-log data

Qi Hu, Kristopher A. Innanen, Marie Macquet

We present a rock physics workflow based on the soft-sand model to convert reservoir properties (e.g., porosity, lithology, fluid saturation, and pressure) to seismic elastic attributes (e.g., velocity, density, and modulus) at the CaMI Field Research Station, Alberta, Canada. This model is selected based on the geological setting of the study region and its visible fit to the well-log data. We use the constructed rock physics model to predict the shallow section of velocity and density logs that are missing. The result shows a good agreement with the local geology. We further carry out sensitivity studies for the estimation of reservoir properties from seismic attributes. This is a nonlinear inverse problem, and we solve it using a directed Monte-Carlo method (neighborhood algorithm). Various input data parameterizations and model parameterizations are considered. We illustrate that most reservoir properties are difficult to estimate when the inversion system is underdetermined with non-unique solutions. To obtain accurate estimates, it is best to include enough input data or focus on limited solid and fluid phases by making appropriate assumptions on the others. Because the rock physics model used in the study is validated using well data, our analysis should be applicable to the regional area centered on the well.