Multimineral log analysis is a quantitative formation evaluation tool for geologic and petrophysical reservoir characterization. Rock composition can be estimated by solving equations that relate log measurements to the petrophysical endpoints of minerals and fluids. Due to errors in log data and uncertainties in petrophysical endpoints of constituents, we have used effective medium models from rock physics as additional independent information to validate or constrain the results. We examine the Voigt-Reuss (VR) bound model, self-consistent approximation (SCA), and differential effective medium (DEM). The VR hound model provides the first-order quality control of multimineral results. We first show a conventional carbonate reservoir study with intervals in which the predicted effective medium models from multimineral results are inconsistent with measured elastic properties. We use the VR bound model as an inequality constraint in multimineral analysis for plausible alternative solutions. The SCA and DEM models provide good estimates in low-porosity intervals and imply geologic information for porous intervals. Then, we present a field case of the Bakken and Three Forks formations. A linear interpolation of the VR bound model helps validate multimineral results and approximate the elastic moduli of clay. There are two major advantages to using our new method: (1) Rock-physics effective medium models provide independent quality control of petrophysical multimineral results and (2) multimineral information leads to realistic rock-physics models.
展开▼
