Data-driven rock physics analysis of North Sea tertiary reservoir sands

Abstract

We have demonstrated an approach for data-driven rock physics analysis, where we first do facies classification using elastic well log data from several wells, followed by facies-constrained regression analysis to establish local rock physics relations for the prediction of VP and VS from geological input parameters. We have applied this approach to a multi-well log data set (53 wells, 40 of which had reliable/useful data) from the greater Alvheim area. We show how we can derive very robust local empirical rock physics relations for the prediction of P-wave and S-wave velocities as well as densities, for given combinations of porosity and clay volume. These locally derived empirical relations are recommended instead of universal rock physics models, even when the latter are locally calibrated. Using elastic facies with geological characteristics (cemented versus unconsolidated; normally compacted versus injectites; homogeneous versus heterogeneous) helps to improve the predictability of the regression models. The local rock physics relations that we obtain can furthermore be used to create training data for AVO classification.