Fine Scale Rock Properties: Towards the Spatial Modeling of Regionalized Probability Distribution Fuctions

Conventional logs and borehole images are common sources of short-scale property data for the characterization of petroleum reservoirs. Short-scale properties are consequential for production performance purposes but are also representative of too small rock volumes for direct full-field modeling. Average, possibly facies-based, properties are generally derived and stochastic simulation methods are used to simulate them throughout the whole reservoir. The regionalization of short-scale property distribution functions is considered here to take into account the short-scale variability of data. The novelty of the proposed approach is to regionalize mixtures of distributions and to correlate them to secondary information. Application to simulate distributions of fracture-frequencies in a naturally fractured reservoir illustrates the approach.