Hybridisation of Geostatistical Realizations Constrained by Well Test Data

This abstract presents the approach of hybridization of geostatistical facies realizations with integration of a dynamic constraint conducted on a turbidite case study. It was observed that local uncertainties explored using multiple stochastic realizations have a drastic impact on the model dynamic response. The hybrid model is constituted of equiprobable realizations selected around wells based on matching criteria to well test data. The hybridization intended to deliver a robust facies model matching well test data and whose early dynamic response is now less sensitive to the facies realization.

The hybridization method implemented here consists in generating multiple equiprobable realizations of the initial facies and petrophysical model. All realizations are then simulated dynamically. Among these, the optimal realization providing the best match to well test data is selected for each well. The corresponding realization is then frozen around each well in a given radius of influence. This serves as hard data for the generation of the hybrid facies model. It was finally controlled that the hybrid model is geologically-consistent and provides a satisfactory match to well test data on all wells.

This hybridization approach allows the integration of dynamic data early in the construction of the static model.