Limitations of Constraining Reservoir Model Facies and Properties Using Seismic Impedance Data

The use of seismic properties from inversion of seismic data has long been an attractive source of information for conditioning reservoir model facies and property distributions. In the context of reservoir modeling both relative and deterministic inversions suffer from limitations related to scale change and resolution and, in the case of deterministic inversion, insurmountable problems introduced by the low frequency model which precludes deterministic inversion from use in reservoir modeling. In contrast, stochastic seismic inversion has several advantages as a means to condition reservoir models with seismic properties, particularly in addressing scale change, spatial variability and as a workflow for introducing the seismic properties and their associated uncertainty to the reservoir model. Although not perfect, stochastic seismic inversion is currently the best tool available for introducing seismic property constraints to reservoir models. A number of geostatistical issues remain problematic for reservoir modeling, whether seismic constrained or not. These include the problem of selecting and parameterising an appropriate spatial model and the stationarity assumptions of the model. There also exists a requirement for a new form of facies model description, one that is flexible enough to incorporate complex facies geometries, but is also computationally efficient. This requirement is summarized and explored.