Accurate identification and portrayal of a channel system is rarely easy. Exploiting it is even harder. The depositional processes and crosscutting relationships do not leave behind simple linear features with welldefined boundaries. Conventional examination of poststack seismic reflections leaves much to be desired. Is it a channel-levee system or an accretionary channel-infill architecture? Are there impermeable shaledrape and mud-plug boundaries separating permeable sands? The answers to such questions have a profound impact not only on drilling success rates, but also on fluid flow and hence exploitation, especially in tertiary production.
When it comes to reservoir modelling, conventional stochastic modelling misses many details, especially with respect to compartmentalization within a complex reservoir. Such compartmentalization often results in the loss of ability to recover large amounts of hydrocarbons. Using traditional poststack seismic, we noticed indications of potential channels within the shallow section of a 3-D seismic volume shot over a deeper Jurassic reservoir in the North Sea. To describe the geologic framework, we first mapped stratigraphic sequence boundaries immediately above and below the channels. These boundaries were used to create a framework for geologic scenario models.
The interpreted system was then mapped with the aid of seismic attributes. Every year, more attributes become available to make the most use of the information in seismic signals. Seismic facies classifications based on the attributes enabled us to detect geobodies, which we then used to guide the distribution of lithologies in a geomodel.
Our modelling technique provided a step change over purely stochastic modelling by basing the distribution of clastic lithologies on geologic processes. Stochastic (geostatistical) modelling methods, in contrast, do not provide the same level of detail for stratigraphic layering and facies relationships. We placed lithofacies and sedimentary structures within the reservoir model using rules based on depositional processes observed in laboratory and outcrop studies. This allowed us to create digital representations of the distribution of rock bodies within the reservoir, ending up with a finely layered model that mimicked the distribution you would see in outcrop and core.
The final goal was achieved. We were able to identify flow barriers, at a sub-seismic scale, associated with changes in mud content, boundary layer shales at the base of a main fluvial channel, and abandoned channel fill. By populating geologic process-guided facies models with porosity and permeability characteristics, we were able to create geomodels that described the distribution of these properties at a scale appropriate for reservoir simulation.
Peter Phillips and Jackie Zhang, Geomodeling