The objective of the work was to improve the knowledge and understanding of sedimentary facies which consist of thin beds, the impact of the cut-offs, and to improve the coherency and quality of 3D models.
Applying a cut-off consists in defining a threshold value on one or more logs to separate the reservoir intervals in which hydrocarbons are mobile from the gross rock thickness. In a thin beds context, this deterministic approach is insufficient because of the resolution of the logs and a limited number of data. The probabilistic approach considers that the relationship between the cut-off value on the selected log and hydrocarbon mobility in the sandy laminae is a statistical function.
The method is adapted to alternating sand-shale thin beds that have not undergone extensive burial, and has been applied to reservoirs in a turbidite depositional context. In such environment, the cut-off on the VClay log is the most discriminating.
The process of upscaling petrophysical properties is performed for each facies using two cellular mini-models. One is built at log resolution; the second at the scale of the cells in the reservoir model. The cut-off is applied in the log-scale mini-model using a probability function on VClay. The upscaling process provides distributions and correlation transforms for PHIE, NTG and PHIEnet which can be then used to populate the reservoir model. Transforms and Mean values differ from those obtained using a deterministic cut-off as the latter tends to underestimate the reservoir volume in shale-dominated thin bed facies and overestimate reservoir volume in sand-dominated thin bed facies.