A new Rock-Typing approach has been applied to a Lower Kharaib reservoir from an in development field of onshore UAE. A High Resolution Sequence Stratigraphy Framework introduces five High Frequency Sequences (HFS) which control the final layering of the model. Transgressive and Prograding intervals display different lithofacies and will have to be considered separately.

Discrimination of the model in half-sequences allows the construction of facies proportion curves within each half-sequence. Use of these proportions in the Geological Model will allow uncertainty assessment.

Rock-Typing approach has been completed following a workflow supported by a dedicated software, TECHLOG™. A statistical and purely objective method was applied, leading to a quantification of the uncertainty on the relationship between Geology (Lithofacies) and Petrophysical Groups at core scale (PGC). Each Lithofacies is in fact a combination of different petrophysical properties.

An important phase of data QC on logs (including normalization) enabled to obtain homogeneous log data and to discard inconsistent data.

The facies log typing used neural network supervised by the core facies. Around 75% prediction success was attained when comparing observed against predicted facies outside the reference data set and almost 85% successful prediction obtained within the training set.

The Petrophysical Groups or RRT assignment has been determined from both core and log data based on log-derived predictions of porosity and permeability. These Petrophysical Groups have been quantitatively defined incorporating also the capillary pressure data with the routine core data at core scale and using supervised neural network derived for the log scale and consequently in non-cored wells.

Finally, taking into account the uncertainties from this study (lithofacies proportion per half-sequence, quantitative relationship between Lithofacies and PGC and/or PGL), an uncertainty analysis could be done at the level of Geological Model by producing multi-scenarios models.

This content is only available via PDF.
You can access this article if you purchase or spend a download.