Within North Kuwait heavy oil fields, integrated reservoir modelling is challenged by inherent reservoir heterogeneities, regional non-stationarity (i.e. trends), asymmetrical well and seismic distributions, and the need to maintain alignment between various the model scales required and multiple purposes for which the models will be used. This paper presents a number of customized workflows adapted to characterize these reservoir architectures and heterogeneities within one field, appropriately at all model scales and in regions with variable well control.

A reliable new rock type classification scheme was derived from cross plot analyses of Gamma Ray and Bulk Density (GR-DENS) logs. Within an initial production area containing over 900 regularly spaced wells, 3D variograms for these lithotypes were estimated, calibrated with 3D seismic and reservoir equivalent surface outcrops. The lithotypes were distributed into full field static models using these variograms and the Sequential Indicator Simulation (SIS) algorithm. An additional declustering step was implemented to express regional trends and account for asymmetrical data distribution. Petrophysical property modeling (shale volume, effective porosity, water saturation) was performed using the Kriging algorithm conditioned to lithofacies.

From these full field models, sector models were created to capture geological heterogeneity at a smaller grid increment. Full-field facies were downscaled onto the sector model grids, and then the Sequential Gaussian Simulation (SGS) algorithm was used to interpolate petrophysical properties, constrained by histograms of the kriged background models. This allowed information from wells outside of sector models to be incorporated efficiently into them.

The facies and heterogeneities represented within the full-field static models have improved upon earlier versions, by being distributed more consistently relative to known seismic and well control, and to outcrop reservoir analogues. Modelled petrophysical properties also show a more consistent linkage with known values derived from core analyses. This consistent set of models can now be used with greater confidence, to answer questions ranging from in-place volume uncertainties to dynamic production forecasting, to life of field development. This has also led to reduced dynamic model run times, and improved reservoir management and operations optimization.

In summary a robust series of full-field and sector models was developed and customized to a North Kuwait heavy-oil field, with information from data-rich areas being elegantly applied to reduce uncertainties in data-poor areas. These nested models can now be matched to the detail required for the model purpose. For example heterogeneities that matter-for-flow in dynamic simulation models can be represented explicitly, whereas for full-field volume estimations property averages can be used.

You can access this article if you purchase or spend a download.