Modeling reservoir architecture with limited data

After successfully developing Cretaceous reservoirs in several northern Kuwait fields, an operator discovered hydrocarbons in deeper, naturally fractured carbonate reservoirs of Jurassic age. High-temperature/high-pressure wells were drilled in six structures, producing commercial quantities of gas-condensate and volatile oil.

At the time, however, well penetration was still limited and wireline logs were unable to capture the full variability of carbonate facies. Also, existing 3D seismic provided inadequate resolution of internal reservoir architecture for two reasons. First, seismic data had been acquired to target shallower reservoirs. Second, inter-bed multiples from overlying salt-anhydrite layers strongly affected the seismic signature of deeper reservoirs.

Existing geological models had been built on conventional layer-cake correlations, leaving considerable uncertainty as to the areal distribution of depositional facies and reservoir properties. To develop these reservoirs going forward, the operator needed a more reliable 3D geological model with a higher level of predictability.

The operator's geoscientists and engineers worked with Schlumberger petrotechnical experts to develop a new high-resolution static model of the largest and most important Jurassic reservoir based on sequence stratigraphic principles.

Building the 3D geological model

The multidisciplinary workflow began by conducting sedimentological studies on approximately 12,000 ft of cores. Typically, a third of the reservoir had been cored per well. In cored intervals, detailed descriptions successfully identified shallowing upward depositional cycles and distinct breaks in sedimentation representing sequence boundaries and maximum flooding surfaces. In uncored intervals, well log responses were tied to core descriptions to help identify these cycles. Integrating log and core data, the team interpreted distinct sedimentary facies.

For each key reservoir unit, gross depositional environment (GDE) maps outlined the areal distribution of facies such as anhydrite, dolomite, multiple types of limestone, and shale in depositional environments ranging from tidal flats to shelf and basin. The best reservoir facies were found in highstand tracts representing progradation of clinoforms toward the basin.

Construction of the 3D geological model took place in three stages. First, the sequence stratigraphic framework was built using the structural trend of a single good seismic reflector, faults mapped from 3D seismic, well tops, and stacked isopach maps of clinoforms. Second, 3D facies were modeled between wells using the 2D GDE maps as guides for each stratigraphic zone within each depositional sequence. Lastly, the 3D geological facies model was used to distribute key reservoir properties such as porosity, permeability, and water saturation.

Validating the 3D geological model in two ways

To validate the model, previously interpreted sequence boundaries, maximum flooding surfaces, and clinoforms were calibrated using stable isotope geochemistry. Also, clinoform slopes and basin geometry were validated by comparison with well-studied carbonate analogs such as the Permian San Andres outcrop in New Mexico and the Triassic Muschelkalk facies of Germany.

Reducing uncertainty in predicting reservoir facies

Following the development of the 3D geological model, the operator drilled four new wells. Prior to drilling, the study team predicted the type and thickness of reservoir facies that would be encountered. Predictions closely matched actual drilling results, confirming the model's enhanced predictive power.