Improve runtime for modeling a complex brownfield

When operating the Heera field, located in the Mumbai Offshore basin, the Oil and Natural Gas Corporation Limited (ONGC), a central public enterprise under the government of India, experienced multiple challenges in performing highly detailed and complex dynamic modeling of the carbonate reservoirs. Accurate and efficient simulation of this brownfield was essential for determining methods for EOR and aiding corporate investment decisions.

The large faulted area (270 km²) contains seven stacked formations, with some reservoirs in communication, some in partial communication, and some with no communication. All of the formations have large variations in terms of petrophysical properties. There are 479 wells with commingled production and injection spanning more than 30 years. Well trajectories are complex and include multilateral and horizontal configurations, and prediction involves simultaneous water-alternating gas (SWAG) EOR. In addition, the dynamic model poses challenges for simulation because of the large number of cells; 21 equilibrium, 9 PVT, and 14 saturation regions; global and local changes in static properties for calibration; and uncertainty and complexity in modeling the carbonate reservoirs.

To ensure that there would not be an impact on the integrity of the model, rigorous testing was performed, including quality checks of input data and results. Certain engineering adjustments were made to original data deck to make it compatible with the INTERSECT simulator. The result was much more comprehensive use of the input data. The INTERSECT simulator uses an efficient operating logic and is scalable when run in parallel, which results in a much more efficient simulation run. When applied to the Heera field, the computing time was largely reduced with no effect to the results. After a successful evaluation, the INTERSECT simulator was used for the reservoir engineering study.

Achieved significant improvement in runtime, improving efficiency with confidence

The INTERSECT simulator significantly reduced the elapsed runtime, up to 20 times faster than in the original base case. As a result of the speedup, many more runs could be performed to refine the history match. Multiple sensitivities could be used to help understand and reduce the uncertainties in a more comprehensive manner. Moreover, the prediction cases could be optimized to provide the maximum recovery scenario. Following this experience, fine-resolution simulation will be applied in future studies for updating the models as the field is developed, preserving the predictability of the models and facilitating operational decision making.

Comparison of case runtime for both the history-match and prediction modes confirmed that the INTERSECT simulator cases were not only faster but also showed good parallel scalability in employing more processors. The INTERSECT simulator is based on a 16-core parallel configuration, so no runs were compared for less than 16 processors. The value of using the INTERSECT simulator was demonstrated as the project was completed with greater efficiency and improved confidence in the results.