Online Journal for E&P Geoscientists

Obeida, Tawfic A.^*1; Al Alawi, Nidhal ¹; Al Shan, Hani ¹
(1) RDD, ADCO, Abu Dhabi, United Arab Emirates.

Pilot objectives usually predetermined before pilot field implementations. To study different pilot performance predictions, dynamic reservoir simulation is used as screening tool for pilot performance cases to meet pilot objectives. A Mechanistic model "box model" which is a limited size simulation model is used to compare different pilot configuration performances on equal bases and save CPUs and simulation time.

A mechanistic model (1 km x 1 km) with refined grid (25m x 25m) was constructed. The reservoir properties, attributes and layering are similar to the geological "static" model to avoid upscaling issues. Transmissibility multipliers were used at the boundaries of the mechanistic model and production/injection rates of the edge wells were constrained to mimic full pattern performance.

Several pilot scenarios were studied in terms of well configurations, well spacing, well type (vertical or horizontal), well completion (reservoir sub-zone), injection /production rates, injected fluid (water, gas), injection continuality (continues injection, WAG) and WAG cycles. All these prediction cases were run under reservoir voidge replacement.

Certain criteria were selected to screen these cases such as recovery factor at breakthrough, ultimate recovery factor, sweep efficiency, injection rates and breakthrough time.

The mechanistic model is a good tool to screen many pilot scenarios in quicker time than running bigger compositional models. The prediction results of the pilot performance can be used to establish uncertainty (Tornado) charts to determine the impact of certain uncertainty parameter on oil recovery, sweep efficiency and breakthrough time. These charts will dictate the pilot design and pilot monitoring.

More than fifty cases were run and screened using compositional model. Handful of these cases which meet the pilot objectives were selected to be run in bigger sector model for comparisons with the mechanistic model. Results indicated good agreement between the mechanistic model and the sector model.