Abstract
The aim of this work is to evaluate the desirability of conducting the History Matching using flow lines instead of traditional simulation. The results are very encouraging. The simulation time was reduced to a 10% from its original value, with a very good match. This was subsequently confirmed with a traditional simulation.
The San Jorge basin (CGSJ), located in southern Argentina, extends over the central part of the Patagonian region. The sedimentary fill of the basin is related to different rift and sag tectonic phases, from Triassic to Cretaceous. During Late Cretaceous-Early Tertiary, a marine transgression from the Atlantic developed. Tertiary sediments completed the basin fill. Sediments from continental rivers form a huge overlying of fluvial-shallow lacustrine units deposited under late sag conditions. These units contain the reservoirs that host the hydrocarbon accumulations of the basin. The study area to be developed by Secondary Recovery has more than 250 reservoirs and 300 drilled well. The drenage radio is about 150 mts. The average production of each well is around 15 Bbl/d.
As part of the Reservoir Caracterization, a high-resolution reservoir model was build. The model contains 5.4 million cells (1.1 are active) with 290 producing layers and 260 wells. Production started mid 1993 (22 years of History). Water flooding started 2002 when 21 wells were converted to injectors. It's a simple grid with 50 to 50 meters by cell. A conventional simulation run takes 12 hours. No assisted history matching was possible, due to the size of the model. With the use of a fast flow simulation technique, streamlines, the time was reduced to 2.5 hours. This makes possible to work over the simulation parameter. With this normal simulation time, enough runs were done to match pressure behaviors and field productions. So, a standard history matching workflow was used with the exception of flow lines to perform the simulation runs.
The use of stream lines reduced nine times the standard simulation time. Thanks to this, it was possible to manually iterate the model parameters, doing each time a new simulation run to evaluate results. This work methodology allowed the correct understanding of Reservoir drainage mechanism. With all this, history match was done in a reasonable time. The absolute difference in liquid production was less than 12%. Finally, a conventional run was performed to verify the consistency between the two methodologies. In this case, the difference between both runs is imperceptible. This confirms that the use of strem lines is an excellent tool to adjust big size models.
