An accurate understanding of the internal architecture of a reservoir is required to improve production forecasts. Due to the lack of information between wells, a geostatistical model is used to generate equiprobable lithofacies simulations between wells. Then, after assigning petrophysical parameters to these lithofacies, 3D fluid flow simulations are performed in these 3D geometries.
Variations of the geostatistical parameters implies differences on the production forecasts. Before giving some comments on the fluid flow behavior, the geological investigations and the main characteristics of the geostatistical model are recalled.
For an actual case, a methodology is proposed for a better integration of the geology into the reservoir evaluation process. The multidisciplinary approach proposed is illustrated on synthetic examples.
It has long been known that the heterogeneities of a reservoir must be taken into consideration for making realistic production forecasts1,2,3. In fact, reservoir heterogeneities are ones of the main factors for wrong forecasts. Generally these heterogeneities cannot be modelized with only well data because of long distances between wells.
The problem considered here is: â??How could we include general geological knowledges in a reservoir fluid flow simulator in order to improve production forecasts in heterogeneous reservoirs ?â?¿
THE CLASSICAL APPROACH
Indeed, when production data are available, the procedure generally followed can be diagramed as follows:A simplified model of what really exists is compiled by considering the reservoir to be made up of different levels with which petrophysical data are associated.The dynamic behavior of this reservoir model is simulated numerically.The computed well behavior are compared with actual one (mainly pressures and/or multiphasic productions).The petrophysical data are then modified to obtain a good historical fit of production on a well by well basic with regard to pressure and then to the flow rate of each phase (GOR, WOR). It is the so-called â??history-matchâ?¿ procedure.Then the reservoir model is used to forecast the reservoir behavior and to compare different production scenario
This procedure gives good results when:The structure of the reservoir is not too far away from that representation i.e. with large lateral homogeneity;Production forecasts concerns the same recover) process.