Efficient Ranking of Stochastic Reservoir Models: A Dynamic Approach Suitable for Natural Depletion.

@inproceedings{MonamicqGM2012,
 abstract = { Stochastic reservoir modeling can generate a huge amount of petrophysical realizations sampling the uncertainties associated with geological, seismic and well data. In this work, we develop a fast method to reduce the uncertainties by eliminating realizations inconsistent with early production data. Based on the work of Hird and Dubrule [1998b] and Hovadik and Larue [2007], we introduce a 3D dynamic connectivity using the fast marching algorithm. In particular, we propose specific strategies to account for material balance concepts and pressure decline during reservoir depletion in setting up the fast marching velocity. This algorithm allows a fast computation, on each static realization, of an estimated drained volume which characterizes the reservoir dynamic behavior in terms of mass balance.
We show that a relationship can be obtained between these estimated drained volumes and observed produced volumes by simulating flow only on a few realizations. An estimation of primary recoveries can be obtained for all the other realizations by computing their estimated drained volume and applying the relationship. Using this estimation and observed data, the likelihood of each realization with respect to observed produced volumes can be assessed. This quantitative ranking of realizations allows a reduction of uncertainties and can also be used as quality control, for instance in the frame of petro-elastic seismic inversion. },
 author = { Monamicq, Pierre AND Gentilhomme, Théophile AND Caumon, Guillaume },
 booktitle = { Proc. 32nd Gocad Meeting, Nancy },
 title = { Efficient Ranking of Stochastic Reservoir Models: A Dynamic Approach Suitable for Natural Depletion. },
 year = { 2012 }
}