Two Procedures for Stochastic Simulation of Vuggy Formations SPE 69663
Casar-González, Ricardo Suro-Pérez, Vinicio
Two Procedures for Stochastic Simulation of Vuggy Formations SPE 69663 - Dallas, Texas Society of Petroleum Engineers 2001 - Casar-González, Ricardo .
Simulated annealing and stochastic simulation based on indicator kriging are used to describe vuggy formations in a Mexican offshore field. Besides heterogeneity associated to these carbonated formations, vug spatial distribution is an important issue since hydrocarbon storage and deliverability to the wells is highly controlled by vugs density and connectivity. From computed tomography images, exhaustive 3D statistics are derived to build 3D stochastic models of those vuggy formations. Multiple point statistics simulated annealing and sequential indicator simulation based on kriging are used to model vugs geometry. The results indicate that both approaches honor the target statistics, and visual inspection of such images confirm good reproduction of connectivity. Further, it is shown that computer performances are different with simulated annealing being the most demanding. Although stochastic images from indicator kriging can be used as initial images for simulated annealing, it is demonstrated that few value is added by imposing multipoint statistics in the framework of vugs modeling. Therefore, it seems that stochastic simulation based on indicator kriging can build, with an acceptable computer cost and good reproduction, 3D images of vuggy formations.
Two Procedures for Stochastic Simulation of Vuggy Formations SPE 69663 - Dallas, Texas Society of Petroleum Engineers 2001 - Casar-González, Ricardo .
Simulated annealing and stochastic simulation based on indicator kriging are used to describe vuggy formations in a Mexican offshore field. Besides heterogeneity associated to these carbonated formations, vug spatial distribution is an important issue since hydrocarbon storage and deliverability to the wells is highly controlled by vugs density and connectivity. From computed tomography images, exhaustive 3D statistics are derived to build 3D stochastic models of those vuggy formations. Multiple point statistics simulated annealing and sequential indicator simulation based on kriging are used to model vugs geometry. The results indicate that both approaches honor the target statistics, and visual inspection of such images confirm good reproduction of connectivity. Further, it is shown that computer performances are different with simulated annealing being the most demanding. Although stochastic images from indicator kriging can be used as initial images for simulated annealing, it is demonstrated that few value is added by imposing multipoint statistics in the framework of vugs modeling. Therefore, it seems that stochastic simulation based on indicator kriging can build, with an acceptable computer cost and good reproduction, 3D images of vuggy formations.
