Caracterização de reservatorios com tecnicas de otimização combinatorial

AUTOR(ES)
DATA DE PUBLICAÇÃO

1992

RESUMO

Stochastic modeling has received increasing attention in the oil industry, being established as a ordinary tool for helping the elaboration of development plans in producing reservoirs. Combinatorial optimization techniques, such as simulated annealing, allow to produce equiprobable models of reservoir variables, reproducing a priori any characteristic that can be stated as an objective function, besides the histogram and the variogram, which can be honored with traditional geostatistics methods. In this thesis, the ability of the simulated annealing algorithm to incorporate geological and production (well test) data, reducing uncertainties in simulated models is reviewed. Recent works resuIts (Pérez, 199] j Deutsch, ]992) are reinforced and extended for a wider range of properties. Stochastic evolution, a newly developed algorithm, is introduced in probabilistic modeling. A comparison between this technique and annealing is performed, taking into account computing times and capacity of reproducing several complexity levels characteristics for different size problems. The quality of simulated models is also studied using some cri teria, such as CPU time, tolerances, correlation coefficients between simulated and real images, and flow performance. The algorithms ability for generating reservoir vertical cross sections with several constraining information levels is verified. The effect of inc1uding global (whole reservoir) variogram, local variograms, local averages and well test derived permeability with two approximations methods is analyzed.

ASSUNTO(S)

petroleo - geologia otimização combinatoria reservatorios

ACESSO AO ARTIGO

Documentos Relacionados