Agradecimentos: The authors are grateful to PETROBRAS, the research network SIGER 3 (Grant Agreement No. 0050.0100204.16.9), the National Agency of Petroleum, Natural Gas and Biofuels (ANP), the Center of Petroleum Studies (CEPETRO-UNICAMP) particularly the UNISIM research group, the Department of...

Agradecimentos: The authors are grateful to PETROBRAS, the research network SIGER 3 (Grant Agreement No. 0050.0100204.16.9), the National Agency of Petroleum, Natural Gas and Biofuels (ANP), the Center of Petroleum Studies (CEPETRO-UNICAMP) particularly the UNISIM research group, the Department of Energy of the School of Mechanical Engineering of the University of Campinas (DE-FEM-UNICAMP), the Foundation CMG and National Counsel of Technological and Scientific Development (CNPq), for supporting this research, and also, to the Computer Modelling Group (CMG) and Schlumberger for software licenses

Reliable reservoir characterization is essential to predict future behavior, however, scarcity of data results in uncertainties and consequently, high variability in simulated data. This variability can be decreased by history matching where uncertain parameters of the reservoir are altered to...

Reliable reservoir characterization is essential to predict future behavior, however, scarcity of data results in uncertainties and consequently, high variability in simulated data. This variability can be decreased by history matching where uncertain parameters of the reservoir are altered to minimize the mismatch between history and simulated data. The most complex uncertain parameters to be treated are characterization of petrophysical properties. Changing petrophysical properties is still a challenge due to the characteristics of the parameter. To reduce uncertainties in the spatial distribution of petrophysical properties, we can use image perturbation methods to redefine the probability distribution function for each property. Previous works present different approaches to perturb a single property of the reservoir, this paper moves further and we present a methodology to simultaneously treat multiple petrophysical properties. These properties can be categorical (facies) or continuous (porosity and permeability) and, together with a multivariate sensitivity analysis, we can identify which regions and attributes affect the mismatched objective function and so reduce the subjectivity in perturbing the model. To test our methodology, we use case study UNISIM-I-H, a complex synthetic reservoir with 25 wells and 11 years of history data. This approach has shown to be efficient and to allow for local perturbations, making possible the match of individual wells without mismatching others. Our methodology guarantees consistency of the matched reservoir model by preserving well log data, variograms and the relationship between different properties. Finally, this work contributes to the area of history matching by facilitating the integration of the process under a framework that takes into account probabilistic approaches and geostatistical modelling

CONSELHO NACIONAL DE DESENVOLVIMENTO CIENTÍFICO E TECNOLÓGICO - CNPQ

Fechado