CT&F - Ciencia, Tecnología y Futuro
versão impressa ISSN 0122-5383
GONZALEZ, Santiago e IDROBO, Eduardo A.. CARACTERIZACIÓN DINÁMICA DE YACIMIENTOS ESTRATIGRÁFICAMENTE COMPLEJOS USANDO ALGORITMOS GENÉTICOS. C.T.F Cienc. Tecnol. Futuro [online]. 2004, vol.2, n.5, pp. 23-51. ISSN 0122-5383.
A novel methodology is presented in this paper for the characterization of highly heterogeneous oilfields by integration of the oilfields dynamic information to the static updated model. The objective of the oilfield's characterization process is to build an oilfield model, as realistic as possible, through the incorporation of all the available information. The classical approach consists in producing a model based in the oilfield's static information, having as the process final stage the validation model with the dynamic information available. It is important to clarify that the term validation implies a punctual process by nature, generally intended to secure the required coherence between productive zones and petrophysical properties. The objective of the proposed methodology is to enhance the prediction capacity of the oilfield's model by previously integrating, parameters inherent to the oilfield's fluid dynamics by a process of dynamic data inversion through an optimization procedure based on evolutionary computation. The proposed methodology relies on the construction of the oilfield's high resolution static model, escalated by means of hybrid techniques while aiming to preserve the oilfield's heterogeneity. Afterwards, using an analytic simulator as reference, the scaled model is methodically modified by means of an optimization process that uses genetic algorithms and production data as conditional information. The process's final product is a model that observes the static and dynamic conditions of the oilfield with the capacity to minimize the economic impact that generates production historical adjustments to the simulation tasks. This final model features some petrophysical properties (porosity, permeability and water saturation), as modified to achieve a better adjustment of the simulated production's history versus the real one (History Matching). Additionally, the process involves a slight modification of relative permeabilities, which have been changed to allow calibration of these properties that also feature a high degree of uncertainty.
Palavras-chave : reservoir characterization; geostatistical modeling; streamline simulation; genetic algorithms; global optimization; automatic history matching.