Facies associations dispersion via rule-based simulation with genetic algorithm and kriging: A methodology conditioned by seismic facies

Abstract

Rule-based simulation of facies is a valuable tool in oil and gas field exploration. The manuscript presents a hybrid methodology for simulating facies associations, using paleobathymetry as the main data and seismic facies as the secondary. A dispersion model of facies, which is based on paleobathymetry and wave energy is proposed, and the parameters of this dispersion are optimized via genetic algorithm, using seismic facies. The relationship between seismic facies e facies associations is quantified in order to construct a fitness function used in genetic algorithm to select good parameters for the dispersion model. In order to refine the model generated by the genetic algorithm, we use indicator kriging to adjust the simulation to the available well data. The methodology was applied to data from the Sapinhoa field-Brazil, focusing on three ages, 113 Ma, 114 Ma, and 115 Ma. The results obtained demonstrate that the simulated facies association maps largely respect the geometries observed in the seismic facies, and at the same time, honor the data from wells along the simulated area. The results demonstrate that the simulated facies association maps largely respect the geometries observed in the seismic facies while honoring well data across the simulated area. This approach addresses the inherent uncertainties and biases in traditional facies modeling, providing a more reliable and automated method for calibrating facies intervals.