Permeability estimation from the joint use of stoneley wave velocity and support vector machine neural networks: a case study of the Cheshmeh Khush Field, South Iran

Accurate permeability estimation has always been a concern in determining flow units, assigning appropriate capillary pressure and relative permeability curves to reservoir rock types, geological modeling, and dynamic simulation.Acoustic method can be used as an alternative and effective tool for permeability determination. In this study, a four-step approach is proposed for permeability estimation from acoustic data. The steps include estimation of the Stoneley wave slowness from conventional logs using a support vector machine neural network, determination of the Stoneley wave slowness in non-permeable zones, calculation of the Stoneley permeability index, and calculation of the Stoneley-Flow Zone Index (ST-FZI) permeability using the index matching factor (IMF). Finally, a comparison is made between the ST-FZI permeability with those derived from CMR log and core analysis. The results of this study show that acoustic method in conjunction with robust SVM neural network can be considered as an accurate tool for permeability estimation in the mixed clastic-carbonate reservoirs with complex pore type systems.