Fourier spectrums clustering for automated facies recognition of field Y

Facies determination is a key parameter for proper modelling of reservoir behaviour. The subject of current research is optimization of interpretation process and decreasing the subjectivity in facies determination through the automated process of facies recognition based on Fourier spectrums clustering of SP logs. Spectral method is based on decomposition of SP curves into Fourier series consisted of basis of periodic functions orthogonal on the interval. Main attributes for clustering are Fourier coefficients, energy, homogeneity degree and slope of spectral density. EM algorithm is applied for clustering including opportunity to estimate the probability of facies recognition. Advantages of method are combination of parameters, which responsible for curve shape as a combination of different scale heterogeneities correlatable in the interwell space. Also the results of clustering allows considering descriptive geology in the mathematical sense, which reduce the interpreter bias and make it possible to correlate the facies of different reservoirs with the same attributes. The automated facies recognition is applied in the Field Y which is situated in Western Siberia, accounts 1774 wells with SP log distributed in the 268 km2 area. Methodology proved itself as a reliable tool for facies recognition with probability of 84% for known facies of reservoir bs11c.