| Popis: |
NMR signals measurements are being widely used to estimate reservoir petrophysical properties, which directly impact calculations of recoverable oil. NMR T2 distribution curves have plenty of information about rock properties, however most of the work done on this matter has been concentrated in studying only few parameters of that curve (logarithmic and geometric averages). This work is focused on the use of the total T2 distribution curve for better prediction of petrophysical properties. This means using all information of the NMR T2 distribution spectra instead of an average of them. The approach consists in the use of data mining techniques for a NMR T2 curves pattern classification. The algorithm used identified several patterns, which correlated with rock "storage" properties, such as porosity, FFI, BVF and T2 cutoff, with no apparent correlation for permeability. Using only short times of the curves, it was possible to, obtain an excellent correlation with the "transport" property of the system (permeability), reducing significantly the correlation with "storage" properties. The same analysis was done using the long times of the curves, with no correlation. These findings showed that the use of the total T2 distribution curve relates to rock "storage" properties and the use of short times of the same curve relates to "transport" properties. This very important result indicates that a specific non-linear algorithm should be used to predict the different rock properties, "storage" (porosity, FFI, BVF and T2 cutoff) vs. "transport" (permeability), from NMR signals. This new approach can potentially improve reservoir characterization and estimation of oil reserves, improving reservoir characterization and exploitation management. |
