This publication presents the calibration of a downhole nuclear magnetic resonance (NMR) log-based oil viscosity correlation with laboratory live oil viscosity measurements. The laboratory data set was acquired from formation tester sampling (FTS) including 37 pressurized single-phase oil samples taken from 11 wells. The FTS oil viscosity range was 1-1,400 centiPoise (cP).
In two Saudi Arabian carbonate fields, the moveable hydrocarbons consists of crude with in-situ oil viscosities of ~1-3 cP. Thick tar mats are located below the oil columns, separating the moveable hydrocarbons from the aquifers. For pressure support, horizontal water injectors are drilled into the heavy oil transition zones, located between the moveable oil and the tar, utilizing real-time logging while drilling (LWD) NMR data and formation tester mobility data for well placement. For optimum water injector placement, accurate NMR log-based determination of the reservoir oil viscosity is critical. The NMR logs are processed using an integrated petrophysical model that subdivides the oil volume into light, medium and heavy components. The in-situ viscosities are calibrated to the relative percentage of heavy-medium components to the total oil volume.
Despite the large geographic distance between the 11 sampled wells, the presented results reveal a remarkable consistency between the in-situ oil viscosity data from the FTS laboratory analyses and the NMR log responses. In this particular case, the well results suggest that one viscosity relationship is adequate for describing a large geographical area containing multiple medium and heavy oil reservoirs. The results indicate the logarithm of viscosity to be a clear function of the heavy-medium oil volume percentage. Two distinct linear segments are sufficient to cover the full 1-1,400 cP oil viscosity range. One of these segment describes the mobile oil column with low heavy-medium oil volume percentage and oil sample viscosity of less than 3 cP. The other segment defines the oil/tar transition zone where the presence of asphaltene aggregate structures leads to a very rapid increase of oil viscosity versus depth covering the range 3-1,400 cP. The robustness of the method is demonstrated by the low statistical uncertainties for the entire viscosity range, when comparing the predicted NMR oil viscosity correlation results with the laboratory results, from the 37 physical oil samples.
The new NMR empirical oil viscosity correlation was built on a previously published methodology, but the existing correlation did not do a particularly good job for the lower oil viscosity range (<10 cP) and for the very heavy oils. The purpose of this new publication is to present a new empirical NMR viscosity correlation with much wider validity range.