In-situ viscous oil viscosity estimation is critical for identifying sweet spots, selecting depths to collect samples, designing well completion and recovery strategies, and production predictions. NMR logging is the only technique that allows an indirect yet continuous assessment of viscosity. Accuracy of NMR based viscosity prediction largely depends on the robustness of the correlation. The existing empirical NMR viscosity correlations are either based on light and medium viscose crude oils, synthetic oils or extremely viscous oils or tar.
The Kuwait viscous oil has viscosity ranging from tens to few thousands of cP. The present study was focused on developing an NMR T2-viscosity correlation through low-field NMR experiments applicable to data acquired by different logging instruments, which operate at different frequencies and have different data acquisition parameters.
In order to limit number of unknowns, a viscosity correlation was developed that implemented the relationship between T2 geometrical mean (T2GM) at 5 inter-echo spacings (TE) to account for the signal loss related to the faster relaxing components of the crude oil. Compared to previously published T2GM or apparent hydrogen index based viscosity correlations, which uses a single TE measurement, the inclusion of multiple TE data reduces the sensitivity of variations due to noise or inversion of the fast decaying components. This allows the correlation to be easily applied to the analysis of logging data, which may be acquired from various logging service providers at different TEs.
The robustness of the viscosity correlation derived from this data was subsequently tested on select unconsolidated viscous oil core plugs. In-situ viscosity determination using the derived correlation was demonstrated by comparing NMR log measurements and oil data at reservoir temperature for different service provider data. Better agreement between the NMR predicted and the rheological viscosities are obtained in comparison to using the known correlations.