Jackson, Richard Robert (Schlumberger Asia Services Ltd) | Zuo, Julian Youxiang (Schlumberger) | Agarwal, Ankit (Schlumberger) | Herold, Bernd Heinrich (Cairn Energy India Pty. Ltd.) | Kumar, Sanjay (Cairn India Ltd.) | De Santo, Ilaria (Schlumberger Oilfield UK Plc) | Dumont, Hadrien (Schlumberger) | Ayan, Cosan (Schlumberger) | Mullins, Oliver C. (Schlumberger)
Viscosity is one of the key reservoir fluid properties. It plays a central role in well productivity and displacement efficiency and has a significant impact on completion strategies. Accurately assessing areal and vertical variations of viscosity will lead to more realistic reservoir simulation and optimal field development planning. Downhole fluid analysis (DFA) has successfully been used to measure the properties of reservoir fluids downhole in real time. DFA has excellent accuracy in measuring fluid gradients which in turn enable accurate thermodynamic modeling. Integration of DFA measurements with the thermodynamic modeling has increasingly been employed for evaluating important reservoir properties such as connectivity, fluid compositional and property gradients. The thermodynamic model is the only one that has been shown to treat gradients of heavy ends in all types of crude oils and at equilibrium and disequilibrium conditions. In addition, fluid viscosity depends on concentration of heavy ends that are associated with optical density measured by DFA. Therefore, mapping viscosity and optical density (heavy end content) is a new important application of DFA technology for use as assessment of reservoir architectures and a mutual consistency check of DFA measurements. In this case study, a very large monotonic variation of heavy end content and viscosity is measured. Several different stacked sands exhibit the same profiles. The crude oil at the top of the column exhibits an equilibrium distribution of heavy ends, SARA and viscosity, while the oil at the base of the oil column exhibits a gradient that is far larger than expected for equilibrium. The fluid properties including SARA contents, viscosity and optical density vary sharply with depth towards the base of the column. The origin of this variation is shown to be due to biodegradation. GC-chromatographs of the crude oils towards the top of the column appear to be rather unaltered, while the crude oils at the base of the column are missing all n-alkanes. A new model is developed that accounts for these observations that assumes biodegradation at the oil-water contact (OWC) coupled with diffusion of alkanes to the OWC. Diffusion is a slow process in a geologic time sense accounting for the lack of impact of biodegradation at the top of the column. An overall understanding of charging timing into this reservoir and expected rates of biodegradation are consistent with this model. The overall objective or providing a 1st-principles viscosity map in these stacked sand reservoirs is achieved by this modeling. Linking DFA with thermodynamic modeling along with precepts from petroleum systems modeling provides a compelling understanding of the reservoir.
Downhole fluid analysis (DFA) has successfully been used to measure the properties of reservoir fluids downhole at in-situ reservoir conditions, in real time during downhole formation testing and sampling operations (Mullins 2008). DFA also has excellent accuracy in measuring fluid gradients which in turn provide valuable inputs to thermodynamic and reservoir fluid modeling. While the cubic equation of state (EoS), (Peng & Robinson 1976) is well now established to model gas-liquid compositional gradients and saturation pressure gradients of reservoir fluids, there had been no thermodynamic treatment to model gradients of dissolved (or colloidally suspended) asphaltenes in reservoir fluids. This deficiency is largely due to the lack of understanding of the molecular and colloidal sizes of asphaltenes in crude oil, and even in laboratory solvents. In recent years, the molecular and nanocolloidal structures of asphaltenes has been clarified and codified in the Yen-Mullins model (Mullins 2010, 2011, Mullins et al. 2012). Recently, important confirmation of this model has been obtained from various sources (Majumdar et al. 2013, Korb et al. 2013, Goual et al. 2011, 2014).