1. “Don’t Forget Viscosity” Dave Bergman BP America July 28, 2004
Core Laboratories and The Petroleum Technology Transfer Council
2nd Annual Reservoir Engineering Symposium.
“From the Matrix to the Market – What You Don’t Know CAN Hurt You”
“Don’t Forget Viscosity” Dave Bergman BP America July 28, 2004

2. Background Viscosity Important to: Reservoir Modeling Flow Assurance
Viscosity Important to:
Reservoir Modeling
Production Rates
Mobility for water Flooding
Flow Assurance
Rates
Heat Transfer
Facility
Pump Design
Equipment Sizing
Pipeline Diameters
Heat Transfer Area
Just about any calculation you make will be dependent on Viscosity.
As Many Correlations as “Fish in the Sea”
Are all equally valid for your data?
Without data, +_50% (1 STD) errors in Prediction

3. Export Pipeline Example:
Pressure Drop by Correlation

4. Neither Correlation was Correct
Review Calculations
Neither Correlation was Correct
BR correlation used LOG(T,F) so Viscosity= Infinity at T=0F
Wrong temperature Dependency over all temperatures, but most noticeable below 70F, resulting is too much change with Temperature
Thodos also had wrong temperature Dependency, under estimating change with Temperature.
New Correlation Needed for extrapolation Versus temperature, <70F.

5. What should New Correlation look Like?
Check viscosity vs temperature for pure compounds.
Beggs and others had plotted Log Log (Viscosity+1) versus Log(T)
Started there.
Best Linear fit obtained with Log Log (Viscosity+1) vs Log(T,F+310)

6. Correlation for Pure Hydrocarbons

7. Dead Oil Viscosity vs Temperature

8. Correlations available for Calculating Viscosity
Over 30 in the Literature
Many developed on limited data set for a given area of the world.
Inputs Dead Oil: API, Temperature, Watson K, Molecular weight.
Compositional Correlations not covered in this presentation

9. Correlations available for Calculating Dead Oil Viscosity

10. Correlations available for Calculating Saturated Oil Viscosity

11. Correlations available for Calculating Under Saturated Oil Viscosity
Kouzel API Data Book

12. Dead Oil Correlations

13. Dead Oil Correlations

14. Dead Oil Correlations

15. Dead Oil Correlations

16. Live Oil Correlations

17. Live Oil Correlations

18. Under Saturated Oil Correlations

19. How Accurate are They? Dead Oil Correlations
Standard Deviation is around +-50% independent of Correlation given a wide data set.
Having 1 data point will significantly reduce your error to about the error in the Data, assuming using reasonable correlation
Live Oil +- 15%
Biggest difference is curvature. Related to Temperature and Gas Gravity
Under Saturated Oil %
Pressure range very dependent on Correlation

20. Expected Errors in Prediction of Dead Oil Viscosity
1 Std
1 Std

21. Expected Errors in Prediction of Live Oil Viscosity
1 Std
1 Std

22. Comparison of Under Saturated Errors: Live Oils

23. Comparison of Under Saturated Errors: Dead Oils

24. Export Pipeline Example:
Pressure Drop with Improved Correlation

25. Well Flow rate vs Viscosity Correlation

26. Viscosity by Correlation

27. Flow Rate vs Viscosity

28. Conclusions
No correlation will always be the best, and none will ever be perfect.
Experimental data near temperature of interest very important to minimizing errors.
Dead Oil correlation determines the accuracy of your results.
Using a better correlation may not give significant improvement
Using a bad correlation can be disastrous
Correlations are very dependent on range of data used in Development.
Compositional Correlations are less predictive than Field Parameter ones.