Predicting Petroleum Phase Behaviour During Secondary Migration and Accumulation

Rolando Di Primio, Brian Horsfield, and Andreas Fuhrmann
GeoForschungsZentrum Potsdam, D-14473 Potsdam, Germany

The possibility to model petroleum composition during hydrocarbon generation as well as the PVT behaviour of the fluids during migration is now available in modern basin modelling software packages. The main problem in determining petroleum compositions from pyrolysis results lies in the fact that while the gas to oil ratio (GOR) and the liquid composition can correctly be assessed, the gas composition, which primarily controls fluid phase behaviour, is always too wet. In the comparison of natural petroleum maturity series and closed-system pyrolysis predictions we have recognised that gas composition is strongly heating rate dependent, hence, kinetic descriptions of the generation of individual gas compounds as measured in the laboratory are invalid for geologic heating rates. Bulk oil and gas generation are, however, not heating rate dependent. Gas composition is systematically related to GOR in natural fluids, whereby correlation coefficients are excellent for fluids from individual source rock organofacies. This observation allows the difference between pyrolysate gas composition and live oil gas composition of samples, related using their respective GORs, to be characterised. Following these results we have trained neural networks to predict natural fluid composition based on closed system pyrolysis results. Integration of compositional predictions from the neural networks and compositional kinetic models for bulk gas and oil generation allow the definition of multi-compound kinetic models which correctly reproduce petroleum composition and phase behaviour. These models have been implemented and tested in basin and fluid flow modelling with excellent results regarding GOR, saturation pressure and API gravity prediction.

AAPG Search and Discovery Article #90039©2005 AAPG Calgary, Alberta, June 16-19, 2005