Fitting a mixed regression model for C7+ characterization of petroleum fluids

Characterization methods of heptane-plus fractions (C7+) have a significanteffect on the phase equilibrium prediction of petroleum fluids and thereservoir simulations. C7+ characterization is encountered with majorproblems such as the limitation of available data and the wide range ofboiling points for reservoir fluids and crude oils. In the present work, anextensive statistical analysis was conducted on 900 experimental datapoints collected from different gas condensate and crude oil systems. Anew mixed linear regression model was proposed to characterize the C7+fractions by SG7+ as only available information with a good accuracy. Also,a different data set obtained from North Sea oil was utilized to evaluate themodel performance.