论文部分内容阅读
Precipitation and deposition of asphaltene have undesirable effects on the petroleum industry by increasing operational costs due to reduction of well productivity as well as catalyst poisoning.Herein we propose a reliable model for quantitative estimation of asphaltene precipitation.Scaling equation is the most powerful and popular model for accurate prediction of asphaltene precipitated out of solution in crudes without regard to complex nature of asphaltene.We employed a new mathematical-based approach known as alternating conditional expectation(ACE)technique for combining results of different scaling models in order to increase the accuracy of final estimation.Outputs of three well-known scaling equations,including Rassamdana(RE),Hu(HU),and Ashoori(AS),are input to ACE and the final output is produced through a nonlinear combination of scaling equations.The proposed methodology is capable of significantly increasing the precision of final estimation via a divide-andconquer principle in which ACE functions as the combiner.Results indicate the superiority of the proposed method compared with other individual scaling equation models.
Precipitation and deposition of asphaltene have undesirable effects on the petroleum industry by increasing operational costs due to reduction of well productivity as well as catalyst poisoning. Herein we propose a reliable model for quantitative estimation of asphaltene precipitation. Scaling equation is the most powerful and popular model for accurate prediction of asphaltene precipitated out of solution in crudes without regard to complex nature of asphaltene. We employed a new mathematical-based approach known as alternating conditional expectation (ACE) technique for combining results of different scaling models in order to increase the accuracy of final estimation. Output of three well-known scaling equations, including Rassamdana (RE), Hu (HU), and Ashoori (AS), are input to ACE and the final output is produced through a nonlinear combination of scaling equations. is capable of significantly increasing the precision of final estimation via a divide-andconquer princip le in which ACE functions as the combiner. Results indicate the superiority of the proposed method compared with other individual scaling equation models.