Modeling apparent viscosity of waxy crude oils doped with polymeric wax inhibitors

Wax precipitation which frequently occurs throughout transportation and production of waxy crude oils results in various pitfalls including wax deposition, gel formation, and flow reduction, thus exacerbating pipeline performance. To mitigate such issues, polymeric flow improvers are mainly utilized in petroleum industry to improve the rheological behavior, and thus mobility and flowability of waxy crude oil. In the current paper, least squares support vector machine optimized with three algorithms including coupled simulated annealing, particle swarm optimization, and genetic algorithm is employed to develop intelligent models for the prediction of apparent viscosity of waxy crude oils doped with polymeric additives based on experimental data from previously published literature. A dataset with 526 datapoints is implemented to train and test the intelligent models. The forecasted values are compared against the laboratory data with the aid of graphical and statistical approaches. The findings illustrate that an acceptable correlation exists between the experimental and predicted values with coefficients of determination equal to 0.999136, 0.998848, and 0.998827 for the developed models optimized with particle swarm optimization, genetic algorithm, and coupled simulated annealing, respectively, based on total datapoints. The constructed models could be made use of to predict the apparent viscosity of waxy crude oils doped with polymeric additives without requiring experimental methods which are inherently expensive, tedious and time-consuming.

Automated summary

Intelligent models were developed using least squares support vector machine and three algorithms to predict the apparent viscosity of waxy crude oils doped with polymeric additives. The experimental results showed a high correlation between the predicted values and laboratory data.