Advanced Machine Learning Applications to Viscous Oil-Water Multi-Phase Flow

The importance of heavy oil in the world market has increased over past twenty years as light reserves have declined steadily. high viscosity this kind unconventional results energy consumption for its transportation, which significantly increases production costs. A cost-effective solution long-distance transport viscous crudes could be water-lubricated flow technology. water ring separates oil-core from pipe wall such a pipeline. main challenge using lubricated system is need model that can provide reliable predictions friction losses. An artificial neural network (ANN) was used study to pressure losses based on 225 data sets independent sources. seven input variables current ANN are diameter, average velocity, density, viscosity, and content. developed backpropagation technique with processing neurons or nodes hidden layer demonstrated optimal architecture. comparison other intelligence parametric techniques shows promising precision model. After validated, sensitivity analysis determined relative order significance parameters. Some parameters had linear effects, while polynomial effects varying degrees