One of the challenges of transporting highly viscous crude oil is to ensure that the flow of oil will be delivered. It is also necessary to keep the operational standards and conditions along sections of pipes and fittings. Today, with low oil prices, it is important to minimize energy losses through the pipelines and accessories. However, new designs are often based on correlations that have not been developed for heavy oil water mixtures and are not frequently reported in the literature. Moreover, conventional calculations do not take into account the presence of accessory lines, or simply consider by empirically adding an extra percentage of energy loss or according to the engineer design test. Even more, the current correlations that could estimate accessory loss do not work well for viscous fluids and are even less suitable for the case of two-phase mixtures. For example, Gardel correlation  was made for water flow through yee type accessories. Applying this correlation to viscous fluids result in high deviations, more than 500% compared to CFD simulations. The present work attempts to predict the fluid dynamics behavior and the energy losses of these viscous fluids and mixtures (oil - water) going through a Yee type confluence.
All simulations were carried out using ANSY CFX® v14.5. Mesh number of elements was optimized using Pipe-It® (optimization software). A grid independence study was also carried out automatically in Pipe-It® to ensure the quality of results. Several conditions have been simulated: angle confluence of 45°–75°, diameter ratio 2–7, oil viscosity from 10 to 105 cP, and water cut of 0–1. As the main result, a correlation that predicts the behavior of viscous mixtures in their passage through yee type confluences was developed using a genetic algorithms technique . This correlation takes into account: viscosity, fluid fractions, input speeds, confluence angle and other parameters that are not normally considered by other authors. Therefore, it may be used in mixtures of water with light and heavy crude oil. Finally, correlations with 10% deviation compared to CFD simulations were obtained.