Hello! We pump the contour of a metal tube with a diameter of 3 mm with oil with a viscosity of 0.0031 kg / m * s (abt. the viscosity is 4 times higher than that of water). We are interested in the pressure for pumping this circuit (at the inlet).
When validating the model in Ansys CFX with k-epsilon default settings, the difference between the result and the experiment reaches 100%, although the mesh is adjusted according to the tested and validated model, only on the water (with a maximum deviation of 10%). The Reynolds number in a pipe with a diameter of 3 mm is Re=2500, that is, we are dealing with a transitional flow regime, and the pipe is not hydraulically smooth in terms of the critical number Re=20d/"roughness".
Tell me, please, maybe for this mode it is necessary to use another turbulent model, for example, "Reynolds stress" or "k-epsilon" set up somehow differently? Maybe there are works in which a similar problem is solved, or there are works with recommendations on the use of turbulence models?