Hi everybody.

I am working on CFD of an internal cooling system with mineral oil flowing inside the casing of an X-ray tube for medical imaging scanners. This is a conjuguate heat transfer problem with strong temperature gradients on the inner boundary walls. The oil is considered to be a non-participating medium as far as radiation is concerned. I am a bit confused as to whether or not I should be using turbulence models, and if so which ones. This is important since I really need to obtain heat transfer coefficients along the inner wall boundary surface (the heat source in the model). The oil's physical characteristics, and in particular kinematic viscosity, are very dependent on temperature (nu=14.2e-6 m2.s-1 at T=20C and nu=5.5e-6 m2.s-1 at T=50C). With an inlet tube diameter of 0.5 inch, I get an inlet Reynolds number between 600 and 1600 (based on tube diameter) depending on oil temperature. Inside the casing, velocity magnitudes are below 0.15 m.s-1 except in in some small-diameter areas where the velocity can reach 0.8 m.s-1. Is my assumption of laminar flow correct, and if not what turbulence model would be best suited to my problem ? I'd like to point out that my calculation mesh comes from somebody else and is definitely not refined near walls. Thanks !

(1). I think it is a good idea to run a few cases starting from a lower Reynolds number case. In this way, you can get a good feeling about the mesh effect near the wall. (2). As you increase the Reynolds number, you can plot the results against the Re. In this way ,you can establish a trend. It is easier to check the accuracy and reliability of the results this way. (3). since it is not easy to cover the laminar/turbulent transitional flow region, you can then approach the calculation from the fully turbulent flow end and work backward, that is reducing the Re for the turbulent flow calculations. This will establish another trend. (4). For the turbulent flow calculations, the standard wall function appraoch is recommended, mainly because of the concern about the mesh density there. At least, wall function approach is simple and repeatable. (5). At that point, you can explore the turbulence modelling effect and the more difficult ( actually very difficult ) subject of laminar/turbulent transitional flow.