Learning Forum

[Nabuzor]

Hello everyone,nI am trying to simulate different pipe flows with some obstacles inside:nstationary ones - these cases are often unable to achieve steady state and transient run is needed, but behaviour of a system is not perfectly repetitve over time;nrotors, which are moved due to forces exerted by fluid flow with use of six do solver - these are of course inherently unsteadynI am dealing with low Re numbers, like 1400, 1900 etc. which are calculated with respect to diameter of a pipe as a characteristic length and this would indicate laminar flow. Medium inside a pipe is water. However, due to unsteadiness of both cases and presence of such obstacles in flow I am not quite sure if I should use a laminar model or tubulent one (maybe even with some transition model turned on?) Or maybe I should calculate Reynolds number with respect to different characteristic length? I would be really grateful for any tips.nThank you in advance for your help.n

[Rob]

If you read the various papers by Reynold's turbulence occurs for Re over about 2300, laminar below about 1700. But, and this is where the theory breaks down, that's for clear flows. Add in disruption and you can easily see turbulent flow at Re of 1500-1700 and lower (I have a lab report showing turbulence at Re=1100ish courtesy of the local bus company). nIn CFD if you use laminar flow and it's marginally turbulent the flow tends to be transient or convergence isn't ideal. If you use a turbulence model and the flow is laminar you may diffuse some of the flow features out as the model promotes mixing. We tend to try and avoid transitional flow as engineers as it's hard to calculate drag (from charts) and oscillations in the flow can cause excessive vibration:heat transfer and mixing are poor too.If you read through the documentation you'll find there are transitional models, if you use these make sure you model sufficient upstream distance to let the solver determine the state of the fluid as there are coefficients to set. Given the turbines etc I'd favour using a turbulence model, probably kw and a very good quality mesh. nAs an aside, for rotors I'd use a sliding mesh and use a UDF to determine rotation speed from the force exerted on the blades. Don't forget about the gearbox losses. n