CFD Online Discussion Forums - Wavy-vortex flow

(1). That's a great idea. What I did was I typed "Taylor-Couette" into Yahoo web search, and I was very surprised that there were so many sites related to this simple flow problem between two cylinders. (2). So, it took me a while to visit some of these Taylor-Couette flow sites. There were sites with vortex pictures, with book reviews, with test set-up, with FEM simulation,....etc...(3). So, if you are tired of running the flow over a square cavity problem, this so-called Taylor-Couette flow is going to be very interesting. (4). When the gap between the two cylinders is equal to the height, you have a square cavity problem in cylindrical coordinates.(with the inner cyliner rotating). This has some applications in turbomachinery in terms of the cavity flows, flow through seals,etc. But I am not going to touch those problems right now. (5). When the height of the two cylinders is much larger than the gap, it becomes a Couette flow in cylindrical coordinates. And if the radii of the two cylinders become very very large, you have almost the 2-D couette flow, that is a moving flat plate over another stationary flat plate leaving a small gap in between. (6). With this background, the interesting thing is people think that there is a relationship between this rotating cylinder system and the real world cartesian system. The only problem is, in the real world cartesian system, the system is open, and the outlet is not looped back to the inlet. Otherwise, it has been used to study the possible cause of the flow transition and turbulence. (7). In the finite system like the two cylinder problem, there is not much one can do. You can only change the rotating speed of the inner cylinder, make it a function of time (oscillating cylinder), feed the flow through instead of two end walls. Anyway, these are things you can change. (8). Under the critical Taylor's number of ( I don't have the definition for it yet) 1708, the flow will be steady state. after that, it will take several states depending upon the rotating speed. And the flow will be 3-D transient laminar first, and eventually it will become turbulent. So, you could be getting into the LES or even DNS domain. (9). But this wavy-vortex (I strongly suggest that readers should visit the Taylor-Couette sites with pictures first) is the second stage of development. So, it is not yet in the turbulent domain. (10). My suggestion is that you must find out the exact range of this wavy-vortex stage in terms of RPM first, and try to do the simulation with the same rotating speed. Whether you can capture this stage depends on several factors, such as the time-step, the mesh density, the numerical scheme used. (If you can't get the answer, you should contact the vendor about the problem and get their opinion as to whether the code is suitable for this type of problem or not).