I am going to model particle fluidization in a pipe using 2D simulation. To compare the Fluent result with experimental data, how can I select an appropriate width an length for 2D model by having the radius and height of the pipe?
When your make a comparison between the experimental and simulation results, or generally speaking, between the model and prototype, please remember the principle of dimensional analysis. With this principle you would know that the non-dimensional numbers of them must be the same. For example, the Reynolds number, the Prandtl number, etc.
I think that you are sure about the axisymmetric condition of your pipe (just as a reminder: the axisymmetric geometry of a cylinder is the necessary condition, not the sufficient condition by itself; the non-asymmetric boundary conditions and body forces can invalidate the accurate 2-D solution!). If so, in the cylindrical coordinate system, the results will not depend on theta, they will only depend on r-direction (radial) and z-direction (axis of the cylinder). So, the 3-D cylinder will be reduced to a 2-D geometry, a rectangle with the width R (radius of the pipe) and length H (length of the pipe). The thickness of the pipe is neglected (if you don't care about the heat transfer in the wall), otherwise, the pipe will be an annular geometry whose 2-D counterpart will be a thin rectangle (representing the wall) attached to a thicker rectangle (representing the interior of the pipe).
Next step is the boundary condition, the axis of symmetry and pipe wall will be assigned to the two long edges of the rectangle (you are free to select); inlet and outlet to the two short edges.
When your make a comparison between the experimental and simulation results, or generally speaking, between the model and prototype, please remember the principle of dimensional analysis. With this principle you would know that the non-dimensional numbers of them must be the same. For example, the Reynolds number, the Prandtl number, etc.
2D means that the flow depends on 2 spatial coordinates. You have essentially two options for 2D:
- planar 2D an axial coordinate x and a transversal coordinate y
-axi-symmetrical 2D an axial coordinate x and a radial coordinate r
Looking at a transition between two flow remiges, such as the transition from laminar to turbulent flow, there is a huge difference between 2D planar and axi-symmetrical. While the onset of instability for 2D planar Poiseuille flow occurs at a finite Reynolds number based on the flow channel height H (Re_H about 8000) for infinitesimal perturbations an axially symmetrical Poiseuille flow can remain laminar up to infinitely large Reynolds numbers based on the pipe diameter D (Re_D>500 000 have been observed experimentally). I would advice therefore to choose an axially symmetric flow simulation. Furthermore the reults might be disappointing if the transition from non-fluidized to fluidized occurs localy in a strongly non uniform pattern. Simulations are useful, but should not be expected to be accurate. Actually the difference between predicted and simulated behavior can be a very interesting source of inspiration for further studies (experimental or theoretical). Comparison between experiments and a model is always much better than carrying only experiments. A simplified model is a magnifying glass allowing you to distinguish between the trivial flow behaviour and interesting deviations from the simple model.
Of course as noted Nazaruddin Sinaga using a dimensional analysis, to determine the key dimensionless parameters is essential. The only relevant results are dimensionless results presented as a function of dimensionless numbers.
Note that in experiments, the flow in a rectangular duct is strongly affected by spurious effects in corners.