CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FloEFD, FloWorks & FloTHERM (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/)
-   -   Rotational peridocity in domain? (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/220614-rotational-peridocity-domain.html)

Rotational peridocity in domain?

1 Attachment(s)
I am wondering the limits of what the domain can be in order to apply periodic boundary conditions for a rotational region. I have seen examples in technical articles where the domain (fan, shroud, etc.) are all perfectly revolved cylinders essentially. If I have a circular fan in a square shroud, though, can I still divide it into "sectors" and apply periodic BC's? I mean, the flow field is essentially repeated at each of the four "quadrants" (see attached picture) correct? I feel I should be able to divide the domain into 25%, but am not sure.

Or can one not apply rotational BC's in FloEFD, only translational?

 Boris_M October 18, 2019 08:06

No, the periodic boundary condition is in the plane you select and basically stretching into infinity. So in your second image, you can see the direction as dashed boxes getting weaker when you set up the periodic for one axis direction. So it is like cutting out a slice of a cucumber, it stretches for some distance into the other directions. You can then say, this slide is just the same as the next and the next and the next etc. which is basically true. But if you cut the cucumber in a 1/4 piece and use periodicity, then it is not the same into the infinity to the two sides. One side like in your image is round and the other is the cut off part. So they are not equal. It is like holding a mirror at the two sides of the periodicity. If you would do that with a cucumber slide as mentioned before, it would be correct expect that after some distance there is an end of the cucumber. If you would do that with your 1/4 piece, the one vertically cut side would be come half of it but then into the same direction the round side would start and the same in the other direction. So it would look like as if you would aling 1/2 sliced cucumbers on a plate into infinity. But that's not correct.
Also the flow is always acting the same at the periodicity BC. So if the flow tries to leave through it, there is basically the mirror telling there is also flow coming from the periodic side and acting against the flow that would want to leave. So like two equal forces acting onto each other and the result is there is nothing happening into the direction of the periodic BC, only parallel to it.

In particular to fans the rotating flow would need to come in from the one periodic face and then leave 90° through the other, but it would not be able as it is like a mirror so there is always a counter flow.

What you need is the sector periodicity or axial periodicity. Here you can specify it as a slice of x° and it is able to have rotating flow. So what comes into one side of the slice is leaving the other side. You would still need to make sure that it conforms to the actual x° periodicity of the model. So in your case, it is 90° and cannot be 20° and not even 45° as one is cut to the corner of the model and the other perpendicular to the side of the square and you always have to be able to let things to come in with the same as it leaves the end. So the 45° is not the same side as the perpendicular side. If you have a round fan and shroud, then the limit would be set by the blades of the fan. The reason is that the blade cannot be cut in half in the middle at the one side of the slice and have the end of the blade tip at the other because the blade tip flow would come in at the half-cut slice which is not physically correct. Ideally, you should have 3 blades anyway or more as you have to expect some numerical imperfections at the inlet directly, so the first blade is not ideal the second is good and the third is not ideal again with regards to the flow directly at the slide sides.

I hope this is clear,
Boris

Thank you, it makes sense. So to enable sector periodicity, I'm assuming the assembly in SolidWorks has to be manipulated prior to opening Flow Simulation correct? So you would basically be performing an extrude cut to eliminate the sections you do not want, leaving only a "wedge" shaped portion of it.

But my next question is, if you are running an internal simulation then these "open" sections that are exposed on either side of the wedge need lids, but the periodicity is not a boundary condition that can be applied to the face of the lids right? It is defined under computational domain I believe. So how would you handle the BC's for each side of the "wedge", because it is not as simple as having an inlet/outlet that opens to atmosphere.

 Boris_M October 18, 2019 14:15

1 Attachment(s)
No, you leave the model as it is as you only change the computational domain which tells what to calculate and how. The model can stay as a whole.

No, it is as simple as that. Only what is in contact with the computational domain is used for the inlet or outlet BC. You only need to consider that you use a portion of the mass or volume flow as this is the overall amount. The velocity doesn't care how big the surface area of the inlet or outlet is.

Regards,
Boris

 All times are GMT -4. The time now is 00:49.