[tobiley]

OJ thanks for your quick reply!

To convert to a 100% open area coefficient the fluent manual provides this formula Kl'=Kl*v(25%open)/v(100%open)

Idelchiks correlations are base on Kl=2*delta(p)/Rho*w1) and w1 is specified as the velocity before the perforated plate which is not influenced, which should be a 100% open area velocity in my opinion?! Therefore the coefficient is not calculated via v(25% open) and I thought I do not have to covert it...

Regards, Tobias

[oj.bulmer]

Okay, I admittedly generalized a situation a bit! I typically work with applications where the Reynolds number is high (in excess of 1e5). In such cases, the contribution due to viscous effects is negligibly small since inertial effects influence the physics. Hence I typically use the value of very high (1e10) so is very small. And I calculate C2 using Idelchik's values, by calculating for 100% open area.

For smaller Reynolds numbers, yes, you need to consider the viscous effects and specify both the factors. But here as well, you can always go for Idelchik coefficients for smaller Re and then use them to calculate your C2 values, thus incorporating both viscous and inertial effects using a single coefficient.

OJ

[fredom]

Hello,

Had Tobiley find anything interesting in your result ?

Which dimension is your model ? 2D or 3D?
Because If your model is in 3D I thought that you could model the perforated sheet metal as a rectangle without hole and introduce the pressure drop as Idelchik's mention in the book. (maybe it's wrong)

Best regards

Fredom

[oj.bulmer]

That is what porous jump does, by introducing a source term of negative pressure gradient in momentum equations. Value of C2 is derived from Idelchik values.

OJ

[fredom]

Ok, thanks to confirm my hypotheses oj.bulmer. I read that porous jump was only used for 2D case, does it true?
I will try to make an study (3D) with heat transfer where the flow is parallel to the perforated sheet...
If I have some difficulties I will create a new thread

Thanks

[fredom]

I found my the answer, porous jump is also applied for 3D cases and it's preferable to use it because it allows a better convergence...

[tobiley]

Hi,

The porous jump model is only a 1D simplification of porous media, are you guys sure that you can use it to solve a 3D problem? I always thought that you can only use it if you have a straight normal flow through your screen where a 1D Model like the Idelchik ones are applicable.

And another question, has anyone experiences with solving a porous problem by using the physical velocity (real hole velocity) instead of using the superficial (velocity of 100% open area)? I read in the Fluent manual that it is more accurate to use physical, but I get wired results with that option. E.g. flow pathlines are not going through the porous zone anymore or the flow pathlines are not turned. If I use superficial velocity the flow pathlines look sensable and if I specify a loss coefficient in all three directiosn they will be turned as well, as you expect.

Thanks and Regards,
Tobias

[oj.bulmer]

Originally Posted by fredom

I found my the answer, porous jump is also applied for 3D cases and it's preferable to use it because it allows a better convergence...

Originally Posted by tobiley

The porous jump model is only a 1D simplification of porous media, are you guys sure that you can use it to solve a 3D problem? I always thought that you can only use it if you have a straight normal flow through your screen where a 1D Model like the Idelchik ones are applicable.

You should be careful here. Yes porous jump can be used in 3D problems, but it is more relevant only when the flow is incident at normal angle. Please understand that regardless of how the flow is entering the perforated plates, it becomes almost perpendicular to its surface after exiting it- at least for reasonably thick perforated plates. But this fact is ignored by porous jump model and the flow continues in the same direction before and after perforated sheet, which is physically wrong. Now, in case of normal angles, again, the flow direction is same, but the flow is normal to the surface, and hence Idelchik's correlation apply. Having said that, I have seen papers which mention that Idelchik's coefficient apply to the angles upto 20-30 deg to the normal of the perforated sheet. But this again depends upon the type of perforated sheet.

Originally Posted by tobiley

And another question, has anyone experiences with solving a porous problem by using the physical velocity (real hole velocity) instead of using the superficial (velocity of 100% open area)? I read in the Fluent manual that it is more accurate to use physical, but I get wired results with that option. E.g. flow pathlines are not going through the porous zone anymore or the flow pathlines are not turned. If I use superficial velocity the flow pathlines look sensable and if I specify a loss coefficient in all three directiosn they will be turned as well, as you expect.

Can you elaborate more, what do you mean by "using physical velocity" instead of "superficial velocity"? I mean, use where and how?

OJ

[tobiley]

Thanks for your reply OJ!

Do you maybe remember the papers saying that you can use the Idelchik Correlations up to 20-30°? I made similar experiences if I compare the results of a simple CFD-Calculation of one hole (hole length is equal to hole diameter) with a passing flow over it (ca. 60% of the flow goes through the hole and 40% are passing) with the result of a porous region with the same mass flow rates (I used the Idelchik value for C also for the two other directions).

You can specify if you want to use the physical or superficial velocity in the Cell Condition Tab if you click on the zone which is specified as Porous zone. http://www.sharcnet.ca/Software/Flue...media-phys-vel

superficial velocity = porosity * physical velocity

Regards,
Tobias

[oj.bulmer]

Do you maybe remember the papers saying that you can use the Idelchik Correlations up to 20-30°?

Guo, B. Y., et al. "Numerical modelling of the gas flow through perforated plates." Chemical Engineering Research and Design (2012).

I made similar experiences if I compare the results of a simple CFD-Calculation of one hole (hole length is equal to hole diameter) with a passing flow over it (ca. 60% of the flow goes through the hole and 40% are passing) with the result of a porous region with the same mass flow rates (I used the Idelchik value for C also for the two other directions).

You mean 60% open area? How can you check the effect of incident angle using only one hole? I don't think it would be appropriate.

You can specify if you want to use the physical or superficial velocity in the Cell Condition Tab if you click on the zone which is specified as Porous zone. http://www.sharcnet.ca/Software/Flue...media-phys-vel

Oh, this is porous media. I thought you were using these values in porous jump! From the documentation, the only difference I see that the formulation using physical velocity (through holes) and superficial velocity (free stream velocity for 100% open area prior to the porous zone) is only differentiated by the factor of , the porosity. Thus the equations being solved are same as that of superficial velocity, with slight adjustments, to get the accurate values of physical velocities. However the overall result should be the same. Having said that, it is important to calculate the C2 and and as if you generally calculate them. Don't make them based on the hole velocity, but base it on superficial velocity. Once you tick the physical velocity option, FLUENT internally calculates the equivalent coefficients.

Moreover, I think the equation 7.2-31 on the link you gave is wrong. Since is a physical velocity here, and since the general equations are solved using superficial velocity ; the last term in this equation should have . But I don't see it there.

OJ

[tobiley]

I just made a simple model with one hole and periodic conditions, which has some open area ratio. Then I compared what is the difference if I have a pure normal flow through that hole and if I have a passing flow over the hole where around 60% of the entry mass flow goes through the hole and 40% leaves the model via another outlet. The amount of mass flow through the hole is the same compared to the case of the pure normal flow.
The result is that the pressure drop rises a little bit in presence of the passing stream and the flow out of the hole is not a pure normal flow anymore, but almost, most of the tangential or axial velocity is "killed". So as you would expect it. I just wanted to check if it will have a massive impact on the pressure drop if I have a passing stream, because Idelchik wrote that it is practically the same, which is probably right as long as the plate thickness is very small and the flow doesn't really have to turn. But if I specify the coefficients in the two other directions in the Porous Media Model as well I get a similar effect. If I only specify one in normal direction I get for sure a different flow distribution because the flow exits the Porous Media Zone normal. That's why I use the Porous Media Model instead of the porous jump boundary condition.
Am I right?



Regards, Tobias

[oj.bulmer]

Your images do not exactly tell your story. Can you clearly show where is the passing stream? Also, for some reason, I think that your second image is a bit dodgy, and unrealistic. Why the velocity at the inlet part so small?

A hole is opaque in both perpendicular direction, and is only open in normal direction. What coefficient did you typically specify in other two directions to make it reasonable according to you? I suspect that even if you specify the same coefficient in all three directions, the fluid will still escape normal to the surface, choosing the path of least resistance.

It is always preferable to use porous media over porous jump, if you want accuracy, since there is a correct energy balance in this method. But often it is very expensive.

OJ

[tobiley]

Hi OJ,

Thanks for your reply.
Sorry for my poor explanation, I attached three other pictures. Hopefully they make it a clearer.

What I meant regarding the coefficients is that if I do not specify all three coefficients only the normal one, I won't see any difference in the flow direction before and after the porous media zone as you mentioned it in your previous comments about porous jump. If I do specify the other coefficients as well I get for sure a different flow direction as it should be (if the hole has a reasonable length) the only question was which values do I need to put in. I am using now the same value as for the normal direction (Ansys even told me to use values up to factor 1000 higher).

Is that understandable?

Thanks and Regards,
Tobias

[oj.bulmer]

Like I said, since the hole is opaque in all two directions and open in only one, it is recommended that the transverse multiplier of 1000 should be used, as ANSYS documentation mentions.

If fairness, I do not believe that you can judge the suitability of porous model for angled flow over perforated sheet, just by simulating a single hole. This is because the top and bottom walls (that is the boundary walls intercepted by the direction of incident flow) will have significant influence over the flow. Even if you specify them as no slip/symmetry etc, they still will enforce no normal velocities, which isn't the case if you have perforated sheet. It is only fair when you use enough multiple holes so that boundaries do not influence the flow. I have done some simulations with inclined perforated sheet by modelling multiple holes in a narrow strip with all boundaries as symmetry. And I modelled at least 20 holes from top to bottom. Yet, I wasn't happy since top and bottom walls were significantly influencing the flow, which isn't be the case in reality. And goine beyond 20 holes means the mesh size would be in excess of 8 million, where my RAM maxes out

OJ

[tobiley]

That's why I wanted to model only one hole but your point about the walls is understandable and yes I tried to minimise their influence by symmetry etc. but indeed these boundary conditions will not enforce a normal velocity!

I didn't want to really judge the porous model. I made the calculation more for myself to get an understanding how big the influence of a passing stream is and if it is qualitative correct to define the perpendicular direction coefficients by factor 1000 higher. And I think therefore it is not necessary to model more holes, do you?
Due to the fact that the hole has a certain area/volume where the flow could flow tangential or axial (although the hole is opaque in the two directions), especially if the ratio between plate thickness and hole is small. Therefore I modelled that model with a ratio of almost one like the real plate will have.
Is that understandable? Or am I talking rubbish?

Thanks and regards, Tobias

Ps: Would you recommend to use the option alternative formlation in the porous media tab if I define the coefficients in the other two directions by factor 1000 higher?

[oj.bulmer]

Let me ask you a simple question, is the following statement right?

"In a real perforated sheet, the fuel can pass in normal direction but not in transverse direction. So the loss in normal direction is finite and the loss in transverse direction is infinite."

Do you think this is true?

OJ

[QCFD]

Using a transverse multiplier of 1000 (essentially restricting all flow motion in the plane of the perforated plate) is valid in my opinion only if the holes are much smaller than the characteristic dimension of the problem you are solving. If the holes are large, the physical flow will have an appreciable amount of momentum in the plane of the plate which will be eliminated in such a porous model. (within a large hole, your flow can very well have a local velocity component in the plane of the perf plate)

That said... if you cannot afford to model the holes explicitly then you have little choice but to assume the plate will kill all your transverse velocity and accept that you are missing some physics

[LuckyTran]

Originally Posted by stuart23

Last time I used porous jump, I had to specify Forcheimer and Darcy coefficients. How are you doing it with one single coefficient?

Not sure if this helps you in any way but setting the Forcheimer term to 0 will give you the simpler Darcy model. And for many problems only the Darcy term is known.

[Mandar saraf]

i want to simulate spray dryer, in that there are two inlet one of liquid spray and one of air.. which is through prefforated(air), this prefforated plate is 3mm thick and having holes of 3mm dia, this hole are arrange in traingular pitch of 5mm, this plate is made up of steel, i trie to model this but its very expensive, please tell me, wether it is possible to solve this problem using porous media?? if so please trell me what parameters required to have equivalent porous media as of this sheet,
dia of sheet is 1290mm, and mass flow rate of air is 2.12kg/s
please reply,
thanks

[LuckyTran]

Originally Posted by Mandar saraf

i want to simulate spray dryer, in that there are two inlet one of liquid spray and one of air.. which is through prefforated(air), this prefforated plate is 3mm thick and having holes of 3mm dia, this hole are arrange in traingular pitch of 5mm, this plate is made up of steel, i trie to model this but its very expensive, please tell me, wether it is possible to solve this problem using porous media?? if so please trell me what parameters required to have equivalent porous media as of this sheet,
dia of sheet is 1290mm, and mass flow rate of air is 2.12kg/s
please reply,
thanks

What you need is the equivalent Darcy coefficient (and/or maybe Forchheimer coefficient). You need some knowledge of the pressure drop vs velocity across the perforated plate (or pressure drop vs flow rate). You can get this info from say, the discharge coefficient or some correlation for your perforated plate. But what you will end up modelling is basically a pressure jump condition in your domain.