CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Perforated sheet, wiremesh - resistance coefficient (http://www.cfd-online.com/Forums/cfx/114670-perforated-sheet-wiremesh-resistance-coefficient.html)

oj.bulmer March 15, 2013 06:29

Perforated sheet, wiremesh - resistance coefficient
 
We have been using Idelchik's "Hydraulic Resistance Handbook" for ages to find the resistance coefficients of perforated sheet, wiremesh and other porous regions. But irony is it is 2013 yet we are using handbook based on compilation of research more than two decades ago. Though there have been reprints of latest edition, the information has changed little.

I was wondering if someone knows about any other database/literature that considers the relatively latest research done in this area? In the era where everything meta-morphs within few years, there must be better ways invented after Idelchik compiled this handbook.

OJ.

ghorrocks March 16, 2013 06:02

It might be old but the resistance of a perforated sheet hasn't changed much in 20 years. So if Idelchik did a good job 20 years ago then the fact it is still being used is a sign of the quality of the work, not that it is old fashioned. Reynold's definition of the Reynold number is about 120 years old and still in frequent use. Why? Because it works. If it works then there is no need to supersede it.

oj.bulmer March 16, 2013 08:50

Well, that is true. But what I wanted to say is that as per his own words, his results are based on compilation of theoretical and laboratory studies (including crude experimentation at times) of research he has seen and done. Which was before 90s. Now, there are things to consider:
  • The theoretical correlations part (which he borrows into the book) may have improved over the years.
  • The data captured relates to the flow primarily perpendicular to the resistances. It is difficult to carve out the coefficients when flow is at an angle, or if the resistances are curved (conical/cylindrical etc), in which case the flow is not always perpendicular. Using the same coefficient in this case can be a significant simplification.

Reynolds number is a definition of properties of flow. The quantification of resistance coefficient is based on our understanding of cause and effect (in different scenarios of flow angles etc). We are better equipped today - experimentally and theoretically, than in 90s. I am a big admirer of Idelchik's work. But it's true somewhere that partly, the reason why engineers (including myself) use his handbook is because there is no other source available. At least, I don't know it, hence the thread. Incidently, I have seen experimental studies that do not agree exactly with Idelchik's predictions.

Don't want to sound cheeky, but Newton's definition of gravity WAS improved by Einstein, to cover a broader range of possibilities and our understanding of universe is better today than the times Einstein's best works were published :)

OJ.

ghorrocks March 17, 2013 05:32

Very true - when somebody does a better job of explaining what is going on then Idelchek's work will be superseded.

But to answer your original question: I am not aware of any better works than Idelcheks'. I have heard of people doing their own resistance measurements or simulations for their precise media, but I do not know of anybody who has brought a wide range of results together.

oj.bulmer March 17, 2013 09:52

1 Attachment(s)
Quote:

...I have heard of people doing their own resistance measurements or simulations for their precise media...
This is what I want to do, especially for the angled flow. When it comes to curved (cylindrical and conical) perforated strainers, the flow is not always perpendicular (refer to attached snap). I was considering simulations to figure out resistance coefficients at different angles. Problem is, if I know resistance coefficients for say, 3 angles (0,22.5,45); and assuming that the flow generally is incident on the perforated sheet within this range, how to figure out the final coefficient of resistance (based on these three) which I can apply to porous interface?

Thanks
OJ

ghorrocks March 17, 2013 18:09

If you can resolve it into directional components then the built in anisotropic porous region model should be able to do it.

oj.bulmer March 18, 2013 08:16

I am having trouble exactly here. I don't want to use porous region, but porous interface, though it would be more physical if I go with former. Making it a porous region is really expensive. There are several such designs of conical strainers with size ranging from 2 inch to 72 inch diameter. Using porous region for them means timelines will suffer. Hence, it has been general practice to stick to interfaces.

I was wondering - if I know the coefficients at different angles, is there any way by which I can come up with a resultant coefficient (as a function of angled-case coefficients) and then use it in my porous interface, such that this will be representative in its entirety of all angled cases? Perhaps some sort of weighed coefficient?

I know, it is on the border line of being ambitious and ridiculous. But I am trying to understand if it is feasible at all. If possible, it would be a breakthrough for my project since the predictions will be even closer to the actual values, without significant increase in computational resource.

Thanks
OJ

ghorrocks March 19, 2013 06:07

OK, if you want to keep it as an interface then I suspect you will need to specify it as your own momentum source term. Then you can specify normal and tangential components and keep the tangential velocity component. Have you considered this?

oj.bulmer March 21, 2013 06:48

Darcy's equation for pressure loss for resistance screen is:
\Delta P =  1/\alpha * \mu v L + 1/2 \zeta \rho v^2

\alpha and \zeta being permeability and inertial loss coefficient respectively. I typically omit first viscous resistance term since Reynolds number are in excess of O(1e5) and only proceed with the second inertial resistance term.

This is how I now approach the problem. I create an interface and use Pressure Change setting, and use following relationship to define pressure drop across the interface.

surfaceNormalV = Normal X *u + Normal Y *v + Normal Z *w
pDrop = 0.5*\zeta*FluidDensity*surfaceNormalV*abs(surfaceNormalV)

Normal X, Normal Y etc are components of normal vector at any point on conical strainer. surfaceNormalV is a dot product of this normal vector and the incident velocity.

Thus, I use normal component of velocity, apply Darcy's formula and thus realize pressure drop. Choosing normal components makes sense since majority of other-angled velocity components are consumed (because the perforated sheet is thick - the hole to thickness ratio is 1). Besides resistance coefficient is defined only for normal direction.

Incidently, this is an empirical relation for pressure drop. Unlike momentum source which would be expressed as negative pressure gradient.

dP/dx = -Cr2*u*abs(u), Cr2 = 0.5*Kloss*density=0.5*\zeta*FluidDensity/Thickness

Point is, I do not see any momentum source tab in interface. While there is a tab in porous region definition where I define above source. How do I implement a momentum source on interface (using pressure change)?

OJ

oj.bulmer March 26, 2013 06:53

Any comments?

ghorrocks March 26, 2013 07:27

Hmm, just looked in the documentation and it looks like momentum sources are not available at boundaries.

Have you tried the direction loss porous stuff? If you can specify the flow direction as a function of space (ie your conical surface) then you can have the normal coefficient going forwards and some multiple of that in the transverse direction.

oj.bulmer March 26, 2013 12:26

That's how I typically model the 3D porous loss cone, and specify the transverse loss multiploer as 1e8 times. But this mandates a 3D geometry. It is slow and time-consuming. Moreover, since the flow doesn't exist in transverse direction, it is just a waste of computational overhead to simulate a thick volume allowing through-flow in only perpendicular direction. Hence I was finding ways to implement this on a surface resistance, which seems ideal in this scenario, along with the "tuned" resistance coefficient that is applicable for both perpendicular and angled incident flow :)

Guess CFX has limitations afterall, when it comes to flexibility with interfaces...

OJ

ghorrocks March 26, 2013 17:54

I see. I do not use the porous models much so am no expert in them - sounds like you know me about them than me. But it does look like you might have to model it as a 3D region and cop the additional overhead.


All times are GMT -4. The time now is 07:24.