Difference between Friction in FDS and Friction from Moody chart

[Réda]

Hi everyone,

 

The aim of this modelling is to confirm  if FDS takes in account the MOODY chart as it 's written in the report  http://fire.nist.gov/bfrlpubs/fire09/PDF/f09029.pdf

"We verify that this is true for FDS by reproducing the Moody chart, a plot of friction factor versus Reynolds number for pipeflow [8]"

 

I am modelling a duct of 100*1*1m lenght*height*width , I put a pressure of 10Pa in the left side and and an OPEN in the right side

I added a roughness of 0.05 on all the walls.

I used this formula  λ = (2*ΔP*D)/(1.2*L*V²) to calculate the friction (ΔP,D=1 (hydrolic diameter), L=100 (lenght) and V (veolocity) are given from FDS using the DEVC)

the Reynolds = (V*D)/v  with v=15.6*10^(-6)

I used two different meshes, one with 10cm*10cm and the results is close the Moody chart, and when I use the mesh 5cm*5cm the results I had :

the friction in FDS is λ= 0.03853 and in the moody chart I found with the same conditions λ = 0.07, it's a big difference!

I don't see where is the problem, do you have an idea where the difference came from ?

 

here is the code :

 

!!! Generated by BlenderFDS 1.1.2 on Blender 2.62 (sub 0)

&HEAD CHID='coeff_frt', TITLE='simulation tube long 100 m ', /

!!! External config file

&TIME TWFIN=80./

&MISC TMPA=20./

 

!!! Boundary condition definitions

&SURF ID='Material', RGB=204,204,204, /
&SURF ID='rugosité', ROUGHNESS=0.05,/

!!! Computational domain

! 100000 cells, size: 0.05, 0.05, 0.05 m
&MESH ID='maillage', IJK=2000,20,20, XB=-50.000,50.000,-0.500,0.500,-0.500,0.500, /

 

!!! Geometry

! pression_entrée_imposé
&VENT ID='pression_entrée_imposé', SURF_ID='OPEN', DYNAMIC_PRESSURE=10,
      XB=-50.000,-50.000,-0.500,0.500,-0.500,0.500, /

&VENT ID='rugosité', SURF_ID='rugosité', XB=-50.000,50.000,0.500,0.500,-0.500,0.500, /
&VENT ID='rugosité.001', SURF_ID='rugosité', XB=-50.000,50.000,-0.500,-0.500,-0.500,0.500, /
&VENT ID='rugosité.002', SURF_ID='rugosité', XB=-50.000,50.000,-0.500,0.500,-0.5,-0.5, /
&VENT ID='rugosité.003', SURF_ID='rugosité', XB=-50.000,50.000,-0.500,0.500,0.500,0.500, /
&VENT ID='sorti open', SURF_ID='OPEN', XB=50.000,50.000,-0.500,0.500,-0.500,0.500, /

!!! Control logic and output

&DEVC ID='capteur_vitesse_sorti', QUANTITY='VELOCITY', XYZ=50.000,0.000,0.000, /
&DEVC ID='capteur pression entrée', QUANTITY='PRESSURE', XYZ=-50.000,0.000,0.000, /
&DEVC ID='capteur_vitesse_entrée', QUANTITY='VELOCITY', XYZ=-50.000,0.000,0.000, /
&DEVC ID='capteur pression sorti', QUANTITY='PRESSURE', XYZ=50.000,0.000,0.000, /
&DEVC ID='capteur_débit_sorti', QUANTITY='VOLUME FLOW', XB=50.000,50.000,-0.500,0.500,-0.500,0.500, /
&DEVC ID='capteur_débit_entree', QUANTITY='VOLUME FLOW', XB=-50.000,-50.000,-0.500,0.500,-0.500,0.500, /
&SLCF QUANTITY='VELOCITY', PBX=-50.000, /
&SLCF QUANTITY='VELOCITY', PBY=-0.000, /
&SLCF QUANTITY='PRESSURE', PBY=-0.000, /

&TAIL /

 

thank you for your time,

 

Réda

[dr_jfloyd]

You are computing using the velocity at the center of the pipe.  Friction loss in the Moody diagram is based on the mean flow. 

[Réda]

Thank you dr_jfloyd for your response,

 

I am using for my calculs the velocity deducted from the mean flow which is constant in all the pipe (DEVC, VOLUME FLOW).

I saw the the differents code that used to verify the values of friction in Moody chart from this link : http://code.google.com/p/fds-smv/source/search?q=Moody-*&origq=Moody-*

 

I tried to compile this one but it didn't work, the problem came from DEFAULT=.TRUE. , in the user guide this function is not mentionned.

However, to set a roughness in all the walls , I created differents VENTs, the thing is I don't know other ways to put this roughness.

also in this code, they used PERIODIC and FORCE_VECTOR to set a pressure, I didn't find the definition of this functions, in my case I used only a DYNAMIC_PRESSURE.

 

&HEAD CHID='z0=p01_dpdx=-1_N8', TITLE='LES of flow through a rough channel' /

&MESH IJK=64,8,8, XB=0,8,-.5,.5,0,1 /

&TIME T_END=100.0 /
&DUMP DT_PL3D=1.E10 /

&MISC FORCE_VECTOR(1)=1.,PERIODIC_TEST=100,STRATIFICATION=.FALSE. /

&SURF ID='rough wall', DEFAULT=.TRUE., COLOR='GRAY', ROUGHNESS=0.01 /

&VENT MB='XMIN', SURF_ID='PERIODIC' /
&VENT MB='XMAX', SURF_ID='PERIODIC' /

&SLCF PBY=0, QUANTITY='VELOCITY', VECTOR=.TRUE. /
&SLCF PBX=4, QUANTITY='VELOCITY', VECTOR=.TRUE. /

&DEVC XB=4.0,4.0, -0.5,0.5, 0.0,1.0, QUANTITY='U-VELOCITY', STATISTICS='MEAN' /
&DEVC XYZ=4.0,0.0,0.5, QUANTITY='U-VELOCITY' /
&DEVC XYZ=4.0,0.0,0.5, QUANTITY='VISCOSITY' /
&DEVC XYZ=4.0,0.0,0.5, QUANTITY='DENSITY' /

&TAIL /

 

 

my questions are :

 

- Is there any other recent verifications with the avalaible functions which work on FDS 5.5.3?

- Do i have to follow this method using PERIODIC and FORCE_VECTOR instead of DYNAMIC_PRESSURE? because I need to verify that the coefficient of friction from FDS is the same in the Moody chart in a real case and not theoric one.

 

Any help is welcome !

 

Réda

[Randy McDermott]

You need to use the periodic bc.  If you use different vents and outflow conditions then you have entrance effects, etc. and you are not modeling the Moody pipe flow experiments.

What is not "real" about set up in the verification guide?  Are periodic bcs not a good representation of a real duct system with a mean pressure drop?  If you don't want to use periodic bcs then you obligated to model the entrance region and make sure the outlet far enough downstream of your downstream pressure measurement.

[Réda]

Thank you very much for your help Randy,

The reason I avoid to use the Periodic BCs (PERIODIC option) because, I don't know how to use them ( I am a new user of FDS and I am learning now, I am trying to understand the files from the repository that you talk about in an old post ) and in the futur I am planning to modelize a ductwork (several air ducts) in a tunnel where i can impose a velocity of a fan in the entrance and see what happens, so I could calculate the singular pressure drop.

This is why, I wanted to confirm that the coefficient of friction from fds is the same in the moody chart and after compare the singular pressure drop with the Idelshik chart for simple cases. 

 

Is there a way to impose a Periodic Bcs and at the same time a special Vent for  imposing a velocity or a pressure ?

 

Réda

 

 I am using the method of the periodic BCs,

[Réda]

Hi everyone,

 

I am using now the methods of PERIODIC BCs with dp/dx=-1 (FORCE_VECTOR(1)=1) , I sumulated two cases, one where the cell size is 0.1m and the other one 0.05m, the results of the mean velocity in the middle of my duct are : 6.20m/s for 1st case and 9.18m/s for the 2nd case, I changed just the size of my mesh, and I didn't touch the text.

I don't understand this difference of the velocity, Do you have an idea where the error come from ?

 

thank you for your help,

 

Reda

 

Ps: please find attached the two files

[Randy McDermott]

Reda,

The dimensional cell size is not the relevant parameter.  For wall models you need to look at the z+, the nondimensional distance from the wall.  This is discussed in the verification guide.  The z+ of you first cell should be within the log layer, but the upper limit depends on the Reynolds number (roughly, your first cell should be such that z+ = O(100) or less).  If you are outside this range, or you have complex duct work, it is not surprising if you see grid size effects in your numerical solution.

It is good that you are scrutinizing FDS with regard to duct flow losses.  But note that the comparisons you make with Idelchik themselves would be considered a validation exercise.  To my knowledge, no one has yet published LES results for such cases (I would be very happy to get these cases into our guide).  That we have published LES results for Moody is itself a first.

If all you care about is getting the pressure drops correct, the best engineering approach is to use Jason's HVAC formulation, pulling the loss coefficients directly from Idelchik.  Complex CFD is not always the better approach.

Best,

Randy

[Réda]

Hello Randy,

 

thank you for your explanation, that explains the problems I faced.

 

I read in details the section of Velocity Boundary Condition in the technichal guide, I tried to understand how to calculate z+ for rough walls, if I understand well, I calculate u+ from u+=(1/k)*ln(z/z0) + B  and then I have z+ from u+=z+ or u+=A(z+)^(1/B) , is it right ? I had z+=187

my question is : do I need to calculate z+ to define the smallest cell size for being accurate?   I understood that z+ is only calculated for the first cell.

 

I would be happy to give you my cases for idelshick but now I am only an intern and i am still learning, if I have time at the end of my internship I ll send you the cases.

 

Yes I am caring about to get pressure drops correct so in that case I can modelize forms that are not defined  in idelshik, this is one of the aims of my internship. I tried to google Jason's HVAC formulation but I didn't find this formulation, do you have any link or documents on it ?

 

Thank you,

 

Reda

 

 14:21:03 UTC+2, Randy McDermott a écrit :

[Randy McDermott]

Reda,

z+ is defined as z/d, where d is the viscous length scale (d=nu/u_tau).  For very rough walls, z0>>d, you can use z/z0 as the nondimensional distance from the wall.  And, yes, when you set up your case make sure z/z0=O(100).  A good discussion of this is given in Pope (2000), Turbulent Flows, Cambridge, Section 7.2.

There is a discussion of HVAC in the User Guide, Section 9.5, and the Tech Guide, Chapter 11.

Best,

Randy

[Réda]

Dear Randy,

 

Thank you for your help,

 

Instead of trying to make verifications with my cases, I tried to verify the cases in the Technichal reference guide  Volume 2 (verification) section 3.4  figure 3.10 the green curve p.24.

 

first, I took the code from this link :

http://code.google.com/p/fds-smv/source/browse/trunk/FDS/trunk/Validation/Moody_Chart/FDS_Input_Files/z0%3Dp01_dpdx%3D-1_N8.fds

 

when I tried to run it , it didn't work, after I modificated it as in the file attached to this message (test.fds) I only added a roughness on the walls with Vents and PERIODIC_TEST=100.

 

After that I tried it with 3 differents meshes for Nz=8, Nz=16, Nz=32 (as it's mentionned in the guide) that corresponds respectively the cell size : 12.5cm, 6.25cm, 3.125cm. there is H = 1 hydrolic diameter, Periodic BCs and with dp/dx=-1pa/m,

I hoped to find exactly the same results but I didn't, the frictions f calculated from FDS are respectvily : 0.036, 0.032, 0.009. they are under the green curve. (the results are in the file attached resultat.ods)

 

With the periodic Bcs, is deltaP = 8pa ? and friction f = (2*H*deltaP)/(1.2*8*V²)?

 

I don't understand how the velocity changes with differents meshes ( I understood that it depends on the z+, and I calculated it as you said in the last post : z+ =z/z0 and z = cellsize/2, roughness z0=0.01 and for the three cases respectivly z+ = 6.25, 3.125, 1.565 < 100 ... so normally I should have a constant velocity)

 

Can you please just check where I made mistakes ? and if there aren't , can you explain me why it didn't work ??

 

Thank you very much for your patience and your help.

 

Reda

[Réda]

Hello,

 

I just need to find an explantion to that.

 

any explanations are welcomed...

 

Réda

[Randy McDermott]

Reda,

The results obtained in my report are using DYNSMAG=T. I think you
are seeing the results of constant Smag, which does not exhibit
convergent behavior.

Randy

On Apr 12, 5:01 am, Réda <reda.belmajd...@gmail.com> wrote:
> Hello,
>
> I just need to find an explantion to that.
>
> any explanations are welcomed...
>
> Réda
> Le mercredi 28 mars 2012 17:24:44 UTC+2, Réda a écrit :
>
>
>
>
>
>
>
> > Hi everyone,
>
> > The aim of this modelling is to confirm  if FDS takes in account the MOODY
> > chart as it 's written in the report
> >http://fire.nist.gov/bfrlpubs/fire09/PDF/f09029.pdf
> > "We verify that this is true for FDS by reproducing the Moody chart, a
> > plot of friction factor versus Reynolds number for pipeflow [8]"
>
> > I am modelling a duct of 100*1*1m lenght*height*width , I put a
> > pressure of 10Pa in the left side and and an OPEN in the right side
> > I added a roughness of 0.05 on all the walls.

> > I used this formula  *λ* = (*2*ΔP*D)/(1.2*L*V²) *to calculate the

> > friction (ΔP,D=1 (hydrolic diameter), L=100 (lenght) and V (veolocity)
> > are given from FDS using the DEVC)
> > the Reynolds = (V*D)/v  with v=15.6*10^(-6)
> > I used two different meshes, one with 10cm*10cm and the results is close
> > the Moody chart, and when I use the mesh 5cm*5cm the results I had :

> > the friction in FDS is *λ= 0.03853* and in the moody chart I found with
> > the same conditions *λ = 0.07,* it's a big difference!

> > I don't see where is the problem, do you have an idea where the difference
> > came from ?

> > **

[Réda]

Thank you very much Randy

 

I start to understand step by step ...

 

Reda