CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > FLUENT

Pipe flow simulation

Register Blogs Members List Search Today's Posts Mark Forums Read

Like Tree1Likes

Reply
 
LinkBack Thread Tools Display Modes
Old   February 14, 2013, 12:41
Default Pipe flow simulation
  #1
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Hello guys!

I'm a final year student trying to complete a project using Ansys FLUENT. One step of which is to try to simulate a 3D pipe flow, then compare the simulation results to Poiseuille's law.

Now, the problem is, the pressure gradients I got from the 3D simulation is nowhere near the one calculated with Poiseuille's equation. What could possibly have gone wrong with my settings?
- I created and meshed a cylinder - with inlet, wall, outlet as boundaries
- Laminar flow
- No-slip wall
- Outlet = outflow, inlet = velocity-inlet (with a known velocity at inlet)
- Solution method: SIMPLE
- The pipe is long enough for the flow to be fully developed before it reaches the outlet

I've read through a few tutorials on pipe flows and such, and managed to follow through the steps, but the problem still persists. So now I'm kinda at a loss.

Any suggestions?
flippy is offline   Reply With Quote

Old   February 14, 2013, 17:22
Default
  #2
Member
 
Join Date: Sep 2012
Location: FL
Posts: 76
Rep Power: 4
victoryv is on a distinguished road
You need to have fine mesh at the boundaries to resolve the boundary layers. Change the residuals accuracy to 10^-6 and monitor other parameters as well along with the residuals.
victoryv is offline   Reply With Quote

Old   February 14, 2013, 18:17
Default
  #3
Far
Super Moderator
 
Far's Avatar
 
Sijal Ahmed Memon (turboenginner@gmail.com)
Join Date: Mar 2009
Location: Islamabad Pakistan
Posts: 3,914
Blog Entries: 6
Rep Power: 38
Far will become famous soon enoughFar will become famous soon enough
Send a message via Skype™ to Far
Quote:
Outlet = outflow
Make it pressure outlet
Far is offline   Reply With Quote

Old   February 14, 2013, 20:26
Default
  #4
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Quote:
Originally Posted by victoryv View Post
You need to have fine mesh at the boundaries to resolve the boundary layers. Change the residuals accuracy to 10^-6 and monitor other parameters as well along with the residuals.
Thanks. I'll try this tomorrow and see how it goes.

Quote:
Originally Posted by Far View Post
Make it pressure outlet
Isn't this only used when the outlet pressure is a known value? That isn't my case, actually; I'm just trying to find the pressure drop between the inlet an outlet. Please do clarify this for me if you don't mind?
flippy is offline   Reply With Quote

Old   February 15, 2013, 00:24
Default
  #5
Far
Super Moderator
 
Far's Avatar
 
Sijal Ahmed Memon (turboenginner@gmail.com)
Join Date: Mar 2009
Location: Islamabad Pakistan
Posts: 3,914
Blog Entries: 6
Rep Power: 38
Far will become famous soon enoughFar will become famous soon enough
Send a message via Skype™ to Far
hmmmmm

I am talking about static pressure at outlet
Far is offline   Reply With Quote

Old   February 15, 2013, 08:54
Default
  #6
New Member
 
Join Date: Nov 2011
Posts: 27
Rep Power: 5
Jabba is on a distinguished road
Hello,

I believe that you are simulating an incompressible flow, so I will proceed from that hypothesis.. you should verify if your mesh has an adequate resolution as someone said before

regarding the numerical schemes that you applied, they seem adequate.. which interpolation methods are you using for pressure and other variables?

if your domain is long enough so you really have a developed flow at the outlet, pressure outlet and outflow should give the same results

you need to specify a pressure reference (Pref) and a static pressure (P) at somewhere, preferable at the outlet since you are using a velocity inlet boundary condition.. in this way, the static pressure at the intlet will be resolved

since the flow is incompressible, SIMPLE and the others methods, such as PISO, will resolve the flow for a determined pressure P+Pref, so I guess it really doesn't matter which Pref you input, you will get the same pressure drop between inlet and outlet (again, reminding that the flow is incompressible)
Jabba is offline   Reply With Quote

Old   February 15, 2013, 11:42
Default
  #7
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Hello Jabba,

Yes it is incompressible flow. I guess my mesh should be at a reasonable resolution, because I set it as close to the limit as possible (the version I'm using is academic version so it has a limit of 512000 cells, mine is pretty close to that).

Regarding the P-ref I've tried:
- setting a value at the inlet
- not touching it
Both give approx. same result which is about 30% off from Poiseuille's equation result.

I've tried doing the same model, but in 2D, and the result is pretty close to Poiseuille's, so I'm guessing there's not much else I can do is there?
flippy is offline   Reply With Quote

Old   February 15, 2013, 11:51
Default
  #8
Senior Member
 
flotus1's Avatar
 
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19
flotus1 will become famous soon enoughflotus1 will become famous soon enough
Poiseuille flow can be simulated up to the limits of computational accuracy of the system.
So the 30% deviation you still have clearly indicate that there is something wrong with your simulation setup.
And even in 3D, you dont need 512k cells to achieve accurate solutions.
flotus1 is offline   Reply With Quote

Old   February 15, 2013, 12:02
Default
  #9
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Quote:
Originally Posted by Jabba View Post
which interpolation methods are you using for pressure and other variables?
I leave them as they are. Pressure was Standard I believe, can't seem to remember what the others are, though.

Quote:
Originally Posted by flotus1 View Post
Poiseuille flow can be simulated up to the limits of computational accuracy of the system.
So the 30% deviation you still have clearly indicate that there is something wrong with your simulation setup.
And even in 3D, you dont need 512k cells to achieve accurate solutions.
That's what I fear...

Well I can't be the only one with these problems can I, so any suggestions to what I should look at for mistakes?
flippy is offline   Reply With Quote

Old   February 15, 2013, 12:20
Default
  #10
Senior Member
 
flotus1's Avatar
 
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19
flotus1 will become famous soon enoughflotus1 will become famous soon enough
Although the solution is quite straightforward, there are of course many possibilities to make bad decisions for a CFD beginner and even some traps for experienced users (see Cylindrical Pipe)

Lets go through the setup:
  1. Check your mesh! Especially the actual size of the domain.
  2. Use laminar viscous model
  3. Check the viscosity of your fluid (dynamic viscosity!)
  4. Check if your domain uses the correct fluid
  5. for the easiest setup, use pressure inlet/pressure outlet boundary conditions. Check that the pressure difference is actually small enough to ensure laminar flow.
  6. Use second order upwind for the convective fluxes
  7. under monitors, untick the "check convergence" boxes for all equations
  8. Initialize with zero velocity or with the expected bulk velocity
  9. Run as many iterations until the residuals level out
  10. If still not satisfied with the solution, use a better mesh
flotus1 is offline   Reply With Quote

Old   February 18, 2013, 10:06
Default
  #11
New Member
 
Join Date: Nov 2011
Posts: 27
Rep Power: 5
Jabba is on a distinguished road
also try to use PRESTO! for pressure interpolation

regards
Jabba is offline   Reply With Quote

Old   February 18, 2013, 11:50
Default
  #12
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
So today I gave the simulation another try and I'm still not getting anywhere near the theoretical values.

Here's my case if anyone wanna try and see if they can get the desired results:
- Pipe dia: 1.6cm
- Pipe length: 50cm
- Flow speed: 0.6 m/s
- Steel pipe, fluid = water
- Models: Viscous - Laminar
- Scheme: SIMPLE
- Gradient: Least Squares Cell Based
- Pressure: Standard
- Momentum: 2nd order upwind
- Monitoring the area-weighted values of pressure at inlet and outlet
- Residual 10^-5
flippy is offline   Reply With Quote

Old   February 18, 2013, 18:49
Default
  #13
New Member
 
Join Date: Nov 2011
Posts: 27
Rep Power: 5
Jabba is on a distinguished road
for these conditions, isn't the flow turbulent? Re ~ 9500?
Jabba is offline   Reply With Quote

Old   February 18, 2013, 20:33
Default
  #14
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Quote:
Originally Posted by Jabba View Post
for these conditions, isn't the flow turbulent? Re ~ 9500?
Oh dear the dimensions were wrong, they were supposed to be mm, not cm, sorry!
Right I'll run the simulations again tomorrow with correct dimensions!

Just to clear things out: Only the one I ran today was with wrong dimensions, the ones before that I did with correct dimensions for laminar flow.
flippy is offline   Reply With Quote

Old   February 19, 2013, 04:19
Default
  #15
Senior Member
 
flotus1's Avatar
 
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19
flotus1 will become famous soon enoughflotus1 will become famous soon enough
Quote:
Originally Posted by flotus1 View Post

Lets go through the setup:
  1. Check your mesh! Especially the actual size of the domain.
  2. Use laminar viscous model
  3. Check the viscosity of your fluid (dynamic viscosity!)
  4. Check if your domain uses the correct fluid
  5. for the easiest setup, use pressure inlet/pressure outlet boundary conditions. Check that the pressure difference is actually small enough to ensure laminar flow.
  6. Use second order upwind for the convective fluxes
  7. under monitors, untick the "check convergence" boxes for all equations
  8. Initialize with zero velocity or with the expected bulk velocity
  9. Run as many iterations until the residuals level out
  10. If still not satisfied with the solution, use a better mesh
Why is nobody listening to me...
flotus1 is offline   Reply With Quote

Old   February 19, 2013, 05:28
Default
  #16
Far
Super Moderator
 
Far's Avatar
 
Sijal Ahmed Memon (turboenginner@gmail.com)
Join Date: Mar 2009
Location: Islamabad Pakistan
Posts: 3,914
Blog Entries: 6
Rep Power: 38
Far will become famous soon enoughFar will become famous soon enough
Send a message via Skype™ to Far
What results you are expecting? What is the viscosity of Fluid?
Far is offline   Reply With Quote

Old   February 19, 2013, 09:01
Default
  #17
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
@Far:
the fluid is water - viscosity 0.001003 kg/m-s = copied from fluent itself.

So I've run the simulation again today (with correct dimensions!!)
- Re is about 960 so it is laminar
- Monitoring: Vertex average of static pressure at outlet - gives a value of about 600 Pa, while the pressure drop from Poiseuille's is about 375 Pa so clearly there's still something wrong, or I'm monitoring the wrong thing (which shouldn't be the case because that's what I did with the 2D model and I was able to get the result as close as 5% to the Poiseuille's value)
- I tried with air as the fluid and the results I get is still slightly different from Poiseuille's (CFD value: about 7 Pa, Poiseuille's: about 6.7 Pa)

One thing I noticed is the volumetric flow rate at the inlet. According to my calculations it should be 1.206e-6 m3/s, while the value used in FLUENT was 1.188e-6 m3/s (monitoring volumetric flow rate at the inlet as well), could this be the reason?
flippy is offline   Reply With Quote

Old   February 19, 2013, 10:50
Default
  #18
Far
Super Moderator
 
Far's Avatar
 
Sijal Ahmed Memon (turboenginner@gmail.com)
Join Date: Mar 2009
Location: Islamabad Pakistan
Posts: 3,914
Blog Entries: 6
Rep Power: 38
Far will become famous soon enoughFar will become famous soon enough
Send a message via Skype™ to Far
I am getting same value of volume flow rate i.e. from Fluent and Analytical. what is operating pressure?
Far is offline   Reply With Quote

Old   February 19, 2013, 10:53
Default
  #19
New Member
 
James
Join Date: Feb 2013
Posts: 10
Rep Power: 4
flippy is on a distinguished road
Quote:
Originally Posted by Far View Post
I am getting same value of volume flow rate i.e. from Fluent and Analytical. what is operating pressure?
~10.1^5 Pa
flippy is offline   Reply With Quote

Old   February 19, 2013, 10:56
Default
  #20
Far
Super Moderator
 
Far's Avatar
 
Sijal Ahmed Memon (turboenginner@gmail.com)
Join Date: Mar 2009
Location: Islamabad Pakistan
Posts: 3,914
Blog Entries: 6
Rep Power: 38
Far will become famous soon enoughFar will become famous soon enough
Send a message via Skype™ to Far
Quote:
Originally Posted by flippy View Post
~10.1^5 Pa
it is 1.01 ^ 5 or 10.1 ^5 pa?
Far is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Blockage in pipe ( using solidwork flow simulation) jchow FloEFD, FloWorks & FloTHERM 1 January 16, 2012 17:03
Periodic Pipe Flow LES dvolkind CFX 5 November 24, 2011 07:04
About Turbulence Intensity (Pipe flow assimilated) gRomK13 Main CFD Forum 1 July 10, 2009 03:11
FDTD Simulation of flow through tapered pipe Jim Main CFD Forum 3 December 25, 2006 11:56
Turbulence in a pipe flow JM Main CFD Forum 4 December 21, 2006 05:04


All times are GMT -4. The time now is 01:54.