Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Non-Newtonian Fluid with Low Reynolds K-Epsilon Single Phase model

Please login with a confirmed email address before reporting spam

Hi folks,

I am trying to validate experimental data through CFD simulation but I am having a bit of hard time...

The fluid is paper pulp with a non-newtonian behavior and I am using single phase Low Reynolds K-Epsilon
to try and replicate the experimental data.

The dynamic viscosity is a function of the fluid mean velocity (see attached file).

The simulation is not converging, and I can't understand why...

I have attached the file.

Hope someone can provide some advice.

Cheers,

Rui.


2 Replies Last Post 23.11.2011, 04:20 GMT-5
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.11.2011, 02:17 GMT-5
Hi

I'm afraid my comments will not solve your issue, but I have a few suggestions:

1) to have a better view of your looong pipe, try to add a 2nd Definition - View and on the "Axis sub-tab, remove the "Preserve perspective ratio", then refresh your graphics. You can then easily switch between the long and compact views.

2) use the units to help you through, but units and fractional powers do not match well, so make the variables set to a fractional power unit-less, by i.e multiplying them by "mod1.spf.U[1/(m/s)]"

3) more important, you have defined an "epsilon = 10^-30 that is really "small" compared to 1.
The smallest binary number COMSOL can differentiate from 1 is "eps" = 2.2E-16, and COMSOL scales all values to be close to 1, at best (but sometimes you must help). This limitation is build into the binary representation of real number and is a/the numerical limit for most physics simulations.

And as most physics are represented by 2nd order functions its rather sqrt(eps) that is the true limit. So I would use a values something like epsilon = 10*eps in your case

Then something I often do: I use a laminar stationary case to set up a flow - pressure initial condition, and run the turbulent simulation from that starting point, or at least I apply a parabolic profile (with the no-slip condition), and use a Poiseille law pressure drop to have a better than all=0 starting point

--
Good luck
Ivar
Hi I'm afraid my comments will not solve your issue, but I have a few suggestions: 1) to have a better view of your looong pipe, try to add a 2nd Definition - View and on the "Axis sub-tab, remove the "Preserve perspective ratio", then refresh your graphics. You can then easily switch between the long and compact views. 2) use the units to help you through, but units and fractional powers do not match well, so make the variables set to a fractional power unit-less, by i.e multiplying them by "mod1.spf.U[1/(m/s)]" 3) more important, you have defined an "epsilon = 10^-30 that is really "small" compared to 1. The smallest binary number COMSOL can differentiate from 1 is "eps" = 2.2E-16, and COMSOL scales all values to be close to 1, at best (but sometimes you must help). This limitation is build into the binary representation of real number and is a/the numerical limit for most physics simulations. And as most physics are represented by 2nd order functions its rather sqrt(eps) that is the true limit. So I would use a values something like epsilon = 10*eps in your case Then something I often do: I use a laminar stationary case to set up a flow - pressure initial condition, and run the turbulent simulation from that starting point, or at least I apply a parabolic profile (with the no-slip condition), and use a Poiseille law pressure drop to have a better than all=0 starting point -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23.11.2011, 04:20 GMT-5
Hi,

No matter if you switch to a low Re, as I already explained to you you have both laminar and turbulent flows because of your velocity dependent viscosity:

www.comsol.it/community/forums/general/thread/23967/

Given ETA=0.0132*V^(-0.1524), Reynolds number becomes: Re=rho*d*75.76*V^1.1524, since at 2000 we have self-sustaining turbulence we can look for the critical value of velocity that causes turbulence in your flow and we get 0.39m/s, that is below this critical value you have laminar flow. There is no turbulence damping, simply laminar flow according to your model of viscosity.


On the other hand I just realized that there is another problem, are you really trying to validate experimental data by means of a simulation? It usually goes the other way round.

Cheers
Hi, No matter if you switch to a low Re, as I already explained to you you have both laminar and turbulent flows because of your velocity dependent viscosity: http://www.comsol.it/community/forums/general/thread/23967/ Given ETA=0.0132*V^(-0.1524), Reynolds number becomes: Re=rho*d*75.76*V^1.1524, since at 2000 we have self-sustaining turbulence we can look for the critical value of velocity that causes turbulence in your flow and we get 0.39m/s, that is below this critical value you have laminar flow. There is no turbulence damping, simply laminar flow according to your model of viscosity. On the other hand I just realized that there is another problem, are you really trying to validate experimental data by means of a simulation? It usually goes the other way round. Cheers

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.