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.
Particle trajectory length result for fluid flow
Posted 13 mars 2019, 08:31 UTC−4 Fluid & Heat, Particle Tracing for Fluid Flow, Porous Media Flow Version 5.3a 3 Replies
Please login with a confirmed email address before reporting spam
Hi all,
The result of my model doesn't sound rational when I do particle tracing for fluid flow. I have a 3D hydrological model which has two different layer. The hydraulic conducity for the first layer is 1E-4 and second (bottom) layer 1E-9. for the flow modeling, I used Darcy's law phisycs which work fine I think! I wanted to release a particles from a falt surface which located in the second layer and see the residence time and trajectory length for each particles. Therefore, I assumed Massless particles and defined a grid surface for the release. Two auxiliary dependent variables were introduced for the particle residence time and trajectory length as it has been explain in particle tracing for fluid flow user guid! I assume the unit for trajectory line is meter cause in auxiliary dependent variable for integration along trajectory the equation is d(rp)/ds=R in which the unit of R is 1/m (I know that I can change the unit of dependent variable to m as well, but its 1 at the moment).
The problem is that the distance between the release pont and the end point is around 9 meter (the model is 10 X 10 X 10 cubic meter); but the result of trajectory length is in rder of 1E4 (i.e., 10000 meter)!!!!
So, there should be a silly mistack in my model which I can't find it! Attcahed is my COMSOL model which I have cleared the mesh and solution to decrease the size . Any suggestion would be really appreciated!
Bests, /Brian
Attachments:
Please login with a confirmed email address before reporting spam
Hi Brian, this seems like a rare case where we use machine epsilon ~2.2E-16 in the definition of the norm of an expression: fpt.Veps=sqrt(u^2+v^2+w^2+eps) and the addition of the eps can no longer be considered small. In fact, in your model, the eps contributes to almost all the trajectory length.
One workaround is to set the source in the equation for rl to: fpt.V/fpt.Veps instead of 1. This means that the correct scaling factor will be used integrating along the trajectory length.
Dan
Please login with a confirmed email address before reporting spam
Thanks for the very quick response Dan. And many thanks for your suggestion. The scaling factor works well for the residence trajectory length. However, using this scaling factor for the residence time, does not change the results of residence time! I think the the residnce time result is still wrong (even when I use this scaling factor). Shall I use a different scaling factor for the residence time? Or what else?
Again thanks for your time and consideration.
Bests, /Brian
Please login with a confirmed email address before reporting spam
Hi Brian, my comments only apply to the case when you are integrating along the particle trajectory. For the residence time, you are integrating with repect to time, so there is no velocity factor included in the equation to contribute any error. Therefore, in that case, you can leave the source as 1.
To me, the residence time values all look correct.
Dan