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 The information provided may be out of date.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.
NaN values for particle tracing using brownian motion
Posted 26.03.2013, 07:42 GMT-4 Charged Particle Tracing, Particle Tracing for Fluid Flow Version 4.3a 1 Reply
Please login with a confirmed email address before reporting spam
Hi, I was wondering if anyone had experienced this problem and/or knows of a fix:
I have a very simple model involving 3D steady-state laminar flow through a rectangular channel with symmetry sidewall conditions (this is a simplified version of a more complex domain). I am using the particle tracing for fluid flow module (along with the drag and brownian force) to track particles through the domain, where one of the walls capture the particles via the Freeze condition. I would like to obtain the positions where each particle "sticks" to both the outlet and the wall condition - the qx, qy, and qz positions of each particle would be sufficient.
The problem is that, when including the brownian force, the positions of the immobilized particles register as NaN. I can plot the positions of these particles just fine (3D plot group, particle trajectories), so I believe the particle positions exist somewhere within COMSOL. I get the NaN values either trying to export to a table or exporting the data set.
When disabling the brownian force, the particle positions are as normal.
Is there any way of obtaining the particle position data other than qx, qy, and qz?
The simple model is attached here.
Thanks for your help,
Nicholas
I have a very simple model involving 3D steady-state laminar flow through a rectangular channel with symmetry sidewall conditions (this is a simplified version of a more complex domain). I am using the particle tracing for fluid flow module (along with the drag and brownian force) to track particles through the domain, where one of the walls capture the particles via the Freeze condition. I would like to obtain the positions where each particle "sticks" to both the outlet and the wall condition - the qx, qy, and qz positions of each particle would be sufficient.
The problem is that, when including the brownian force, the positions of the immobilized particles register as NaN. I can plot the positions of these particles just fine (3D plot group, particle trajectories), so I believe the particle positions exist somewhere within COMSOL. I get the NaN values either trying to export to a table or exporting the data set.
When disabling the brownian force, the particle positions are as normal.
Is there any way of obtaining the particle position data other than qx, qy, and qz?
The simple model is attached here.
Thanks for your help,
Nicholas
1 Reply Last Post 26.03.2013, 12:31 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
OK, I think I have found the problem here.
The Peclet number of the simulation was appx. Pe = 100. On the inlet there were a number of particles that were diffusing upstream of the main axial flow direction (along the top and bottom surfaces of the channel) and out of the domain through the inlet. There does not seem to be any restriction on the inlet B.C. that keeps these particles in the domain, and they become essentially lost.
The problem can be corrected by extending the fluidic domain in the direction opposite to that of the main flow, where the length of that extension will be a function of the diffusivity of the particle as well as the smallest time step of the time-dependent solver.
The Peclet number of the simulation was appx. Pe = 100. On the inlet there were a number of particles that were diffusing upstream of the main axial flow direction (along the top and bottom surfaces of the channel) and out of the domain through the inlet. There does not seem to be any restriction on the inlet B.C. that keeps these particles in the domain, and they become essentially lost.
The problem can be corrected by extending the fluidic domain in the direction opposite to that of the main flow, where the length of that extension will be a function of the diffusivity of the particle as well as the smallest time step of the time-dependent solver.