Hi all,
I am using the functionality in the GUI related to monitoring points to observe the evolution of the solution over time at certain points. I notice that, when opening the probes_coords.csv file, the coordinates that I entered were slightly modified. My question is, do they automatically change to the position of the nearest cell center? Is there a way to configure it to maintain the entered position using some type of interpolation?
Automatic change in monitoring point
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
Yvan Fournier
- Posts: 4157
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Automatic change in monitoring point
Hello,
Yes, you can set part of this behavior in the GUI ("one probe per cell"), and can have finer grained controls in user-defined functions.
Check the Doxygen (html) documentation for cs_probe_set_t. The default behavior will also depend on whether the mesh is fixed (where we snap coordinates to cell centers) or moving/deformable. The distance between requested probe locations and matching cell centers is also summarized in run_solver.lig.
Regards,
Yvan
Yes, you can set part of this behavior in the GUI ("one probe per cell"), and can have finer grained controls in user-defined functions.
Check the Doxygen (html) documentation for cs_probe_set_t. The default behavior will also depend on whether the mesh is fixed (where we snap coordinates to cell centers) or moving/deformable. The distance between requested probe locations and matching cell centers is also summarized in run_solver.lig.
Regards,
Yvan
Re: Automatic change in monitoring point
Hi Yvan,
I’ve encountered an issue related to this topic. The mesh cells in the area are 30x50, and I’ve placed 100x100 monitoring points with "activate interpolation" and "one probe per cell" enabled. However, the instantaneous results from the monitoring points appear discontinuous, as shown in the attached velocity results.
It seems the interpolation method being used is based on a gradient-based first-order Taylor expansion, and I suspect this might be the cause of the discontinuities. Is there any method or available code to implement a higher-order interpolation technique that could help improve the continuity of the results?
Best regards,
Ximeng
I’ve encountered an issue related to this topic. The mesh cells in the area are 30x50, and I’ve placed 100x100 monitoring points with "activate interpolation" and "one probe per cell" enabled. However, the instantaneous results from the monitoring points appear discontinuous, as shown in the attached velocity results.
It seems the interpolation method being used is based on a gradient-based first-order Taylor expansion, and I suspect this might be the cause of the discontinuities. Is there any method or available code to implement a higher-order interpolation technique that could help improve the continuity of the results?
Best regards,
Ximeng
- Attachments
-
-
Yvan Fournier
- Posts: 4157
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Automatic change in monitoring point
Hello,
Yes, the interpolation is based on a 1s-order Taylor expansion. There is currently no better option. The interpolation is actually designed so that a user could pass their own interpolation function, but no such function is available.
One solution would be to use cell to vertex interpolation than vertex to point interpolation, but although this is feasible in a specific user context, it is difficult to handle in an optimized manner (i.e. avoiding extra computation or memory use) without prior knowledge of the fields to interpolate, as this can require a cell to vertex interpolation using all cells associated with the probes.
Best regards,
Yvan
Yes, the interpolation is based on a 1s-order Taylor expansion. There is currently no better option. The interpolation is actually designed so that a user could pass their own interpolation function, but no such function is available.
One solution would be to use cell to vertex interpolation than vertex to point interpolation, but although this is feasible in a specific user context, it is difficult to handle in an optimized manner (i.e. avoiding extra computation or memory use) without prior knowledge of the fields to interpolate, as this can require a cell to vertex interpolation using all cells associated with the probes.
Best regards,
Yvan