Cannot update geometry via design points
-
-
Hello,
I believe I have a similar issue to this thread here, though it's not clear if the recommendations there apply to my case. /forum/discussion/16298/geometry-dont-change-by-solving-different-design-points?
I'm attempting to perform a parameter sweep, modifying the dimensions of a pipe with multiple material layers. I am doing this using driving dimensions in Spaceclaim (see attached image).
I set all of my dimensions row by row in the parameter set dialog. If I set every row to "retain" for every case, then "set current", open up the fluent mesher, manually re-import the geometry and remesh, open fluent, and recalculate, everything works fine. However, if I click "update all design points", or if I just "set current" and update from the workbench, it keeps the geometry of the first design point and I get this error message:
Parameter update incorrect: Attempted = 5.1, Assigned = 4.6, Name = tss1
This only happens with parameters related to the geometry (fluid inlet temperature does not cause this problem). For even a modest parameter sweep, this will take weeks what should be doable overnight, so this workaround is really impractical. Is this a limitation of the student license or am I doing something wrong? Is there a way to force it to re-import the geometry and remesh as part of the run?
.In the attached images, I have one variable, "tside", shown for two different cases (Figs. 1 and 2), as well as "ttop" (Fig. 3) and "tss" (Figs. 4 and 5). There is an additional set of variables that are adjusted together with tss in such a way that the geometry remains properly continuous and the red region has a constant thickness of 0.2 mm.
Each of these is associated with a pull command. I have a fluid flowing through the center (green) in the y-direction (into the screen) and a heat flux on the surface in the positive z-direction (left). In particular, I am interested in the maximum temperature of the red region.
If I take two design points, changing tside from 0 to 1 mm (comparing the cases from Figs. 1 and 2), highlight them both, right-click, and "update selected design points", I get 827.364 K in both cases (other output variables report exactly the same results as well). If I manually reload the mesh and run directly in fluent, I get 827.36 K and 817.59 K, respectively.
On further review, it seems that my issue is nearly identical to this one here: /forum/discussion/7134/doe-parametric-optimization-solution-is-different-than-fluent-stand-alone
Like both Flow and Antoinel, I am getting very different results when using the design point interface vs. opening fluent and running from there. Flow claimed the problem went away with the R12020 version, but that is what I'm using. I don't think automatic case modification can help me, as the variables that are causing problems aren't really internal to Fluent, but instead should modify the geometry that is sent to the fluent mesher.
There doesn't seem to be an obvious pattern to this - sometimes it correctly updates the geometry, sometimes it does not. This can happen with any of my geometric variables; tside and ttop do not give me the error message, I suspect because if one of the radial pull variables (like tss) fails it can result in overlapping geometry or a gap.
Here are five cases (the arrow corresponds to the boldened variable):
ttop=0, tside =0, tss=4.6 mm
ttop=0, tside=1 mm,tss-4.6 mm
ttop=0, tside=0, tss=4.6 mm
ttop=0, tside=0, tss=4.6 mm
ttop=0, tside=0, tss=5.1 mm
.
.Thank you very much. That did clear up that part of the problem, but apparently that is not the root issue. Design Modeler makes the changes without issue, but if I run this from Workbench, I can see in the fluent solution field that the changes made in Design Modeler did not propagate down to fluent.
For example, I start in a configuration with a wide inner pipe (see first image), then check the fluent solution set (second image), I get a comparable geometry. If I retain a design point with a narrow pipe that I run from the workbench using "update design point", then set that point as current, I find that the DesignModeler geometry is properly updated (see image 3), but the fluent solution geometry is still identical to image 2 (see image 4).
I also have a report from fluent that simply measures the area of that innermost face (yellow). It measures the same area (25.13 mm^2) in all cases. If I manually change a new design point to current, refresh the DM point, and open fluent meshing, I find it still has the old geometry. (Image 5)
If I then click on import geometry -> revert and edit -> update, it loads in the new geometry (Image 6) and from there everything will work as intended. (The thin interlayer between the yellow and green is still there - it's just difficult to see).
.
It seems to be skipping the step of importing the updated geometry into fluent meshing when I run from workbench. Viewing 9 reply threads- The topic ‘Cannot update geometry via design points’ is closed to new replies.
-
You are navigating away from the AIS Discovery experience
