How do I stop a meshing issue?

I'm too new to this to be able to properly explain the problem, but the basic scenario is that I'm trying to model wind flow around/through an open greenhouse (really more of a very large gazebo with plants inside, if I'm being honest). I made the model in inventor because it was the modelling software I knew how to use, imported it into workbench as a step file, build an enclosure around it as pictured (the huge size along the sides is because I'm trying to model variable wind flow direction, and I calculated that based on the angle, that side distance would mean that wind flow could travel in a straight line from the corner of the enclosure to several meters past the corner of the greenhouse).

But whenever I try to mesh the thing it gets stuck halfway for ages and then tells me meshing was unsuccessful. I've rebuilt this model from the ground up in multiple ways, but it keeps happening, so I know that's I'm making some basic amateur mistake, probably in the meshing process. Any advice on what might be causing the meshing issue (and also how to do this better; I'm having trouble finding good tutorials for this particular goal) would be greatly appreciated. Please send questions if you need more information.


  • AniketAniket Posts: 1,411Ansys Employee
    1. what error do you get?
    2. Where are you meshing this model?
    3. From the looks of it, it looks like the outer covering of the greenhouse is a surface Have you subtracted the inner volume of the greenhouse from enclosure?


    How to access Ansys help links

    Guidelines for Posting on Ansys Learning Forum

  • Larz_JohnstonLarz_Johnston Posts: 11Member
    1. "Update failed for the Mesh component in Fluid Flow (Fluent). Error updating cell Mesh in system Fluid Flow (Fluent)." and "(DP 0) The mesh file exporter does not support overlapping geometry in named selections. Please resolve the issue and try again. You can use the option “Show Geometry in Overlapping Named Selections” (when you right click on the Mesh branch of the tree) to help diagnose the problem." I'm working on the second one, and it says the overlapping geometry is the outer covering. I've also gotten "(DP 0) The mesh file exporter could not resolve cyclic dependencies in overlapping contact regions. Please try to Repair Overlapping Contact Regions" before, which I think is the same, but am not sure.
    2. I've tried to mesh it in both the basic Ansys Meshing tool and the Fluent with Fluent Meshing.
    3. There's not an inner volume per see; this greenhouse is distinctly unusual to model due to not having walls: as I mentioned, it's basically a gazebo with plants inside. Unfortunately, it couldn't be simple to model like those greenhouses with a single inlet/outlet setup, and wind can enter or exit the greenhouse from all sides. as for the roof being a surface, that is correct: I have another version of the model where it is a 0.2 mm sheet if that would work better, or I could change the base model to have the surface o.2 mm away from the structure instead of constrained to being directly on it.
  • Larz_JohnstonLarz_Johnston Posts: 11Member
    edited November 24

    slight update for more information found: it works just fine when I don't have it attached to anything like this:

    It'll update quickly and not give me any trouble. but when I attach it to a Fluent tool (either of them) like this:

    It takes longer to update and gives me an error message. I can change it when it isn't attached however I like, and I changed the settings on the mesh for CFD, but that did not fix it

  • RobRob UKPosts: 11,770Forum Coordinator

    It will work if not connected to Fluent. Look in Meshing and resolve the overlapping NS, I suspect it'll work then.

  • Larz_JohnstonLarz_Johnston Posts: 11Member
    edited November 25

    I've tried that, and assuming that fixing that issue will finally let me do my final year project the issue is that the supposedly overlapping geometry isn't overlapping. Or at least it shouldn't be. my first draft for this model had a thin sheet (0.2 mm) which I built to run exactly against the top layer of the metal struts, but when that failed I can to a similar conclusion and somewhat rebuilt my model. As a result I then had a sheet of the same thickness displaced for a small distance from the support structure. That also failed, so I tried to replace the thin sheet with a surface (which created its own issues, so I abandoned it). Just today I made a new one with a displacement of five millimeters (so a 5mm gap between it and the support structures) so I'm sure that there is no overlap. Now meshing is failing to either generate or update even when Fluent is not attached.

    On the bright side, no more error message telling me of overlapping geometry, so that's nice, but I'm technically further away from a complete model now, which somewhat sours my better feelings about getting rid of an error message.

    EDIT: New error message in the mesh editor:

    Just getting the basic "Update failed for the Mesh component in Mesh. Error updating cell Mesh in system Mesh. View the messages in the Meshing editor for more details" in workbench

  • RobRob UKPosts: 11,770Forum Coordinator

    To confirm, you have a solid body 0.2mm thick? How big are the struts? What's the overall domain scale? Remember if you have a body 0.2mm thick you must has one cell across the gap, so the overall cell count may be rather high.

  • Larz_JohnstonLarz_Johnston Posts: 11Member

    ah, now that information contains the sort of basic thing I think I may have been missing (as I said in the original post, I'm very new to this).

    Yes, the sheet is 0.2 mm thick. Meanwhile, the struts are 120mm filled-in square posts and repeat down a length of more than 50 meters (they were originally pipes, but filled in squares seemed faster to model based on some tutorials. but the size of the domain is probably the big one: because of how I'm approaching this (with variable wind direction) I've made the enclosure very big, so that a straight line from the corner on the steepest angle will still completely miss the greenhouse. That size may be affecting the model itself. It's also possible that it's a problem with the import method: I'm building the model (sans enclosure) in inventor, with the cover constrained to exactly match the arc of the struts, then exporting as a STEP file and importing. I've noticed that the arc of both parts is much more jagged than it should be (more like a polygon than an arc). when it is too close it overlaps and causes issues, or creates an area too small for simulation resolution if it's moved just enough to stop intersection.

    So if at least one of those is the issue, what advice would you give for solving each problem:

    • how can I have a variable air flow direction with a smaller enclosure while still fully affecting the greenhouse with wind?
    • How can I import the model in a way that preserves the shape and does not create these intersections?

    It probably doesn't help with figuring this out, but I just experimented with brute-forcing a working simulation (not a good one, just one that might solve the issue) by making the covering much thinner and giving the curved section a relatively large separation from the curve of the arc. Unfortunately, it still seems to fail.

  • RobRob UKPosts: 11,770Forum Coordinator

    OK. This makes more sense. In CFD we model the fluid side, and on occasion the solid parts if they heat up. To do this we must resolve the features. With the scale you're working at it's very likely that CAD tolerances are causing the two side of the "roof" to overlap: the faceting isn't refined sufficiently.

    Taking a big step back. What do you want to learn from the model? Assuming it's just flow, with some heat transfer we can throw away the roof solid: it's a thin surface. We can also lose the posts as they don't really do much to the flow. Have a look at the coupled wall option in Fluent (Boundary conditions chapter, under wall I think).

  • Larz_JohnstonLarz_Johnston Posts: 11Member

    Model goals then? alright. The goal of the model is to model temperature inside a greenhouse, and in this case the input data is the wind speed/direction, and the solar loading. those rectangular rows in the middle represent the plants, and will be modeled as porous medium with heat flux along the dies of the walls that cools the greenhouse based on the amount of water the greenhouse uses. I'm using this method instead of modelling humidity to try and simplify the model, though I may have to change to using humidity anyway if the results are inaccurate enough.

  • RobRob UKPosts: 11,770Forum Coordinator

    OK. So the "roof" can be a wall and we don't need a solid. Greenhouse sides can be open (interior surface) or porous jump if the supports effect the flow. Not sure whether you'll need humidity or not, and you may want to consider solar radiation too.

  • Larz_JohnstonLarz_Johnston Posts: 11Member

    I figure I'll use solar ray tracing for the solar aspect, yes, and you are probably right that the supports won't massively affect the flow. if results from simulation vary from experimental results by too much I will probably only need to model the section from the roof to the ground anyway. the most important factor realistically is the plants: several spaces of a porous medium that cools the interior thanks to evaporation, the rest is only as necessary as needed to improve model accuracy.

    What steps and changes are you proposing, though? I think what you're saying is that the external enclosure is not necessary and that I can model just the interior of the greenhouse, but I'm not sure how to make my list of times and wind directions/speeds work with all the angles and enter each side properly.

  • RobRob UKPosts: 11,770Forum Coordinator

    No, if the sides are open you still need "outside". However you don't need a thickness of material - the Fluent solver allows "thin" walls to have a thickness. We're not permitted to offer much "how to" advice so you need to figure out what to do and we can advise from that. I do know that the Technion and Ben Gurion Universities in Israel have published on green houses and wind through shrubs so it'll be worth doing a literature review to see what they did.

Sign In or Register to comment.