3 Replies Latest reply on Mar 27, 2017 7:57 PM by Bill McEachern

    Mesh Sanity Check?

    Scott Lelievre

      Howdy folks -

       

      I've been off the Simulation horse for a few years and am getting back on with a very simple analysis. We are using FLOW to look at heat flow through a very cold chunk of ice that, for the purposes of the model domain, I've modeled as a 1/4 cylinder with symmetry conditions on the XY and ZY faces. We're interested in the average heat flux at the inner bore hole over time, and given our initial conditions we get an initial large spike that quickly tapers off to a long tail.

       

      Given that heat flux through the inner cylindrical surface is our goal, I've called for a locally tight mesh on that surface, and two more local meshes as the heat moves out radially into the bulk of the model.

       

      The mesh seems to look good on the flat surfaces but I get some high aspect ratio elements on the inner surface. I'm just not convinced that this is a good mesh for this surface and I'm curious what I'm doing wrong. Is there a better way to mesh curved surfaces like this? I've played with most of the mesh settings with little luck. I thought the Curvature Criterion under Advanced Refinement might get me there but not so far. Hopefully you can view the larger image and see what I'm talking about.

       

      JCS Mesh.JPG

       

      Any general guidance would be appreciated.

       

      Justin Smith

      Contract ME

      Stone Aerospace

      Del Valle, TX

        • Re: Mesh Sanity Check?
          Bill McEachern

          nothing there looks alarming to me. You could up the cell count in the axial direction by specifying a larger number in the basic mesh layout. To really see what the mesh looks like turn off the CAD geometry. Going higher on the curvature refinement can get you a smoother mesh on the surface. If you check the CAD area of the surface and compare it to the Flow Sim area (reported in surface parameters) the difference is indicative of the mesh refinement..

            • Re: Mesh Sanity Check?
              Scott Lelievre

              Thanks for the reply Bill. I believe I tried cranking up the basic mesh but the model domain is fairly large in an attempt to not provide any undue forcing on the solution, and the element count gets really high if I do that. Can you explain what you mean by increasing the curvature refinement - do you mean the Curvature Criterion thing in Advanced Refinement?  Interesting little trick to check the CAD area versus the Flow Sim area...

                • Re: Mesh Sanity Check?
                  Bill McEachern

                  it looks like a conduction problem so its linear so  mesh density is not something I would be too concerned about. You may have material non linearities but for normal materials they tend to not change too abruptly unless of course you are trying to model severe discontinuities.