6 Replies Latest reply on Jul 2, 2014 3:20 PM by Jared Conway

    SW’s automatic meshing vs. ANSYS ICEM

    Andrew Castelino



      I was wondering why some other larger simulation suites doesn’t adopt the almost automatic meshing procedure of Solidworks Flow Simulation?


      For example the top part of the attached image is from http://en.wikibooks.org/wiki/ICEM_CFD and is typical of ANSYS ICEM, which I have seen lots of video tutorials and found really time-taking; in addition it is not automatic and relies on the user’s input for the initial mesh grid sizes.



      Whereas the bottom image was from the blog post of CADVision (only partial of the image, please follow link to the original page) in which we see that after refining, the original rectilinear mesh grid acquires finer resolution near the body interfaces.


      There are some comparisons:


      I have heard many around saying that creating mesh takes about 50% of their simulation project time. With SW if we don’t check “manual gap size” then the size is determined automatically which makes the task like a breeze.  This is obviously very convenient and appeals to users both at amateur and professional level: but simply why ANSYS, Gambit and other software doesn’t take it? I tend to believe that they they do have made thorough and comprehensive considerations, and what is their rationale to not adopting this approach, after decades of developing CFD products?


      Q1: ANSYS ICEM allows Tetra, hexa, prism and many sophisticated controls of the mesh, and its solver is also apparently math-intensive. It is quietly unlikely that the lack of automatic meshing (or perhaps I am incorrect with on this point since ANSYS also have Meshing program tightly integrated with the Workbench, but allows very limited operation, far less than that of ICEM; I don’t know if it is provided for automatic meshing) is due to their incapability in developing that. Then if they actually could, why they don’t implement that in like ANSYS?


      The automatic (when “manual” deselected in Initial Gap in SW) meshing actually also has the additional advantage in preventing wrong user input in the most cases. If a project of sophisticated shape variation is all manually meshed by the user, there are quite some chance that at certain areas the meshing would be either inadequate or having other problems; with automatically meshing such cases are unlikely.


      Q2: Also, the parametric study of SW allows changing geometry dimensions across simulations, and for each different dimension the new mesh is generated automatically. Can ANSYS achieve the same if it requires manual meshing, or it actually has a workaround to generate new meshes during parametric study?


      Q3: In future releases is there any planned expansion of SW meshing controls, for example, including Tetra, Hexa instead of the current all-rectilinear design?




        • Re: SW’s automatic meshing vs. ANSYS ICEM
          Ellemieke Henquet

          I'm very curious about the answer, because this is a very big disadventage!

          • Re: SW’s automatic meshing vs. ANSYS ICEM
            Jared Conway

            answering in reverse


            q3, unlikely, the method is accurate, it is mostly efficient, rarely runs into meshing issues, and mostly makes sense to most users. they have put a lot of r&d into the method and expect them to continue this.


            q2, it is very likely that other tools have to remesh like flow simulation. however some may be able to adjust the boundaries and make the mesh change around the change in the geometry. something that could certainly be investigated with the mesher in flow. that being said, the meshing part is usually not that bad in flow, so its kind of moot point.


            q1. i think this goes down to where you spend your development money and time. flow simulation solves a lot of different problems for a lot of different people. but it doesn't solve everything. that being said, its also reasonable for most people to pick up and use without needing to be phd's in CFD. other tools cater to different types of users which mean that they want more controls over their mesh for example for the 1% improvement in accuracy and they are ok with the fact that it takes them time. and instead of making the mesher different or more effective or the interface better, they focus their efforts in other areas like focused analysis types or physics...etc. in the end, most people have taken an approach and stuck with it. in the end, go with what works for you, meets your budget and gives you accurate results.



            what is the scope of the research that you're doing on these differnent tools and the comparison with solidwroks flow simulation?