9 Replies Latest reply on Oct 7, 2014 5:02 PM by Jared Conway

    The case of the disapearing point goals in Flow Simulation.

    Matthew Brown

      Has anyone ever encountered the problem where point goals invalidate themselves when the geometry is modified?  For example, I have an assembly that had a part that contained only a surface and was suppressed.  I decieded I wanted that feature so unsuppressed it and I applied a thicken command in the part to obtain a 3D solid.  However, when I go back into my Flow Simulation model all of the previously defined point goals have been invalidated.  I get a message stating something to the effect of 'cannot find vertex associagted with this point goal'.  When this happens, the only fix I have found is to delete and reapply the the point goals.  However, my current model has something like 80 some odd point goals, with several more equation goals based on these values.  It is a huge pain in the rear to delete and reassign these every time this happens and it has happened several times so far.

       

      Does anyone have a work around or can tell me what I am doing wrong to cause this?  It seems like a very annoying bug to have goals disappear if you add or modifiy geometry.

       

      -Matt

        • Re: The case of the disapearing point goals in Flow Simulation.
          Jared Conway

          version and service pack?

          what workflow was used to create the point goals?

          how are the point goals related to what you added to the comp domain?

            • Re: The case of the disapearing point goals in Flow Simulation.
              Matthew Brown

              Solidworks Premium 2014 x64 Edition, SP4.0

               

              The workflow was roughly:

                 1. Create geometry including sketches with point locations.

                 2. Use Wizard to create project.

                 3.  Setup and mesh model.  Final step in setup was to create point goals from previously defined sketches.

                 4.  Run model, no problems encountered.  Stopped analysis to add additional feature.

                 5.  Applied thicken command to sheet body and activated in component control.

                 6.  Return to Flow Simulation tab and error appears, but only highlights the first point goal in the list as being a problem.

                 7.  Remove, rebuild...  Then the next one is the issue.  And so on....

               

              The point goals are defined from sketches that exist in both the assembly file, and some of the associated part files. Spacially, they are no where near the part that was added.  It's not like these are suddenly falling inside a solid body or something like that. I have even tried removing external references, but that didn't seem to make a difference.

               

              -Matt

            • Re: The case of the disapearing point goals in Flow Simulation.
              Jerry Steiger

              Matt,

               

              If you moved this thread to the Flow Simulation sub-forum, I wouldn't be distracted by it. Plus, you might actually get more people who are really interested in it looking at it. Move should be available to you under Actions in the upper right part of the page.

               

              Jerry S.

              • Re: The case of the disapearing point goals in Flow Simulation.
                Matthew Brown

                Hi all,

                Thanks for the replys.  Unfortunately, I have had to abaondon Solidworks Flow Simulation all together.  My model's were pushing the limits of what was possible with this software and when it started taking a day to do 5 iterations for a model with only 2.5M cells, I had to make other plans (namely Star-CCM+).  I still have one model running which has been going for close to a week now.  When it finishes I will try to do a simple bug check to see if I can duplicate the issue with a simplified model and come back here to post an update.  I have nothing against FS, its just that my models are too big and complex to model effectively in this software and it's affecting my productivity.  I will still use it for simple trade studies and preliminary type analyses, I just need a more robust approach for my work.

                 

                -Matt

                • Re: The case of the disapearing point goals in Flow Simulation.
                  Matthew Brown

                  I would have to respectfully disagree about that being good. I have 32GB of ram available and every FS simulation I had run before this was of comparable size, but limited scope. These all ran out to several hundred iterations in just a few days or less. I do not have a schedule that allows for 4.5 hrs/iteration on this new project as I have 20 conditions to look at and only a few weeks to do it in.

                   

                  The analysis is not transient either. However, it has a ducted internal fan which seems to always slow things down. Then when you add in the thermal conduction in solids it seems to take even longer. Part of the reason it has taken so long is that I have to go in every couple of days and update the fan curve so I can get the mass flow rate I am trying to target. I don't have actual data on the fan as it is essentially a representation of a buried engine I am using to set mass flow through the inlet duct of an aircraft. The models are huge since my duct flow is really dependent on the external flow around the AC and not just what is happening in the duct via the fan.

                   

                  I have tried to use mass flow inlet/outlet BC's inside the duct to do this, however, I started running into odd results and unstable models. When I say odd results, I am talking near zero density and Mach numbers approaching 10^3. Despite numerous attempts to refine the mesh, move the BC's, modify geometry and so on, I kept getting obviously bad results.

                   

                  I can't speak directly to why star-ccm+ is doing a better job. I can tell you that it is much more efficient with resource usage. A 2.5M cell model takes at least 1 to 2 hours to mesh in FS. Star-CCM+ does it in just under 15min and doesn’t bog my machine down to the point I can’t do other things. I have also not been able to successfully mesh anything larger than 5M cells in FS (computer usually freezes) while in Star-CCM+ I can easily achieve 15M cells. Then there are the obvious advantages of their meshing capabilities which are light years ahead of FS, I am sad to say. Polyhedral mesh, extruded mesh, prism layers, cylinder mesh, etc... They also have options and lots of controls for their turbulence models which is my biggest complaint about FS. K-e models are great, but aren't always the best choice and FS doesn't give you any other options. I would say that another issue for me would be the absence of proper solution controls. Perhaps I have missed these but being able to control relaxation
                  factors, especially independently of each other (continuity, momentum, turbulence), is a very nice feature.  Also, being able to ramp various parameters up gradually can help control problematic models.

                   

                  Flow simulation is a nice tool, but it isn’t appropriate for every situation.  Just like Star-CCM+ isn’t the best choice all the time.  I still foresee a need, it’s just not going to be my only option or my first choice for large complicated models.

                   

                  Best Regards,

                  -Matt

                    • Re: The case of the disapearing point goals in Flow Simulation.
                      Jared Conway

                      2.5 million cells should not take 1-2hrs to mesh unless there is something really strange about your geometry. i don't know if i expect 15mins like ccm. so you might be fighting some experience issues, some computer issues and maybe some software issues. you may want to connect with your VAR to see if they can help. we do this in my group for customers all the time. we had a customer that had a 5millon dof problem that took 2 weeks to solve. we brought that down to a few hours per iteration just be reviewing the model and the way he was setting up the analysis. if you're looking for some help in that respect, don't hesitate to drop me a line at jared@hawkridgesys.com.

                       

                      as for the rest of your comments about star ccm vs flow, i think you have a lot of valid stuff there but like that you see that there is a use for each tool. i'm just surprised that they aren't "equivalent" for problems that are reasonably the same and have reasonable physics.

                    • Re: The case of the disapearing point goals in Flow Simulation.
                      Matthew Brown

                      I will say this, the finite element approach used in FS
                      makes it easy to get quick answers on simple models.  I got very good correlation to experimental
                      data on 2D cylinders and spheres.  I also
                      got great results on a fairly simple ejector manifold analysis (branching pipe
                      flow).  It has taken much more effort and
                      greater fidelity to achieve these results in Star-CCM+.