5 Replies Latest reply on Oct 26, 2014 11:24 PM by Jared Conway

    using porous media to represent floor grating

    Dan Hofstetter

      I'm performed some experiments in a ventilated room recently, and I'm trying to validate my simulation using measured gas decay data.  Below is an image of my model:Room.jpg

      The Pit Fan supplies ventilation air to evacuate pit gases, and he barn outlet fans exhaust the room air.  There is also a perforated ceiling (not shown) with 1/2" diameter holes on a 6" x 6" grid that I represented using a perforated plate and environmental pressure.  The "totally-slotted floor" shown is actually made of 18" x 24" sections of grating with ~ 1/2" x 3" slots and 42% open area.  The grating is less than 1/2" thick.  To simplify my model and see if the trends come close to matching observations, I replaced the grating with long rectangular slots that have the same % open area.  Initially, there is 100% air above the slotted floor, and 150 ppm of contaminant gas below, separated by plastic sheeting to control the initial conditions.  At first glance, a manually-refined mesh produces reasonably correct results when I compare the point parameters from the simulation to the measured data at the same locations 12" below and 6" above the slotted floor.  However, there are a few places where the simulation greatly over or underestimates the gas concentration over time.  To try to improve the simulation results, it was suggested that I try to use the actual grating geometry, shown below:

      Grating.jpg

      The floor area covered by this grating is 9 ft wide by 36 ft long, and the grating is 1/2" thick in my model, so placing the real geometry in the model would result in a very large, fine mesh in the region of the flooring.  I set up a simulation using one section of grating in a wind tunnel with ideal walls and ran a parametric study to get the pressure drop across the faces of the grating, then defined a one-direction porous media to represent this flooring.  I then ran another simulation replacing the grating with a solid body with porous media, and got reasonably close results when comparing the pressure drop across the media.  Thinking the porous media would greatly reduce the mesh density in my larger model, I replaced the simplified flooring with a solid body that was 1/2" thick, and inserted porous media for that body.  When I meshed the model the first time, I noticed the number of mesh cells increased dramatically, probably due to the number of small cells needed to mesh that 1/2" thick body.  I wound up inserting a local initial mesh for the porous media body to make sure the mesh in that region appeared reasonable.

       

      There isn't much guidance about actually using porous media in the help documentation or in the Flow tutorials.  Do I need to do anything special with regard to meshing the model when using porous media like this?  Really, a perforated plate would be much better for this model, but it isn't possible to use perf. plates except at the computational domain boundaries.  Any help is greatly appreciated!

       

      Thanks,

       

      Dan

        • Re: using porous media to represent floor grating
          Jared Conway

          i'm not sure i understand the problem definition

           

          you built a porous medium based on the real geometry because meshing the real geometry would be a challenge > this is a good fit for the porous medium

          you checked the porous vs real geometry > they match > this is good

           

          now it looks like you have 2 issues, one about the way the mesh is handled in and around the porous medium

          and the results when put into a larger solution?

            • Re: using porous media to represent floor grating
              Dan Hofstetter

              Jared,

               

              You are correct - I am looking for some guidance about actually using porous media in this application.  I've seen a few nice tutorials and blog posts about using porous media, but none of them go into the fine details about actually defining the medium in the engineering database or how to handle meshing.

               

              I started a simulation last night where I set the local mesh level inside the porous media body to level 4, which appears to result in nearly the same thickness cell as the floor grating.  But I was wondering if anyone had any other advice or rules of thumb regarding the number of mesh cells required across a porous medium.  i.e. if the medium is 6" thick, can the mesh cells inside the medium be the full 6" thick and give correct results?  Or should I have more cells across the thickness in the direction of flow?  Would a 12" thick mesh cell result in the resistance to flow being applied to the entire mesh cell?  I'm just trying to learn more about how porous media gets handled and what I need to do to use it correctly.

               

              Regarding the results when put into a larger solution:  I logged gas concentration data every second at sixteen points inside the room while running ventilation fans, so I am trying to validate the CFD model versus the measured data.  I'm looking at the porous media as one place where I might be able to improve the simulated results.

               

              Thanks,

               

              Dan

                • Re: using porous media to represent floor grating
                  Jared Conway

                  there are no hard set rules. just like the rest of CFD you should iterate on cell size until the solution stops changing. i wouldn't put only 1 thickness of cells in the porous medium but theoretically it shouldn't make a difference since it isn't modeling the channels or anything along it.

                   

                  my guess is that if you overall solution isn't working the way you expect it is probably more the assumptions made by porous medium that are coming into play than mesh. for example you've chosen one directional porous medium but that isn't exactly how it will behave. the only way to really check this is to run with the grate modeled.

                    • Re: using porous media to represent floor grating
                      Dan Hofstetter

                      I think you're right.  I changed the model around and made those long, narrow openings in the simplified grating 1/2" wide with 1" linear pattern spacing.  Since the slots are orthogonal to the mesh, it really isn't too bad in terms of overall number of cells.  While thinking about this yesterday, I was really wishing there was a way in Flow Simulation to create geometry from the mesh - the solid body representing my floor grating is not perfectly aligned with the mesh, which results in lots of partial cells.  Since the offset distance is so small, I think moving the solid body so that it was perfectly aligned with the mesh wouldn't affect the solution very much, but it would make the mesh much more efficient.  Maybe a tool to turn off a certain percentage of mesh cells in a defined region between two control planes, similar to linear patterns.  It would make modeling grating very simple.

                       

                      Anyway, I'll re-run the simulation with the revised grating and see how the results look.

                       

                      Thanks!