8 Replies Latest reply on Sep 18, 2012 2:18 PM by Chris Michalski

    Defining a non-zero starting velocity for natural-convection models

    David Maxham

      When running a natural convection model, especially with many cells, the solver takes a long time for the early iterations.  Once the iteration number reaches say 30 or so, the solver proceeds much faster.  I think that if I were to set a non-zero starting velocity for the fluid cells the solver would iterate faster.  The ability to set a starting velocity is default in Flotherm, where the starting velocity for natural convection models is 0.2 m/s.


      Can a starting velocity be set with Flow Simulation?  If so, how?  Please note - I'm not asking about setting an initial velocity, which is where there is a prescribed flow field, but how to set a starting velocity so that the solver can iterate faster for natural convection models.



        • Re: Defining a non-zero starting velocity for natural-convection models
          Chris Michalski

          David -


          when you first define the simulation model, there is the initial conditions on the last tab of the general settings.  If you insert a vertical initial velocity here it will set all of the Z-velocity vectors to this value to start.

          • Re: Defining a non-zero starting velocity for natural-convection models
            Bill McEachern

            Well, this is not what you really want to hear and there but you could run a simulation and solve for the 0.2 m/s and use it as an initial condition in a subsequent study. Not exactly elegent but it would work. You would use the take previous in when you invoke the solver.

              • Re: Defining a non-zero starting velocity for natural-convection models
                David Maxham

                Chris, Bill,


                Thanks for your help. I should have pointed out that this is a steady-state external flow.  I initially thought that the initial condition would be persistent as Bill mentions above, but I'm not sure based on what I found in the help (see below).  So maybe when I do set the initial vertical velocity to 0.2 m/s, it is actually just a starting or initial condition and not persistent.  I'm going to run the model both ways, first as I'm doing now with no initial velocity set, and secondly with an initial condition equal to 0.2 m/s.  I'll report back with the difference, if any, in results.


                Here's what the help section says.  They are clear that if you want to solve an internal flow faster, use the initial condtions, as Chris suggested above.  Although they don't specifically say so for the external flow, it seems to be implied.  We'll see.




                Initial and Ambient Conditions

                Specifying Initial (for an Internal    analysis) or Initial and Ambient (for an External    analysis) Conditions means specifying values of Thermodynamic parameters, Velocity parameters, Turbulence parameters, Solid parameters (to solve "Heat conduction in solids") and Concentration (for more than one fluid).

                •   If you want to solve a steady Internal problem in a shorter time, we recommend that you use Initial conditions (initial values of the flow parameters) that are closer to the assumed solution than the default initial conditions. 
                • If you solve a steady External problem, specifying Ambient conditions means specifying initial conditions within the Computational Domain and boundary conditions at the Computational Domain boundaries. The specified thermodynamic and velocity parameters are considered as parameters of the undisturbed external flow and used to define the initial flow state

                  • Re: Defining a non-zero starting velocity for natural-convection models
                    Chris Michalski

                    yes, because you have an external problem setting an "initial" condition actually creates a boundary condition with these values.  So the bottom of your bounding volume will always have an incoming flow at 0.2m/s. 


                    You would be better to use the "check geometry" and "create fluid body assembly".  Then save that part and insert it into your model and use it to define the initial velocity (then disable it).  This will eliminate the continual boundary condition that Flow creates for an external simulation.  This is how you define wind or some other artificial velocity for an external simulation.