i am running an external air flow with conjugate heat transfer thru my 3-component model. From what I learned thru Trimech distributor's tech support, SW2016 Flow Sim only allows for 3d solid, 1/4 symmetry modeling of my problem. This limited capability yields 100k+ cell count models & long run times on my desktop 2-processor, 12-core computer with auto-time stepping enabled (on the order of 2e-07s).
You could model a small slice, then set an ideal wall boundary condition (i.e. tangential flow on the walls).
Flow simulation doesn't like much models with faces that are not parallel to global cordinates system, therefore there will be some mesh refinement on angled faces where you slided the model.
Alex
100K cell count is not your problem. The default time step is very small and is the principal contributor to the long processing time that you are experiencing. Try increasing the time step to 1-60 seconds depending upon the realities of your actual problem.
Thanks for response, David. As you said, I had used the default "auto" time-step feature in FW2016 and refined the meshes/cells in fluid, solid, and fluid/solid regions to "level 3". I will try manual input of a relatively large time step, as you recommended, to see if cpu time decreases. Does overriding the default time step affect accuracy of the solution and somehow disregard the cell refinements i initially made?
thanks again,
john
John, I don't think that lengthenning the time step will affect the accuracy of the flow analysis. But with an external analysis you may need to increase your mesh in the area of interest. I usually never get it right the first time until I get a reasonable analysis that doesn't take a week of CPU time. Just start with a coarse mesh and a long time step, and then tighten up the mesh and decrease the time step as you see appropriate to see the impact on your result.
David: when I tried to re-run my flow sim with a manual time step input of 1s (replaced the "auto" method in the Solving tab of the Calculation Control Options), the solver reduced it to the very small value...see attached image. Thoughts?
thanks,
john
John, Just to review with you:
In Calculation Control Options:
Second Line down "Time Step" tab right and select "manual"
Further scroll right and select time (seconds)
Sounds like what you were doing but the program reset to automatic time step. Could be an issue with your selection of heat conduction in solids or conjugate heat transfer. I vaguely recall having this issue as well, but do not recall the issue.
You could model a small slice, then set an ideal wall boundary condition (i.e. tangential flow on the walls).
Flow simulation doesn't like much models with faces that are not parallel to global cordinates system, therefore there will be some mesh refinement on angled faces where you slided the model.
Alex