This content has been marked as final. Show 4 replies
The goals are not used for convergence in a transient analysis but I wouldn't call them useless. You can manually control the time step. Just look under convergence control options. Without knowing more of the specifics various options are available such as flow freezing but that would be dependent on whether the convection cells in the "water" cavities is considered important as opposed tomodeling hte water as a "solid" only considering conduction. You can always develop a model for the water so that you could model it as a solid (as far as heating it up goes. Natural convection and high turbulence are typically not compatible with flow freezing.
Flow freezing is there to aid problems where the convective processes have short times to stabalize and the diffsuive process take a lot longer. When flow freezing is enabled the advection terms are held constant during the freezing interations.
With manual time stepping you would need to have fairly short time steps in the beginning so that the flow develops properly and then you could increase them once that occurs. However if you have natural convection in the water cells and it is deemed important then short cuts might be difficult to come by.
In my case, the convection cells within water are not considered, therefore I used a solid "water" instead.
The "flow freezing" option is also a nice feature that works well.
I could compute my case with a manual time step of 1s from 0 to 1000s, with the warning that the automatic defined one would be something around 0.0001 s. What is the risk of taking such a larger step in my case?
temp.JPG 115.2 KB
run the flow freezing using periodic and vary the travel point at which starts till the answers converge.....25, travels, .5 travel, .75 travel, 1 travel etc. That is the confidence building approach for this one.
Ok, thanks for your help !
but because this is an explicit time integration, I thought that -like in some other softwares- I could specify a small time step for the beginning and then floworks would "auto-step" automatically and increase the time step to speed up the computation... (I don't know if my explanation is clear... )
Because specifying a too large time step would introduce an error...and next step will even exagerate this divergence...