Hi Wing Shing, the option to run concurrent simulations was removed in 2013 due to some limitations when used with new features. However it will be back in 2013 SP2.0 according to the SolidWorks technical support team.
I'm having a bit of trouble understanding what you're looking to do. If you want to add some more detail I might be able to help out. Right now your options are:
1. setup multiple studies and use the batch run to run them all in sequence. (when the option is added back in 2013 sp2.0, you'll be able to run more than one of those at a time)
2. use the parametric study mode. in this mode you setup the problem once. tell it what input parameter you want to change and what output parameters that you want and it will automate the creation of the studies you would have had to create manually in option 1 and then run them and provide you an output.
3. use the parametric optimization mode. 2012 and previous versions had this. if you had a specific goal that you were interested in finding (for example shooting for a specific pressure drop), you could have the software change the input volume between X and Y and it would find the optimal value to find your pressure drop.
Jared, you are right. My VAR just told me the same thing. Hopefully SP2.0 won't have too many bugs after this option has been re-implemented.
My company makes air diffusers, and my job is to study the air flow pattern of our products such as velocity and pressure drop. I typically have my system run 3 simultaneous CFD studies at a time just to speed up the process. Typically, I need 3 days to analyze one product in a single configuration. Without the simultaneous option it will be a hassle for me due to much longer time is required.
The parametric in 2013 no doubt is easy to use, but it also runs the test in sequential order. The only advantage I see with the parametric study is that I can save time on the meshing process.
are they separate files? fire up multiple sessions of solidworks and run them from another session. i've done that many times. i actually prefer that method over having multiple files open in one session of solidworks and using the batch run procss to run multiple studies at one time.
no additional licenses will be necessary in case you're concerned about that. at HRS we've tested both standalone and network licenses to confirm that.
even if they aren't in separate files. just do a pack and go to make 3 copies and you should be good to go. that way you don't have to rely on sp2.0 if you're comfortable with sp0.0 or sp1.0.
all this being said, from a computational perspective, it would be interesting to see how much faster you can solve 3 studies simultaneously vs sequentially. the reason being that the only part i think you're going to be saving time on is the meshing because you can mesh multiple projects at one time and spread that across multiple cores when you'd be stuck using 1 core sequentially for the others. from the solving phase, if you have less than 6 cores, breaking up the solving of 3 projects over 6 cores probably isn't as effective as solving each with 6 cores.
personally i only use the simultaneous option if i need to watch studies get started and i'm pressed for time. once i get them started and know they are ok, i usually stop and let them each use all the resources on their own.
I have a a 12 core system (24 under HT mode) I normally have 8 CPU to solve each simulation. Mesh size is around 3~4Mil typ.
I shall try your approach with different files to start the simulation project. but it could take more memory to do so.
I'd keep HT disabled so that you can get the highest possible core speed to the meshing process.
Choosing 8 per study means that you have studies with overlapping CPU requests. If you have 12, you might have more success with 4 cores each. However I haven't tested whether that will thread as we expect. Similarly, you might want to just chooose 12 for each and let the solver and CPU sort things out.
3-4 million cells seems like a lot. I'd double check you need that many cells to get the same level of accuracy. (mesh convergence) Also, you might want to review your model to make sure there can't be any additional simplifications made.