25 analyses in one
 

HI,

I have to determine some kind of flow characteristic. To do that I have to perform about 25 analyses. The different between those analyses is only pressure at outlet. It is mean that pressure outlet in first analysis is let’s say 6bar and in the last one 1bar and in every analysis it change about 0.2bar. Is there any solution to do it automatically, to don't change outlet pressure after each analysis, and to have possibility to check flow properties to all pressure steps.

Thank you

Mateusz K.

If you have to change the Outlet Pressure 25 times, what's the problem of doing it and saving 25 different *.def files? It's gonna take much less time than the Solver will take to run each case and less time than you'll need to spend in Post.

Or are we missing something?

save the outlet pressure as a dx parameter (or whatever its called nowadays in v12) and use workbench to change the parameter within the range you want to test. either that or what rui said and perhaps run the simulations through a batch excecutable as glenn sugests here
http://www.cfd-online.com/Forums/cfx...atch-mode.html

Hi,

If I was doing this I would have a single def file, a CCL for each configuration and run it through a batch file such as:

cfx5cmds -write -def Definition.def -text run01.ccl
cfx5solve -def Definition.def
cfx5cmds -write -def Definition.def -text run02.ccl
cfx5solve -def Definition.def

etc.

If you want to get snazzy you can use perl and batch/script file looping stuff to do it nicely, but the simple approach is easier for small jobs like this.

Of course the whole idea of Workbench 12 is so you can do it easily in WB without this lower-level stuff but I am an old-timer and would rather avoid the overhead of the whole WB environment - but that's up to you.

Glenn Horrocks

thank you for your help. I have created batch input file and it works, as ghorrocks suggested in his previous post http://www.cfd-online.com/Forums/cfx...atch-mode.html

Hey Glenn, do you have any idea on how big the overhead is running Workbench compared to running standalone? We have linux on our cluster and there will be a workbench version for linux later this year (at least that's was ansys is saying...), but the question is if it is worth having a fancy-pancy interface with a big overhead.

Cheers
Lance

IMO No. unless you currently or intent to process the geometry through workbench (which is highly unlikely as you are using linux)

I dont think that setting the case in workbench has any advantages or purpose as anything more complicated that the standard ANSYS presentations in which have been carefully pre-tested so the simulation wont crash (or would require user input for tuning)

oh.. and if ansys wont fix the automatic update for the whole simulation setup if for example you do something in design modeler and then by accident the geometry needs to be regenerated... take my advice and remove any sharp objects from your desk prior the above exercise as they soon will become lethal projectiles....

This is my personal opinion obviously... :rolleyes:

If you are running on a 64bit machine with a decent amount of RAM, the overhead of the Workbench environment is minimal, and it does have many advantages for building complex coupled analyses.

Having said that, ck is absolutely correct. If you have a complete analysis and accidentally change one little thing, even if you change it back, the whole analysis will go stale, and will want to run again. This is not really a problem for CFX, as the default is to restart from the previous results file; you can do this and stop it after a few time steps and you're fine, but it is a certainly a hassle. If you have your model (geometry, mesh, CFX setup) complete and are making a large number of runs that vary only in ways that can be handled via CCL changes, the batch method is certainly a great option. However you lose access to all of the great Workbench optimization tools, which can be very nice for such analyses.

Too late, I've already commited hara-kiri three times with a stapler... :mad:

Seriously, Ansys got a few things to fix before I would use WB on a daily basis.

surprisingly the current version was released after in house beta testing... :rolleyes:

makes ya wonder sometimes.... ;)

Hi,

Well, WB is focussed on a different market.

WB lowers the entry point to FEA and CFD, making it easier to get started for simple analysis. Some complicated analysis is also much easier done on WB, such as FSI or optimisation.

However WB has a big failure in that it does not work well where you need traceability of input files, analysis and results. Once you have done a run and want to "lock it in" - that is archive it for later reference and never change it - well this is not easy in WB. If you want to do another analysis based on that old run but changing a parameter you either have to fire up the old project and add another analysis (then you modifying an archived simulation which is a big no-no) or you have to copy the whole analysis and do your modifications (then you have multiple copies of components of the setup which is another big no-no).

For traceable simulations you really need things to be driven by a command line with specified input and output files. Then these input and output files can be traced.

That is why most of my stuff is run from the command line. You set up the input files, run it from the command line and get the results. Then the input files, command line and results files are archived as a simulation run. Maybe not as pretty and slick as WB but traceable and all input have known histories and archived simulations are not modified.

I would be interested to hear other peoples thoughts on this issue.

Glenn Horrocks