Learning Forum

[kiran.purushothamakeshavan]

Hi,

In ANSYS Fluent, I have two .inj files containing randomly generated particle injection data from a surface. I want to alternate between these files at each time step or Fluent run. Is it possible to automate this switching using a journal file or through a UDF? If so, what would be the recommended approach?

Have you ever come across something like this before?

Thanks,
Kiran

[Rob]

It's probably possible but the IO overhead for switching files may be a problem. Is there a reason for alternating rather than just having two injections running all the time? 

[kiran.purushothamakeshavan]

I'm not currently concerned with alternating injections. My main goal is to use multiple injection files at separate stages of the simulation.

For instance, each .inj file contains particles initialized at random locations, and I want them to act as independent injections. Suppose I have two .inj files  I need the first one to inject particles into the domain, allow those particles to move and reach a specified boundary, and only after that, the second injection file should activate and release its particles into the domain.

[Rob]

You can just set two injections with different start & stop times: so injection one is 0-0.001s and injection two is 10-10.001s for example. 

[kiran.purushothamakeshavan]

If let's say I have to use 20-30 different injection files to randomise the inlet position, is there a way to automate this rather than upload 30 files?

[Rob]

You can use a journal, and that would set up the injections. I'm still unsure why you need everything to be so random: won't that prevent repeatability of the experiment?

[kiran.purushothamakeshavan]

Great, I found a way around this using a journal file like you said. 
But this is for particle data, where particles pass through a sensor and I don't quite know the locations of these particles so I'm guessing a randomised injection using multiple files might represent what happens during the test.

[Rob]

It might, but if the input isn't repeatable neither will the result be so you're randomly changing conditions to see if it gives the same result as an experiment. 

[kiran.purushothamakeshavan]

I see that you have a great point. Do you think I can compare the recorded results in terms of Particle size distribution at a particular surface using this randomised particle at the inlet?

[Rob]

You can, but I'd just run a single (or some) fixed injections and report that. I tend to favour some injections each of a fixed size as it's easier to post process visually: how can you easily see where a track is going if it's masked by many of various sizes?  Remember you're tracking parcels in Fluent so if you have too few the statistics become meaningless.