[Noix_V]

Hello.

I have a scheme (.scm) script which I would like to run at the end of each calculated time-step. I know I can do this using the "Execute Commands" option, however I don't know how to name the script and the documentation is rather scarse...

If I use the "Read" option it justs executes the script right away, isn't there a way to make it run only when the end of a time-step is met?

Thanks a lot.

[obscureed]

Hi Noix_V,

If executing the .scm file does what you want, then you can put a TUI command in Execute Commands that executes the .scm file: for example

/file/read-macro "example.scm"

It does not matter whether the .scm file does or does not define macros -- this is a suitable TUI command either way.

Another approach is to write your .scm file so that it defines a function but does not execute it. You read in the Scheme file at any convenient moment. Then, in Execute Commands, you execute the already-read Scheme function. The difference between the two Scheme files is that, in this second approach, you insert lines at the start and finish

(define (do-stuff)
   (existing-command1)
   (existing-command2)
    ;;; etc
)

Then the line in Execute Commands is

(do-stuff)

The second approach is probably a more "Scheme-like" way of working. However, I often prefer the first approach for three reasons: (1) The first approach allows you to reopen the same .cas in a new Fluent session without needing to update. In the second approach, the Scheme function will be forgotten in the new session, so you must remember to re-read the .scm file before proceeding. (2) When the Execute Commands explicitly reads the .scm file every time, this makes it clear where the command is located/documented. (3) When the Execute Commands explicitly reads the .scm file every time, you have the opportunity to tweak the .scm in the middle of a run without interrupting Fluent. OK, so documentation and recorded history becomes messy, but this is often very handy.

You can combine the two approaches: make the .scm file define a function, and then read and execute in Execute Commands:

/file/read-macro "do-stuff_v001.scm"  (do-stuff)

And now you get the benefits I mentioned, but you can also run (do-stuff) at other moments.

As you say, documentation is sparse, so it would be useful if you could post a working example when you have one.

Good luck!
Ed

[Noix_V]

Ed:

Thanks a lot for your quick response! I haven't been able to test my .scm yet because of another problem in my UDF, regardless of that I'm trying to do a boundary change condition based on an UDF trigger, like in this thread: Is it possible? Changing BC from wall to pressure-outlet with an UDF trigger

If I get it to work I'll be sure to post my final versions. But I have one question, the path

/file/read-macro "example.scm"

Must be exactly the same as the working folder? or it has to start from the C: drive? What I mean, if my simulation is called "Test" would this be acceptable?

/Test_files/user_files/read-macro "example.scm"

Thanks a lot for your time.

[`e`]

The /file/read-macro is the command for the TUI, and "example.scm" is the filename of your Scheme script within your Fluent folder.

If you were reading a Scheme script from another location on your computer, then you could use an absolute path such as:

/file/read-macro "C:\Users\Noix\Documents\Scripts\Fluent_Scheme_Script.scm"

[obscureed]

I agree with 'e'. Well, except that I think that using full paths can sometimes bring unwelcome surprises: you move the project to a different computer, and things stop working; or you duplicate the project on the same computer, and the new project uses the old Scheme file. So, in general, I try to keep .scm files in the working directory, and if not to use paths relative to the working directory.

Oh, and the backslashes make me nervous. Should they be escaped\\doubled inside quotes? I have a vague memory that Fluent will sometimes accept Unix-style forward slashes and replace them with backslashes if required by the operating system at the time.

[`e`]

True, a local path would be easier for loading the script on another computer. But either way, you have the option of either specifying a local or global path in the TUI.

In Microsoft Windows, the folder location of the script can be copied and pasted from the file explorer which provides this format. Executing the TUI with the format of these backslashes worked for me. A Unix-like system uses forward slashes instead. If the file is not found by Fluent then checking the filepath would be a good place to start troubleshooting.

[Noix_V]

Edit: Nevermind this post, Fluent was crashing on me since I was searching for an ID that didn't exist. Still working on this!

[Noix_V]

Okay, now I have a real problem.


As I said before, I'm trying to run the scripts in this thread: Is it possible? Changing BC from wall to pressure-outlet with an UDF trigger (last post)


However, the value set on Fluent ("pressure_on_wall") does not seem to be updating with each time step.


I believe this may be due to me using a Paralell solver. I know there are some commands like host_to_node to send info, how can I use them? Or is that not my problem?


Thanks a lot for your time.

[AlexanderZ]

DEFINE_ON_DEMAND macro works only when you manually click it to be executed

Use, for instance, DEFINE_EXECUTE_AT_END instead.

logic you should release in parallel code:
1. find max pressure on computational nodes
2. send pressure data from node_to_host
3. send max_pressure from host to scheme

You may use Ansys Fluent Customization manual for more details

best regards

[obscureed]

Hi Noix_V,


It is certainly true that the code you linked to will not work as desired in parallel. AlexanderZ has outlined the steps. I'm in a helpful mood, so I have suggested a parallel version of the code: Is it possible? Changing BC from wall to pressure-outlet with an UDF trigger


Good luck!

Ed

[Noix_V]

Alexander: thanks!

Ed: wow, your help has been amazing! Just one more thing tho, I already tried the host_to_node command before, but I'm getting a rather undocumented error on two lines:

line xxx: mpnode_to_host_double_1: no function prototype
line xxx: MPT_gdhigh1: no function prototype

Needless to say, I didn't change the name on any of those functions and they also have their semicolon ( ; ) after them. Any ideas to why may that happen?


Edit: It seems to be because I need to compile my UDF rather than Interpret it. Any tips how to do this? I've never done it before...


Edit2: I'm downloading Visual Studio to compile UDF's.

[AlexanderZ]

I suggest you to use google, type there:

"how to compile udf in fluent"

best regards