I'm a long-time user of CFX4 and I heavily employ the user FORTRAN. It appears that CFX5 is now the top of the line product. What is the status of user Fortran in CFX5? Considering that CFX5 is an unstructured solver, will user FORTRAN in 5 ever be as functional and accessible as it was in 4?

Thanks,

Worth

Long time ago I also used CFX 4 and had to use Fortran for the difficult stuff. Nowadays I use CFX 5 and have never used Fortran up till now. For some reason I can do without.

Astrid (back again)

Hi Astrid and Worth,

(Welcome back Astrid!) CFX5 has CEL code which allows you to define most expressions which CFX4 required fortran to define. CFX5 does have fortran subroutines, but their scope is limited and you cannot access everything you could in CFX4.

Just like Astrid, I used lots of fortran in CFX4, and am now setting the entire simulation up in the command file of CFX5. I don't even have a fortran compiler anymore.

Glenn

Neither have I.

Astrid

Hey Guys, There is still some need for the old fortran - its not dead yet. For areas such as moving mesh - something we have been working on, Fortran is still necessary. We also use fortran to interrogate volumes within a model and use the fortran to export particular factors from the volumes and even stop the simulations once a certain criteria has been reached. I'd say that the fortran is not as easy to use as in CFX4, but the Guys at CFX are very helpful, and they will advise you as to the best way to tackle your particular problem. Like Glenn and Astrid said, you may even be able to get away from using the fortran all together. Bob

There are two flavors of FORTRAN as well.

1. User CEL functions 2. Junction Box routines

The User CEL functions are MUCH easier to code and I've written several for calculating complex boundary conditions and extracting needed transient data. You simply write your routine as a function that can be called directly from CEL and stored in a CEL variable. Any variables accessible from CEL can be passed in for use in the routine.

The Junction Box routines are harder to use but I've not found a need for them yet. Nothing gets passed in or out. You do everything by accessing the MMS (memory management system) arrays, which is anything but straight forward or well documented.

There are examples of both in at the end of the Solver Models manual and a pretty good power point on FORTRAN on the community site.

Jeff

Like the previous contributions, I also find CEL very useful for most applications. I do however think that CEL has limitations, especially if you need to access gradients of varibles for your computations. As far as I know, you are not allowed to access gradients in CEL? - Or am I missing something? What are your experiences with that?

Jesper.

You're right, Jesper

Gradients are still inaccessible through CEL. But it is easily accessed by User Fortran.

I believe that gradients can be available in CEL in CFX-5.7.

Ribeiro.

Nope, no gradients in CEL for 5.7. You will still have to use USER_GETVAR for that.

Neale

This is really not hard. There's a perfect example of doing this through a USER CEL function at the end of the 5.6 Solver Models manual. I do this on a regular basis. You have to be careful with additional variables, however, which are not solved as transport equations. Unless the algebraic variable is used as a source for another equation, the gradients don't appear to be available.

Jeff