[schwaral]

Hello,

does anyone know if it is even possible to change the UDS Inlet Typ from "UDS Value" to "UDS Flux" by using an UDF?
My problem is, I have an oscillating flow. So my Inlet becomes the Outlet and the other way arround.

With my UDS I want to have at the UDS-Outlet always a gradient = 0. That is equal to UDS Flux = 0.

At the Inlet I want to have always a specified "UDS Value" (to adress the relativ humidity)

So when my velocity changes from positiv to negativ my UDS-Inlet becomes my UDS-Outlet and I need to define a UDS Flux=0 instead a certain value.

I hope my problem is undestandable.


PS: Or is there any other possibility instead of stopping by hand the Simulation and switching the UDS_boundary_types?

[AlexanderZ]

you can change type of Boundary Condition from UDF,
but it is possible with scheme + TUI commands

if I were you, I would make:
1 flag variable in scheme
2 Read it in fluent
3 Apply value to flag (depends on your criteria)
4 Get value of flag in scheme
5 Change/not change boundary condition

[schwaral]

I think I figured out something but I don't understand, why this scheme if statement is now not working. Like you can see my variable (veldirect) that I want to use for changing the BC is printed and calculated correctly. But scheme seems not to use this variable correct. It thinks it is always 0.

reversed flow in 20 faces on pressure-outlet 11.
900 7.1974e-03 4.2344e-05 2.3118e-06 1.3848e-16 1.0680e-05 0:00:00 1
veldirect: 1

reversed flow in 20 faces on pressure-outlet 11.
901 5.9673e-03 4.3454e-05 2.0997e-06 1.3159e-16 6.8316e-06 0:00:00 0
(ti-read-journal "testjournal.jou")
Reading journal file testjournal.jou...

> (cond
((= (%rpgetvar 'veldirect) 1)
(ti-menu-load-string "define/boundary-conditions/velocity-inlet/9 yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 yes no 0")
)
((= (%rpgetvar 'veldirect) 0)
(ti-menu-load-string "define/boundary-conditions/velocity-inlet/9 yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 no yes no 0.8")
)
)
define/boundary-conditions/velocity-inlet/9(inlet)
yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 no yes no 0.8#t
#f

I have tried diffrent ways to use if statement in scheme but it is always the same result. It thinks veldirect = 0. Why does this happen :/ The only thing I can see is weird is that veldirct is not calculated for the last iteration. I calculate it with a DEFINE_ADJUST UDF

[schwaral]

when I use (rpgetvar 'var_name) to check the Value it shows me 0. So I think I have to use the UDF in a diffrent way, that calculates veldirect.

DO you have any Idea how can I set the value so my journal gets the actual calculated value and not somehow a 0? My UDF is

DEFINE_ADJUST(adjust_UDS_bc, d)
{
int zone_ID = 9;
Thread* t = Lookup_Thread(d, zone_ID);
if (fnew >= 0)
{
RP_Set_Integer("veldirect", 0);

}
else
{
RP_Set_Integer("veldirect", 1);

}
}

[AlexanderZ]

try this code

(if (= (%rpgetvar 'veldirect) 1)
	(ti-menu-load-string "define/boundary-conditions/velocity-inlet/9 yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 yes no 0")
	(ti-menu-load-string "define/boundary-conditions/velocity-inlet/9 yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 no yes no 0.8")
)
(ti-menu-load-string "define boundary-conditions velocity-inlet 9 () yes yes yes yes udf velinlet::libudf no 0 no 1 no 0 no 300 no yes no 0.8#t #f ")

DEFINE_ADJUST is macro which is executed before iteration (or timestep if transient)
DEFINE_EXECUTE_AT_END is macro which is executed after iteration


if you've defined somehow criteria for condition yor udf will looks something like this

**** calculated fnew here  ****

node_to_host_int_1(count); /*transfer data to host, put it inside inside macro | if you are using v19.0 or above*/

DEFINE_ADJUST(adjust_UDS_bc, d)
{
#if !RP_NODE /*execute on host or in serial */
if (fnew >= 0)
{
RP_Set_Integer("veldirect", 0);
}
else
{
RP_Set_Integer("veldirect", 1);
}
#endif
}

[schwaral]

Thanks for the response. My UDF looked like that except these two lines down below. I added them now and my UDF is identical with yours.

Originally Posted by AlexanderZ

**** calculated fnew here  ****

node_to_host_int_1(count); /*transfer data to host, put it inside inside macro | if you are using v19.0 or above*/

#if !RP_NODE /*execute on host or in serial */
#endif
}

Unfortunately I received an error because of the last line in your Scheme. I deleted that and now it works fine. So I think the magic was the correction of my UDF.

I noticed some significant preformance reduction that I think is because of using parallel on 0 nodes. When I switch to serial the change of BC does not work anymore. Do you have any ideas for that?

[schwaral]

Originally Posted by AlexanderZ

try this code

**** calculated fnew here  ****

node_to_host_int_1(count); /*transfer data to host, put it inside inside macro | if you are using v19.0 or above*/

DEFINE_ADJUST(adjust_UDS_bc, d)
{
#if !RP_NODE /*execute on host or in serial */
if (fnew >= 0)
{
RP_Set_Integer("veldirect", 0);
}
else
{
RP_Set_Integer("veldirect", 1);
}
#endif
}

Is there any possibility to make this work in parallel mode on local machine?

When I start in parallel my variavle fnew is computet correctly but this adjust_udf ist printing the wrong value. 0 instead 1. In serial mode is everything fine. I tried already different versions and without #if !RP_NODE. So the result in parallel is at the moment, that the BC are not switching at all.

[AlexanderZ]

try to add line

node_to_host_int_1(fnew )

after you've calculated it

[schwaral]

I added this line and changed it to node_to_host_double_1(fnew ) because my fnew is of the typ double.

Unfortunatelly now my BC are all the time the same and not switching. My rpvar is now always = 1. Fnew seems to be calculated correctly.

Do I need to prepare the UDF in which fnew is calculated for parallel usage further more?

Or is it a problem that I have define fnew globally? If it helps, in the File below is my Code for calculating fnew. It is basically a simple Fourier Trafo.