[panasonic18]

Hey,



i've real big problem using an UDF for velocity inlet and can't solve it. It's a Mesh over an Airfoil created with ANSYS ICEM 15.0. When i load the Mesh into FLUENT 15.0 everything is fine, no errors.



Hybrid Initialisation is well without using the UDF, so no Errors without using the UDF.



But when i want to implement the UDF and press Hybrid Initialisation, FLUENT says this:



Error: %prf-set-var: invalid flonum
Error Object: -1.#ind


I can't figure out, where the problem is. Also never heard this error before and searching via google wasn't succesfull. So maybe someone coul help me? Here my defined UDF:




#include "udf.h"

#define c_max 30.88 // unit m/s
#define delta 0.04
DEFINE_PROFILE(inlet_velocity, t, i)
{
real x[3];
real y;
face_t f;

begin_f_loop(f,t)
{
F_CENTROID(x,f,t);
y = x[2];

if(x[2] <delta)

F_PROFILE(f,t,i) = c_max * pow(x[2]/delta, 0.2);

else

F_PROFILE(f,t,i) = c_max;
}
end_f_loop(f,t)

}


I appreciate every kind of help.



Best regards

[teguhtf]

Dear panasonic18

I think there are mistakes. You define real x[3] but you do not use x[3] in your whole code. you have already written y = x[2] so I think no need to write again x[2] in if(x[2] <delta). Just write if(y <delta). The same case can be applied also in pow(x[2]/delta, 0.2).

For the comment in

#define c_max 30.88 // unit m/s

you should write

#define c_max 30.88 /* unit m/s */

Hope it helps

[pakk]

The line with x[3] is correct, don't change it to x[2]. It signals that x is a three-dimensional array. Better is to change it to x[ND_ND].


I think the real problem (apart from the comment-problem after #define) is that you might take a non-integer exponent of a negative number



If you have a cell with negative y-coordinate, for example y=-0.04, then you calculate:

pow(-1, 0.2)

A negative number to a non-integer exponent results generally speaking in a complex result. C does not handle this, and gives an error. Here, because 0.2=1/5, it happens to mathematically give a real result, but still you get an error.


If this is really what you want, then you should help the compiler:

if (x[2]<0) {
  F_PROFILE(f,t,i) = -c_max * pow(-x[2]/delta, 0.2);
} else {
 F_PROFILE(f,t,i) = c_max * pow(x[2]/delta, 0.2);
}