[Azim07]

Hello Guys,

I want to define the specific heat and viscosity of the material by following UDF and it is showing 'Floating Exception Error'. How can i Fix this? And what kind of precautions i should take to avoid any further Error?


#include "udf.h"
DEFINE_SPECIFIC_HEAT(cp_pcm,T)
{
real cp;
if (T<303)
cp=1250;
else if (T>=303 && T<=304)
cp=125000;
else
cp=1250;
return cp;
}

DEFINE_PROPERTY (cell_viscosity1,c,t)
{
real mu;
real temp = C_T(c,t);
if(temp<303)
mu=100000000;
else if (temp>304)
mu=0.008;
else
mu =(15200000000-(50000000*temp));
return mu;
}

[AlexanderZ]

use small enough time steps to get convergence

best regards

[Светлана]

(Sorry remove this messageplease, posted in a wrong thread)

[Azim07]

@AlexanderZ
i am already using the time step size 0.05... In this When i take initial temp as 303 K then it does not converge and goes not simulating.
but when i take initial temp as 306 and 308 K then it shown floating error.

What What should i check now?

[AlexanderZ]

#include "udf.h"
DEFINE_SPECIFIC_HEAT(cp_pcm,T, Tref, h, yi)
{
real cp;
if (T<303)
cp=1250;
else if (T>=303 && T<=304)
cp=125000;
else 
cp=1250;
*h = cp*(T-Tref);
return cp;
}

DEFINE_PROPERTY(cell_viscosity1,c,t)
{
real mu;
real T = C_T(c,t);
if(T<303)
mu=100000000;
else if (T>=303 && T<=304)
mu =(15200000000-(50000000*T));
else
mu=0.008;
return mu;
}

because you are using abnormal properties your time step should be small
size depends on your heat rate. How fast your system is heated up on 1 K?
divide this time at least 20 times (I recommend more) -> this is your time step

best regards

[Azim07]

So you mean i should use the code sent by you?
and i should use the more less time step size? You mean 0.005 or less??

[AlexanderZ]

these are mine recommendations

best regards

[Azim07]

It is still showing floating error... Please suggest me the following criteria i should use.

1. Time Step size
2. Max. Iteration
3. Methods to use for Scheme, Pressure, Momentum, Energy, and Gradient
4. Any other suggestion you like to say

Thanks & Regards

[Светлана]

You could do a dumb simplification so that your UDF always returns the same constant. For instance, replace "125000" with "1250" so that it always uses 1250. Does it converge then?

[AlexanderZ]

go to material properties and use piecewise-linear option for properties you are interested in.

best regards