[sangeet]

Hello,
I am currently trying to use a convective boundary condition with spatially varying heat-transfer coeff and Tinf. I have both of these as 9th order (please dont ask me why) polynomial function of non-dimensional radial distance from origin. In the code below, D is used to non-dimensionalize the radial distance which has been calculated by using pos().x and pos().y. My Tinf is Taw here. k and kg are thermal conductivites. thetaAW is the non-dimensionalized Taw. h is the heat transfer coeff. The p and q are just constants of the polynomials.

EDIT: I am using laplacianFoam.

target
    {
        type                groovyBC;
    variables        
"D=0.0063;
k=51.9;
kg=0.0323;
r=sqrt(pow(pos().x,2)+pow(pos().y,2))/D;
h=($p1*pow(r,8)+$p2*pow(r,7)+$p3*pow(r,6)+$p4*pow(r,5)+$p5*pow(r,4)+$p6*pow(r,3)+$p7*pow(r,2)+$p8*r+$p9)*kg/D;
thetaAW=($q1*pow(r,8)+$q2*pow(r,7)+$q3*pow(r,6)+$q4*pow(r,5)+$q5*pow(r,4)+$q6*pow(r,3)+$q7*pow(r,2)+$q8*r+$q9);
Taw=$Tinf+($Tg-$Tinf)*thetaAW;
f=1/(1+k/(h*mag(delta())));";
    value            uniform 294;
    valueExpression        "Taw";
    gradientExpression    "0";
    fractionExpression    "f";
    }

The problem is that this results in a completely blown up solution, if i instead use a constant Tinf i.e, by

valueExpression        "294";    //say

then the solution does not blow up. Can anyone help me out here please?

[gschaider]

Originally Posted by sangeet

Hello,
I am currently trying to use a convective boundary condition with spatially varying heat-transfer coeff and Tinf. I have both of these as 9th order (please dont ask me why) polynomial function of non-dimensional radial distance from origin. In the code below, D is used to non-dimensionalize the radial distance which has been calculated by using pos().x and pos().y. My Tinf is Taw here. k and kg are thermal conductivites. thetaAW is the non-dimensionalized Taw. h is the heat transfer coeff. The p and q are just constants of the polynomials.

EDIT: I am using laplacianFoam.

target
    {
        type                groovyBC;
    variables        
"D=0.0063;
k=51.9;
kg=0.0323;
r=sqrt(pow(pos().x,2)+pow(pos().y,2))/D;
h=($p1*pow(r,8)+$p2*pow(r,7)+$p3*pow(r,6)+$p4*pow(r,5)+$p5*pow(r,4)+$p6*pow(r,3)+$p7*pow(r,2)+$p8*r+$p9)*kg/D;
thetaAW=($q1*pow(r,8)+$q2*pow(r,7)+$q3*pow(r,6)+$q4*pow(r,5)+$q5*pow(r,4)+$q6*pow(r,3)+$q7*pow(r,2)+$q8*r+$q9);
Taw=$Tinf+($Tg-$Tinf)*thetaAW;
f=1/(1+k/(h*mag(delta())));";
    value            uniform 294;
    valueExpression        "Taw";
    gradientExpression    "0";
    fractionExpression    "f";
    }

The problem is that this results in a completely blown up solution, if i instead use a constant Tinf i.e, by

valueExpression        "294";    //say

then the solution does not blow up. Can anyone help me out here please?

As soon as I hear "polynomial of 9th order" I run for the hills: that is bound to be unstable https://www.wikiwand.com/en/Runge%27s_phenomenon

If you have no better approximation and mujst use them:

• have you tried plotting the polynomial in the expected range and checked that it looks "sane"
• are you sure that you're not using the polyomial outside of the data range used for fitting (because that will make sure that it explodes)
• try writing every time-step and check when and where the "explosion" starts

Bernhard

[sangeet]

Thank you so much for your reply. Really helped me, I just shrank the domain to within my fit interval.

Btw thank you for making swak4Foam. It has been an immense help.