[lcchaos]

Hello

I'm trying to calculate the force on a surface to update the moving-wall boundary condition of the surface. The gradient of U is needed for calculating viscous contribution of forces.

The code is based on the example on wiki
http://openfoamwiki.net/index.php/Co...ectionalForces

    
sphere
    {
	type		groovyBC; 
  	variables 	(
			"pressure_force=sum(p*normal()*area());"
			"viscous_force=sum(-snGrad(U)*area()*0.01);"	//nu=0.01
			"force=pressure_force+viscous_force;"
			"acceleration=force*1000/33.51;"	//density=1000;mass=33.51 SI
			"torque=sum((pos()-vector(0,0.4,0))^(p*normal()*area()-snGrad(U)*area()*0.01));"
			"angular_acceleration=torque*1000/0.5362;"	//moment of inertia=0.5362
			);
	valueExpression	"(time()>1.0) ? U+(vector(acceleration.x,0,0)+angular_acceleration^(pos()-vector(0,0.4,0)))*deltaT()*0.5 : U";
	value		uniform (0 0 0);
    }

I noticed that snGrad(U) is not the inner product of normal vector of face and the deformation tensor. So I change snGrad(U) into grad(U)+grad(U)T (T means transport matrix). However, I got an error

--> FOAM FATAL ERROR: 
 Parser Error for driver PatchValueExpressionDriver at "1.1-4" :"field grad not existing or of wrong type"
"grad(U)"
  ^^^^
--|

Thanks for your help!

Regards,

[RodriguezFatz]

Hi!

Originally Posted by lcchaos

I noticed that snGrad(U) is not the inner product of normal vector of face and the deformation tensor.

But what is snGrad(U) else?

[lcchaos]

Originally Posted by RodriguezFatz

Hi!

But what is snGrad(U) else?

Hi, snGrad(U) is the gradient of U in the surface normal direction.

[RodriguezFatz]

Well ok, this does not account for the moving wall... or what is the problem?

[lcchaos]

Originally Posted by RodriguezFatz

Well ok, this does not account for the moving wall... or what is the problem?

Philipp, I want to impose a moving wall boundary condition on the surface of a sphere. The motion results from the hydrodynamic force . Therefore, the gradient of U at the surface must be obtained to calculate the force. The problem is that an error appears when I use grad(U). Using snGrad(U) is incorrect physically.

[gschaider]

Originally Posted by lcchaos

Philipp, I want to impose a moving wall boundary condition on the surface of a sphere. The motion results from the hydrodynamic force . Therefore, the gradient of U at the surface must be obtained to calculate the force. The problem is that an error appears when I use grad(U). Using snGrad(U) is incorrect physically.

You need the gradient in tangential direction? There is no built in implementation for that in OF (and therefor not in swak). Best way to get this would be to calculate grad(U) (in a seperate step. groovyBC can't do this) on the interior and then get that value from the cells next to the patch (with internalField). But this is probably rather inaccurate.

On the other hand: as you probably know the U of the sphere it is probably better to calculate grad(U) in tangential direction analytically

[lcchaos]

Originally Posted by gschaider

You need the gradient in tangential direction? There is no built in implementation for that in OF (and therefor not in swak). Best way to get this would be to calculate grad(U) (in a seperate step. groovyBC can't do this) on the interior and then get that value from the cells next to the patch (with internalField). But this is probably rather inaccurate.

On the other hand: as you probably know the U of the sphere it is probably better to calculate grad(U) in tangential direction analytically

Hi Bernhard,

Thanks for your advice! I' m trying to do the calculation by modify the source code. My general idea is adding a new equation for the tensor grad(U) after the momentum equation is solved. Does this way can avoid the inaccuracy?

Regards
Chao