Discussion Forum

Hi, i am a newcomer in comsol, early had to do with LS-Dyna + Hypermesh.
I try to model a spherical piezoelectric layer in air (polarization occurs perpendicular to sphere). Additionally is there a structure with an axial symmetry. Therefore the model is axial symmetric. I started with a 2D - axial symmetric coordinate system (z,r). I have a trouble with a definition of polarization direction and need your estimation of my logic.

In comsol is assumed, that polarization must be along of the third coordinate: z-coordinate in a the global cartesian coordinate system (x,y,z).

As i understood it from your example "Radially_Polarized_Piezo", an another coordinate system with the last coordinate in the polarization direction it must be construct for it.

But my current system is not (x,y,z) as in the example and i have only two coordinate (z,r).

In the spherical radial direction R (system {theta, phi, R}) unit vectors have the following components:
by R: {sin(theta)*cos(phi), sin(theta)*sin(phi), cos(theta)},
by theta: {cos(theta)*cos(phi), cos(theta)*sin(phi), -sin(theta)},
by phi: {-sin(phi), cos(phi), 0},
where
theta=pi/2 - atan(z/r); phi=atan2(y/x); R=(r^2+z^2)^0.5;

In my current axial symmetric system (z,r) is phi=0. Therefore unit vectors have the following components:
by R {z,r}={cos(theta),sin(theta)},
by theta: {z,r}= {-sin(theta), cos(theta)},

In 4.2 comsol - version a new coordinate system is defined as base vector system as followed:

x1{r,z}={sin(atan(z/r)), -cos(atan(z/r))}
x2{r,z}={cos(atan(z/r)), sin(atan(z/r))}.

The normal component of the acceleration on the boundary between air and the piezoelectric layer is defined:

an= pzd.u_ttR*cos(atan(z/r))+ pzd.u_ttZ*sin(atan(z/r))

Is this definition of the polarization vector and of the acceleration component correct?

And yet a wish of a newcomer:

In Hypermesh and CAD - Software is there a possibility to divide a 2D- or 3D-volume by a plain or different surfaces. For the fem-model creation is it normally enough the volume dividing with simple surfaces as plane or cylinder. Is it difficult to allow this possibility in comsol too?

Thank you in advance.

Vladimir