0
$\begingroup$

I am trying to solve Lid driven square cavity flow problem of Stokes equation using finite element method. I have boundary conditions for velocity as zeros on every boundary but u=1 on top boundary. can you please help me with the boundary conditions of stream function and how to obtain ? especially on top boundary.

I have gone through many literature but why most of them considered as zero on every boundary.

$\endgroup$
1
$\begingroup$

The velocity and stream function are related by $$ u = \psi_y, \qquad v = - \psi_x $$ This can be related to the vorticity $$ -\Delta \psi = \omega = v_y - u_x $$ On left and right, we have $(u,v)=(0,0)$ $$ \psi = const, \quad \psi_x = 0 $$ On bottom $(u,v) = (0,0)$ $$ \psi_y = 0, \qquad \psi = const $$ On top $(u,v)=(u_0,0)$ $$ \psi_y = u_0, \qquad \psi = const $$ You cannot enforce two boundary conditions. Let us use Dirichlet bc $$ \psi = const $$ on all boundaries. Since $\psi$ must be continuous (assuming $\omega$ in $L^2$ atleast), the constant must be same on all sides which you can set to zero.

$\endgroup$
  • $\begingroup$ I had some doubt that psi_y =u0 does not mean psi = constant, it can be psi=u0y $\endgroup$ – Eda Suresh Reddy Aug 23 '18 at 12:16
  • $\begingroup$ On top boundary $\psi = const$ implies $\psi$ does not depend on $x$. $y$ is already fixed on the top boundary.. The other condition says $\psi_y(x,y_{top}) = u_0$ but you cannot integrate this as you seem to do, because that condition holds only at $y=y_{top}$. The thing is you have two bc on the top, a dirichlet and neumann. I am suggesting to use dirichlet bc. The solution you get must satisfy the other bc, maybe only in a weak sense. $\endgroup$ – cfdlab Aug 24 '18 at 3:27
0
$\begingroup$

For the streamline formulation the divergence equation for the velocity $\vec{v}=(u,v)^{T}$ $$\mathrm{div}\,\vec{v}=0$$ holds identically if $$ \vec{v} =(u,v)^{T}= \left( \frac{\partial \Psi}{\partial y},-\frac{\partial \Psi}{\partial x}\right)^{T}\tag{*}$$

For boundaries in which the velocity is $\vec{v}=\vec{0}$, it is easy to see from $(*)$ that $\Psi=constant$ and $\vec{n}\cdot\vec{\mathrm{grad}}\,\Psi=0$. Without any loss of generality you can set this constant to $0$.

On the other hand, on the top of the cavity in which $\vec{v}=(u_0,0)^{T}$ the streamline function $\Psi$ must fulfill: $$ \frac{\partial\Psi}{\partial y}=v_0\qquad \frac{\partial\Psi}{\partial x}=0 $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.