I'm considering a topology optimization problem where the design domain is a rotating annulus (in $\mathbb{R}^2$). One of the adjoint equations of the Navier-Stokes system whose solution is to be used in the sensitivity analysis has a weak form with righthand side: $$\int_{\Gamma_{\text{out}}} \rho(r\times u) \cdot \omega)n\cdot v + \int_{\Gamma_{\text{out}}} \rho(r\times \omega \times r)\cdot \omega)n \cdot v + \int_{\Gamma_{\text{out}}} (\rho(\omega\times r)\bigotimes u)n\cdot v,$$ where $v$ is a test function, $r$ is "perpendicular to the axis of rotation" (I've assumed that this is a position vector, see below), $\omega$ is the angular velocity (which is constant, say $|\omega| = 500$), $\Gamma_{\text{out}}$ is the outflow boundary (the outer boundary of the annulus), $n$ is the outward normal to $\Gamma_{\text{out}}$ (i.e. $n(x) = \frac{x}{||x||}$), $u$ is the velocity computed by solving the Navier-Stokes equations, and $\rho$ is the mass density (scalar). I'm wondering how I can formulate this problem for a 2D problem. What I'm doing now is writing $$r(x) = \begin{bmatrix} x_1 & x_2 & 0\end{bmatrix}^\top,\\\omega(x) = \frac{500}{||x||}\begin{bmatrix} -x_2 & x_1 & 0\end{bmatrix}^\top,$$ and so on, treating every 2D vector as a 3D vector in the $xy$-plane, doing the computations by hand and then coding them explicitly (since of course the PDE software I'm using doesn't know how to interpret a cross product of vectors in $\mathbb{R}^2$ either). I'm wondering if this approach is correct or there's some other way of formulating this problem that I'm missing. I'm also not sure that I'm interpreting $r$ or $\omega$ properly, so if there's anything I'm missing there I'd be grateful for some feedback. I'm not an engineer or physicist by training, so I'm not aware of the conventions for these things!


Your Answer

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

Browse other questions tagged or ask your own question.