mesnardo / petibm-decoupledibpm

Decoupled Immersed Boundary Projection Method with PetIBM

Flow over a stationary circular cylinder ($Re=40$ and $100$)

Figure: Vorticity contours around the cylinder at Reynolds number $40$. (Contour levels between $-3D/U_\infty$ and $3D/U_\infty$ with increments of $0.4$.)

Figure: Vorticity contours around the cylinder at Reynolds number $100$ after $200$ time units of flow simulation. (Contour levels between $-3D/U_\infty$ and $3D/U_\infty$ with increments of $0.4$.)

Figure: Pressure coefficient along the upper and lower surfaces of the cylinder at Reynolds number $40$. We compare with the results from Li et al. (2016).

Figure: Pressure coefficient along the upper and lower surfaces of the cylinder at Reynolds number $100$. We compare with the results from Li et al. (2016).

Flow around an inline oscillating circular cylinder ($Re=100$)

Figure: Contours of the vorticity field around an inline oscillating cylinder at different phase angles ($\phi = 2 \pi f t$): $\phi = 0^o$ (left) and $\phi = 288^o$ (right). (Contour levels between $-20 U_m / D$ and $20 U_m / D$ using $30$ increments.)

Figure: Contours of the pressure field around an inline oscillating cylinder at different phase angles ($\phi = 2 \pi f t$): $\phi = 0^o$ (left) and $\phi = 288^o$ (right). (Contour levels between $-1 \rho U_m^2$ and $1 \rho U_m^2$ using $50$ increments.)

Figure: Profile of the velocity components ($u$: left, $v$: right) at four locations along the centerline for various phase angles $\phi$.

Figure: History of the drag coefficient of the inline oscillating cylinder obtained using different algorithms. We also show zooms at early and developed stages.

Figure: History of the drag coefficient obtained with Algorithm 1 for different time-step sizes and different grid sizes.

Figure: Variations of the $L_\infty$ and $L_2$ norm errors of the streamwise velocity as a function of the computational time-step size.

Figure: Variations of the $L_\infty$ and $L_2$ norm errors of the streamwise velocity as a function of the computational grid spacing.

Figure: History of the drag coefficient using Algorithm 3 with force-prediction scheme 3. We compared the history obtained with different Lagrangian mesh resolutions: $N_b = 500$ Lagrangian markers on the boundary and $N_b = 202$ markers (the latter one corresponding to the same resolution as the Eulerian background grid).

Flow around an impulsively started circular cylinder (Re=40)

Figure: History of the drag coefficient of the impulsively started cylinder. Comparison with the analytical solution of Bar-Lev & Yang (1997) and the numerical results from Taira & Colonius (2007).

Figure: Vorticity contours around the impulsively started circular cylinder at $t=1.0$ (left) and $t=3.5$ (right). Contour levels between $-3 \omega_z D / U_o$ and $3 \omega_z D / U_o$ with increments of $0.4$.

Figure: History of the recirculation length measured in the reference frame of the impulsively start cylinder at Reynolds number 40 and for different time-step sizes.

Three-dimensional flow around an inline oscillating sphere ($Re=78.54$)

Figure: Contours of the pressure field in the $x$/$y$ at $z=0$ at three phase angles. Contour levels between $-2 p / \rho U_m^2$ and $2 p / \rho U_m^2$ with $30$ increments.