SearcharxivSearch

arXiv subjects

C. Airiau

Publications and source records attributed to C. Airiau.

2 recordsLinked to original sources

Stability and dynamics of the flow past of a bullet-shaped blunt body moving in a pipe

The flow past a bullet-shaped blunt body moving in a pipe is investigated through global linear stability analysis (LSA) and direct numerical simulation (DNS). A cartography of the bifurcation curves is provided thanks to LSA, covering the range of parameters corresponding to Reynolds number $Re = [50-110]$, confinement ratio $a/A = [ 0.01 - 0.92]$ and length-to-diameter ratio $L/d = [ 2-10]$. Results show that the first bifurcation is always a steady bifurcation associated to a non-oscillating eigenmode with azimuthal wavenumber $m=\pm 1$ leading to a steady state with planar symmetry. For weakly confined cases ($a/A<0.6$) the second bifurcation is associated to an oscillating mode with azimuthal wavenumber $m=\pm 1$, as in the unconfined case. On the other hand, for the strongly confined case ($a/A>0.8$), on observes destabilization of non-oscillating modes with $|m| =2,3$ and a restabilization of the $m=\pm1$ eigenmodes. The aspect ratio $L/d$ is shown to have a minor influence for weakly confined cases and almost no influence for strongly confined cases. DNS is subsequently used to characterize the nonlinear dynamics. The results confirm the steady bifurcation predicted by LSA with excellent agreement for the threshold Reynolds. For weakly confined cases, the second bifurcation is a Hopf bifurcation leading to a periodic, planar-symmetric state in qualitative accordance with LSA predictions. For more confined cases, more complex dynamics is obtained, including a steady state with $|m|=3$ geometry and aperiodic states.

physics.flu-dyn

Adjoint computations by algorithmic differentiation of a parallel solver for time-dependent PDEs

A computational fluid dynamics code is differentiated using algorithmic differentiation (AD) in both tangent and adjoint modes. The two novelties of the present approach are 1) the adjoint code is obtained by letting the AD tool Tapenade invert the complete layer of message passing interface (MPI) communications, and 2) the adjoint code integrates time-dependent, non-linear and dissipative (hence physically irreversible) PDEs with an explicit time integration loop running for ca. $10^{6}$ time steps. The approach relies on using the Adjoinable MPI library to reverse the non-blocking communication patterns in the original code, and by controlling the memory overhead induced by the time-stepping loop with binomial checkpointing. A description of the necessary code modifications is provided along with the validation of the computed derivatives and a performance comparison of the tangent and adjoint codes.

physics.comp-ph