On bifurcations and traction forces on an obstacle in incompressible flow
Jakub Cach, Karel Tůma, Jan Blechta, Sebastian Schwarzacher
TL;DR
The paper investigates bifurcations in 2D incompressible Navier–Stokes flow past a confined cylinder (Schäfer–Turek benchmark) up to Re = 1000, showing that changes in steady traction profiles on the obstacle correlate with long-time transitions such as Hopf bifurcations, symmetry breaking, and the emergence of multiple steady states.Using a combination of deflated continuation to uncover multiple steady branches, linear perturbation theory to hint at instabilities, and direct simulations to identify long-time attractors, the authors establish that traction diagnostics on the upstream obstacle provide a computationally inexpensive indicator of regime changes that precede and reflect unsteady dynamics.While linear perturbation theory reliably detects the first Hopf bifurcation and predicts associated frequencies, its predictive power diminishes for later transitions, whereas traction profiles consistently reveal localized, upstream-initiated changes aligned with the observed bifurcations and attractor transitions, including the onset of a Kármán vortex street at high Re.
Abstract
We present a systematic numerical investigation of bifurcations in the two-dimensional incompressible Navier-Stokes flow past a confined circular cylinder. The results indicate that there is a qualitative correspondence between changes in the traction profiles of the steady Navier-Stokes equations and bifurcations of the long-time behavior of the unsteady Navier-Stokes equations. The bifurcations include the appearance of symmetry breaking, oscillations, and multiple steady solutions. The well-known planar Schäfer-Turek benchmark is considered with Reynolds numbers up to 1000. For the analysis of bifurcations and traction profiles, several numerical strategies are applied, including a duality method for computing traction profiles, deflation methods, and linear stability analysis. Long-time flow behavior is often explored through direct numerical simulation of the unsteady equations; an approach that is computationally demanding. The relations presented here indicate the possibility of a computationally inexpensive strategy to detect critical Reynolds numbers.
