Benchmark Reynolds-averaged Navier-Stokes study of a generic marine rudder’s static stall characteristics

When assessing manoeuvring performance it is common to perform steady Reynolds-Averaged Navier-Stokes computations for a range of turning conditions. It is then critical to reliably and accurately predict rudder forces. A recent collaborative benchmark study showed inconsistent capturing of stall effects on aft control surfaces for a generic submarine. To investigate this issue, we present a new generic marine rudder and the influence of the numerical setup on the prediction of its static stall characteristics. We demonstrate the existence of a static hysteresis loop, which we compute with Menter’s k - w SST turbulence model at a full-scale Reynolds number of 7.44 x 10⁶. One branch shows a light trailing-edge stall with a high lift coefficient, and the other a deep stall with a low lift coefficient. We show how the initial conditions and solution methodology affect which solution is achieved. Notably, even 1° below the low limit in the hysteresis loop, an incorrect deep-stall result can be obtained when using a free-stream initialization, due to poor iterative convergence. In contrast, potential-flow initialization and unsteady RANS methods effectively produce the correct high-lift solution below the hysteresis loop. A discretization analysis at α = 24°shows that six flow solvers produce consistent results to within a small discretization uncertainty.