Computation vs. experiment for high-frequency low-Reynolds number airfoil pitch and plunge

Abstract

For a set of pure-pitch and pure-plunge sinusoidal oscillations of the SD7003 airfoil, phase-averaged measurements using particle image velocimetry in a water tunnel are compared with computations using an Immersed Boundary Method and an unsteady Reynolds-Averaged Navier Stokes solver. Re = 40,000 and Re = 10,000, based on free stream velocity and airfoil chord, were chosen as representative values for, respectively, a case where transition in attached boundary layers would be of some importance, and a case where transition would not be expected to occur in attached boundary layers. The two computational approaches were compared for capacity to resolve shed vortical structures near the airfoil and in the wake, and for capacity to resolve the mean streamwise momentum balance in the wake. The plunge and pitch were each at reduced frequency k = 3.93 and with kinematically equivalent amplitudes of effective angle of attack, with the pivot point at the quarter chord. For the plunge cases, agreement between computation and experiment was qualitatively excellent and quantitatively acceptable, but for the pitch cases, the wake structure in the experiment was markedly different from that predicted by both computations, which were however similar among one another. This result was not appreciably altered by whether or not the test section walls were modeled in the computation. In all cases, Reynolds number effects were found to be negligible. Experimental-computational agreement for plunge, but lack of agreement for pitch, is presently left unresolved.