To further understand the underlying unsteady large-scale structure in low Reynolds number separated flows, numerical simulations over a curved surface and a flat plate geometry were conducted in the present study. In the curved surface study, the universality of the previously found unsteady large-scale structure in laminar separation bubbles was evaluated by numerically determining the boundary-layer development over an Eppler 387 airfoil in low Reynolds number regime (Re <​ 500,000). The wall curvature effect on the unsteady separated flows was then isolated by comparing the vortex structures developed on a curved surface and a flat plate geometry, where both surfaces were subject to the same inviscid surface pressure. The present computations were performed by solving the unsteady incompressible Navier-Stokes equations using the method of pseudocompressibility.

After an initial transient, all computations in both geometries, started from free-stream conditions and without any perturbation, asymptotically reached limit cycle shedding. The resulting limit cycle vortex shedding exhibited behaviors reminiscent of a free shear flow. These phenomena include a lifted shear layer near separation, vortex roll-up, and subsequent vortex transport. Vortex pairing, one of the major mechanisms responsible for the shear layer growth, was also observed. After the flow reached limit cycle shedding, time-averaging of unsteady structure over a few shedding periods resulted in a separation bubble which was strikingly similar to experimentally measured laminar separation bubbles. The time-averaged results of the computations showed a region of nearly constant surface pressure followed by an abrupt increase in surface pressure just prior to reattachment. The vortex shedding was shown to be caused by an inviscid instability wave which is due to the inflectional velocity profile.

In the airfoil study, the separation and reattachment points measured in a low-turbulence level wind tunnel were closely reproduced by the computed time-averaged separation bubble. The computed time-averaged surface pressure coefficients also compared favorably with the experimentally measured data. Since the airfoil surface turns away from the shed vortex, the shed vortex trajectory in the airfoil travels along a higher course measured from the surface than the shed vortex trajectory in the flat plate. In addition, due to the turning away of the airfoil surface, the resulting free shear layer on the airfoil is less confined than the resulting free shear layer on the flat plate. The boundary layer on the airfoil surface is therefore more unstable and vortex shedding takes place earlier and more vigorously, resulting in a shorter time-averaged separation bubble. However, the separation point and separation angle of the time-averaged bubble are comparable in both flat plate and airfoil studies.