Flow separation is an important phenomenon in fluid dynamics because of the effect it has on lift and drag on immersed bodies. Areas of swirl within a separated flow region may have a distinct effect on the surface forces, modifying the lift and drag characteristics. A correlation between the passage of vortical structures and the surface pressure can be used to determine locations on the surface most affected by separation and swirls in the flow. These locations can be used to place sensors to detect any variations in flow patterns that can be used to control lift and drag.

Unsteady separated flow over a square cylinder and a thin airfoil at high angle of attack are investigated using large eddy simulation. A full three-dimensional simulation is performed using high performance parallel computing. The flow Reynolds number is on the order of 10E+4 in both cases. At this Reynolds number, both flows contain separation and periodic vortex shedding over the surface of the object. The effect of these vortical structures in each flow is analyzed using different vortex detection techniques.

Four methods of vortex detection are investigated and compared: (i) the eigenvalue method (lambda_2), (ii) eigenvalue of the Hessian of pressure (lambda_p), (iii) the Gamma function, and (iv) Gamma_p, which is the Gamma function applied to the rotated pressure gradient. Both lambda_2 and lambda_p detect vortical structures by locating local pressure minima and use gradient fields. The Gamma function is the area averaged circulation around each point in the flow. The Gamma_p function shows locations in the flow where the pressure gradient is strongest based on the area integral of the rotated pressure gradient.

The eigenvalue methods tend to detect the vortex cores and small scale features in the flow because these methods are based on derivatives of flow variables, so are most sensitive to changes on the order of the grid size. The features detected with lambda_2 are similar in size and location to those detected with lambda_p. Both Gamma and Gamma_p tend to locate large swirls and group small flow features into larger regions of swirl. They are integrated approaches most sensitive to changes on the order of the size of the integral area. However, Gamma_p tends to identify more individual features than Gamma because it is based on pressure derivatives, so is also sensitive to changes on the order of the grid size. All vortex detection methods tested are used to track flow structures over time.

A two-point covariance between surface pressure and flow swirl is found to be periodic and linked to the oscillatory nature of the flow. In the mean, the correlation is shown to be strongest in regions where the time-averaged Gamma magnitude is the highest. These results can be expanded to other immersed bodies, with the future goal of developing a control scheme for flight.