"Another area in which XFOIL differs from experiment is in the prediction of turbulent separation. The location of turbulent separation predicted by XFOIL is significantly further aft than that observed in the wind tunnel. For most cases tested, the XFOIL predictions of the chordwise location of turbulent separation differ from experiment slightly as the trailing edge flow first separates, with the discrepancies becoming greater as the angle of attack is increased. This tendency of XFOIL to under-predict the extent of turbulent separation results in a stall angle of attack that is greater than that seen in the wind tunnel data."
This is from AIAA 2003-4067, Surface Oil Flow Measurements on Several Airfoils at Low Reynolds Numbers, McGranahan and Selig.
Although not symmetrical airfoils, the phenomenon is the same.
AIAA 97-0511, Boundary Layer Trips on airfoils at Low Reynolds Numbers, Lyon, Selig, and Broeren, has some good trip data, especially 2D vs 3D trips.
Frank