Rudder Centre of pressure with PMB Foil ??

Real data does not support small trailing edge angles for real sized hydro bodies at moderate (5-20 knt) speed, especially at high angles of attack. If you are using Xfoil, at what Rn did you set the turbulent transition for. If you used the default (9x10^6) that is too high by at least a factor of 3 for real applications (and we won't even get into the idea of crossflow inflow caused by waves, roll, etc). There is a real problem of trailing edge rollover if the trailing edge angle is too narrow. This causes extreme flow disruption and oscillating flow/pressure/energy over the body. The EPH shape on the other hand with the bluffer aft body and flat spots (yes, there are designed in flat spots that I've had to go down to the loft and make them put them in) which stablizes the mass flow prior to the trailing edge and actually stablizes the trailing pressure vortex, captures it, binds it stabily to the trailing edge, and bounds it in an energy well, i.e. it doesn't have the energy to cause flow disruption. Remember, in real conditions, the water is not moving and the body is; therefore energy flow in the fluid is in the direction of travel. Real flow is forward and inwards at the trailing edge

Absolutely not messing with you...the body is actualy "streamlined" by two bound vortices, see Hoerner FDD chpater 3. Now that I have Blevins here go look at Table 10-18, figures 1,3,18,and 19. Correcting for how "D" is measured in each figure and using a real Rn of say 2x10^6 the drag is not as different as you would think.

Here's a good one...make a spade rudder out of an exposed shaft and a flat plate. Shaft diameter is 12% of span located 30% of span back from the leading edge; plate thickness ahead of the shaft is 5% span; plate thickness at the trailing edge is 2% span; normal aspect, sweep, and taper. Now make a NACA 0012 with identical aspect, sweep, and taper. Mr Jackson says that when tested at real Rn's (say 5x10^6) the results between the two will be within experimental error and the results between the two will closer than between the NACA and the Xfoil prediction.