I working on design of hydrofoil assisted catamarans. During testing of scaled models, the hydrofoils suffer from laminar separation due to the low Reynolds number I am forced to use at model scale. The increase in drag due to the separation is fairly easy to determine using empirical methods. I need to estimate the loss of lift due to separation.

I have tried to model the problem using 2D boundary layer theory such as the XFoil program from Mark Drela at MIT, with no success.

Does anybody have any experience with a problem such as this? I am thinking about applying CFX to the problem.

My foils are circular arc sections, 8% thickness, Re=1e5, a.o.a. = -1° to 4°.

Hi, which kind of turbulence model do you intend to use? Does the flow reattach after separation? I am using low-Re k-eps models with some success in calculating transitional laminar separation bubbles, i.e. laminar separation followed by turbulent reattachment. It has shown that separation can be calculated without using empirical correlations. Maybe you should think about using these models, though I do not know if they are offered by CFX. Regards, Jens.

I was also thinking about using a low Re k-eps model. They are available on CFX.

There is no turbulent reattachment after separation. That is the main problem for me, the separated region is quite large.

Did you compare your calculated lift with experiments?

Hi, my main interest was focused on short laminar separation bubbles that have only minor effects on the main flow surrounding the bubble. Pressure distributions could be computed to a satisfying degree with some deficiancies in the recovery region after reattachment. Bubble bursting or massive separation is harder to compute, though in your place I would try the low-Re k-eps models. Alternatively, the Spalart-Allmaras one-equation model seems to be a good candidate for flows involving massive separation. Maybe you should try this model in its low-Re version. Regards, Jens.

Jens, What were the Reynolds number of your calculations? Did you get good correlation with the length of the bubble as well as the separation point?

Next week I am on my way to the "Institut für Schiffbau" in Hamburg. I will start these calculations in about 6 weeks on my return.

Regards

Günther

(1). The only suggestion I have is to use real airfoil section, instead of a circular arc shape. (2). The reason is, there is so far no evidence that flow likes the circular arc shape. I am not sure whether Wright brothers airplane uses the circular arc shape as the wing shape. (3). So far I could not find any aircraft or formula-1 car or turbomachinery using the circular arc shape airfoil. (4). The only place is the cage-type radial fan, which has many small circular arc shape blades in it. (but then in that type of application, the flat plate blade is also equally effective ) (5). So, the circular arc shape is not a good airfoil. Even the birds are not using it.

Out of a aero- or hydrodynamic point of view you may be right. There definitely are more efficient profiles out there.

The reason for their use is partly due to manufacturing reasons. These foils are used on very large vessels, a 14m span hydrofoil carrying 80 tons of load is not uncommon. It is easier to manufacture these out of rolled steel plates, than to try and manufacture a complicated shape.

basically I am stuck with circular arc profiles...

(1). Well, in that case, you can try the flow separation control methods. (2). Surface mounted vortex generators have been effective in the early days of high speed airplane to contorl the boundary layer separation on the wing. (3). The similar devices have been used on the B-1B bomber , and engine inlets.

This is purely a model scale problem. The problem lies in that there are to dimensional paramters effecting drag of hydrofoils under the free surface: Reynolds number and Froude number. We test at equivalent Froude numbers, forcing one to correct for Reynolds number effects either empirically or theoretically.

The prototype ships don't suffer from any separation as the Reynolds numbers are high enough. Circular arc type sections have been in use on hydrofoil craft for decades, even though they are not the most efficient, they provide a practical solution when you balance all the factors concerned.

(1). Yes, when you test a scaled mode (geometric model) with the Reynolds number much lower than the full scale Reynolds number, you are going to run into the laminar flow separation problem. (2). The better way to do is to simulate the full scale flow with cfd approach. (if it is available).(3). Thank you for the information.