hard wing vs soft wing?

It's possible for a turbulent flow to separate and then reattach, thus forming a "turbulent separation bubble." This could be caused by a local surface imperfection (to give just one example).

But laminar flows are more interesting (at least to me), because laminar separation often triggers transition to turbulent flow, which in turn can help the flow to reattach. Turbulent separations have no such built-in mechanism for reattaching.

As in all questions involving boundary layers, the details depend strongly on the Reynolds numbers & pressure distributions.

Edit: I didn't see DCockey's post before writing mine, so this is just to say I basically agree with him.

 

Thanks a lot DCockey & Doug,

I was aware of LSB and the associated "Transition Ramp", used to trigger transition, especially in the case of Stratford Recovery.
It seems to be the "Built in Mecanism for reattaching" you mention.

The idea which is difficult to understand for a rookie is the apparent contradiction between:
1-Transition from laminar to turbulent is helpful for recovery as turbulent BL are more robust to address devaforable pressure gradient (AFAIU).
2-Despite this apparent robustness to resist separation, reattachment of the turbulent separation is not granted, or at least does not seem to be manageable like LSB & Transition Ramp ?

Cheers

 

A few comments for the uninitiated:

"Recovery" typically refers to the increase in pressure on a foil aft of its minimum value, as it approaches its trailing-edge value.

A flow with increasing pressure has more tendency to separate, and is said to have an unfavorable pressure gradient (not devaforable).

"Stratford Recovery" is a theoretical pressure distribution for the recovery region that corresponds to a flow that is just on the verge of separating. This was exploited in the design of Liebeck airfoils to find foil sections with extremely low drag & high lift/drag. These are described in AMO Smith's 1975 Wright Brothers paper, and many other places. (Not that it's important, but Bob Liebeck's desk was about 10' from mine in the Douglas Aircraft Aero Dept. around that time, so I got to eavesdrop on many of his discussions with AMO & others).

Turbulent flows are more resistant to separation than laminar flows, so transition can sometimes prevent separation. But that's only one factor. Whether the flow separates or not, and whether it reattaches or not depend on the details & can't be predicted without going through more complicated calculations.

Don't worry that your intuition isn't enough to predict it.

 

Thank you Doug

I guess you make reference to BL Parameters like the H parameters, I still need to dig here as I am not very comfortable with Momentum Thickness and others.
Cheers

 

The whole problem with keeping the flow attached is slowing the flow down. The velocity at the trailing edge will be near free stream velocity, so the flow has to decelerate from the maximum velocity to there. I liken it to driving in winter. You can go fast and the ride is smooth on ice, but you can only decelerate very gently without the tires letting go. Driving on packed snow has more resistance, but better traction so you can decelerate harder.

An adverse pressure gradient means the pressure is higher ahead of a blob of air than it is behind it, which is what slows the blob down. In the boundary layer, the pressure is essentially constant across the thickness of the layer. As the increasing pressure pushes back on the flow, the slower speed near the surface gets affected relatively more than the faster speeds in the outer part of the boundary layer. If the pressure is enough to actually push the flow backward next to the surface, there's a problem. You have the flow coming down the section meeting flow that is being pushed back, and both flows come to a halt. The air can't go forward because of the increasing pressure. It can't go backward because of the momentum of the oncoming flow. It can't go through the solid surface. That only leaves one direction the flow can go - away from the surface. That's when you get flow separation. It really is that simple.

The mean velocity profile of a turbulent boundary layer has a more full shape - the velocities near the surface are higher and the profile shape is more curved - than a laminar boundary layer. As you move away from the surface, the velocity in the turbulent boundary layer increases more rapidly than in a laminar boundary layer of the same thickness. The turbulent boundary layer has this kind of profile because of the mixing of higher velocities from higher in the boundary layer with the fluid that's been slowed by contact with the surface. This steeper gradient at the surface means there is more shear and more skin friction. But it's the higher velocities near the surface that make the turbulent boundary layer better able to withstand an adverse pressure gradient.

Technically, a Stratford pressure distribution has a constant margin from separation. If you set the margin to be low, then it is on the verge of separating everywhere. That is the shortest, steepest way to decelerate the flow. A Stratford pressure distribution is quite concave. It has a very rapid deceleration at the start, when the boundary layer is fresh and can withstand a lot of push-back. It flattens out as the boundary layer gets more tired.

From a design perspective, if the inviscid pressure distribution is steeper than a Stratford distribution, then there is no hope of having attached flow. When the flow separates, the pressure tends to be nearly constant. After all, the fluid is basically just flowing straight back and not following the curvature of the surface. One model of a laminar separation bubble has the pressure being constant from the point of separation until transition. Then the pressure increases like a Stratford pressure distribution. If the pressure intercepts the inviscid pressure distribution, reattachment occurs and you have a laminar separation bubble. If the pressure does not intercept the inviscid distribution, then there is no reattachment and you have pure laminar separation, as with a leading edge stall.

The purpose of a transition ramp is to apply a gentle adverse pressure gradient that is enough to either get the flow to transition through instabilities (a high Reynolds number approach) or to get the laminar boundary layer to separate (a low Reynolds number approach). If the flow separates, the gentle slope of the pressure distribution over the transition ramp ensures the post-transition pressure increase will intercept the inviscid design pressure of the ramp. This ensures the flow is attached and turbulent before it encounters the steep increase in pressure at the start of the recovery zone. Since the ramp is just ahead of the start of the pressure recovery, the turbulent boundary layer doesn't have much time to sap the boundary layer of its energy, so it starts the pressure recovery quite fresh. If transition doesn't occur by the end of the ramp, though, you have a situation like a ski jumper flying down the mountainside. The Stratford-like pressure increase may intercept the concave design pressure recovery and reattach, but it will be a long time in doing so, and there will be a long separation bubble with an attendant increase in drag. Not to mention the risk of a massive stall if the two curves never intersected.

One drawback of the transition ramp is it tends to be fixed at one location along the chord. If all you care about is high lift, then that may be OK. But if the section needs to be efficient at low lift coefficients, then the transition ramp may be giving up a longer run of laminar flow at low lift that could reduce the drag. That is an argument for a somewhat convex shape to the pressure distribution that results in transition being well back at low angles of attack, but moving smoothly forward as angle of attack increases. The concave shape to the pressure distribution after laminar separation and transition will very quickly intercept the convex inviscid design pressure distribution and the bubble will be short. The convex pressure recovery may give up some maximum lift, but it may be a better choice for all-round performance.

You can have a purely turbulent separation bubble, too. What you need is an outer flow that decelerates and then accelerates again before the final deceleration. The flow can separate at the first deceleration, but when the constant pressure associated with the separated flow intersects with the accelerating pressure distribution, the flow will reattach. I've used this strategy in fairing in the track of a self-tacking jib. There was no avoiding the flow separation due to the track and I couldn't put a fairing ahead of it because the sheet needed to be led forward and would sweep through the volume occupied by the fairing. So I put a hump behind the track. The flow accelerated over the hump and led to reattachment of the wake of the track. That resulted in a comparatively short turbulent separation bubble.

 

Great explanation Tom. You should write a book!

 

I think I just did.

 

Thank you very much Mr Speer for this "Private Lecture"
Your explanation regarding the use & design of bubble ramp is very interesting, and new for me, I will resume playing with XFOIL.
Long time ago on this forum I read : The art of wing section design is mostly :Boundary Layer management.
And your msg is a clear demonstration of this adage, at least to me.

Thks again for taking time to write such long & insightfull msg

Cheers

 

I should add that I've become a fan of flat section shape in the recovery region. This results in a mildly concave pressure distribution that promotes short laminar separation bubbles and stall that begins at the trailing edge and works its way forward as the angle of attack increases. The maximum lift may be less than for a Stratford recovery, but it is still pretty good, and better than a convex pressure recovery. It is also easy to build because the trailing edge and aft part of the section can be laid flat on a table, or constructed from flat panels.

 

One of the advantage of a hard wing vs a soft wing (teardrop mast + batten sail), as explained by Joseph Ozanne for BMWOR big tri, afair, was during tack & gybe:
The flow would separate and took time to reattach on the other tack with a teardrop mast and a full batten sail, while it would remain attached when a 2/3 elements hard wing sail changes side/tack.

The question it could raise is how the new AC75 mast+ double skin sail would work during tacks & Gybes:
1-A bit like a hard wing with minimum separation during the tack change
2-More like a teardrop mast + sail, with significant separation somewhere.
3-May be not that relevant because of the jib

Happy New Year

 

What would cause this difference?

 

The teardrop mast/ sail junction is not very streamlined when changing tack

 

I think it may be more that the sail luffs when it tacks and a hard wing does not.

One of the advantages of the hard wingsail is the crew can control when it pops through to the other tack. That's the purpose of what looks like a sheet to the clew of the main element on the AC72 wingsail.

With regard to the AC75, I think it would depend on whether or not the sail luffed, or if it held its shape and then snapped through to the shape for the other tack.

When I analyzed the jib section with mast and sail sections of the BOR90 trimaran, I didn't see much effect of the jib on the boundary layer characteristics of the mast and mainsail. The gap was too large. And jibs are used with hard wings just as they are with wingmast/sail combinations. I think the jib is as much about maneuvering, balance of the helm, and the spanwise load distribution as it is about influencing separation on the mast and mainsail.

 

Agreed, a jib makes tacking faster and more certain, among other advantages.

-Will (Dragonfly)