hard wing vs soft wing?

[DCockey]

There is a "concept" or CFD phenomenon, called : "Dissipation" I suggest you to investigate.

My understanding is JavaFoil and XFoil use panel codes to solve the inviscid flow field, and boundary layer type calculations to model the near field surface effects. Dissipation (aka numerical viscosity) is not an issue for panel codes such as those used by JavaFoil and XFoil. Panel codes only solve for the flow values along the boundaries, not through out the entire flow field. (The velocity and pressure can be calculated anywhere in the flow field after the boundary values are obtained.

Dissipation (aka numerical viscossity) occurs in finite difference / finite volume (and similar methods) which solve the field equations thoughout the entire flow field. My recollection is it generally is of interest for inviscid models, and is related to the stability of the numerical methods.

 

[Erwan]

Thank you very much for the clarification

With rookie's words can one says that :
Dissipation is not a natural fluid dynamic phenomenon,
but a "useful trick" for modelisation?

Cheers

 

[DCockey]

I would not call it a "trick". It is side effect of a stable numerical method. Some methods have dissapation introduced into the method explicitly. In other methods it arises from the details of the numerical algorithm.

 

[tspeer]

That Javafoil flowfield plot doesn't look reasonable to me. It appears to show the flow separating at the leading edge of both the main element and flap, yet following the overall contours of the section with a large and near constant displacement. I don't see how this would be physically possible.

Like I posted earlier, Javafoil is not capable of predicting the flow physics that lead to wake bursting, as with this configuration. Javafoil will give you an answer, but it will not be valid.

 

[DCockey]

That Javafoil flowfield plot doesn't look reasonable to me. It appears to show the flow separating at the leading edge of both the main element and flap, yet following the overall contours of the section with a large and near constant displacement. I don't see how this would be physically possible.

It looks to me like a illustration of the limitations of a potential flow plus boundary layer displacement model.

 

[Erwan]

Thanks for the explanation DCockey.

 

[Doug Halsey]

It looks to me like a illustration of the limitations of a potential flow plus boundary layer displacement model.

It doesn't help that the JavaFoil runs are at Reynolds # = 1E6, while the MSES runs are at 3E6.

But you're right - any calculations near CLmax are probably questionable for potential flow plus boundary layer methods. (That might also be said about the Euler plus boundary layer methods, but probably to a lesser degree.)

 

[DCockey]

(That might also be said about the Euler plus boundary layer methods, but probably to a lesser degree.)

For speeds where compressibility is not significant I would expect differences between an Euler plus boundary layer methods and a potential flow panel method plus boundary layer methods to be due to details of the bondary layer calculations , differences in modeling the interaction between boundary layer and flow field and flow field solver implementation, and not the difference in flow field equations. My understanding (perhaps significantly out of date) is Euler equation methods become preferred at transonic or near transonic speeds.

 

[tspeer]

For speeds where compressibility is not significant I would expect differences between an Euler plus boundary layer methods and a potential flow panel method plus boundary layer methods to be due to details of the bondary layer calculations , differences in modeling the interaction between boundary layer and flow field and flow field solver implementation, and not the difference in flow field equations. My understanding (perhaps significantly out of date) is Euler equation methods become preferred at transonic or near transonic speeds.

It depends on whether you're talking about single element sections or multi-element sections. For a single element section, you are right.

However, for a multi-element section, things can happen in the middle of the flowfield that are not captured by boundary element methods. As the wake from the main element of a wingsail passes by the flap, it experiences increasing pressure over most of the flap chord. This can cause the flow to back up in the main element wake, leading to wake bursting and separation. An Euler code like MSES can predict this, while a potential flow code like Javafoil cannot.

Here's an MSES prediction of such a case. MSES uses an adaptive streamline-fitted grid, so the grid lines show the flow direction.

You can see that the flow through the slot is attached to the flap lee side, while the wake is separated. All the telltales would be lying flat, but the wing would be stalled!

I can't put my hand on it just now, but I have run across a report of wake bursting being observed in the flight test of a STOL research airplane that used boundary layer control. Wake bursting really does occur - it's not just a numerical aberration.

 

[Doug Halsey]

...However, for a multi-element section, things can happen in the middle of the flowfield that are not captured by boundary element methods. As the wake from the main element of a wingsail passes by the flap, it experiences increasing pressure over most of the flap chord. This can cause the flow to back up in the main element wake, leading to wake bursting and separation. An Euler code like MSES can predict this, while a potential flow code like Javafoil cannot.
...

There's also the possibility that the wake from the forward element can merge with the flap's boundary layer, creating an even larger viscous region.

I don't know whether MSES has been applied to such cases, but I suspect that's a condition that would require Navier Stokes codes, not just Euler/boundary-layer.

 

[tspeer]

That case would only happen for small flap deflections, and MSES won't converge then. It would require an NS code. My approach has been to assume those cases lie in the linear range and extrapolate the lift, drag and moment from the cases I can get to converge.

 

[Doug Halsey]

That case would only happen for small flap deflections, and MSES won't converge then. It would require an NS code. My approach has been to assume those cases lie in the linear range and extrapolate the lift, drag and moment from the cases I can get to converge.

That's very interesting! In addition to being a problem for the code, wouldn't it also be an undesirable condition for the actual flow? Meaning you can't flatten the sail beyond a certain amount. Or is it not any worse than just having a single-element wing in those cases?

 

[tspeer]

I was talking about a rigid wingsail with slotted flap. And, while I couldn't analyze the section when flattened beyond about 10 degrees, it didn't stop the crew from flattening the wing when sailing!

I think it would be like a single element wing with some separation bubbles on each side. But I don't have any data to compare to establish that assumption.

 

[Erwan]

A rookie question regarding the separation bubbles
1-Are separation bubbles always "Laminar separation bubbles" or does some other form of bubbles exist ?
2-Is it correct to assume Laminar separation bubbles can only appear with a laminar boundary layer ?
so in case of high level of natural turbulences in the free flow: no laminar BL so no LSB can appear ?

Cheers

 

[DCockey]

Separation can occur when the boundary layer is either laminar or turbulent. By definition a laminar separation bubble has a laminar boundary layer upstream. Separated flow can either remain detached and merge into the turbulent wake, or reattach and form a separation "bubble". An everyday example of a separation bubble with a turbulent boundary layer upstream is the airflow at the base of most automobiles' windshields.

As noted in a previous post turbulent boundary layers are more resistant to separation than laminr boundary layers. In some situations it may be possible to eliminate a separation bubble by tripping the boundary layer upstream to force a transition from laminar to turbulent.