# Flowfield calculations in javafoil

Discussion in 'Hydrodynamics and Aerodynamics' started by johan gronvall, Dec 30, 2014.

1. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

I'm conducting a 2D parameter study in javafoil to understand the key parameters of sails affecting lift and drag. I test with 2 simple NACA profiles, where I can set the very thin foil thickness and vary nose radius to mimic a main with a mast in front.

Firts I studied jib and main separately and secondly I put them together to understand if and how there was a synergy of the combination.

When testing the combination I notice that the sail to a great extent affects the the flowfield also upstream. The angle of attack of the flow in the vicinity of the boat differ significantly from the far field - in the range of 20 degrees.

My questions: Is this natural? Is there a modeling mistake or does the invicid flow give such results in general?

#### Attached Files:

• ###### flowfield jib + main.jpg
File size:
130.4 KB
Views:
817
2. Joined: May 2008
Posts: 2,598
Likes: 367, Points: 83, Legacy Rep: 1082
Location: Beaufort, SC and H'ville, NC

### philSweetSenior Member

Yes, this is normal. The ideal fluid flowfields around several simple planar geometries can be calculated directly. As long as you are working with a system of linear equations, the solutions to simple cases can be superimposed to model much more complicated ones. The updraft and downdraft are fine in the image.

There is quite a big difference between the 3D solution and the 2D for typical low aspect ratio sails. The link below shows some 3D models on one of our member's (Mikko's) website. There is also a big difference between real world and ideal flow in the realm where sails operate.

All bodies that experience an induced lift from an ideal fluid flowing around them (with Kutta condition applied) will perturb the velocity field of the flow. The perturbation will be a circulation about the body. The velocity of this perturbation decays as the distance from the body increases. So you have the original flow field, then a circulation superimposed on this, which is strongest at the body's surface and goes to zero at infinity in all directions.

3. Joined: Oct 2009
Posts: 5,212
Likes: 611, Points: 113, Legacy Rep: 1485
Location: Midcoast Maine

### DCockeySenior Member

The flow with two sails will differ from the superposition of the flow around the individual sails. While the equations and boundary equations are linear, the effect of perturbations of the velocity field by each sail on the zero normal velocity condition of the other sail has to be taken into account.

The circulation is part of the perturbation of the flow field but it is not the entire perturbation.

4. Joined: May 2006
Posts: 571
Likes: 79, Points: 28, Legacy Rep: 258
Location: Finland

### Mikko BrummerSenior Member

You can find a 3D example at http://www.wb-sails.fi/Portals/209338/news/99_1_AeroShape/Aero.htm. This is inviscid, too, analysed with the vortex lattice method.

5. Joined: Feb 2002
Posts: 2,319
Likes: 303, Points: 83, Legacy Rep: 1673
Location: Port Gamble, Washington, USA

### tspeerSenior Member

Javafoil is good for what it does, but you need to be aware of its limitations. First, it is only calculating the two-dimensional flow, which corresponds to the flow about an infinitely high rig. When a lifting surface has a moderate to high aspect ratio, you can approximate the three-dimensional flow in two planes. The first plane is at right angles to the mast, and this is what Javaflow calculates.

The second plane is at right angles to the apparent wind, and is located far behind the trailing edge. This plane is used to calculate the flow in the wake of the rig, and handles the fact that the rig actually has a finite span. Javafoil does not handle this aspect of the problem at all. You need to get the results from both planes in order to estimate the drag of the three-dimensional rig.

The second big limitation of Javafoil has to with its boundary layer calculation. It uses potential flow, which is about the simplest possible approximation of the physics, to represent most of the flow field, and it uses a functional approximation (call an integral method) to calculate the boundary layer's skin friction. The boundary layer approximation is only good for fully attached flow. So it really can't tell you what the maximum lift will be. All the telltales have to be laying flat for Javafoil's approximation of the physics to be valid.

These approximations are pretty decent for something that is high aspect ratio and operates at low to medium angles of attack, say < 10 deg, like a glider's wing. But most sail rigs are low enough in aspect ratio that the separation of the 3D problem into two 2D problems is starting to be kind of iffy. And the kinds of angles you're trying to use will probably have some separated flow associated with them.

There are also some flow effects that Javafoil can't predict at all because they happen in the middle of the flowfield where Javafoil thinks everything is nice and smooth.

So it's a great way to get a qualitative understanding of what's going on, and to get a feel for some of the magnitudes. But it's going to be way optimistic with regard to maximum lift, and the drag will be too low. You're not going to see any of the separation in the flowfield that will actually exist for a high-lift configuration like a sloop rig.

If you could actually achieve the lift that Javafoil predicts, then yes, you'd see those kinds of flow angles. But you're probably not going to get that much lift.

6. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

Thanks for help and additional reflexions

Hi all and thanks for your support.

I have thought of going to 3D, but I hesitate because of the the effort of modelling the geometry. Are there modeling tools within XFLR5 and SAIL7? I saw that there are addins for lofting in google sketchup - have anyone tried it and is the geometyry useful for the calculation tools?

/J

7. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

More thoughts

Ironically I just came across the articles published by Arvel Gentry back in the 70ies when he did about the same numerical experiments http://arvelgentry.jimdo.com/articles/ to understand jib - main interaction.

Maybe you can also comment on a couple of reflexions and observations from my little parameter study;

1. A jib alone does not behave as a assymetrical wing foil with a straight CL curve from CL0 to stall at 10-15 deg. The CL curve for jib seems to have a CL0, a short straight curve and quite an early stall at 3-5 deg

2. If you put a circular object in front of the sail like a main - It doesn't either behave like the assy wing profile. The main CL curve tend to be almost zero to about 10 deg, increases rapidly to stall at about 15 deg.

3. If you add thickness to the sail, CL increases and it starts to behave like the assy wing profile with quite high CLmax (GT 2)...;-)

4. Combining jib and main, it was hard to find sheet angles where jib and main lift curves cooperated (by turning on/off integration of jib and main respectively you could figure out jib and main contribution respectively. Typically you get two CL peaks; jib CL peak at low alfa and main at higher) as changes in sheet angles affected the perturbation/upwash significantly. Finally I came to a tighter sheetangle for the main and a more open for the jib than expected - from analysining the sails stand alone.

5. Main twist is most often explained by wind shear effect, but my reflection is that the difference in flowfiled perturbation/upwash in the jib - main region versus a main alone with less cord length contributes at least in the same magnitude as the wind shear effects to main twist

8. Joined: May 2006
Posts: 571
Likes: 79, Points: 28, Legacy Rep: 258
Location: Finland

### Mikko BrummerSenior Member

Actually, Arvel's experiments, as groundbreaking as they were, were done with an anlog plotter, they were not numerical. They suffer from the same symptom as Javafoil, the 2D-only, as explained by Tom above, so not too too many conclusions about sail trim should be made based on them.

1. A single sail, be it a jib or main, does behave similar to a more general cambered airfoil (with thickness), only there's always separation present on either side at the luff - Javafoil just doesn't handle this well, and also 3D effects are important.

2. Same applies as above, the mast would not change this much.

3. Actually, thickness probably decreases the power (CL) of the sail, if the sail is trimmed at a correct camber for the wind given angle (ideal angle of attack).

4. The interaction is clearly beneficial, especially at higher angles of attack (apparent wind angle), compared to isolated sails.

5. Agreed.

9. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

Hi again,
Maybe I overinterpreted Arvels' text as he referred to the possibilities with computers back in 73 ;-)

Never the less, I got the results summarized in the attached picture. I used the same NACA (12400) profile ensuring lee curve was the same, but varying thickness and nose radius. I put the Clark Y like profile in as a reference (which have what I mean typical CL curve behavior - CL0 + 2PI*ALFA until stall region at 10-15 deg).

Would going to 3D provide different conclusions?

Has someone made similar parameter studies using 3D?

I really like the stuff on the WB sail homepage, but as I understand the SW is proprietary. Sail7, an XFLR5 addon - any experience? How do you create the geometry?

#### Attached Files:

• ###### wing section study.jpg
File size:
42.7 KB
Views:
452
10. Joined: May 2004
Posts: 5,372
Likes: 256, Points: 93, Legacy Rep: 3380
Location: Italy (Garda Lake) and Croatia (Istria)

### daiquiriEngineering and Design

There's something wrong with the curve of the bottom-right airfoil. It is not possible that it gives a lift coefficient equal to zero up to 10-12°.
It should have a positive Cl at 0° AoA, and then rise with as the AoA increases, just as the others do.
Cheers

11. Joined: May 2006
Posts: 571
Likes: 79, Points: 28, Legacy Rep: 258
Location: Finland

### Mikko BrummerSenior Member

Also, are you sure the jib profile is the same as the lee (upper) surface of the "plane", or is it rather the camberline (mid line between upper & lower surface)? If it was the lee surface, it should give more lift than the profile "plane" with thickness. Else the jib lift curve is typical for a (2D) thin profile: at less than -1 or = deg there is less lift due to a separation bubble on the inside (windward side). The kink in the curve corresponds to the ideal angle of attack (where telltails would stream nicely on both sides of the luff), and the next hmp corresponds to the occasion when a separation bubble bursts on the leeward side.

However, in 3D the the separation bubble is stretched into a separation "tube" or "vortex": The sluggish air on the surface of the sail is sucked into the "drain" formed by the tip vortex up the leech, and the effect of the separation is much more complex. I will try to post you an example when I find the time.

I haven't tried SAIL7, but it is probably very similar to my own MacSail VLM-code. I don't know if it has any separation prediction like Macsail, but even so it would suffer from the inability of potential flow codes to predict separation. In Macsail I checked for separation in chordwise profiles along the span, and if detected separation I adjusted the chordwise pressure accordingly. This is, however, a very crude method and does not represent 3D separation well with the low aspect ratios of sails. You can look at our simulation in the you tube to get the picture.

Sorry, I meant "tjock stor" with what I called "plane" in the above.

Last edited: Jan 8, 2015
12. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

Hi again,
Yes I had messed up the effects of camber line and thickness....

I re-made the foils being cautious of the lee curve. It showed to be difficult to obtain the same lee curves, but I managed to get all the lee curves through the same "max" camber point at 40% cord, se the attached picture.

I also set the nose radius for the "wing" profile the same as for the "thick" profile.

The same patterns remains, but the "main" profile still behaves a bit strange, maybe caused by separation behind the "mast" at alfa < 10-15 deg, maybe the very high Cd is related to this?

In one way I feel that the "main" profile results makes sense; When you tack without a jib, it takes quite some turning until you gain power in the sail. My interpretation is that a main with a mast does not provide lift until higher AoA. What do you think?

Regarding the high CL of the "thick" profile - doesn't that explain why some uses fixed/rigid wing sails, e.g. AC?

#### Attached Files:

• ###### sailing angles_3.jpg
File size:
118.6 KB
Views:
582
13. Joined: Feb 2002
Posts: 2,319
Likes: 303, Points: 83, Legacy Rep: 1673
Location: Port Gamble, Washington, USA

### tspeerSenior Member

It looks like you've used the same coordinates for both sides of the sail. This may be causing some numerical problems in Javafoil. I suggest adding a small amount of thickness to the sail, say half a percent. That will still give you a thin sail, but you won't have singularities at exactly the same place.

14. Joined: Feb 2002
Posts: 2,319
Likes: 303, Points: 83, Legacy Rep: 1673
Location: Port Gamble, Washington, USA

### tspeerSenior Member

Here is how you can enforce the proper symmetry for a wingmast:
1) Generate the starting coordinates for the lee side and leading edge.
2) Draw a straight line between the trailing edge of the mast and a point on the leading edge such that the line meets the leading edge along a local perpendicular.
3) Rotate the section about the leading edge/mast chord intersection until the mast chord is horizontal.
4) Reflect the lee side mast contour to the windward side.
5) Rotate the section back to its original orientation
6) Add a small amount of thickness to the sail portion

This process will ensure that you have a symmetrical mast and a numerically tractable sail. It can be implemented in a spreadsheet. Once you have it set up, you can load coordinates from an airfoil design code, apply the symmetry modification, and save it as new coordinates to analyze in the airfoil code. You will need to do this step after each design iteration, because the airfoil design codes don't have a provision for enforcing symmetry over just part of a section.

Another way to get the same effect would be to modify the camber line so it is straight between the leading edge and mast trailing edge.

15. Joined: Dec 2014
Posts: 7
Likes: 0, Points: 0, Legacy Rep: 10
Location: Sweden

### johan gronvallJunior Member

thickness

Hi,
I used 0.1 % thickness in the example shown.

I also tested to vary thickness. The same pattern remains, but as thickness increases, the CL peak moves to lower AoA, eventually you get CL0 > 0 and a linear CL- AoA curve until stall....

#### Attached Files:

• ###### main var t.jpg
File size:
56.3 KB
Views:
438
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.