Any help appreciated: Airfoil design and using xfoil

Hi everyone I was just wondering if anyone could steer me in the right direction. I am currently helping with the design of a landyacht. We are using a wingsail and expect winds of around 20km/h.

I have chosen a NACA 9512 as the airfoil section due to its high lift and low drag in light winds... any other suggestions or advice?

I am using Xfoil to calculate the lift coefficients and then converting these to lift forces for various angles of attack to try and gauge whether enough force is going to be generated.
Firstly Im kind of unsure how to use Xfoil, all I have done so far is load up the NACA 9512 airfoil section then turn viscous fluid on then run the programe and recorded all the Cls it produces. Quite often it says that Xfoil does not converge, I then just push ! repeatidly until it does converge and gives me an answer, is this correct?

Also can anyone give me some relative lift forces of sail types. For example once I figure out how much force is gonna to be generated I would like to know if it is comparable to say a 2 meter high windsurfing sale..

Any help much appreciated.

 

Hi
NACA9512 is good a choice.
The procedure that you made for calculating lift force is correct, but you must repeat it more than 50 times. I seggest you to use polar curves for this foil. They are more correct and the calculating is more fast.http://www.nasg.com/afdb/list-polar-e.phtml.

Example:
4.7m2 windsurf sail at 20km/h and surfspeed 40km/h in direction perpendicular to race have a lift force about 170 N or 17kg. But this depend also from speed of surf
partenov@yahoo.com

 

Thanks for that.
Ok in regards to repeating 50 times does that mean for say AOA 3 degrees i run xfoil tru 50 times and record Cl each time making sure they arent varying at all, or am i missing something? Can u set it to run 50 times ?
Also im a little unsure about polar curves... could anyone breifly explain these what they do and how to use them? On that link u gave i could not find the polar curve for NACA 9512.

Thanks for your reply much appreciated.

 

I typically use 30 iterations and an angle of attack increment of 0.2 degrees. I've found that if a point doesn't converge by then, I need to do something special. This might include changing the angle of attack increment, or re-initializing the calculation on the other side of the troublesome spot. Sometimes one has to approach the bad spot from the opposite direction.

A section polar curve is just a plot of lift vs drag coefficients. It is produced by calculating the section characteristics at a variety of angles of attack (or lift coefficients). For example, the first figure, below, shows the NACA 9512 at -1.15 degrees angle of attack, producing a lift coefficient of 1.0, which is right in the middle of its minimum drag range. At the bottom is the airfoil shape with the boundary layer displacement thickness added to the upper and lower surfaces. This is the effective shape of the section because the outer flow has to go around the slower flow in the boundary layer, like cars moving to an inside lane to avoid slow traffic in the outer lane.

Above the section shape is a plot of the pressure coefficient, Cp, along the upper and lower surfaces. The dotted line shows what the Cp would look like if you ignored the effect of the boundary layer and allowed the flow to follow the actual airfoil contour. The solid lines show the Cp due to the flow following the effective contour defined by the boundary layer displacement thickness. The bumps you see in the Cp plot near 70% chord on the upper surface and 10% chord on the lower surface are due to laminar separation bubbles. Where the slope of the Cp plot starts to differ significantly from the slope of the dotted line indicates that the flow has experienced laminar separation and is no longer following the airfoil shape. But the flow rapidly transitions to turbulent flow and reattaches, indicated by the steep increase in Cp and the resumption of the trend indicated by the no-boundary layer case (dotted line). This is the mechanism by which the flow transitions from laminar to turbulent at this Reynolds number. The transition from laminar to turbulent flow, and the location of turbulent separation, is critical to determining the section's performance.

When you reproduce this calculation at a number of different angles of attack, you can produce the polar diagram in the second picture. This is actually a presentation of five different characteristics simultaneously. I've circled the points that correspond to the condition of the first figure. In the middle is a plot of lift coefficient, CL, vs angle of attack (alpha) - this is the line that slopes up to the right. It arcs over at the top, indicating a maximum lift coefficient of 1.85 at an angle of attack of 10 degrees.

The plot on the left shows drag coefficient on the horizontal axis and lift coefficient on the same vertical axis as used for the center plot. The drag coefficient of 0.00786 looks to be pretty much the minimum drag for this section with natural transition at this Reynolds number. I've drawn a red line on the plot that shows the maximum section lift/drag ratio occurs at a lift coefficient of 1.5 and 3.73 degrees angle of attack. There's a small dip in the drag polar in this region that is explained by the plot on the right.

The right-most plot shows the location of transition as a function of lift coefficient. The line running up and to the right shows where transition is located on the lower surface and the line running up and to the left shows the location of transition on the upper surface. You can see that at a lift coefficient of 1.0, transition occurs at 10% chord on the lower surface and 70% chord on the upper surface, corresponding to the two laminar separation bubbles that left their signatures on the pressure coefficient plot of the first picture.

In the middle lift range, as the angle of attack increases, the transition point moves aft on the lower surface, increasing the amount of laminar flow and decreasing the drag. But on the upper surface, the transition point moves forward as the angle of attack increases, increasing the drag. These two effects largely counter-balance each other, producing the wide flat drag characteristic of the drag polar shown on the left.

But above roughly 3 degrees angle of attack, the transition point on the lower surface moves rapidly to the trailing edge, and for all higher angles of attack the flow on the lower surface is fully laminar. This is why the drag polar dips right at Cl=1.5. The Cp plot for this condition is shown in the third picture. The laminar boundary layer on the bottom surface is very thin and the laminar separation bubble on the upper surface has moved forward to near 60% chord.

As the angle of attack is increased further, the laminar separation bubble moves farther forward and the increase in pressure to the trailing edge on the upper surface gets ever steeper. At 8 degrees angle of attack, shown in the last picture, the stagnation point has moved to the underside of the leading edge, causing a leading edge pressure spike as the flow moves around the tight corner. At the trailing edge, the pressure distribution is flat. This indicates that the flow has separated. This is turbulent separation this time, occuring at 90% chord, and there's no reattaching of the separated flow before it reaches the trailing edge. The airfoil has started to stall. You can see the boundary layer lifting away from the airfoil contour near the trailing edge, making it look as though the airfoil had deployed a spoiler flap. On the drag polar, the drag has started to shoot up because of the separated area and the low pressure on the aftward-facing trailing edge contour.

This is a long post, but I hope it helps you to tie everything together. The polar accumulation command, PACC, is used to start collecting the polar data into a disk file. The ASEQ command will compute a series of angles of attack, plotting the cp curves on top of each other and saving the data in the polar file. You can also execute the polar one point at a time with the ALFA or CL commands. PPLO will display the polar plot so you can see if you need to go further with the angle of attack to get the whole polar. Finally, the PACC command is used again to turn off saving the data to the polar file. The polar data file is an ASCII text file that you can read into Excel or other programs if you want to plot the data or pick out specific numbers.

 

Some Rhino Tools

I'm not sure if you are using Rhino or not for design of the landyacht, but this is a very interesting thread since I am working with XFoil more myself, and I assume someone using Rhino might find these very useful.

The enclosed scripts are tools I wrote for drawing foils directly from labeled foil files created with either XFoil, DesignFoil, or the UIUC database foils, and also to export NURBS curves of foil sections out for analysis by XFOIL or DesignFoil using Rhinoceros.

This version of the output script does a Cosine weighted distribution of points. My other script for export I posted in the Moth Foiler thread had only a straight distribution of points- this gives programs like XFOIL a really hard time particularly at the leading edge.

you might ask, why create a foil using Rhino? I don't usually, I use this feature to try to analyze shapes that I get from others, such as already designed hydrofoil sections I only have as a drawing or surface model.

I also use it for doing CFD for shapes that aren't good foils to start with, such as variations in leading edge radius, fairing techniques, etc. for say a flat plate foil such as used on the older Scows and many older sportboats like the Lightning, etc. These thin, typically small leading edge radius sections under the best of circumstances cause major headaches in XFOIL.

 

ncrit

Tom,

I found that back in 2004 you were discussing some XFOIL design issues specific to hydrofoils with Mark Drela. Should I use NCrit=3 for general submerged hydrofoil analysis?

Also in that thread was the duscussion of peoples general attitudes toward circular arc sections and/or small leading edge radii for surface piercing struts. I used a thin (8%) ProFoil section designed with Michael Seligs WWW code for our MM56CX hydrofoil testctraft's first set of foils with great success. I dod not modify it's leading edge- it's pretty fine already.

Then we ran a client's strut design using a large leading edge radius section, I think it was actually a NACA 6 series. I will say the Profoil foil with the smaller leading edge radius caused much less spray, and worked very well. The large leading edge had a profound effect on pressure distribution of the foil section below. The purpose of the study was to verify codes developed by CSULB and result from runs made with USAERO.

It no doubt does create earlier separation thereby inducing ventilation, if you are to get much incidence on it, but to redue spray less area exposed to total dynamic head right at the surface does work at high speed.

 

I'll leave it to Dr. Drela to answer the "should" part of the question. But ncrit=3 is what I use for hydrofoils.

Ogival sections work fine as long as you can stay within the operating range of the leading edge. It makes sense to me that they'd be suitable for struts on a powered hydrofoil. My interest is in sailing hydrofoils, and they need a much wider operating range. I suspect that much of the ventilation experienced by many sailing hydrofoil builders can be traced to their use of sharp leading edges. So no doubt my opinion with regard to ogival sections is heavily biased by the particular application I have in mind.

The only data I've seen on spray drag of different sections is Chapman, Richard B., "Spray Drag of Surface Piercing Struts," NUC TP 251, Naval Undersea Research and Development Center, 1971. It's on the IHS AMV CD#1. They found the drag of an ogival foil (max thickness @ 50% chord) could be approximated by:

Dspray = 0.011 q c t + 0.08 q t^2

And for a NACA 66-series section the drag was approximated by:

Dspray = 0.036 q c t - 0.03 q t^2.

for thickness ratios in the range of 12% to 20%.

For sections that were 12% thick, this would mean CDt = 0.17 for the ogival section and CDt = 0.27 for the NACA 66-012; 37% more. At the thick end of the range, 20% thickness, the ogive has CDt = 0.14 and the NACA 66-020 has CDt = 0.15, for 11% more spray drag. (CDt in this case being Drag nondimensionalized by q t^2) Hoerner suggests CDt = 0.24 for spray drag, dropping to 0.12 for sections above 20% thick, also based on tank tests of what might have been ogival sections. These numbers are in the same ballpark and provide a good cross-check of the NUC results.

To further put these numbers in perspective, assume the strut has the same chord as an aspect ratio 6 lifting foil, whose planform area is used as the reference area for nondimensionalizing the drag. The drag coefficients based on thickness then become the following drag coefficients based on planform area: CDspray = 0.00041 for 12% ogive, 0.00065 for NACA 66-012, 0.00093 for 20% ogive, and 0.00100 for the NACA 66-020. The difference is around 2 counts per strut.

So the tank data are consistent with your operational experience of less spray drag for sharp-edged sections. The question for the designer is, "What do I have to give up for a reduction in spray drag?" If you only have a range of +0.1 CL about your operating point without cavitating at the leading edge or ventilating the strut, vs +0.2 CL for a rounded leading edge (Hoerner), is that an acceptable trade? It probably is for a powered hydrofoil, but not for a sailing hydrofoil. Personally, I'll eat the two counts of drag if it reduces the chances of suddenly ventilating the foil.

 

Fantasic answer- I have both the IHS AMV CD's and will review Chapman's report. Agreed on all points, avoiding suddenly ventilating the foil is very desireable!

Our testcraft indeed did not suffer from ventilation or cavitation on the struts or foil when using sharper leading edge sections, whereas the sailing problem is definately much different. Lateral forces were only required for turning, and then strut loading was also mitigated by turn coordination via the control computer and joystick using the foil itself to supply most if not all of the neccessary centripetal force in the turn as with an aircraft.

 

Xflr5

I recommend you try XFLR5. See my posts in the Foiler Design Thread.

 

http://www.miswislandiceyacht.com/

Site with good info on some of the practical problems associated with developing a land yacht.