Stealing Industry Secrets? Deriving Foil Data with a cell phone

[mercury_ballistic]

As a challenge I thought it would be an interesting thing to test:

1. Learn about some of the basics of fluid dynamics and
2. Extract some real world data from the stuff I use on the water.

Read this knowing that I am not a CFD pro, I am simply someone with a desire to learn and tinker. Mistakes I have made, omissions, whatever please point them out so I can learn, just be cool about it.

Something that has troubled me for a while is most of the foils in this sport are sold with very few specifications. Area and aspect ratio is the only one they all seem to publish. But span, chord, cross section, performance specs, etc, are all usually omitted. Even area can vary from reality where the name of the foil and its true measurements can differ. I wished there was a way I could collect objective data on various foils as I don’t like it when I’m told a new product “feels” “loose” or “fast.” I want to see some numbers and how they compare to others. I admit that there is still a subjective feel and many other variables at play such as board, rider mass, mast position, fuse length, stabilizer size/angle, etc. Lets start with just foils since they are what we all care about most.

I spent a while debating how I could test foils for performance data. Towing? A Tank? The obvious answers were not practical or very expensive. Then I learned about phyphox. It is a free app that allows you to log sensor data from your smart phone. And most modern smart phones have a surprising amount of sensors. Accelerometers, 3 axis gyros, magnetic compass, GPS, temp, barometric pressure, etc. With these sensors I can now make my phone log data that can be used to derive pitch and velocity from gps. I set up an experiment in phyphox to log the data I needed, and taped my phone to the deck of my board behind my back foot. Pro tip, blue tape works way better in salt water than duct tape. I also used a screw in a footstrap hole to tie my phone case to the board should the tape fail.

I started logging and got in the water and went for a ride. I varied speed from barely on foil/about to stall, to moderately fast, and went back and forth to average our any current. After a short ride I came back to the beach, and stopped logging. That was it for the data collection. For reference I was using an Armstrong 880, 60 cm fuse and 180 stab. For a second run I used a 480, with the same fuse and stab.

When I got home I exported the data and used google colab (google’s version of a jupyter notebook) and did some vibe coding in python to clean up my datasets. I trimmed the first and last few minutes to omit me carrying the board and getting up on foil. That left me with data where I was on foil and moving. Colab/Gemini helped me derive pitch and velocity from my raw data. I then took that data and smoothed it a bit. There is a lot of little variations as you ride from chop, the thought was to clean up the signal here. I now had a chart of Pitch angle vs velocity. I did not have a good way at the beach to determine zero pitch so I decided to calibrate my data where my max velocity would be at 0 degrees pitch and everything else would be relative to that. So I know my numbers are not accurate, but the relative values from max to min should be correct, maybe?. Anyway, I got data that made some logical sense. When I’m barely on foil I am at a higher pitch angle which can also be construed as Angle of Attack. As I speed up, that angle gets smaller.

Here is the chart I generated from the raw data. You can see the comparison of minimum foil speed for each foil.

With this I can also approximate my stall speed/angle

With this I can now bring in the Lift Equation

Ceofficent of Lift is a dimensionless number representing how efficiently the wing shape and angle of attack generate lift.
ᵨ is water density (1025 kg/m3)
V is velocity in m/s
S is the wing area in m2

Since I know the Lift is equal to my mass plus that of my gear board, etc I can measure that with a scale and convert to Newtons.
For me at 75 kg with the board. Converted it is 735.75 Newtons

Now I can take all this and plot my Angle of Attack vs Lift Coefficient, and here it is:

So it seems like these values are within reasonable ranges, but now I need to go out and replicate the experiment and see if I can get similar numbers.

There are also issue in my methodology I want to address. I want to determine the true angle of attack, not set max speed to zero. For now I can at least be consistent in how I define zero by using the same mast, stab, fuse, and board.

I want to collect more data, longer sample periods. More data means I might be closer to the true values since there is a lot of chop and small motions when I am riding that add error.

I want to have data collects with multiple riders on the same system, so I can see how the plots compare for different ranges of lift.

I want to see how the data looks when I am pumping.Can I get anything of use from it?

Can I derive drag with the tools at hand? I think I can infer it by doing tests where I speed up and depower and coast to a fall while maintaining pitch angle.

So, after all that, thoughts on this? What am I doing wrong? Right?
Does this data have any value or is it just a curiosity?
If any brands want to give me gear, I will test it for you and report back.

[I_draw_boxes]

Very cool. I’m interested if those results are repeatable.

I’ve wondered if any brands are using testing rigs or strictly rider feedback.

A six axis f/t sensor as a plate between the mast and board would be key. Think bike power meter in more dimensions.

RTK GPS would resolve position to about 5 cm and do much better on height.

Between force / torque, GPS position and IMU data probably 95% of what needs to be known can be answered directly from the data just by riding. Objective measurement for most qualities and rider feedback on the subjective feel between similar foils or how to put together a range of complimentary designs. Would also be interesting to assess pumping technique to maximize distance per watt/h - could put a live readout on the front of the board and see the result of changes immediately, e.g. moving your feet back.

• Glide
• Watts to pump / projection
• Roll force required
• Pitchyness
• Low end
• Drag at all speeds

[Matt]

@mercury_ballistic

taped my phone to the deck of my board behind my back foot

I have been meaning to do this for years! I would want raw accelerometer data and pitch and yaw to validate some pump physics

[Maayan]

Just downloaded the app. Super interested to test it out and see what I find.

Regarding AOA; I doing know that actual 0* is important as long as you’re consistent with how you set it. Each wing is set with a specific AOA on the fuse and many boards also have rocker or deck concave which changes the effective AOA relative to your feet. To actually figure out the AOA you would need to use an inclinometer to compare the board deck to a selected tangent of the wing surface. T the bottom of the fuse would be more consistent though.

[mercury_ballistic]

Initially I mounted the phone and set the board upright with the foil flat on a level surface. I could tey applying this number to the raw values.

[mercury_ballistic]

I will be testing again to see if results can be replicated

[artsurf]

Your list is indeed extremely important, and these data for a straight line are the first basic need, but i think we need more data, especially about the characteristics in turns: When the tip comes out of the water in a turn, at what speeds and angles do you have a stall. And about stalls in all angles, how suddenly do they occur, can they be felt and corrected, do they let the board nosedive or is it a more gradual move downwards. How easy is it to carry speed through a turn, what types of turns does this foil like. How bad is a foil affected by bumpy water (superhigh aspect foils tend to make the ride more tiring if there is bumpy heavy chop). And what type of testing is needed to get these data from every foil. Maybe a round tank can be used to simulate what happens in turns, measuring lift at a standard radius with a device that rotates and measures everything (?) And what radius should that be: foils that are made for hefty waveriding are more happy with short turns and higher aspect foils have more glide in long fast turns. Maybe others can think of other aspects that are important. (?) How can we standardize these characteristics for true comparisons…

[Maarten88]

This is a fun exercise. As someone designing foils (as a hobby), I’m looking at it from the other direction and I think you are not yet stealing industry secrets but more like stating the obvious. Some observations:

• Defining the top speed angle as zero is fine, there is no need to somehow get a “real” zero angle. Many foils use twist (so the zero angle varies along the span) and more importantly they use sections that provide a different amount of lift at their theoretic zero angle. A cambered foil is mounted more downward to the fuse compared to a neutral or reflexed foil. I think most designers will set the mounting angle to zero degrees a top speed, this is what I do.
• You did measure AoA, it matches my designs. Foils having stall angles below 12 degrees are very sensitive to start.
• You are leaving the stabilizer out, which is too simplistic. I’m pretty sure the Armstrong 480 section does not have CLmax over 1.6. What you are probably seeing is the stabilizer (1/3 the surface of the 480) is helping takeoff. This effect is smaller for the 880, because the relative size of the stab is smaller. On the other and of the speed range, the effect is reversed: the stab creates significant downforce that the main wing needs to compensate.
• The really interesting thing to measure would be drag. As a designer I’m always looking for less drag, the best foil is the one that has the least drag inside some speed range and characteristics bracket. Measuring that would be really interesting, but I guess you’d need some towing setup.

[mercury_ballistic]

@Maarten88

• You are leaving the stabilizer out, which is too simplistic. I’m pretty sure the Armstrong 480 section does not have CLmax over 1.6. What you are probably seeing is the stabilizer (1/3 the surface of the 480) is helping takeoff. This effect is smaller for the 880, because the relative size of the stab is smaller. On the other and of the speed range, the effect is reversed: the stab creates significant downforce that the main wing needs to compensate.
• The really interesting thing to measure would be drag. As a designer I’m always looking for less drag, the best foil is the one that has the least drag inside some speed range and characteristics bracket. Measuring that would be really interesting, but I guess you’d need some towing setup.

I assumed the stab had influence on the performance of the foil so for now I just kept it the same. Maybe going fwd Ill use a smaller stab with a smaller foil but then I am introducing more variables.
I suppose I could use the same front foil and swap stabs and see how that changes the data.

I think I might be able to get some data on drag by depowering and tracking my deceleration while maintaining pitch. I have not yet figured how I would detect this in the data and isolate it for analysis.

For measuring data on turns, Im nowhere near the point of capturing turns with tips breaching in a reliable and repeatable way.

Another factor may be my wing providing lift to reduce my downforce when I am trying to flirt with stalling. At speed the force is more lateral but at slow speeds its much more vertical

[DrKnowNothing]

Great work!

So much to measure before we even talk about turning…
Just going straight, without pumping hasn’t been measured adequately.
Just covering speed range at varying heights is enough of a beginning challenge.

For measuring zero angle I propose:
Have 2 foot-shaped support bars, raised a bit off the level floor. Flip the entire board/foil setup upside down and support the board on the “feet” where your feet would be while riding.
Then shim the board/baseplate connection until either the mast is plumb, or the fuse is level (hopefully this would always occur at the same baseplate angle, for most manufacturers? Not Alpine from 10 years ago, etc…)

The next big instrumentation upgrades could be load cell between board and baseplate, and/or an ultrasonic sensor on the bottom of the board that measures height above water surface?