Auto-Flight
I think it might need a little more.
Perhaps.
My mental model of events goes like this: ama lifts, sensor comes out of the water, loading increases, sensor rotates, more of the blade comes out of the water, sensor rotates more, pops out entirely, power off.
Not how I mean for it to work at all. Model is; helmsman hardens his end of the sheet until the sensor/actuator comes “off the peg” and into the middle of its power zone. Helmsman stops adjusting. Presuming there’s enough wind, the ama comes up, and the load *decreases,* as less and less of the sensor/actuator is immersed at fixed angle of attack, and the ama comes back down. There is no “pops out entirely, power off,” due to the taper. There is also no marked increase in AoA, as leeway is set by other board(s), elsewhere n the boat
Tom Spear rightly points out that for this to happen in this manner, the auto-flight must not be the only source of leeway prevention; if it is, then your model holds, and the actuator’s load/sq inch will increase (as leeway increases, providing a constant lift from the (only) board, until it either stalls or pops out, and the power slams off, as you say.
This would not happen if the boat’s roll angle responded faster than the sensor, but the sensor has the lower moment of inertia and should respond faster. So either introduce some damping, which probably would make the sensor too slow to prevent capsize, or introduce some negative feedback instead of positive. I can think of two ways to introduce negative feedback, which could be used separately or in combination.
It is critical that the sensor/actuator respond faster than the boat’s roll angle. Optimal would be instantaneous, but “pretty darn quick” is OK too. (I’m trying to use the term “sensor/actuator” for the blade only to remind us all that it serves both functions; changes to one must not compromise the other)
Instead of the rotation axis being parallel to the longitudinal axis, give it 40 to 50 degrees toe in. Then rotation increases the angle of attack. Then two sensor blades would be needed, one for each tack. Giving the sensor blade some twist might be safer than tapering, to be sure the foil doesn’t stall when it needs to sheet in again.
This is clever, and Tom suggested similar—though he suggested tilting the other way, in order to *decrease* AoA as the board rises, to avoid a high-load stall. Neither scheme is necessary if the major sideforce generation is provided by some other source—rudder boards in the main hull for instance.
OTOH, if the auto-flight is the sole source of leeway prevention, then your model holds, the increasing leeway will be increasing the AoA, and you wouldn’t want more, or the blade will stall. Twist instead of or in addition to taper is a good idea and would certainly help, but I chose the taper only as it is so much simpler—and you only need the one.
Move the rotation axis inboard, so that rotation of the sensor immerses the blade more. Or use the sensor ama.
This could be done to slow the response if desired (as above, it is NOT desired, I believe), but as the sensor/actuator moves farther from the ama, two things happen; the system is de-sensitized and de-powered (there is less change in sensor/actuator force per distance unit that the ama rises). Second, the sensor/actuator is working in water farther distant from the ama, so mightn’t accurately sense to what’s happening there—think waves.
Thanks for the insight!
Dave
Model is; helmsman hardens his end of the sheet until the sensor/actuator comes “off the peg” and into the middle of its power zone. Helmsman stops adjusting. Presuming there’s enough wind, the ama comes up, and the load *decreases,* as less and less of the sensor/actuator is immersed at fixed angle of attack, and the ama comes back down. [snip] There is also no marked increase in AoA, as leeway is set by other board(s), elsewhere n the boat
That is the scenario I assume. Because leeway remains constant, the load on the sensor blade decreases as the ama rises. What stops it from rotating entirely out of the water? I see only two sources of negative feedback. One is the ama falling again, and we agree that is a slower response than the roll angle of the sensor blade, so the roll angle of the boat can’t supply the negative feedback to keep the sensor blade in the water. The other source (which I didn’t think of last time) is the angle of attack of the sail. That has less rotational inertia than the whole boat and should respond a bit quicker, but I would prefer to have something that responds to the attitude of the boat, which is what is supposed to be regulated. After all, your ama might rise because a gust has increased wind speed and angle of attack. Then you can’t rely on the sail losing power to make sure the sensor blade doesn’t pop out.
Move the rotation axis inboard, so that rotation of the sensor immerses the blade more. Or use the sensor ama.
This could be done to slow the response if desired (as above, it is NOT desired, I believe), but as the sensor/actuator moves farther from the ama, two things happen; the system is de-sensitized and de-powered (there is less change in sensor/actuator force per distance unit that the ama rises).
I meant to move only the axis of rotation, not the sensor blade itself. Think vector foil, but the sensor blade can be vertical, because the downforce is supplied by the sheet. Though if you want the sensor to work when there is little or quite variable side force (broad reach), the sensor ama would be a better bet.
Second, the sensor/actuator is working in water farther distant from the ama
It would remain right where you drew it. Do you like that better?
I grant it takes a bit to get one’s head around just what this gadget actually does: It has nothing to do with controlling the course of the boat, and it contributes nothing to lifting the ama against gravity. It simply and only defines the upper edge of the performance window. The helm maintains complete control of all other aspects within that window at all times. It’s a bit like a self-braking car that slows itself if the road ahead is obstructed. The driver can do whatever he pleases at all times, except to direct the car to crash into something. Similarly, the auto-flight equipped helmsman can direct the boat to do anything at any time, except to capsize via over-powering the main.
Because leeway remains constant, the load on the sensor blade decreases as the ama rises.
This is precisely what is wanted. The ama rises due to sheet tension, set by the helmsman and allowed by the gadget increasing drive form the sail. As the ama rises, the immersed blade area decreases, blade drive decreases and the blade rotates (it is “preloaded” by the mainsheet). The sheet is eased, the sail is depowered and the ama comes down. As the ama comes down the blade is more fully immersed, its drive increases and the blade rotates. This causes the sheet to be hardened and the sail power increases. The ama rises…
Recall that I recommend building the blade oversize in early iterations so that the main never overpowers the gadget.
You understand that AoA is fixed, right? Adjusting it may be necessary but only as a pre-set. I’m presuming that an asym section and the boat’s leeway will give it sufficient working AoA that it may be set and left at zero at all times. I may be wrong, but make the point so you don’t misunderstand what I mean by “the blade rotates.”
What stops it from rotating entirely out of the water?
Simple answer, nothing at all. The system does not require constant immersion of the blade to operate correctly. The gadget shouldn’t pop out as it has been approaching “hard over” thus depowering the mainsail as it has been approaching popping out, but should it pop out completely, it simply continues its travel to hard-over, additionally depowering the main. I fear I have been confusing many; I have been calling this phenomenon “positive” feedback when it is of course negative feedback—self-limiting. Apologies.
Are you perhaps exploring the potential effects of rotational inertia? Ie: what if the ama comes up and its inertia is such that, even with dumped mainsheet and fully depowered main, angular momentum takes the boat over the top? I don’t want to guess, so will await your clarification before responding.
Dave
I meant to move only the axis of rotation, not the sensor blade itself. Think vector foil, but the sensor blade can be vertical, because the downforce is supplied by the sheet.
OK, I understand the setup, but not the advantage. Recall that the only thing we want from this gadget is varying tension and sufficient throw on the mainsheet. I’m not understanding your phrase “downforce is supplied by the mainsheet.” Maybe a sketch?
Though if you want the sensor to work when there is little or quite variable side force (broad reach), the sensor ama would be a better bet.
Now this is pretty cool. It is hard to fly a hull off-wind, let alone control that flight. Forces are minimal and variable, as you say. A drag-type device, as opposed to a lift-type, might more reliably develop its drag, it’s true. The device could then be a sensor ama (like the one on Harryproa here: https://au.groups.yahoo.com/neo/groups/harryproa/files scroll down to sheet release.jpg) or the simple bendy wand/flapper that moths use, and you will get more reliable control in hairy conditions. The downside is now we’re using drag for primary control rather than lift. The blade might have a L/D of 10, the ama/flapper only 1, so potentially 10 times more drag created. Still, if the system were sufficiently optimized… Maybe with a Balestron rig?
A fast boat will draw its apparent wind forward. Off-wind apparent courses at mainsail forces high enough to fly a hull are not common, worse, such courses have potential pitchpole issues. I’d rather not use ANY automatic controls under such conditions, but that’s just me. 😉
Dave
I think it might need a little more.
Perhaps.
My mental model of events goes like this: ama lifts, sensor comes out of the water, loading increases, sensor rotates, more of the blade comes out of the water, sensor rotates more, pops out entirely, power off.
Not how I mean for it to work at all. Model is; helmsman hardens his end of the sheet until the sensor/actuator comes “off the peg” and into the middle of its power zone. Helmsman stops adjusting. Presuming there’s enough wind, the ama comes up, and the load *decreases,* as less and less of the sensor/actuator is immersed at fixed angle of attack, and the ama comes back down. There is no “pops out entirely, power off,” due to the taper. There is also no marked increase in AoA, as leeway is set by other board(s), elsewhere n the boat
Robert is correct. The vertical paddle will act only as an on/off device. Whether it is tapered on not, it will simply hold the sheet on until (due to heeling) the sheet load becomes greater than the load generated by the paddle, at which point the paddle will ‘let go’ and rotate to it’s full extent.
I would like to be able to fly the ama at a relatively constant height, and to do that you need horizontal separation between the hinge point and the sensor. This allows the sensor to track the water surface. The sensor ama setup provides this.
I understand why it would be nice to use a paddle or foil, firstly it will suffer less from wave impact drag, secondly you can derive some useful force from it. For this reason I earlier suggested using a hinged ‘vector fin’ type of device. The object here is to replace the ama part of the sensor ama setup with a ‘vector fin’ or hook foil. The foil is not used to prevent heeling (as proposed by Fritz Roth), but is used to track the water surface. This would reduce the wave impact drag that might be imposed on a sensor ama, and should provide some useful downforce, whereas otherwise you would have to rely solely on the weight of the sensor ama.
No, wait! That’s not correct. What was I thinking?. The sheet load would actually tend to force the sensor ama to sink, so the ‘vector fin’ would not need to provide any downforce, only enough lift to counter the sheet load. So you could probably just replace the sensor ama with a buoyant foil.
Mal.
I disagree with your hypothesis about the sensor just letting go of the sheet.
The ama rises, the immersed sensor area decreases, so the lift of the foil decreses. This means the sensor rotates, and starts letting out the main a bit. It would only be an on/off mechanism if the force from the foil would be an on/off thing as well, but it’s not, in the operating region there’s a linear relationship, or with a tapered blade a quadratic relationship, between the heel angle and the mainsheet tension.
Does anyone actually have a small proa on hand who’s willing to just build the darned thing in an afternoon or two? Or are all of us just armchair designers for the time being? :D Comon guys! Who wants to try and build the first autoflight mechanism in multihull history? 😊 If I had a boat on hand I’d jump at the opportunity! 😊
Marco
Does anyone actually have a small proa on hand who’s willing to just build the darned thing in an afternoon or two?
I have a scale-model that i could try a auto-flight device on.
I need some rough numbers on appropriate size of the different parts. I have never built anything like this, so i have absolutely no clue how to size the sensing foil. Should i start with sensing foil 10% of the sail-area?
Cheers,
Johannes
Fantastic, I love your get up and do it attitude! 😉
I would recommend going for the variant where the sensing foil produces its lift to windward, since the boat might sideslip quite a bit (no other boards), which would interfere with the working of the mechanism. If you want to sensor to lift to windward, you still need something above the ama to route the mainsheet around.
I’d recommend making the attachment points for the mainsheet to the boom, and for the mainsheet to the spar which is attached to the sensor, adjustable so you can play with them.
I’m working on getting you an estimate for the the size of the board…
I would recommend going for the variant where the sensing foil produces its lift to windward, since the boat might sideslip quite a bit (no other boards), which would interfere with the working of the mechanism. If you want to sensor to lift to windward, you still need something above the ama to route the mainsheet around.
I will try the most simple version first. When working with scale-models there is more friction compared to the forces in the mechanism. Every bend in the line connecting the sensing foil with the boom adds huge amounts of friction.I will have to use some kind of rubber-band to hold the foil down in the water (its wood and it will try to float).
The side slip is not excessive. The deep-v hull has a good “grip” in the water, and will resist being pulled sideways. It is very apparent when trying to pull the proa back to the beach. I have tried a lot of different hull-shapes and even some pure foils and the deep-v has a lot of lateral area and a very sharp keel-line (creating huge vortices of the crossflow) resisting side slip.
I am adding a platform out to lee, to be able to add some weight to lee for easier ama-flying in light winds.
Cheers,
Johannes
I found out that Johnny Cash likes auto-flight proas too, so with his help we found a suitable piece of plywood, gave it a very scientific shape (this kind of looks right to me - give me the belt-sander Johnny!) and to the tune of “I got Stripes” i shaped it to what you see in the pictures below.
I have no idea if this is anywhere near right size or shape, and i will probably build several foils before i get stable flight. I do have to start somewhere, and i had this piece of plywood laying around…
Next foil will probably be made of aluminum.
I’m working on getting you an estimate for the the size of the board…
Thanks Manik!!
Cheers,
Johannes
Hey Johannes,
I’ve attached a photo of the formula and it’s derivation (if anyone wants to check it) for the relationship that needs to hold between the lengths. Your point about the friction in the model is a very good one, so I’d go for ‘high’ loads in the sheet and mechanism, by placing the sheetpoint near the end of the boom. Furthermore I’d recommend ensuring that the mechanism’s range of travel covers a reasonable amount of the sheeting spectrum, the range of 30° to 60° between the boom and the centerline would be good for close-hauled sailing. You’ll have to select some operating range of angles for the sensor foil as well, say +/-30°. From that you can figure out your L_rod—how long does the rod have to be so that when it travels from +30° to -30°, the sheet moves enough to get the boom from 30° off-center to 60° off-center?
For the velocities I’d guestimate a model like that might make something like 0.25 - 0.50x the windspeed to windward, so you can plug 0.25^2 = 0.0625 in for v_sail^2/v_foil^2. Then you just need to set your height and you are all good to go. Note that there’s a slight mistake in the diagram there: L_height needs to be the height from the pivot point to the center of effort of the sensor foil, and not the foil’s waterline. The CE of the sensor foil will be something around about 0.4 * the depth of the foil beneath the waterline.
I don’t know how accurate this formula will be in practice; even if the derivation is correct, the real values are probably going to be quite different due to some effects not considered, or one of the many approximations made here. I hope it at least provides a reasonable point to start though.
What sort of a ratio do you get if you work out and plug in the values Johannes?
Cheers,
Marco
With the 800-fold difference in density, the ratio of sail area to sensor area is bound to be pretty enormous, so something like 10% of the sail area as the foil area is probably way too big.
Robert is correct.
Need to draw this and play with it a bit; synapses are slowing. Still confident you’re both wrong, but that the two of you not only disagree with me, but agree with each other, is compelling. 😊
Couldn’t your version be pretty simply tested by altering the “simple paddle” sensor/actuator to incorporate an “S-bend” (though with sharp corners) like this:
__| rather than this: |
| |
(set your emailer for constant-width font)
Dave
I found out that Johnny Cash likes auto-flight proas too, so with his help we found a suitable piece of plywood, gave it a very scientific shape (this kind of looks right to me - give me the belt-sander Johnny!) and to the tune of “I got Stripes” i shaped it to what you see in the pictures below.
I think you may need a longer throw and less force on the mainsheet, meaning a taller “top” to the blade. Soon, however, you will know. 😉
I agree with the remainder of your comments. Though Marco is correct regarding reversing the mechanism, so are you regarding friction—and simplicity—in a model. Regarding my last post before this one, it may be very easy to test Mal’s and Robert’s concept as well.
Thanks for the hands-on testing!
Dave