Watch me build my autonomous solar 20-footer...

[gonzo]

Get some lead blocks, install them in the model where the batteries should go, put panels of proportional weight to the solar panels, and float it in your bathtub. You'll see really fast if it is self righting or not.

[Rick Willoughby]

Originally Posted by amaurer

....
I wrote a script that simulates the hull with voxels (volumetric "pixels") and rolls it through various angles while calculating the resulting waterline. The deck is shown in red voxels and the keel in blue. Green voxels fill the area that lies beneath the waterline:


So, reading that, by 5 degrees of roll I have 100ftlbs of restoring moment... which to my untrained eye seems like a lot... but what do I know. Moreover since all the moments are positive, the design is self righting from any angle.

Andrew
Your little script is producing a reasonable result. The only aspect I question is the kick up in righting moment between 160 and 180 degrees. The moment should be steadily reducing from the peak to 180 degrees.

The boat is very stable. So long as it has positive moment up to about 160 degrees it will be self-righting. It will have very strong tendency to roll upright from the inverted position.

Any boat this small on the open ocean will get rolled around and surely turn turtle in the worst conditions. The likely worst thing will be that you rip off a solar panel. Could be serious if you lose all of them of course or the wiring is damaged.

As far as Freeship goes you should be able to produce the cross curve for design displacement similar to what I have attached. This goes a long way toward determining the righting moment. You will note the similarity to what you have already produced.

Rick W

[amaurer]

Originally Posted by Ad Hoc

If you can 'roughly' model the hull, that you think looks about right, print off a set of hydrostatics. Check this against your own little macro, the two should be very close...if not, there is an error. Keep doing until the two values are very close.

Once you ahve done that, use metric!!...sorry don't understand imperial units.

The CoG...is that a guess or a calculation?

Once you ahve the CoG, plug it into the program, and run from 0 to 90 degrees. Then post the output on here.

One step at a time.

The CoG I posted was just a guess, but I do have realistic numbers from my CAD model: 13.5" to the CG (from the lowest point), 840lbs total.

I doodled up the model into Delftship and ran the hydrostatic report - it turns out my version can't do the neat crosscurves function that Rick's can, I guess thats only for Pros - and got transverse metacentric height = 1.51ft. Is there another package that can plot the roll data?

[As for my units, I prefer metric as well, but everything from my tape measure to the ShopBot use imperial. C'est la vie, until I own my own... everything... sigh.]

Your little script is producing a reasonable result. The only aspect I question is the kick up in righting moment between 160 and 180 degrees. The moment should be steadily reducing from the peak to 180 degrees.

The boat is very stable. So long as it has positive moment up to about 160 degrees it will be self-righting. It will have very strong tendency to roll upright from the inverted position.

Any boat this small on the open ocean will get rolled around and surely turn turtle in the worst conditions. The likely worst thing will be that you rip off a solar panel. Could be serious if you lose all of them of course or the wiring is damaged.

As far as Freeship goes you should be able to produce the cross curve for design displacement similar to what I have attached. This goes a long way toward determining the righting moment. You will note the similarity to what you have already produced.

Please just ignore the data near/above 170 deg of roll in my plots, my data processing is to blame. Generating the plot of restoring moment means taking a numerical derivative of the CG height - since my step size is still fairly large, differentiation is "noisy", and I use a smoothing function to keep the moment readable, but obviously this has trouble near the ends of the dataset. I'll run it again with a finer step size near the edges of the set and it should help.

I'm certainly not trying to avoid turtling all together, I think thats unavoidable, but I do want to be stable enough that the damping provided by viscous friction is enough to keep roll oscillation from building up in "normal" seas. I'm sure you guys have a word for this (stability margin, perhaps?) but I don't know it.

The panels will always be a weak link if the weather is bad enough to risk turtling. Even assuming my mounts are robust enough to stand the forces on them when getting hammered, the cover glass will always be at risk. It'll be interesting to see how they hold up... how exciting!

[Ad Hoc]

amaurer
"..but I do have realistic numbers from my CAD model.."
With respect, this is also a guess. A guess by the computer. What I should have said is, have you done a proper weight and centres estimate based upon the calculated or known weights to obtain a KG (CoG) of your boat?...i think your reply is a no

Ricks plot....no idea what hull it is and the KN values are very different yours too, a difference of some 25~30%, not minor!

"...got transverse metacentric height = 1.51ft..."
Sorry but where did this come from?...looks like a guess again

as for:
"..but I do want to be stable enough that the damping provided by viscous friction is enough to keep roll oscillation from building up in "normal" seas.."

You're barking up the wrong tree. No need to consider such things. This is also a 'different' branch (keeping the tree pun going ) of stability. You're referring to dynamic stability which is a result of the radius of gyration, added mass, inertia of waterplane area etc etc...so, just stick to the simple basic stuff that you need for now.

As i said before, if you can post the output from your program, we can see how much data is right and wrong and take it from there.

However, the Q i ahve is, from your comments above, are you just interested in how far your boat can heel before she capsizes?..and when she does, can she self right?...if so, as noted above, ignore all the fluff about viscous damping.....woof woof (pun intended again )

Finally, the ability to use a program (and produce fancy plots) in no way implies one being a pro!

[Rick Willoughby]

Originally Posted by amaurer

....

I doodled up the model into Delftship and ran the hydrostatic report - it turns out my version can't do the neat crosscurves function that Rick's can, I guess thats only for Pros - and got transverse metacentric height = 1.51ft. Is there another package that can plot the roll data?

...!

Andrew
The cross curve function is in Freeship. It is a high cost option for Delftship. There are different version of Freeship still around the web. You should be able to produce a Freeship compatible file from Delftship. There are some odd versions around though that are not compatible.

Your RM curve looks just about right apart from the kick up as noted previously. If you look through your script you will probably find a simple error.

Your hull will not capsize easily. The hard chine improves roll damping over a rounded chine. You can add a keel to further increase damping. It could be long with a slight taper from say midship to stern to provide protection for the shaft and prop. Set it lower than the prop so any floating trash is guided below the prop. It will add a little more drag but nothing like a plastic bag wrapped around the prop.

I cannot remember if I pointed out that Kelly Controls has a nice little 800W PMSM for USD149. They have compatible controllers with current limit and reversing for USD99. I think you will need this sort of capability to have any hope of freeing trash from the prop.

Rick W

[amaurer]

Originally Posted by Ad Hoc

amaurer
"..but I do have realistic numbers from my CAD model.."
With respect, this is also a guess. A guess by the computer. What I should have said is, have you done a proper weight and centres estimate based upon the calculated or known weights to obtain a KG (CoG) of your boat?...i think your reply is a no

I'm not following you here. The estimate comes from the computer, yes, but from the solid (CATIA) models that I used to design the structure itself, not from a secondary analysis program. The assembly models contain actual weights and volumes for the materials and components (batteries, panels) that I'm using. I don't have any other way of producing estimates at this point.

Originally Posted by Ad Hoc

Ricks plot....no idea what hull it is and the KN values are very different yours too, a difference of some 25~30%, not minor!

"...got transverse metacentric height = 1.51ft..."
Sorry but where did this come from?...looks like a guess again

As I said, thats what DelftShip's "Design Hydrostatics" function produced for me. I'm not doing model testing, so if you consider all computer output to be "just guesses" then yes, its a guess. The full report is attached. I did not use tremendous care setting the waterline and hull shape, however, so you'll notice a 60lb discrepancy in displacement.

Originally Posted by Ad Hoc

as for:
"..but I do want to be stable enough that the damping provided by viscous friction is enough to keep roll oscillation from building up in "normal" seas.."

You're barking up the wrong tree. No need to consider such things. This is also a 'different' branch (keeping the tree pun going ) of stability. You're referring to dynamic stability which is a result of the radius of gyration, added mass, inertia of waterplane area etc etc...so, just stick to the simple basic stuff that you need for now.

As i said before, if you can post the output from your program, we can see how much data is right and wrong and take it from there.

However, the Q i ahve is, from your comments above, are you just interested in how far your boat can heel before she capsizes?..and when she does, can she self right?...if so, as noted above, ignore all the fluff about viscous damping.....woof woof (pun intended again )

I'm primarily interested in static stability, but if its marginal I'll want to worry more about dynamic stability... so both.... maybe

[amaurer]

Originally Posted by Rick Willoughby

Andrew
The cross curve function is in Freeship. It is a high cost option for Delftship. There are different version of Freeship still around the web. You should be able to produce a Freeship compatible file from Delftship. There are some odd versions around though that are not compatible.

Your RM curve looks just about right apart from the kick up as noted previously. If you look through your script you will probably find a simple error.

Your hull will not capsize easily. The hard chine improves roll damping over a rounded chine. You can add a keel to further increase damping. It could be long with a slight taper from say midship to stern to provide protection for the shaft and prop. Set it lower than the prop so any floating trash is guided below the prop. It will add a little more drag but nothing like a plastic bag wrapped around the prop.

I cannot remember if I pointed out that Kelly Controls has a nice little 800W PMSM for USD149. They have compatible controllers with current limit and reversing for USD99. I think you will need this sort of capability to have any hope of freeing trash from the prop.

Rick W

Rick yes it appears my code has an error when Roll = 180deg, which then my smoothing functions propogate into the nearby data. I'm fairly certain this is because the trig functions in my code flip signs when they hit 180deg.

I've been speaking to Castle Creations for motor controllers - they've tentatively agreed to sponsor my project's controller needs, and yes I've asked for units with reverse.

[Rick Willoughby]

Originally Posted by amaurer

Rick yes it appears my code has an error when Roll = 180deg, which then my smoothing functions propogate into the nearby data. I'm fairly certain this is because the trig functions in my code flip signs when they hit 180deg.

I've been speaking to Castle Creations for motor controllers - they've tentatively agreed to sponsor my project's controller needs, and yes I've asked for units with reverse.

Castle make nice RC stuff and I think offer Neumotors, which are also well regarded. However I believe most of this stuff is made for model plane applications where power to weight is the main consideration. This is not the case with your boat. Durability will be important.

If you use a PMSM to avoid the wear of brushes then the motor really needs field tracking to get good break out torque. This would be the situation if you were trying to free weed. If the motors have field tracking they develop rated torque at zero speed similar to a DC motor.

This shows the Kelly 800W PMSM:
http://www.newkellycontroller.com/pr...roducts_id=509
Compare that with the RC stuff. Compare bearings and size for rating.

Also I have had great success with a folding prop to shed weed. There are folding saildrive props that might be worth a look for you. These are highly developed and quite reliable. They will shed rubbish much easier than fixed blades.

Rick W

[Ad Hoc]

amaurer
"..The estimate comes from the computer, yes, but from the solid (CATIA) models that I used to design the structure itself, not from a secondary analysis program..."
Have you done any hand calculations to verify the output from CATIA is correct?..if not, how do you know what is what?..that is the point I'm making. You need to validate your 'computer' model. Since how does the program know it is correct, it doesn't. That is where the designer, ie you, must review the output.

"..The assembly models contain actual weights and volumes for the materials and components (batteries, panels) that I'm using..."
Sounds like you have some kind of breakdown of items then?...where did these values and list of items come from, a guess or known weights.
have you added a margin too?..you'll need one, around 10% at the final estimate.

"..I don't have any other way of producing estimates at this point..."
This is much easier than you can imagine.
All you do is list everything that is on board. Then guess, calculate or establish exactly, the weight of each item. Then from your GA, you locate the midships, or transom, or a datum you are happy with, then do a simple weight and centres, ie a summation of all the moments to obtain the overall CoG, or KG, vertical, or LCG, horizontally.

If you are unsure what will go in the boat, then either
a) make a guess and add some nominal weight
or
b) if you don't know, who else does?

"...I'm not doing model testing, so if you consider all computer output to be "just guesses" then yes, its a guess..."
That is not what i am saying.
As noted above, how do you know the output is real??...you must validate it by an independent method, ie hand calculations!...not the whole lot but just a few quick rough and ready calc's...if the difference is say 5~10% then it is probably ok. But if the difference is say 150~200% then clearly there is a mistake somewhere.

So...have you or can you produce a list of the items, all, including the structure and then obtain the levers for the KG and LCG?..once you have done that, we can address the stability.

[amaurer]

Originally Posted by Rick Willoughby

Castle make nice RC stuff and I think offer Neumotors, which are also well regarded. However I believe most of this stuff is made for model plane applications where power to weight is the main consideration. This is not the case with your boat. Durability will be important.

If you use a PMSM to avoid the wear of brushes then the motor really needs field tracking to get good break out torque. This would be the situation if you were trying to free weed. If the motors have field tracking they develop rated torque at zero speed similar to a DC motor.

This shows the Kelly 800W PMSM:
http://www.newkellycontroller.com/pr...roducts_id=509
Compare that with the RC stuff. Compare bearings and size for rating.

Also I have had great success with a folding prop to shed weed. There are folding saildrive props that might be worth a look for you. These are highly developed and quite reliable. They will shed rubbish much easier than fixed blades.

Rick W

Oh, I was referring to motor controllers only (sorry, missed the "PMSM" acronym). Castle only distributes the Neumotors (they tell me) so they can't help me there.

For motors I'm going with a toothed belt drive, on the order of 6:1 reduction. This lets me use a hotter motor to help with the fieldless starting issue. However, I don't trust any of these motor's bearings, especially under radial loading from a belt, so the gearbox (belt box?) will support every shaft with its own bearings - the motor bearings will be loaded as lightly as possible this way.

I figure a spring loaded idler and low belt tension will spare the belts if the prop gets knocked. I've read your (very helpful) posts on belt efficiency and I have plenty of power budgeted for those loses.

All that said, I used a Castle Controller once upon a time on a research project with NASA (thats the source of the zero-gravity photo on my site). I was using a low kv (about 60) sensorless motor driving a shaft with a double oring gland seal, it probably took 20ft-lbs to start spinning. Adding a some high-power throttle pulses to the firmware whenever the motor was to start up proved to be a miracle solution, nothing could stop that motor from starting.

[amaurer]

Originally Posted by Ad Hoc

amaurer
"..The estimate comes from the computer, yes, but from the solid (CATIA) models that I used to design the structure itself, not from a secondary analysis program..."
Have you done any hand calculations to verify the output from CATIA is correct?..if not, how do you know what is what?..that is the point I'm making. You need to validate your 'computer' model. Since how does the program know it is correct, it doesn't. That is where the designer, ie you, must review the output.

"..The assembly models contain actual weights and volumes for the materials and components (batteries, panels) that I'm using..."
Sounds like you have some kind of breakdown of items then?...where did these values and list of items come from, a guess or known weights.
have you added a margin too?..you'll need one, around 10% at the final estimate.

"..I don't have any other way of producing estimates at this point..."
This is much easier than you can imagine.
All you do is list everything that is on board. Then guess, calculate or establish exactly, the weight of each item. Then from your GA, you locate the midships, or transom, or a datum you are happy with, then do a simple weight and centres, ie a summation of all the moments to obtain the overall CoG, or KG, vertical, or LCG, horizontally.

If you are unsure what will go in the boat, then either
a) make a guess and add some nominal weight
or
b) if you don't know, who else does?

"...I'm not doing model testing, so if you consider all computer output to be "just guesses" then yes, its a guess..."
That is not what i am saying.
As noted above, how do you know the output is real??...you must validate it by an independent method, ie hand calculations!...not the whole lot but just a few quick rough and ready calc's...if the difference is say 5~10% then it is probably ok. But if the difference is say 150~200% then clearly there is a mistake somewhere.

So...have you or can you produce a list of the items, all, including the structure and then obtain the levers for the KG and LCG?..once you have done that, we can address the stability.

I assumed you were speaking of some obscure and mystical naval architecture technique. The CATIA output is reliable - I weighed a sheet of my ply to ensure my densities were right, the battery data comes right from the mfg, etc.

[Rick Willoughby]

The shaft gearing will get you into a shaft speed enabling a decent prop. You could certainly look for a small folding prop.

If you do not use a folding prop then consider a full shield or guard for the prop. It will also be a safety precaution for approach of people.

My folding prop is a very simply affair but achieves high efficiency because it uses small chord, high aspect blades. The saildrive props are not as efficient but are considerably more robust and durable.

You can use JavaProp to determine your the efficiency for various props. Have you tried it yet?

Rick W

[Ad Hoc]

"..The CATIA output is reliable."

If you say so...!

[amaurer]

Originally Posted by Rick Willoughby

The shaft gearing will get you into a shaft speed enabling a decent prop. You could certainly look for a small folding prop.

If you do not use a folding prop then consider a full shield or guard for the prop. It will also be a safety precaution for approach of people.

My folding prop is a very simply affair but achieves high efficiency because it uses small chord, high aspect blades. The saildrive props are not as efficient but are considerably more robust and durable.

You can use JavaProp to determine your the efficiency for various props. Have you tried it yet?

Rick W

I'm a little bit torn on the propeller, I'm not convinced a folding prop is a slam-dunk solution. I'm no expert, but I'm concerned about the reliability of the joints and mechanisms - is a folding prop really designed to motor for months on end? Sure, the power level is low, but that might also mean that the joints end up moving more since the blade forces are lower.

And on top of that, it obvious to me that a folding prop is a winner against solid debris, but against tangling/fouling debris, it is really that much better? I think the latter is the bigger risk for open ocean usage. Even if IS a little better, I imagine it will affect my reverse thrust, which might be nice to have for a detangling maneuver.

I dunno, in short.

I have also been considering the Torquedo prop. They're cheap so I was going to by one and see how strong it is - I understand they're composite but would like to get my hands on it to get a feel for the elasticity and strength. It would also be within my abilities to mold and recast it in bronze.

[Rick Willoughby]

It would be worth your while to locate a nice folding prop and take a close look. They last a long time. They require very little power to fully deploy. You are right about poor reverse thrust but they will give reverse thrust if revved up.

Once the blades are deployed they do not suffer wear on the pivot pins. You would need to cycle once in a while just to clear any fouling. Maybe something like twice a day.

Some have geared blades so they are always in balance. This would probably be the best choice.

Suggest you look at guards and folding props so you know the options.

Actually the idea of a flexible prop is not that silly. The Hobie flappers produce reasonable thrust and these use rubber with a spring steel stiffener. You are dealing with very low forces. Rubber may be less prone to fouling with sea life.

You have lots of challenges to get reasonable life without attention on the open sea for months at a time.

Rick W