Expert seeking expert to bounce prop design ideas off

Discussion in 'Props' started by Chris Drake, Jun 12, 2020.

  1. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    I've released a Fusion360 add-in called "airfoil tools" (does hydrofoils too) which makes use of 8 years worth of CFD preprocessing to insert "ideal" foil sections for users.

    I've completed the foils and struts, and I'm overjoyed at how well my turbines function.

    I a now up to the point where I do propellers. They're a lot hard than turbines (which always want to extract as much power as they can, which the betz limit allows us to calculate).

    Is anyone on this board obsessive and knowledgeable about optimal design?

    My goal is to interrogate the user so I know how they plan to use their prop, then to use that with math to give them the one best suited - kindoflike "props for dummies" I guess: I'm not asking for reynolds numbers or fluid densities etc - none of that make sense - I'm asking for things people know (speed ranges, temperature, size, medium, etc).

    The work I'm about to embark on involves power and torque and size constraints etc, so I can give my users the reduction ratios, blade counts, and the completed design in return.

    Ideas I feel like I need to talk over with other experts include tip wiglets/ducting, trailing edge uniformity, variable pitch considerations, computing disc fluid velocity at chord stations while taking tip loss into account, cavitation thoughts, hub-station energy recovery, and so on...

    To whet you appetite, see attached chart. If you run a few hundred thousand hours of genetic design on 1440 operating regimes optimising for CL/CD over a bell-weighted +/-6-degree AoA target, you can then draw an interesting table. (I wonder if anyone has ever done that before? Do let me know if you've seen it!! - clue - what is the best Reynolds Number ? Answer - see screenshot :)

    [​IMG]

    My add-in is here, (assuming I, a new user, can post links): Add-in announcement: Hydrofoil and Airfoil Tools. Seeking your feedback / ideas - Autodesk Community https://forums.autodesk.com/t5/fusion-360-api-and-scripts/add-in-announcement-hydrofoil-and-airfoil-tools-seeking-your/td-p/9453985
     
  2. gonzo
    Joined: Aug 2002
    Posts: 16,790
    Likes: 1,714, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    If hard data does not make sense for you, is the request for anecdotal information only?
     
  3. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    It does not make sense to ask, for many different reasons, not least being that there is no one answer. Other reasons include: the user doesn't know, the user thinks they know but they're wrong, the complex non linear range of potential answers doesn't lend any simple way to convey an attempt at a correct answer, or even if we found a user who actually totally understands everything - how are they supposed to tell us what their goals are, and what operating regimes are of no interest?

    The way I deal with that was to ask them what they do know - then I can work out the hard stuff properly:
    [​IMG]
     
  4. DCockey
    Joined: Oct 2009
    Posts: 5,229
    Likes: 634, Points: 113, Legacy Rep: 1485
    Location: Midcoast Maine

    DCockey Senior Member

    What type of CFD have you used?

    From the announcement linked to above:

    What do the percentages represent - differences in lift to drag ratios or something else? What does "all NACA shapes" mean? Is it every shape that NACA ever studied or all members of particular series?
     
  5. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    A custom build of xfoil with a range of bugs removed

    For props, it's CL/CD. For Struts, it's 1/CD. I used a bell-curve-weighted target of +/- 6 degrees AoA for those comparisons - it's likely to be the same for +/- N of either Re or AoA (I'll confirm that one day - for now - I want to get prop creation working first.).

    All 9999 4-series shapes, about 2000 5-series shapes, and every other NACA shape that has found its way into the large assortment of online database profiles and so on that I collected before embarking on this journey (several hundred more).
     
    DogCavalry likes this.
  6. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    p.s. I spent some time with OpenFOAM and was seriously underwhelmed - my general observation of CFD for airfoils is that 99% of time, it is useful for nothing more than confirming that the results are close to some experiment or other. The work that we are never privy to, are all the discarded mistakes they made along the way. That's significant, because they're using the answer to adjust their question, without telling us. What does that mean? When they don't know the answer to begin with, the reliability of the simulation is so poor that the results are all but meaningless. That's my opinion based on reading about 100 papers on the topic.

    My solution to overcome that problem is to replace CFD with what I call AFD: "Actual Fluid Dynamics". I'm using my SLA printer to produce real foils and turbines (and, coming soon, props) and an automated test rig that measures power, speed (rpm+forward), torque, thrust, noise, pressure (around the shape, plus ambient), and flow. From here, I'm able to use genetic design to optimise the performance, knowing that my results are 100% free of mathematical error (because they're actual real results, not modeled theoretical ones). It might, after the fact, be possible to then train an AI to speed up the solution finding step - but for now, I'm able to be confident that my results perfectly translate into reality - because the results are real in the first place.
     
    DogCavalry likes this.
  7. johneck
    Joined: Nov 2011
    Posts: 253
    Likes: 17, Points: 18, Legacy Rep: 117
    Location: New England

    johneck Senior Member

    Sounds interesting, but why do you think that optimizing Cl/Cd of a blade section will produce a better propeller? It seems to me that more critical factors are providing an appropriate radial circulation distribution and making sections to avoid cavitation issues at various points on the blade. Unsteady forces and moments can also be significant factors.
    I also think even if you did want to optimize sections Cl/Cd, +/-6 is not a big enough envelope for any practical propeller design. But I do have quite a bit of propeller data available if you would be interested in talking send me a PM.
     
  8. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    Hi! I think you almost answered your own Q there? cavitation begins from low pressure, which is caused by the wrong section shape+angle for the speed it's travelling. To achieve optimum Cl/Cd, one requirement is that the pressure distribution remains ideal (that's true for either best Cl or best Cd alone though, so Cl/Cd isn't specifically required, but I can't think of any logic where accepting worse lift or worse drag makes sense - so we may as well go for both?).

    My free product has had a great uptake from end users so far, and in it, I've got a "vote" option for the feature I've not finished yet. top-score of what users want is propeller design (30%) and second-top is analysis output (20%) [that's the polars and pressure charts etc]. The second part is done, but needs making available, and I'm still struggling with the first.

    The +/- 6 was a restriction I put in to prevent allowing the selection of any optimal shape on the cusp of it's own failure point (e.g. stall). It's not actually a usage envelope per-se - but that said - I've not paid attention to what those envelopes are yet, so it could be +/- 30 degrees, or +/- 7; and it's no-doubt different for each of the 1440 shapes in my database...

    I've just finished printing my first ducted water turbine ( Fusion https://a360.co/30h8gSz ), and have plans for torque-testing thisafternoon. So far in the pool, it's looking awesome: my design thought-process went as follows:
    1. A turbine can't extract more than the betz limit ever, so, let us use that as the design parameter for the amount of power we're taking out.
    2. make the trailing edge a straight line, so the pressure leaving the blades is as mostly-in-the-same direction as we can get.
    3. use a duct, so I can ignore tip loss issues.
    4. maximise the Re (higher is more efficient I found in my design process), so I increase the blade cord as much as possible to the tips (without exceeding the prop height).
    5. respect bernouli (shape the duct appropriately for the expanded-flow from power-loss after our turbine has removed it from the flow).
    6. use many blades - we know we can't take more than Betz tells us, so my theory is that many blades should help here. I've printed a pair for testing - one with 3 and one with 9 blades, to hopefully verify my theory (seems good so far in the pool - the 9-blade can be felt to push harder than the 3 blade at my 1 m/s design speed)

    All that's great for a water turbine, where we know the much-more-restricted-than-a-propeller speed and power envelopes.

    What comes now, is the prop theory - where both power and speed are very different beasts...
     

    Attached Files:

  9. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    Maybe my forum membership is too new to use PM? [I can't see how anyplace] - my email is
    fusion360 [hyphen] support [at] OceanHydro.com.au (i.e. the customer-support email address on my product web page)
    Airfoil Tools | Fusion 360 | Autodesk App Store https://apps.autodesk.com/FUSION/en/Detail/Index?id=5447707798035545266&appLang=en&os=Win64

    Bottom line - yes please!!
     
  10. gonzo
    Joined: Aug 2002
    Posts: 16,790
    Likes: 1,714, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    That is a wrong assumption. You should use calculations for Kort nozzle applications instead of turbines.

    Also wrong, there is no pressure leaving the blades. If that were so, the thrust would go to zero.


    Another wrong assumption. There will be losses at the interface between the propeller tip and the duct; also turbulence.

    The blades you show are not in a duct, but have a ring around them. The ring will generate a large loss due to friction. I don't know what you mean it can be felt to push, it is a quantifiable force, so show the measured results.
     
  11. Chris Drake
    Joined: Jun 2020
    Posts: 10
    Likes: 2, Points: 3
    Location: Australia

    Chris Drake Junior Member

    1. My example was a turbine - not a propeller. Turbines cannot take more energy out of any flow than the Betz limit dictates (like you said - the flow would go to zero (bypass the turbine entirely) if you took out all the energy)... buy you're on the right path - I'm looking to figure out how to amend my turbine code for propeller usage.

    2. It's a turbine, so there must be pressure leaving the system, otherwise it would be stalled (the pressure is lower of course - we extracted energy)... my bad here though - I meant velocity, not pressure.

    3. There's no gap :) yes - the duct rotates with the turbine. There's a reason for that curious decision - the target turbine is 1MW in the ocean, so needs to generate 600,000 volts DC in order to transport electricity 100km to the shore. In order to reduce maintenance and avoid reduction gearing and a seabed-powerstation, my plan is direct-generation of the 600kv through series arrangement of coils and magnets around the perimeter of the turbine. The lost energy is more than offset by complexity and maintenance reduction... and yes, I do know a fair bit about HVDC.

    That said - "large loss due to friction" is a stab-in-the-dark guess; I'd wager you never measured a spinning duct before, right? How do you know any loss it makes it "Large"? Do you know whether that loss is more or less than the gain obtained from tip-loss reduction? Here's a trick question for you - if the duct spins faster, does it make more or less friction loss ? (hint: Kármán–Schoenherr formula)

    4. I don't have a waterproof scale yet - my work bench is strewn with the bits for far more than that though - load cells for torque and thrust/pressure, servo-mounted pitot tube for spanwise analysis, RPM/volts/amps (watts) power measurements with adjustable load, and an MCU to manage all that so I can automate my tests, and feed all that back into my genetic optimiser to produce literally measurable improvements. There's only so far one can go with CFD, and 90% of what exists is garbage (it's almost all fake - they "validate their CFD with experiments", but they never report their initial failures, which is a classic "time traveler" cheat - they adjust their CFD to match the experiments after-the-fact. Bottom line - if they didn't do the experiment, the CFD would produce utterly wrong answers, so if you're trying to use CFD to produce something that works form scratch... you never can. solution: don't use CFD - use "AFD" (*Actual* fluid dynamics) - 3D printing and real tests, instead of fake CFD simulations based on sketchy guess numbers...)
     
  12. gonzo
    Joined: Aug 2002
    Posts: 16,790
    Likes: 1,714, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    I'll take that wager before answering. How much do you propose?
     
  13. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,152, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    Chris, welcome to the forum.
    Way too much for a forum. Yes, you need a mentor, the two designs you showed in post #8 are..."less than optimum". Realistically, you should not be using anything from AutoCAD for complex fluid dynamics. (I'm a user from ver 2.0 up ( 15 minute redraws anyone?) and AutoCAD leaves a lot to be desired on many subjects). You need to get a copy of Fluid Dynamic Lift and Fluid Dynamic Drag, both by Heorner (and both I believe are available on the net). A quick guided tour through those tomes will show you why turbines make very poor propellers, and why ringed props do not look like your models. I will say this, every unit of root/tip intersection has more drag than the entire blade...and putting the longest intersection at the tip is great for a power extraction turbine, but sucks for a propeller.
     
    rxcomposite likes this.
  14. Mr Efficiency
    Joined: Oct 2010
    Posts: 10,386
    Likes: 1,042, Points: 113, Legacy Rep: 702
    Location: Australia

    Mr Efficiency Senior Member

    Sounds like some good advice from jehardiman, attempts to re-invent propellors probably have similarities to trying to re-invent the wheel. The chartreuse colour does have its appeal, though ! :D
     

  15. Mr Efficiency
    Joined: Oct 2010
    Posts: 10,386
    Likes: 1,042, Points: 113, Legacy Rep: 702
    Location: Australia

    Mr Efficiency Senior Member

    What can be said pretty categorically, is that extra surface area. and particularly surface area that is not applying thrust, adds up to diminished performance. "Ring" propellors have been touted as way to avoid mangling props, or swimmers, but something revolving its little heart out, like the external ring, is suffering frictional losses, even surf life saving rescue boats don't use them, so obviously they don't save swimmers either.
     
    jehardiman likes this.
Loading...
Similar Threads
  1. Bruce P
    Replies:
    9
    Views:
    2,969
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.