sail aerodynamics

Discussion in 'Hydrodynamics and Aerodynamics' started by Guest, Mar 21, 2002.

  1. Packeteer
    Joined: May 2005
    Posts: 68
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: S33 E151

    Packeteer Junior Member

    I'm not sure how correct your information is, or it could just be ancient history. Intel has traditionally always been CISC, not that it has mattered for at least 10 years, as we're well past RISC/CISC in cpu design.

    And more importantly, anyone doing massively parallel computational work should be using GPU's. Even Amazon offers this as a service!

    https://en.m.wikipedia.org/wiki/General-purpose_computing_on_graphics_processing_units
     
  2. Skyak
    Joined: Jul 2012
    Posts: 1,194
    Likes: 25, Points: 38, Legacy Rep: 152
    Location: United States

    Skyak Senior Member

    I see great application of pattern recognition in aerodynamics. The pattern is smart people with vast resources warning that simple approaches get complicated so fast that any chance of insight is lost -followed by great time and effort spent recreating what was done long ago. I use this pattern recognition to chose my battles and weapons wisely.

    I suppose today (april 1) is a bad day to ask, but has anyone actually designed a processor cell specifically for CFD? They made special CPUs for bitcoin mining. I would expect that CFD would be higher value.
     
  3. tspeer
    Joined: Feb 2002
    Posts: 2,236
    Likes: 198, Points: 63, Legacy Rep: 1673
    Location: Port Gamble, Washington, USA

    tspeer Senior Member

    When I was studying CFD in the 1980's there was a lot of talk about silicon compilers. The idea was you'd have high-level software that would create a processor specifically designed to execute the CFD stencil - the pattern of neighboring points whose data are used in executing the equations for a particular point. So you'd write your code in a high-level language, and the result would be both a chip design and the machine code to run on it. Then you'd ship your chip design off to a silicon foundry to have it constructed. In a couple of weeks, you'd get your chip, plug it in, and you'd be in business. But I haven't heard much about that approach these days.

    Today, the hot trend is to use graphical processors. They have lots of cores per processor, so you can do massively parallel computations. Here's an example using 64 Tesla 2090 GPUs (which each have 512 cores) to simulate a billion particles in a wave hitting an oil rig. It required almost 92 hours to simulate 12 seconds of real time.
     
  4. Skyak
    Joined: Jul 2012
    Posts: 1,194
    Likes: 25, Points: 38, Legacy Rep: 152
    Location: United States

    Skyak Senior Member

    Your 80s example is WAY beyond what I was considering. I was wondering if CPUs might benefit from complex instructions specifically for CFD. Maybe even just using FPGAs to for tasks that take too many cycles.

    I am aware of the move to massively parallel GPUs based on phone and graphics card cores -makes sense to follow the money for top tech. Have they solved the challenge of programming for massively parallel or do they still lose efficiency for each core added? I suppose it is hidden in the cloud provider's operations.

    It is interesting that in the 80s the separation of semiconductor design from fabrication business was THE big thing. Today cloud computing is THE big thing. It is not surprising that CFD is following the economics trends of the time, getting the best deal it can on power, but it makes me wonder because these technology companies are producing at a loss and justifying investment with growth. Intel purchased a big FPGA company last year. Maybe there is a new paradigm developing.
     
  5. PI Design
    Joined: Oct 2006
    Posts: 673
    Likes: 21, Points: 0, Legacy Rep: 328
    Location: England

    PI Design Senior Member

    What is the actual point of that oil rig simulation? 92 hours of run time to solve 12 seconds, using hugely powerful computers is wholly impractical for even an oil company, and that excludes the set up time meshing. And then the results are only for one simplistic and isolated wave at one particular wave height, in one direction. And do you learn anything useful from the result?
     
  6. DCockey
    Joined: Oct 2009
    Posts: 4,411
    Likes: 213, Points: 63, Legacy Rep: 1485
    Location: Midcoast Maine

    DCockey Senior Member

    I assume you are referring to this example https://www.youtube.com/watch?v=B8mP9E75D08

    A 64 GPU system is not a large system for serious number crunching. The cost is probably several hundred thousand dollars. Assume it cost a half million dollars, has a useful life of three years, and crunches numbers 80% of the time. The cost per day of calculations would then be $570. The simulation shown in the example ran for 92 hours or a little under 4 days four years ago. With the assumptions above the cost would have been $2200. Even if the cost of the calculations was several times that amount it would still be a very, very tiny part of the cost of engineering an oil platform.

    Point of the calculations - probably to provide predictions of loads from very large waves on the structure and other exposed parts of the oil platform. A number of simulations with waves from different angles and different characteristics were probably run using the same mesh model.

    Ratio of calculation time to "real" event time is meaningless. Auto crashes occur in less than a tenth of a second while simulations can run for days. But such simulations are now the standard tool for the analysis used in designing autos. Physical tests are only used to validate the simulations.
     
  7. PI Design
    Joined: Oct 2006
    Posts: 673
    Likes: 21, Points: 0, Legacy Rep: 328
    Location: England

    PI Design Senior Member

    Oh come on, a billion particle simulation is big in any book. Yes, computer power means it is possible, but is it necessary? Oil rigs are not designed down to the ounce and if you use that simulation to prove your rig is strong enough, it will surely fail when a rarer but bigger wave comes along. Oil rigs can be designed quite satisfactorily using rules - the design may not be optimised but it works.
     
  8. Skyak
    Joined: Jul 2012
    Posts: 1,194
    Likes: 25, Points: 38, Legacy Rep: 152
    Location: United States

    Skyak Senior Member

    I strongly disagree. Simulation can save huge amounts of overbuilding and give great incite into the nature of the loads. Think of things like frequency response. My view of the huge computational expenditure of this problem relates more to trust and emphasis -they want to be certain that the answer is THE ANSWER! and will not be undermined at some later date. You can't design by experience when the requirement is zero failures. In automotive, we had more computational power simulating crashes than in the hands of designers. The reason was that crash testing could only be done on production line vehicles, after a $billion or two had been spent and before the cars could be sold. The cost of a delay at that point would be devastating.
     
  9. powerabout
    Joined: Nov 2007
    Posts: 2,903
    Likes: 61, Points: 48, Legacy Rep: 719
    Location: Melbourne/Singapore/Italy

    powerabout Senior Member

    I agree, they are very good at building rigs and vessels to last 5 years and 1 day these days and designed to only work in certain areas.
    The financial plan is to get return in 5 years then sell for scrap
     
  10. brian eiland
    Joined: Jun 2002
    Posts: 4,788
    Likes: 157, Points: 73, Legacy Rep: 1903
    Location: St Augustine Fl, Thailand

    brian eiland Senior Member

    So now I am surprised that no one bothered to reply to this posting I made







    So now I am 'Thinking One' and looking at this new alternative aftmast concept I posted over here:
    Alternative Aft-mast Rig
    View attachment 106147
    Looks like it might be fair to say that it presents two nice headsails to drive forward with, and a controllable flap by way of that aft mainsail/wingmast arrangement? (I might even be tempted to term that mainsail a mizzen, it is mounted so far aft).

    I even wonder if that aft mainsail/wingmast might be replaced with a more brief MastFoil arrangement operating as a flap device. :idea:

    In either case I think I would prefer this more brief arrangement on the after portion of my sailing rig than an over-tightly sheeted traditional mainsail trying to act as the flap. I've experienced over sheeted mains, and mains pulled up pass the centerline, and sails with leech control lines, and in most cases I did not see big improvements with my windward sailing,...in fact most cases less performance.


    PS: I forgot to add a link to the website (blog) of the rig's creator
    http://www.vizible.co/p/p.html
     
  11. brian eiland
    Joined: Jun 2002
    Posts: 4,788
    Likes: 157, Points: 73, Legacy Rep: 1903
    Location: St Augustine Fl, Thailand

    brian eiland Senior Member

    Flaps on Airfoils, the MastFoil concept

    No replies to this posting as well??


     
  12. CT249
    Joined: May 2003
    Posts: 1,228
    Likes: 65, Points: 48, Legacy Rep: 215
    Location: Sydney Australia

    CT249 Senior Member

    Brian, any over-sheeted sail, or poorly set sail, can be expected to provide poor performance. To say that you have experienced over-sheeted mains etc providing poor performance and therefore implying that mains are bad is like saying that because you've seen a slow trimaran or a badly trimmed jib, all tris and jibs are slow.
     
  13. brian eiland
    Joined: Jun 2002
    Posts: 4,788
    Likes: 157, Points: 73, Legacy Rep: 1903
    Location: St Augustine Fl, Thailand

    brian eiland Senior Member

    I think you missed my point. I was referring to some of these postings which seem to indicate that the oversheeted aft sails (the mainsail most often) can result in generating higher lift.

     
  14. CT249
    Joined: May 2003
    Posts: 1,228
    Likes: 65, Points: 48, Legacy Rep: 215
    Location: Sydney Australia

    CT249 Senior Member

    A main that is actually making the boat go faster and higher (which is what those posts were referring to) is not what most of us would call "oversheeted".

    What those posts are referring to are simple facts - obviously Mark, Mikko etc have not been getting it wrong all this time. In plenty of classes boats undeniably perform better when the mainsail's leach angle is past the centreline - we know that because when we do it, we sail past the boats that don't.

    In at least one class and time (the IOR half tonners of about 1984) the top boats didn't even bother with leach telltales because they never streamed because the leach was so tight - and those boats won world titles so obviously they were doing it right. Tight leaches are less forgiving than loose ones generally (as Jamie notes) but having the leach so tight that the lee flow regularly stalls is a standard trimming technique in a huge number of classes - because it works really well.
     

  15. brian eiland
    Joined: Jun 2002
    Posts: 4,788
    Likes: 157, Points: 73, Legacy Rep: 1903
    Location: St Augustine Fl, Thailand

    brian eiland Senior Member

    Guess I'm just old school. I always referred to this as pinching, and never found it that productive. Maybe that's just a multihull thing,..where we foot off and GO.
     
Loading...
Similar Threads
  1. Federico Ferretti
    Replies:
    1
    Views:
    277
  2. schakel
    Replies:
    4
    Views:
    487
  3. Yull
    Replies:
    10
    Views:
    678
  4. Manfred.pech
    Replies:
    7
    Views:
    1,360
  5. andysailor
    Replies:
    4
    Views:
    872
  6. CocoonCruisers
    Replies:
    73
    Views:
    4,417
  7. mustafaumu sarac
    Replies:
    25
    Views:
    1,293
  8. revintage
    Replies:
    9
    Views:
    743
  9. Dolfiman
    Replies:
    0
    Views:
    539
  10. Dolfiman
    Replies:
    3
    Views:
    856
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.