Design Optimisation

Discussion in 'Sailboats' started by Andrew Mason, Oct 30, 2010.

  1. Andrew Mason
    Joined: Mar 2003
    Posts: 396
    Likes: 17, Points: 0, Legacy Rep: 206
    Location: Perth, Western Australia

    Andrew Mason Senior Member

    For those interested in developments in yacht design technology, I have uploaded a copy of my PhD thesis on "Stochastic Optimisation of America's Cup Class Yachts" to this page.

    Access to the thesis had previously been restricted at the request of the Alinghi team, but with the selection of catamarans for the next America's Cup, Alinghi have kindly waived this requirement.
     
    1 person likes this.
  2. CutOnce

    CutOnce Previous Member

    Thank you for posting your thesis. I will enjoy reviewing it and learning. It is academic sharing that allows the world to progress.

    --
    CutOnce
     
  3. Doug Lord
    Joined: May 2009
    Posts: 14,129
    Likes: 123, Points: 73, Legacy Rep: 1362
    Location: Cocoa Beach, Florida

    Doug Lord Flight Ready

    ===============
    Thanks very much,Andrew!
     
  4. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 145, Points: 0, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    Congratulations! And thanks for uploading the thesis.
     
  5. Landlubber
    Joined: Jun 2007
    Posts: 2,642
    Likes: 123, Points: 0, Legacy Rep: 1802
    Location: Brisbane

    Landlubber Senior Member

    ...thanks Andrew, it is a very interesting article and even though i will not understand half of it, what I have read is inspirational and thought provoking in the methods utilised. Ta, John
     
  6. Michael Y
    Joined: Oct 2010
    Posts: 29
    Likes: 1, Points: 0, Legacy Rep: 14
    Location: Pennsylvania

    Michael Y Junior Member

    I very much enjoyed the short paper, and will read the thesis. I'm a budding boat builder and designer, which is one connection to your work. My PhD advisor was Dr. Simpson (you referenced a paper of his) so there's another connection.

    Give this a look...might resonate with your work.
    http://www.atsv.psu.edu
     
  7. Andrew Mason
    Joined: Mar 2003
    Posts: 396
    Likes: 17, Points: 0, Legacy Rep: 206
    Location: Perth, Western Australia

    Andrew Mason Senior Member

    Thanks Leo

    Although you are not directly cited in the thesis I do owe you a debt of gratitude for providing some of the initial inspiration for the work.

    And I should also congratulate you on your PhD, I have your thesis on my iPad and am slowly working my way through it.

    cheers

    Andrew
     
  8. Andrew Mason
    Joined: Mar 2003
    Posts: 396
    Likes: 17, Points: 0, Legacy Rep: 206
    Location: Perth, Western Australia

    Andrew Mason Senior Member

    Thanks Michael, I hadn't heard of the Trade Space Exploration approach before, it looks quite interesting. I'll have to do some reading.

    Dr. Tim Simpson has done some excellent work that I used extensively, I really appreciated his ability to sort the wheat from the chaff when it comes to sampling and surrogate modelling. What was your thesis topic?
     
  9. Leo Lazauskas
    Joined: Jan 2002
    Posts: 2,696
    Likes: 145, Points: 0, Legacy Rep: 2229
    Location: Adelaide, South Australia

    Leo Lazauskas Senior Member

    Thanks, Andrew.

    I think our approaches diverged a long time ago - you headed for CFD, I do everything I can to keep away from the beast. :)

    I also wanted to do some work on ship motions in waves and added resistance, but when I suggested it to Ernie Tuck he replied, "I never want to touch that subject ever again!".
     
  10. Michael Y
    Joined: Oct 2010
    Posts: 29
    Likes: 1, Points: 0, Legacy Rep: 14
    Location: Pennsylvania

    Michael Y Junior Member

    Neither of those two. :rolleyes:

    My thesis topic was on fast algorithms to identify the Pareto Frontier from large multi-dimensional data sets. The algorithm is the one that operates in ATSV. One of my cohorts did his work in using Krieging to model uncertainty in order to allow for rapid uncertainty propagation in complex models.

    I teach a course in flight test right now, and have occasional thoughts about trying to teach one in yacht design. Have to learn it first, though. Super interesting subject.

    The trade space stuff is useful if you have multiple criteria you are trying to satisfy. In your case, the tournament seems to provide a nice, neat scalar number to optimize. But since what you are actually doing is comparing distributions of outcomes between yachts, maybe it would be interesting to visualize not just the expected values but the median, mode, standard deviation, etc. Also, you could consider min-max solutions and look at that space.

    One question I had is, for all the different yachts you are racing in your tournaments, would the tactics you'd use for one yacht be different than the tactics you'd use with another? Would you sail them differently to extract each one's best performance? Or would you sail yacht A different when racing yacht B as opposed to racing yacht C? I think the answer is no, since they are close to each other in design. But if the answer is 'yes', are you doing some sort of search over the tactics space too?
     

  11. Andrew Mason
    Joined: Mar 2003
    Posts: 396
    Likes: 17, Points: 0, Legacy Rep: 206
    Location: Perth, Western Australia

    Andrew Mason Senior Member

    The variance in the win ratio is directly related to the diversity in the population, hence the need for techniques that promote diversity such as fitness sharing, niching, hall-of-fame etc.


    If the intent is to rank designs then you need identical tactical models, as what you are effectively doing is two boat testing, keeping both boats in as similar a set of conditions as possible.

    If the intent is to simulate races to investigate the effects of different tactical models, or the effects of different design parameters on maneuverability, then yes, you might look at sailing the boats differently.

    Keep in mind that while the ACC V4.0 rule allowed a wide range of displacements, sail areas and lengths, the V5.0 rule effectively fixed these three key design parameters. The result was that the boats could be considered to have negligible differences in maneuvering characteristics, particularly as we did not vary keels and rudders. So eliminating all tactical differences between the boats was a reasonable assumption for the purposes of comparing hull designs.
     
Loading...
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.