ORC
There’s no better illustration of the nuanced benefits of a change in the transom immersion and aft run than the use in some of the 2008-generation Imoca designs of adjustable aft flow control. This is the 2008 Farr design Virbac of Jean-Pierre Dick which featured a two-segment split transom flap system which could be raised or lowered from the cockpit. This allowed a little more rocker than would otherwise be desirable for high-speed sailing so better light and medium-air performance with the flaps raised
Old dog new tricks
In these strange times, held in the grip of the coriander virus, once you’ve tidied your sock drawer for the third time and seen all your favourite movies twice, we do still have some time to reflect on how we got to where we are. The development of the ORC VPP might not be everyone’s first port of call for rumination, but it is an inter- esting case study of how the field of Artificial Intelligence (AI) has infiltrated our lives. When you look at it with the benefit of hindsight we can recognise
now that at its outset the IMS VPP-based handicap was a hubristic step into the unknown. The belief that you could reliably predict the performance of a fleet of random boats using a very simple para- metric description of their hull and their sailplan was bold indeed. The aim was surely altruistic: to escape the type forming of the rating rules in place back then and establish a new rule where you could race any type of boat on a level playing field. For the ORC today the goal is still the same: we try to produce
handicap polars that reflect the performance of any boat based on the measured dimensions of the hull and the sailplan. To do this we need to create force models based on the dimensions and calculated quantities of weight, sail area, keel area and so on. Some of these are easy: having bigger sails, a taller rig and more stability all make a boat sail faster. But the nuances of hull resis- tance and sail forces are much more tricky. When IMS started in the mid-1980s a lot of engineering calcu-
lations were still done using a slide rule, and the early desktop com- puters like the Commodore PET. Computing power of course is now many orders of magnitude greater than back then… ask a modern engineer about log tables and he will direct you to the nearest IKEA. The VPP started off using tank test results taken from 20 or so
models and a crude representation of rig performance. It was a very capable VPP, sensitive to changes in displacement, length, sail area, rig height and draft. To design a boat, it was a perfect tool. Remember how those first-generation IMS boats were much faster and easier to sail than their IOR cousins of the same size? However, if you wanted to handicap a racing boat designed to exploit this early VPP, it was less than robust. Over time it became
way too easy to design a boat that had a lot of slow features but sailed faster than the VPP thought it should. These rule beaters, along with complex scoring, gave IMS a bad reputation outside some small well-funded groups in the Med who were thriving on new rule exploitations. The International Measurement System had morphed into the Italian Measurement System (although truth be told there were plenty of Spanish teams doing the same). At ORC we have been working to widen the appeal of our product
for grand prix racers and club sailors alike, and slowly the system has improved and the fleets are growing. One of the tools being used to enable this improvement is AI. With ready access to better and faster computers, for hydro
modelling we have gone from having only 20 hull types to cover all the boat types in the world, to being able to run 1,000 hulls in Computational Fluid Dynamics (CFD) computer codes. On the aero side we have gone from having a handful of wind tunnel tests to being able to run several thousand ‘virtual wind tunnel’ simulations. And to analyse all this data we have moved to an M & S (Modelling & Simulation not Marks & Spencer) approach. And to make sense of all this data we have had to adopt AI. What this means is that we have a huge data set of cause and
effect from which to derive some sensible conclusions. For example, we now have 3,000 wind tunnel tests with different wind angles, sail trims and traveller positions. Some of these are ‘good’ tests, and a lot of them are ‘bad’ tests in terms of making the boat sail fast. But that doesn’t matter, because in the AI process the machine needs to learn what is fast and what is slow, and it can’t do that just by having all the runs being close to perfection. In fact, what we are trying to do is to take all sorts of sail trims and identify what’s fast. To do this we take our 3,000 CFD results, each one characterised
by a set of variables, say, 10 things: main camber, jib camber, boom position, jib lead position fore and aft, and position in and out, jib twist, main twist, mast height, overlap and fractionality. This is our ‘training set’, and the computer then ‘learns’ from all the interactions between these trimming and shape parameters so that the perfor- mance of the sailplan, trimmed in the best way, can be calculated
SEAHORSE 33
w
JEAN-PIERRE DICK
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118
