Sailonline.org - Online Sailing

Facebook

Login

Support Sailonline

If you haven't already - join the SAILONLINE YACHT CLUB!

Please also consider making a donation - all amounts are greatly appreciated!

Tweets by @sailonline

Board » General Discussion » DTF calculation

Posted by NZL_Scotsman Posts 249 Registered at 2008-10-04 07:59:40	2012-05-01 09:57:51 - DTF calculation
	One of the most frequently asked questions on chat is how the distance to finish is calculated. I've tried to explain that, based on my understanding. If anyone knows differently, please add to the thread. The first two plots here are sketches of a race similar to our SWR Leg 6. That is a good example of where the DTF is not entirely intuitive. The third plot is the mark to mark distance, which many think should be the DTF. The distance to the next mark is determined by the closest distance to the line that goes out from the mark. This line (the mark 2 rounding line) is drawn at right angles to the line between marks 1 and 3, starting at mark 2 and continuing infinitely. From that point, the next part of the DTF is the distance to the next mark. Then mark to mark to the finish. So looking at the attached plots. Boat A sails close to the rhumbline, so its DTF is very similar to a mark to mark course. Boat B, however, is sailing wide of the mark. From B1 we find the closest point on the rounding line (point X). From there to mark 3. All the distances from the various positions are shown on the plots. --- Last Edited by NZL Scotsman at 2012-05-01 10:11:12 --- Attachments 1-existing.JPG (199.6 KB) 2-existing.JPG (260.3 KB) 3-mark2mark.JPG (191.4 KB)
Posted by NZL_Scotsman Posts 249 Registered at 2008-10-04 07:59:40	2012-05-01 10:15:28 - Re: DTF calculation
	If you have another look at the mark to mark system, the main problem with this is that when you approach the mark rounding line wide - then the DTF includes the distance to this mark until you pass it. Then it jumps to the next mark. For example Boat B at B4 jumps from 43nm to 35nm as soon as the line is crossed. The existing system has a similar problem, but it is much less severe. The problem only really occurs when you have a situation like B3, where the mark 3 rounding is very wide and therefore the mark 3 to 4 distance is much greater than compared to boat A. (35nm compared to 17nm).
Posted by NZL_Scotsman Posts 249 Registered at 2008-10-04 07:59:40	2012-05-01 10:25:24 - Re: DTF calculation
	Here is an alternative method of calculating DTF that might avoid some of the issues the current version creates - those rare occasions when the course shape results in longer DTF than is reasonable. The attached plot will help with this description. Instead of waiting until the mark is rounded before the next mark is considered, DTF is determined by going to the next mark as soon as the extended bearing is crossed. If the boat crosses back, then simply revert to distance to the mark. For example B1 has crossed the line, therefore DTF is distance to Mark 3 plus 3-4 distance (141+17=158nm). This compares to 166nm with current method. Now compare B2 and B3. The boat has only moved about 15nm. DTF for B2 = 108nm, B3 = 98nm. Sounds reasonable? The existing method B2 - 112, B3 - 111. I think this method will be easy to compute, so it will not add to the server requirements. I cannot think of any disadvantages, but might be worth testing it with some more examples. Attachments 4-alternative_1.JPG (195.5 KB)
Posted by AGage Posts 411 Registered at 2008-12-27 10:31:57	2012-05-02 06:36:49 - Re: DTF calculation
	My personal solution is to ignore ranking and sail, all will be sweet when we cross the line Your method of calculating is slightly off. Example 1 shows best (at our current situation on VOR6) is how it is currently done, but rather than orthogonal to the line 1-3, the 'scoring line' that extends from 2 to score roundings bisects the angle 1-2-3. OK, in your diagrams this does not make a lot of difference - but shift 2 along the direction of line 1-*3 and you can see it can... This is the BRG value on the NOR - Brg from ship to mark at the instant the server scores your rounding. Just some advice from experience (regarding alternative methods) - you can draw 100 examples and it all works fine, but there is often some edge case where things go nuts... Diagram 4 does appear a better method of computing DTF, but not of scoring roundings (B3 should not be considered to be on Leg3 for example, and could flip from Leg2 to 3 many times on its way to the finish? sim. B1) and we still need the line we currently have extending from the mark to score roundings (and to detect 'unwindings') This would be where is creates more load to compute - 2 lines crossings per mark instead of 1 - actually 1 line crossing (score) and 1 'which side of the line are we' (dtf) (not saying is a bad idea, just commenting because you did about server loads) It's great that others are thinking on things like this though - when you are doing such examples though, try pulling out GE and drawing LONG courses, as these are really the only ones where the ranking ever becomes an issue for some - when a GC path significantly changes where on the line the scoring point is, compared to a Loxo (and also the scoring line itself, being GC) - small distances like in your diagrams don't show this effect. Also GC can change the edge cases a bit - remembering back to the ranking fiasco on Leg 5 while the scoring point was the south pole for pretty much every boat. Of course, the other question is "Is this a good allocation of development resources?" - which always needs to be asked I help develop the client interface for the best online ocean racing sim there is... __/)/)_/)__
Posted by Amigo_do_Boteco1 Posts 4 Registered at 2008-01-18 11:07:00	2012-05-04 01:14:52 - Re: DTF calculation
	Thank you NZL Scotsman This is a very complicate way to measure the DTF. I think its not that hard to implement a way to calculate the DTF based on a point to point measure: - point to point for "mark to mark", and - point to point for "boat to next mark". Its a suggestion to SOL programers. Fair winds --- Last Edited by Armando Faria at 2012-05-04 02:33:41 --- Boteco1 - Make Bore Impossible
Posted by AGage Posts 411 Registered at 2008-12-27 10:31:57	2012-05-11 06:49:06 - Re: DTF calculation
	It is point to point for future legs?? It is actually point to point for all legs, just for the first 2 the exact location of that point is not fixed... It isnt really any more complicated than just computing boat to mark to mark etc, but it is more realistic. It is exceedingly difficult to write an algorithm that can most accurately handle all possible course configurations, sizes, and locations (for example, had the VOR Leg5 been translated N so that it crossed the Equator, but otherwise remained the same shape, we would not have had the ranking issues with the South Pole) If we change it then boats that stay close to the rhumbline will most likely figure higher in the rankings than they are. Personally I would rather appear lower in ranking than I really am. There is no realistic way to have a consistently accurate ranking across the whole race - given the very different configs/sizes/locations of SOL races and the variety of possible routes. Rank calculations and visualisations like you see on the AC45s and VOR inport races only work for pure Windward/Leeward courses, and even then as the wind becomes more shifty they become less accurate. I don't really see there is much wrong with the current system I help develop the client interface for the best online ocean racing sim there is... __/)/)_/)__

Please login to post a reply.

Races

Next Race: 00d 00h 00m

Current Races:

Garagiste's Test Lab

From space Qeshm Island in the Persian Gulf resembles a dolphin. The Question of Qeshm is do we helm our First 27 solo north or south of the island in this 79 nm race. This is a TIMEDrace so you may RE-REGISTER HEREto try again after finishing a run. You will have 13 days and 11 hours to test your skill and decision making after the race opens.
Race #2029
INFOby brainaid.de
First 27 solo Particulars
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking:
TRQ2 - TRCH - SYC
RACE CLOSE: Saturday,
2 April at 23:00 UTC
Race starts: Apr 19th 12:00 Registration will open soon

Facebook

Login

Support Sailonline

Races

SYC Ranking

Series

Mobile Client