The 2006 race from Newport to Bermuda was slowed by minimal wind, with two possible assists from the Gulf Stream that offered the possibility of carrying a boat across patches of light air. To the east of the rhumbline, the main stream of current could carry the boat southeast at 4 knots, through winds as low as 2 knots. Somewhat to the west of the rhumbline, the east side of a rapid cold eddy provided a 4 knot southerly flow through even lighter air. Both options presented the difficulties of trying to enter and exit the stream, and maintain orientation of the boat, with almost no boat speed. Routing solutions for the westerly route were faster, but depended on a very delicate balance of wind and current to make it work. If the wind forecast was off just a bit, the boat could be stranded in next to no air and outside the current. The easterly route looked to take longer, but with a better chance of at least some steady wind.

Not surprisingly, the fleet split in two early on, with almost all of the larger, professionally-crewed yachts taking the eastern route, and the rest taking the western route. As it happened, the western route panned out, with slightly better winds than predicted, leading to an upset first-to-finish victory by the 66-foot yacht, Bella Mente.


The graphic below shows the routing solution from the start off Newport, using the 7-day wind prediction file from Local Knowledge (0.5 deg resolution) and the LK gulfstream current model. The calculations were performed with LK's routing engine, implemented in LK's "Advantage" tactical software. This router is unique in providing quantitative comparison solutions in addition to a quickest "optimal" solution. The user can control how the comparison solutions are constructed, but in general they show the quickest route in various sectors to either side of the rhumbline. Here the comparison routes were restricted to sectors up to the halfway point, and from there were free to find the quickest route to the destination (northeast approach to Bermuda). They give some perspective to the optimal route and are a guide as to what can be expected from other boats that are not following the router's optimal route. (See article about this router in Seahorse Magazine.)

Routing solutions depend critically on the performance characteristics of the boat ("polars"), partly because different boats perform differently under the same wind conditions and wind angles, and partly because the wind pattern is constantly evolving and can change by the time a slower boat arrives at a given point. The computations below were done using Blue Yankee polars (Bob Towse's new 66' boat which sailed in this race).

The graphic below shows an optimal (red) route west of the rhumbline, and comparison routes which take longer. The times to complete each route (in hh:mm:ss) are shown in the "Best Course Comparisons" box, color-coded to match the route. The calculation indicates that the red route west of the rhumbline will take about 97 hours, about 5 hours faster than the blue route east of the rhumbline. (The "black" route produced by the router is the quickest route in the vicinity of the rhumbline, and in this case happens to coincide with the optimal red route.) [Clicking the "Winds" button displays the minimum and maximum wind on each route.]

byall1.gif (68315 bytes)
Another unique feature of the LK Router is the ability to display predicted winds at intermediate points (and times) of the optimal and comparison routes in a single graphic, shown below. The wind vectors are all comparable at the beginning (north end) of the routes, in the 12-13 knot range, but vary significantly therafter. All of the routes pass through a region of light air (short arrows) roughly in the middle of the course. By clicking the 'Winds' button on the 'Best Course Comparisons' box (above), you get the alternative data shown below, listing the minimum and maximum winds encountered along each route. The minimum value for the optimal red route is shown as zero, but may in fact be some tenths of a knot (the values are rounded to the nearest integer). Of course, there can be local conditions which produce light and variable winds at any given point that do not show up on the overall forecast.

All of the routes clearly pass through very light air, but the blue route does somewhat better, including winds of at least 2 knots all the way. Despite the lower minimum wind, the red route is predicted to be faster because of how it utilizes the gulfstream currents. The question is how sensitive the route is to accuracy in the wind forecast and how feasible it is to handle the boat in a 4-knot current with next to no boatspeed.

The LK router also includes the ability to specify minimum (and maximum) acceptable wind speeds for the routes. The minimum setting can be useful to eliminate routes too close to "holes" in the wind pattern, which can make the result very sensitive to the accuracy of the forecast. When this routing is repeated for a minimum wind of 2 knots, the "optimal" route is found to be identical to the blue route (this is not true in general, but this coincidence simplifies the discussion below).

byall2.gif (84226 bytes)
The red (or western) optimal route is able to take advantage of the south-flowing edge of a very strong cold eddy, located just south of the main gulfstream current (see image below). The blue vectors in the image are proportional in length to the current, which exceeds 4 knots in both the eddy and the main stream. The blue (or eastern) route utilizes part of the main stream to take the boat southeast. The red route stays closer to the rhumbline (so the boat sails a shorter distance) and makes better use of the current, but takes the boat through what is predicted to be lighter air. The blue route is safer in the sense that the boat is more likely to find consistent wind, but the boat must turn south out of the stream to avoid sailing even greater distance and does not utilize current as effectively.

In the graphic, the animation ("A") button on the 'Best Course Comparisons' form was used to bring up the controls which enable the user to step along the routes and display predicted boat positions at intermediate timepoints (shown as black dots along the colored routes). The intermediate time shown is 2018 hours on 6/17, at which the "boat" on the red route is approaching the northwest edge of the eddy. The numbered boxes on the left show reported positions of Bella Mente at various times. At 2018, she had reached box 6 (shown in yellow). Her rating of 433.2 is slower than that of Blue Yankee (403.3), so the graphic suggests that she is several hours ahead of predicted location. This was also the case for Blue Yankee, whose positions in this graphic are shown by the numbered boxes to right of the blue route. This was apparently due to winds in the early portion of the race being somewhat higher than predicted.

bybm.close1.gif (66186 bytes)
Despite the strong eddy current, the boat on the optimal route is predicted to take almost a day (some 22 hours) to sail to a position south of the eddy. The black dot on the red route in the graphic below shows predicted position at 1757 on 6/18. The wind arrows at intermediate points show very little wind (less than 1 knot) at the southwest edge of the eddy. In comparison, winds on this portion of the blue route are light, but are at least 2 knots. Despite this, by sailing lesser distance and getting a bigger boost from the gulfstream, the boat on the red route is predicted to gain about 40 miles during this period. This is derived from the right-hand column of the 'Best Course' box, which shows distance to the destination from the black dot on each route. At 2018 on 6/17, the boat on the blue route is about 3 nm behind (394.1 compared to 391.3), but by 1757 on 6/18 the difference is 293.2 compared to 250.9.
byall3.gif (83937 bytes)
An overall view of predicted wind at the same timepoint (1757 on 6/18) is shown in the graphic below.
byall4.gif (79682 bytes)
By the fourth day, a change in the wind forecast allowed Bella Mente to take a more direct route to the finish. The graphic below shows the original routing solution using the forecast available at the start. This indicated a region of very light air along the rhumbine to Bermuda (e.g., from Bella Mente's position "13" on the graphic), causing the program to route around to the south of it. The graphic shows predicted winds at 2222 on 6/19, at the time Bella Mente reached position "18." The new forecast for the finish is shown in the next graphic.
byfinpredict.gif (72586 bytes)
The updated forecast on 6/19 was much more favorable, and allowed a direct final approach to Bermuda. This gave an additional boost to Bella Mente and the other western-route boats, compared to Blue Yankee and the eastern-route fleet, which had been expected to close somewhat at the end. In the final analysis, the riskier but high-gain western route paid off, for Bella Mente and many other boats that went west.
byfinactual.gif (90958 bytes)


Next time you do the Bermuda Race, take "Local Knowledge" aboard ...