Our latest column by Kevin Hall and brought to you by Mauri Pro Sailing
It’s blowing 20 knots, offshore toward Santa Cruz Island. There’s a 2-3 foot swell and a decent beach break, mostly rights. A few people are out surfing. The kids are playing in the sand, and I am mesmerized by the pelicans, as I have been since I was a kid.
I would be surprised if they are born knowing how to do it. I suspect they learn by following, by copying, and by practice which might include some experimentation on days after good fishing. If there is a any swell, they can fly down the coast at an astonishing pace, without flapping their wings once. For those who haven’t seen it, they fly inshore of and very near the crest of the clean part of the wave with their wingtip nearly touching it. The moving air there provides enough lift that they can be perpetually “descending”. The first trick for them is to abandon a dying wave with enough speed to pull up and bank steeply offshore, to drop in to the next one. The remaining tricks must involve refinement of this technique, and refinement of reading the waves ahead and offshore, to perfect their timing. The kids let me watch for an hour straight – Merry Christmas, dad.
Recent America’s Cup fans are familiar with the process of the pelican : copying, then practice and experimentation, then the development of speed through refinement. Consistent upwind foiling was not born overnight. That said, the commitment made by Oracle Team USA while on or near match point, to setup the boat for foiling and race that way, may have been. I would like to try to shed light on how lots of data and computer modeling could potentially help an America’s Cup team learn to foil upwind.
First, let’s look at some strategic cons of taking on foiling upwind during the relatively limited time each team had before the Cup. In hindsight, seeing how crushing a technique it is, we must ask the obvious question : why didn’t Oracle come out of the blocks doing it? The main potential cons I see are known in advance : 1. it might be faster in spurts, but hard to maintain and hence still slower for an entire beat, 2. it might require a special rudder lift setting, and hence be too impractical for the techniques and settings we have developed for downwind and gybing, 3. it might be impossible to prove it is faster with one-boat testing, 4. it might be impossible to prove it is faster with two-boat testing, 5. even if we prove foiling yields a better VMG, it will probably be lower and faster and make us bounce off the boundary getting dizzy with tacks.
Two-boat testing is my chance to mention a few relevant lessons from America’s Cups past. I was on-water testing manager for OneWorld Challenge leading up to 2003, and for Emirates Team New Zealand for most of the campaign leading up to 2007. It’s possible some of those lessons bit teams in the ass again. This is, perhaps, excusable. Maybe they don’t apply once you add a hull and subtract twenty tones of lead.
Cup teams used to tow out, hoist sails, and take off in a straight line upwind, swapping sides to make it fair. Guess what? Sail entries got flatter and flatter, targets evolved toward a high mode because tests were always won in height, and when it came time to race (which included starting, tacking, footing, and some pinching), that team with it’s flat-entry sails and higher and slower targets was ill-prepared for battle. Downwind testing was harder. You could test for a few months and send your data off to be analyzed by super geniuses, and then get back a statistical study peppered with words like “regression analysis, stochastic error, non-gaussian…”. The study might prove that the boat with less wind won every test. Hmm. Or, you could ask the sailors at the end of any day of testing, and learn that the boat that had more wind in any given two-boat testing lineup, won the test. The sailors usually knew which boat they would pick for a race, though, and they were usually right. They often knew why, too.
In short, the testing wasn’t a faithful model of the racing. The map wasn’t the territory. Back to foiling upwind. We learned a long time ago, from Frank Bethwaite in his book High Performance Sailing, that if a lower, faster mode gives equal VMG, it’s usually better. We carry more speed into the lulls so we have more apparent wind to get through them, and we carry more speed into the tacks. On the new AC course sailing lower hits the boundary sooner, and it was believed for most of the lead up to the cup by most of the teams, that tacking an AC72 was extremely costly. As a quick aside, I fault Oracle for not picking up on ETNZ’s tacking technique much sooner.
Why might it be hard to prove foiling upwind is faster than low riding on an entire 3-5 tack beat? When we’re sailing upwind, leeway is a much bigger deal than when we’re sailing downwind. It’s still key to the automatic ride height process, but it takes us away from the top mark and also makes the apparent wind go forward which makes us aim lower again. For an impressively clear explanation of how ride height works on the AC72s, and leeway’s contribution to the equation, see Dario Valenza’s article here.
OK. We need to set our rudder foil angle of attack (AOA) before the race. If we add more rudder lift by changing the angle of the rudder horizontals to the horizon, our takeoff speed will be lower, but the amount of lift they produce, and the amount of associated drag, will be much higher at high speeds. So the boat will behave differently downwind to what we are used to. We probably saved all our data from learning to foil downwind. There may be days where we came in and the main conclusion was something like : too much rudder lift, impossible to sail downwind.
Suppose we have an extremely accurate 6-axis motion sensor. I don’t think there are any good ones that are cheap, but anyway. Those axes are here (the bow of our boat is coming out of the page and headed to the right). On a TP52, it’s great to have a really good heading, heel is pretty important, and we use pitch a fair bit to help with crew weight through the range (but flip the sign and call it trim). On our AC72, accurately knowing our heel will help – more heel adds “effective cant” (which determines the vertical component of lift of the vertical part of the daggerfoil. If the daggerfoil is set to its minimum leeway position, all-the-way vertical relative to the horizon, and the boat isn’t heeled, the vert produces little upforce, and the amount it does produce depends on how straight (vs. curved or deformed) it is. If it is perfectly horizontal (theoretically only because it can’t go that way in the boat), it produces all upforce.
While sailing, the vert part of the daggerfoil is somewhere in between those two extremes relative to the horizon, determined by the combination of its position in the boat and the boat’s heel. More heel at the same speed causes the front foils to produce more lift. However, more heel also reduces the vertical component of rudder lift until the windward rudder comes out of the water, when the force pushing the back of the boat up drops to half what it was when both rudders were below the surface in clean water. All that to say, heel somewhat affects the pitch stability of the boat at a given speed.
“Effective pitch” of the daggerfoil determines (when combined with boatspeed through the water), the amount of lift the horizontal part produces. Like effective cant depending on heel, it is a combination of the foil’s position in the boat, and the boat’s pitch relative to the horizon. When the boat is level, as in the middle of a gybe, the amount of forward lift of our system includes the effective pitch of the opposite daggerfoil when it hits the water. Note also that in a tack or gybe, for an equal effective pitch the foil in the hull on the outside of the turn is moving faster through the water than the foil on the hull in the inside of the turn and so producing more lift. Turn rate is one of the things that helped the rolling tacks work so well.
Right, so we’ve got boatspeed through the water, roll (heel), and pitch (trim). Why “boatspeed through the water”? If there’s no current, our boatspeed through the water equals speed over the ground (SOG), which we can measure with the GPS. What about if there’s 3 knots of current? The foils don’t care about SOG, they produce pressure gradients relative to their movement through the water, their BSP. So our SOG might read 20 knots in max flood, or 26 knots in the ebb for an actual takeoff speed of 23 knots. If we have an excellent current model, we could display BSP on the wing, not just SOG. This would require running a current model onboard. We’d use our instantaneous position and time to look up the current, then reverse-engineer our BSP and HDG from SOG & COG. This might help our sailors learn the boat more quickly because takeoff speed would always be the same number. Except the times the model wasn’t perfect, of course. We’ll come back to models.
Part 2 tomorrow…