Photo by Tim Wright

Sailing Through Cyberspace

Fred Hapgood

In one of his texts the great sailmaker Tom Whidden takes a moment from the details of construction and upkeep to list the reasons why people sail. He mentions the paradoxical balance of the speed rush with the glow of serenity; the thrill of moving inside the body of the wind, of merging with a force that can only be felt, not seen; the historical resonance in a technology that is six thousand years old. But what Whidden seems to treasure most is knowing that every moment is a corner around which lies a new tangle of forces, a situation never encountered before, either by the sailor or possibly by anyone. For the master sailmaker it is the sense of feeling each moment flower and renew itself that makes sailing so absorbing. Every voyage is a great novel with no last page.

However, it is a fair wind indeed that blows no ill, and another consequence of this fathomless complexity is that it has forced the science and (therefore) the engineering of sailing to develop very slowly. Doing science means controlling all the factors but one, manipulating that one in a controlled fashion, and then reading the results. Controlling all the variables in sailing -- the pitch of the waves, the character and conditions of the sail fabric, the flexions of the spars, the circulating and intermeshing patterns of turbulence in air and water, to mention a few, is hard enough to manage in tow tanks and wind tunnels, let alone tow tanks in wind tunnels; that level of control is a sheer fantasy out on the ocean itself. This is why sailboat designers still find themselves blindsided from time to time by developments like Australia II's winged keel.

Horst Richter, professor at the Thayer School of Engineering, is exploring a second path into the field: virtual boats that sail through cyberspace. In theory in a computer every detail can be controlled and every change read. All the forces and geometries that are invisible in the real world -- the streamlines and vortexes, the points of flow separation and the regions of turbulence, the pressure differentials and velocity gradients -- can be made as visible as the boat itself. Once made visible, they become grist to the extraordinary pattern-finding and recognizing powers of the human brain. Again in theory, such illumination should accelerate the science and engineering of sailing prodigiously; the downside is that the computational resources required lie somewhere between enormous and astronomical, and for all the progress our machines have made they might not be up to this yet.

A specialist in thermodynamics, Richter started vectoring in on this subject in 1992, when he was volunteered to teach a course in the science of sailing. After surmounting a few obstacles -- nothing on the topic had appeared since the 30's that the librarians at Baker deemed worth collecting -- ENGS 2, The Technology of Sailing, was launched and Richter started assigning research projects to students. Examples might be the 1993 study by David Tabors '93 on the loss of boat lengths while tacking, or the computer program written by Alex S. Goldenberg '95 that modeled the whole profile of velocity changes during a tack.

Computer modeling has been done on hull architecture for some time, but less often on sails, which are more complicated by orders of magnitude. (Though apparently the first use of sail simulators was in the early 70's, in a collaboration between sailmaker Lowell North and fluid dynamics specialist Heiner Melder.) In 1996 Peter Neiman '96 added to this short history by writing a simulator and comparing the results with a suite of tests run in wind tunnels using real sails. A good overlap in results would have suggested that simulators could be taken seriously as an instrument in sail science, but the information from the tunnel tests was too limited for conclusions to be drawn either way.

In 1997 Richter and Jeffrey D. Shoreman ('97), then Captain of the Dartmouth sailing team, acquired better data and ran a more detailed comparison. These results agreed to within 5%, a number interesting enough to attract the attention and then the sponsorship of Young America, a consortia of six American yacht clubs (New York, Detroit, St. Petersburg, Chicago, and Anapolis MD) fielding one of the six American entries in the next America's Cup Challenge, which was to be defended by the Royal New Zealand Yacht Squadron in 2000. (www.youngamerica.org)

Ever since the first race in 1851 (The America's Cup is the oldest competitive trophy event in sports) the Cup has been about pushing the edge of what could be done with a sail in a boat. If anything, that character has intensified in this century. Young America was wrestling with the chronic problem facing racing sailboat designers: lots of open questions and no good way to test the possibilities.

"We have two boats we can race side-by-side," says Duncan MacLane*, Design Technologies Project Manager, "but if you run them too close they interfere with each other and if you run them too far apart the sailing conditions are no longer the same." MacLane says there are no settled answers as to the right camber and twist (three-dimensional structure) to give sails, or the optimal arrangement of the rigging, or the design of the deck surfaces. In theory the tenth of a knot that would decide where the Cup sits in the opening years of the next century might lurk in any of these, or (more likely) in some combination of them. The consortium hopes to use the Richter project to drive the fog out of these questions.

To do that Richter, two graduate students (names?), and Olin Stephens, one of the great American yacht designers, have built a quasi- realistic sailing environment in software. (Quasi- in that the sea is flat; contemporary computers are not fast enough for wave action or boat pitch to be incorporated in the models. "Maybe in ten years," Richter speculates dreamily.)

There is much to do even so. Perhaps because so few learn sailing in a classroom even skilled sailors often misunderstand how complicated sailing is. Sails are like wings only on the most abstract level; they do generate lift, but in nothing like the same way.

When wind blows into a sail it piles up into the belly, compressing itself. If the boat carrying the sail was lying on the beach the compressed air would spill around on all sides of the sail into the lee side, equalizing the pressures. A moving boat puts a cork in this process. The air in the belly finds it hard to move up around the leading, forward, edge of the sail, since that means pushing against the currents flowing toward the stern. The compressed air can spill around the rear, trailing edge easily enough, but once it has it has much the same problem in flowing upstream into the (relatively) low pressure area -- it has to do so against the air moving down over the lee. This latter current forces the air moving around the rear edge back upon itself, creating what is sometimes called the 'starting' vortex. (While this cannot be seen in the atmosphere, an analogous process happens with the keel, and starting vortexes are routinely visible in the water as they shed off the boat's track.)

Blockaded from taking the shortest route, the compressed air in the belly shifts to plan B, creating a second vortex (the 'bound' vortex) that goes around the sail in the other direction. The two vortexes fit together like gears in a gear train. (Rotational energy is one of the features nature likes to conserve, as if one clockspring could only be used to wind another.) If we imagine a boat sailing from right to left with the wind blowing up from the bottom then the starting vortexes will flow counter-clockwise and the bound vortexes clockwise. The flow from the bound vortex flows into the low-pressure area on the lee while exerting a forward force on the windward side of the sails. It is this that pushes the boat.

The task of virtual sailing is to enter this story into fluid modeling software (Richter uses a program from Fluent, Inc., one of New Hampshire's rising technology concerns) and then run a specific sail design inside it. The simulation is built up out of cells, computational entities that capture the smallest possible unit of relevant physical change. (Each cell has a size, since changes take place in space, but that size changes with the importance of the change it contains. A cell representing the flow of air some distance from the boat might be a cubic meter; a cell capturing the flow of air around the edge of a sail might a cubic millimeter.) Each cell calculates its own characteristics (mass, momentum, temperature, and so on) from equations that combine the innate physical properties of the material and the behavior of the cell's six -- in the case of cubic cells -- neighbors. When the computation is finished, each cell sends the results to its immediate neighbors, which they use to figure out their condition during the next tick of time. Richter works with hundreds of thousands of these cells and would very much like to work with millions.

Every day Richter's team enters a sailing situation -- a sail geometry, a weather condition, a hull design -- into the computers and then waits a day or so for a statement representing one second or less of actual sailing to emerge. Sometimes these situations come from the client and sometimes from each other, from what Richter calls the 'crackpot ideas' that come in the shower. (He may call them crackpot but he refuses to discuss them till the race is over. Then, he says, he will have a lot of write about.)

Whatever the importance of sail simulators in the Cup race they might have profound effect on sailing itself. Eventually personal computers will get powerful enough to run these programs right from the desktop (or a company might sell access to a mainframe running a sailing simulator over the internet). From there it is a short step to letting everyday sailors import digitizations of their own boats into these simulations and enter their own local sailing conditions. When that happens they will not only be able to order sails optimized for their boat and locale; they will be able to see the steamlines and vortexes and regions of turbulence. They -- we -- will be see what happens and what matters, and from those landscapes learn the arts of sailing that up till now have been locked in the intuitions of its few real masters.

For instance, many sailors believe that the flow of air speeds up between the main and the jib, creating lift and generating power. This causes them to yell at crew for 'standing in the slot'. In reality, the 'slot effect' is a myth; the two circulation fields cancel each other out. This is instantly obvious in a computer simulation. No sailor who sees this simulation once (and who believes it) will worry about blocking the flow through the slot again.

The outcome of races depends on the design of the boat, the skill of the skipper, and luck -- what the winds give or withhold. Simulators will narrow the gap between boats and the gap between skippers, thereby increasing the proportionate role of the random accidents of wind and wave. You might think sailors would object to this, but the results of an informal and completely unscientific poll I have been taking over the past few days are to the contrary. Perhaps sailors are so egotistic that they think they will win anyway, or perhaps they feel that accepting the favors of nature, whatever they are, is really the essence of the craft.