Section 8  The problem of gathering test data on sails

 

I wrote a first version of this section and posted it on the Boat Design Net Forum. The response made it essential to redraft it. I learned a great deal that I could not gain by experience. Unfortunately, as this forum seems to operate on everyone being incognito, I cannot acknowledge contributions.

 

I think that the most important statement made was that the commercial use of sails as a means of propulsion of surface vessels is not seen to be sufficiently viable to attract funding for any scientific investigation and that, in fact, sails are used exclusively for leisure activity. These two facts taken together have led to a situation where there is no agreement on how sails actually work, no structured test data, and, at the same time, lots of people keen to design and build their own boats. Inevitably some people will design in the total absence of any theory or data and others will make every effort to understand the way in which sailing rigs actually work and find any data that is going in order to enjoy the whole design experience more. Most will lie somewhere between these extremes.

 

In Section 1 of this website I analysed the way in which the single soft sail works to drive a boat in the best way possible and went on to extend this analysis to the Bermuda rig for the same purpose. In retrospect I should have looked at the soft sail in more general use. Hence this extra section.

 

I watch leisure sailing boats and it is often evident that not much effort is going into sail setting. Some people appear to set the rig to get the boat on the move and then just steer. They can do this because the soft sail is not particularly responsive to the swinging of the wind and its continual variation in speed. (See my section on wind shifts to understand this feature of the natural wind.)  The sail behaves in this way because it is always stalled and, if it is set at a high angle to the mean direction of the relative wind, changes in the speed and direction of the wind just increase or decrease the force on the rig without any sudden changes in the way that the rig is working. (See the section of wing sailing to see what happens with a rigid wing.)  Then the hull with its very non-linear resistance to motion does not speed up as much as one might expect when changes in the wind increase the force on the rig and conversely it does not slow down as much as one might expect when the speed drops and the angle of attack increases. So idling along is what a sailing rig does best. If the rig is to be raced then the sailing rig must be trimmed continuously because the greatest drive comes just before the sail collapses and the wind is never steady. Furthermore there are two or more sails to be made to interact if the boat is to go close to the wind.

 

So the sailing rig is used for one of two purposes, idling and racing and the racing use of sails has led to a desire to have some scientific explanation of how sails work, that is, to a “theory” in fact.

 

From my earliest contact with the “theory” of sailing it has seemed to be hopelessly confused. I could find nothing that I recognised as mainstream science. Originally I had taken it for granted that there would be a body of knowledge like that for aerofoils but there is not. It was only recently that I wondered why this should be so and decided to find out. I came to the conclusion that understanding sails is really about understanding sailing rigs and that it is possible that an engineer is best placed to do this. I will try to explain why.

 

Aerofoils and sails present scientists and engineers with two quite complicated devices to put to use. There are obvious differences between them, the most obvious being that the aerofoil is rigid and the sail is made of fabric and floppy. Mostly aerofoils work in steady winds as for example the wings of aeroplanes or the blades of turbines where streams of fluid flow steadily over the aerofoils. They find application in hydrofoils of submarines, propellers, the keels of yachts and so on. The one thing that is common to all of these applications is that the aerofoils are worked so that the fluid flows smoothly over both faces of the aerofoil and so it is not stalled. However there is one important application of the aerofoil that is not in this group and that is the wind-powered turbine. Here the aerofoil is rigid but the flow of wind is not steady in either direction or speed from instant to instant nor is it average speed and direction steady from hour to hour. The aerofoil may then have to work in both the stalled and un-stalled condition. By comparison sails always work in the stalled condition. In many ways it is much better to work either in the stalled condition or in the un-stalled condition and not in both. It is the blade of the wind-powered generator that is the troublesome application not the wing or the sail.

 

There is one other serious difference between aerofoils and sails. The aerofoil normally works within a range of angle of attack between 0° and, say, 12° and stalls at, say, 12° to 16°. It is almost never worked beyond the stall. Normally the stall is reached by increasing the angle of attack. The sail is always stalled and it works best, ie for racing, from the smallest angle of attack for which the sail actually inflates up to 90°. This is a range from 30+° to 90° and at the smallest angle of attack we get the largest force to drive the boat from the curved flow over the sail and at 90° the sail is simply obstructing the flow of air and the force to drive the boat is wholly drag. Clearly these two devices operate in quite different ways.

 

So we have the un-stalled aerofoil to use and the stalled sail to use, one commercially and the other for leisure. Engineers need to understand how aerofoils and sails actually work if they are to use them in design. Their normal approach to such problems is to start with observation and to go on to an analysis of the mechanics of the device to create something that I call physics and then to gather test data in a way that makes the physics as easy to use as is practical. Then the test data is used in a quantitative design process to find dimensions and speeds and powers and so on to permit the mechanical design of a practical device. It is not a once through process but a backwards and forwards iterative one. An observation leads to theorising and to experiment, more observation and refinement of theory and so on.

 

I think that there is some sort of agreement about how aerofoils work amongst those who study these things. The strategy then adopted by scientists and engineers in order to find ways to use aerofoils is to gather data on the performance of particular aerofoils that is so accurate that it is unquestionable and then to use the ordinary methods of engineering design including judgement to make use of the data. For a variety of reasons this strategy works well and we are very fortunate that it does.

 

Regardless of the application it is central to any process of gathering data for science or engineering that the test conditions are refined to the point where anyone can understand them. Sadly much data is published and used without this essential information. Aerofoils are tested in wind tunnels to gather data on their performance. The flow in the rectangular working sections of the wind tunnels is very nearly uniform in every way, the velocity is almost the same across the working section and the flow is very nearly free from fine grain turbulence. Anyone can understand the test conditions and these conditions are not very different from the actual air in which an aeroplane flies. The sections tested are wide enough to span the tunnel, big, rigid, made to very high standard, and are two-dimensional. The shape can be specified by some coding system and then any other similar aerofoil can be made using the code and one length, typically the chord. These sections are tested singly over a range of angles of attack of say -15° to +20° and the lift (the upward force) and the drag (the horizontal force) are measured. Whole families of aerofoils are tested and anyone can understand what was tested and how it was tested. When all the data is acquired it is stored in a way that depends on the use of simple expressions, combined with the use of Reynolds’ number for both storage and retrieval and it is only because of this system that the data is condensed sufficiently to be manageable and even so it is still a lot. (See my section on aerofoils in my web site to see how this data is used in a systematic way.) Data from all tests can be added to this information bank provided that they are of the same standard.

 

Sails represent a quite different challenge. They operate at low speed compared with aerofoils. They operate at low altitude where the natural air is full of swirls and eddies and larger rotations and close to the disturbed surface of the water as well. The uniform flow in a wind tunnel is not representative of the natural wind and it would be difficult to take the difference into account. The sail is made of fabric and cannot have a shape that is defined by some two-dimensional mathematical expression as is commonly used to describe the shape of aerofoils. A mainsail is not one device but two, the sail and the mast to which it is attached and, in use, the sail varies its position relative to the mast. The methods of attachment vary, the grooved mast creates no gap between the sail and the mast but the use of mast rings or a jack line creates a gap of unspecified size. So, where we had a code and a length for an aerofoil, for a sail we have some uncertain and possibly changing shape and at least four unrelated dimensions to contend with. This alone makes systematic testing hopelessly unwieldy and there is nothing comparable to Reynolds’ number to reduce the bulk. 

 

Unlike the aerofoil the sail operates at only two significant angles of attack. The first is about 30+° and is the angle at which the sail first fills to take its working shape; the other is 90° when the sail is simply blocking the wind. The sail is interesting in the small range of angle of attack around its inflation point. If one observes a sail on a boat in irons and notes what happens as the boat falls off the wind it is evident that the sail flaps in a random way until an angle to the wind is reached when the flapping dies away and the sail inflates to take its working shape.[1] This is clearly an event in the relationship between angle of attack and the performance of the sail. A further slight increase in angle of attack beyond the angle for inflation makes the fabric of the sail go “hard” as the pressure pattern on it is created. This is the angle at which the force on the sail is a maximum and is therefore of interest. If the sail is allowed to go to a higher angle of attack and then the angle reduced so that the sail reverts to the flapping state the onset of the transition is evident as a tiny flutter at the leech of the sail and the flutter increases until the sail starts just to flap. On large racing yachts designated members of the crew spend all their races altering the angle of the sails to keep them just on the point of fluttering because this is “known” to be the angle of greatest drive. The second condition of just blocking the wind is not much good for running before the wind but it is an option. Usually, if the boat is to run, a massive additional sail is set, the spinnaker, just to get a very large area.

 

Clearly if one were to undertake tunnel testing of sails the behaviour of the sail in the range of angle of attack that extends either side of the point of inflation would be the first thing on the test programme. This could not be done using a rigid substitute for a sail. The notional sail to be tested in a tunnel can be derived by comparison with the models of aerofoils used in wind tunnels. This notional sail would be a rectangular piece of fabric attached to a dummy mast and, when inflated, still be rectangular between the side walls of the tunnel with no leakage, of the same profile from side to side, yet capable of flapping at lower angles of attack. It is not possible because, in addition to all the other problems, no real sail can be two-dimensional. Indeed it is difficult to think of any way in which a sail can be tested in a wind tunnel. There is no way that a sail can be set up for testing so that the test conditions are such that anyone can understand them.

 

I have seen pictures of complete model yachts being tested in wind tunnels. No doubt there are things of a comparative nature that can be determined from such testing but I can think of no way that data from such tests can be fitted into any coherent strategy of testing based on some sort of physics. There are far too many variables for the results from the model to be related to any other yacht.

 

One might think of testing curved metal plates with suitably shaped leading edges to simulate a mast or a wire but the range of angle of attack will be from +25° to +90° and this calls for a large tunnel to accommodate, with sufficient space above and below, this large sail held square to the flow. When these results are gathered and published they will always be suspect because the model is tested in a steady uniform wind and the real sail is used in the natural wind, and the engineer will have a serious measure of uncertainty in his baseline data. However one must never suppose that the outcome will have no value. Until you try you cannot find out and I am sure that this will have been done.

 

So one must conclude that there is no body of data for sails that is remotely comparable with that for aerofoils and that it will never be possible. This absence has led to all sorts of explanations of the action of sails and some of these explanations fly in the face of accepted science. No wonder that there is such confusion. Another consequence is that if you cannot test a sail or a sailing rig in a tunnel and if you choose to design and build a boat with a new type of rig you cannot know whether your new design is successful and, if you claim that it is successful by some yardstick of your choosing, no one can dispute your claim.

 

This accounts for the extraordinary amount of experimenting that goes on in sailing boats and the plethora of explanations of sail action. This is the situation that exists and, perversely, it gives people a great deal of pleasure. People like backing hunches.

 

If we cannot have any controlled testing of sails, books that set out to create a “theory” of sailing based on mathematics are unlikely to be successful. The most that can be expected is to apply ordinary mechanics to a sailing rig and to its fin etc. to show how the ideas of lift and drag from aerofoils can be used to explain the action of sails and the action of keels and rudders. Then the book can be fleshed out with data for the resistance to motion of actual boats that have been tested and lots of attempts to apply mathematics to this totally intractable problem.

 

This opens the question of what can be done to gain some working knowledge of sailing? I think that one must start with what we can know. In the first section of this web site, http://www.ivorbittle.co.uk, I offered an explanation of the action of a single sail and of its limitations. From that I developed an argument for using two sails together to improve the upwind performance of a yacht. I do not think that anyone can question the fact that fore-sails are used in this way regardless of how they think that sails might work and they were used on square-rigged ships for a very long time simply because the fore-sails pulled out of proportion to their size. However a fore-sail can only function properly if it can work in the air that has been diverted by the sails behind it. This means that the fore-sail must be close enough to its larger, following sail for this interaction to take place and be small enough to fit into the region in which the air is being diverted. Generally it is better, and possibly necessary, to use a rig made up of several sails working together rather than to use just one sail of the same total area. However, many rigs are made up of several sails some of which do not work with the others especially when someone tries to design a new rig. I like to classify rigs as seamanlike and un-seamanlike. In seamanlike rigs sails are not fitted in every space created by the spars and the stays just because they are there. They are fitted in those spaces where the sail can contribute effectively to the overall performance, especially upwind. The boats that are most likely to have seamanlike rigs are working boats ranging from windjammers through tea clippers to coastal fishing boats of every description. In these boats a sail that is not contributing enough to justify setting it is simply not used. It is not worth stowing it and looking after it if it does not pull its weight. If you look at seamanlike rigs often enough a single shape for rigs when beating emerges and it becomes increasingly apparent how a rig must be set up to beat properly and then how it should be set up to reach.

 

This means that, regardless of the fact that these rigs are operating in the natural wind, there is repeatability in their behaviour.

 

Once a sail has been made its shape can only be changed within fairly small limits by its arrangement on the standing rigging. Then there will be a best angle for the sails to be set for beating and, as the crews of racing yachts continuously adjust the sails to keep them just off the point of collapsing, we must suppose that this is the best way to operate the sails. Now we can go on to suppose that the shape of each sail when set up for beating is also the best shape of these sails for all other points of sailing so there is no need to keep changing the shapes, we need only to move the sails to suit the angle of the wind.

 

So even if we do not know how sails work we do know how to make them work.

 

From this point engineers would try to find out how a sail actually drives a boat. The first goal would be to analyse the most simple version of a sail with all the important parameters considered but not things like sail twist and heel. They would look at a two-dimensional sail or perhaps a small part of a sail that can be taken to be two- dimensional. I have done this in the first section of this website and, using realistic coefficients of lift and drag just as one would for aerofoils, it is possible to derive the mechanics for a single sail and for two sails acting together. This gives a workable understanding of sail action that matches the behaviour of the rig. The analysis shows why a spinnaker is necessary for running before the wind.

 

That analysis also shows that the drag has a serious effect on beating close to the wind and other points of sailing into the wind. This could not come so easily from experience in using the rig. The drag is in two parts, the drag of the sails and the drag of other items like the rigging and the hull. The drag of the sails is in part caused by the production of lift but it is also caused by the way that the mechanical parts work together. The figure shows the relative positions of a sail and its mast at several positions relative to the centre line of the hull. They are all for valid points of sailing. No one can pretend that this is ideal yet the mechanics of a rotating mast are very formidable. The performance of this mast-sail combination could be tested in a tunnel using a rigid plate instead of the fabric sail. So could other arrangements like a jack-line or mast rings.

 

It follows that it is possible to test in a wind tunnel to find out how to reduce the drag from the rigid parts of the boat.

 

When a theory, like this one for sail action,  is not supported by testing, we must look for other support and, in science, this comes from using the theory for prediction. I found that I was able to predict the way in which sails are made to work together for rigs that were new to me like say that on Team Philips.

 

I also found that some rigs are set up by people who appear not to fully understand how they work. I have pictures of two square-riggers on the same point of sailing with sail twists of opposite hand. One of them must be wrong. It is quite possible to accept that the speed of the wind is least just above the water and increases with height in some non-linear manner. Then the speed and direction of the wind and its direction changes relative to the boat with height and sails must be set to take this into account. We often look at seamanlike rigs that are badly set and these confuse the issue and one must learn to discriminate so that only the rigs that are both seamanlike and set up properly are studied.

 

There is another outcome of this problem of not being able to test. People look round optimistically for mathematical approaches using “known” physics for some vaguely related system. The outcome of such activity is often very doubtful but in the absence of indisputable data the work is likely to persist. Such mathematical approaches are made in the design of hulls in the heeled attitude, for wind gradient and the twist of sails and the performance of sails on a heeled boat and so on. They become difficult to displace like the tap explanation for the modus operandi of the Genoa. Treat them all very carefully, they may be wrong.

 

The fact remains that the modern racing yacht is an astonishing achievement to have come about by evolution and not by theory-led design.