BMW ORACLE Racing used STAR-CCM+ to design world's biggest wing

By now, almost everybody even remotely interested in the America's Cup is well aware that BMW Oracle's yacht USA is powered by a solid wing sail instead of a conventional sail. CD-Adapco, the company that makes the CFD software used in the design of the wing, which happens to be the world's biggest, sent today an interesting interview with BMW Oracle's CFD Manager Mario Caponnetto. There is the obvious sales pitch in the interview but Caponnetto gives a number of interesting technical details.

Why this choice of a rigid wing on your trimaran, instead of a conventional sail? Isn’t this a radically new and risky choice?

[Mario Caponnetto] Rigid wings are not really radically new in yacht racing. They have been used in high performance catamaran races and others racing boats for many years. By the way, a rigid wing first appeared in America’s Cup in 1987. What is radically new is its size: the wing, with its 57 meters above deck, is the largest wing ever, 80% larger than a 747 aircraft wing. No one in our team had designed anything like this before, and this scared us a little bit at the beginning. Starting from white paper and evaluating pros and cons, we decided to move forward and quickly in the project. This project came true thanks to the enthusiasm of our chief designer, Mike Drummond.

What are the benefits and the shortcomings (if any) of a rigid wing with respect to a conventional sail?

[Mario Caponnetto] The main advantage of a rigid wing is shape control. In other words, depending on the angle and the velocity of the wind, there is an optimal sail geometry that in turn optimizes the aerodynamic pressure field. This makes it possible to extract a maximum propelling power from the wind or in other terms to maximize efficiency. On a conventional sail material works, from the structural point of view, like a membrane and shape control is difficult. Some specific shapes are impossible to obtain and the final shape is a compromise. With a rigid sails, shape is much easier to control without compromises. Furthermore, during navigation there is always a feedback between imposed shape and achieved shape, whereas with traditional sails it is already an issue to identify the sail shape during navigation.

I guess the rigid wing benefits have its downside in terms of weight?

[Mario Caponnetto] Not quite so! A conventional sail supports only traction loads and not bending loads. The wing, having a thickness, makes it possible to distribute loads on the two sides of the structure that at the end results to be very light. To sum up, the rigid wing weight is comparable to a conventional mast/sail system. With a one-dimensional analogy, we should think of a sail as a rope supporting a weight (the wind pressure) at its center. If one wishes to reduce its sag, tension will increase; therefore its thickness and weight should be increased in order to avoid breakdown. If we replace the rope with a cantilever, the weight of the structure will be smaller, given the same displacement. Let us think that huge forces are required to put into tension a conventional sail, to the point of stressing the boat structure itself. In comparison, a finger is enough to control the rigid wing…

What are the aerodynamic benefits of the rigid wing?

[Mario Caponnetto] Once again one of the main benefits is shape control, aiming to control lift forces and to reduce drag forces. To do so, the wing is made of a front rotating element and eight independently rotating flaps. This makes is possible to change the vertical aerodynamic load. Between every flap and the frontal element lies a slot that favors air flow between the two sides of the wing. This makes it possible to delay the stall and to dramatically increase the maximum lift. In practice, the wing is able –even with light wind- to lift the central hull of the trimaran out of the water and reduce its resistance, even though the wing lateral surface is less than half of a conventional sail. The wing horizontal sections are more aerodynamically shaped than a thin sail. A sail profile is efficient at a certain angle of attack, more or less when the flow is tangential to the frontal edge of the sail. At smaller or larger angles, a flow tends to separate from the sail, thus reducing its efficiency. The rigid wing, with its rounded front edge, is much more tolerant to variations in the angle of attack. Even at a small angle of attack, the wing will still create lift and push the boat, whereas the sail will beat like a flag and restrain the boat. This is a noticeable advantage during maneuvering, in particular when tacking, and is one of benefits that are most valued ones by our team’s sailors.

How did you develop the wing project?

[Mario Caponnetto] It was developed during a very few months, in house. The project was headed by Joseph Ozanne who linked aerodynamic, structural, electronic and shipyard engineers. The entire aerodynamic project has been based on numerical simulations without wind tunnel. CFD work has been carried on by Francis Hueber and me. In a very short time, the optimization work of the wing profile has been carried on with the STAR-CCM+ CFD code by our partner CD-adapco and exploiting a remote supercomputing cluster.

For us, it was very important that the CFD code was able to give indications on the wing behavior as far as stall is concerned. That behavior was later validated during sea trials. Furthermore, we created a database of optimal wing shape based on all the possibly encountered wind situations. The database is installed on board and allows optimizing, at any moment, wing efficiency.

What really impressed us, during the very first trials, was a better wing performance with respect to conventional sails. Therefore at the end of the testing phase at our San Diego base, it was decided to use the wing for the next America’s Cup matches. This shows the goodness of the project we carried out.

Could you please give us more details on the aerodynamics simulation aspects?

[Mario Caponnetto] STAR-CCM+ is a finite-volume approach to CFD. This is really nothing new at all, its theory can be found in textbooks. What interested us was the practical implementation.

First of all, we exploited the “client-server” architecture of the CD-adapco software. We could use a remote supercomputing cluster facility located in Italy. While sitting in our offices in Valencia or San Diego, we could check in real time the progress of the simulations running on the cluster. This happened thanks to a lightweight client -or if you like the final user- based on a Java interface, and a C++ server –or if you like the supercomputing cluster.

Second, of course, usage of the supercomputing cluster leveraged the STAR-CCM+ capability to scale well, i.e. to exploit the capability to divide the processing tasks between several processors in parallel. This was necessary since computational meshes for aerodynamics can reach several million elements.

The third success factor was process automation. STAR-CCM+ includes a CFD simulation engine (the solver) but also all the preprocessing phase (including construction of the computational mesh) and post-processing. This means we could build one complete workflow, or pipeline, and implement it over and over again during our optimization studies.

So, CFD is a tool for the happy few?

[Mario Caponnetto] Situations like America’s Cup or Formula 1 require a tremendous accuracy and detail since the engineering situation is pushed to the limit, and the optimization requirements for quantities like aerodynamic drag can be orders of magnitude more sensitive than in mass production boats or cars. I think that A.C. will continue to be one of the best benchmarks for CFD tools that can, in industrial situations, be applied in standard design offices based on small clusters or even PCs. Nowadays, all CFD processes should be automated in industrial situations, whereas A.C. pushes the application of the code to its limits in terms of physics, computational mesh or hardware resources. This creates a feedback process between the STAR-CCM+ developer, CD-adapco, and CFD teams in like America’s Cup or Formula 1, and the feedback has a positive fall on other sectors.

For instance, we evaluated several models representing turbulence, from the standard k-e to k-w SST to almost direct simulation via LES, whereas in repetitive industrial automotive or marine simulations just k-e or k-w will be adopted as daily model.

Could you disclose to the public some tips and tricks you implemented in your CFD activity?

[Mario Caponnetto] What I can disclose is that we used the STAR-CCM+ technology for automatic meshing. Both arbitrary, isotropic polyhedral and Cartesian (oriented) trimmed cells are usable. There is no absolute rule on using the former or the latter. Polyhedra may be preferable to capture vortex phenomena whereas the Cartesian grid underlying trimmed cells may be preferable when a preferred flow direction is present. In both cases, a special treatment is used for boundary layer phenomena.

Coming back to sea trials: what were the changes for your sailors?

[Mario Caponnetto] Several changes! It goes without saying that America’s Cup sailors are among the best. Especially when talking about trimmers, we talk about people who developed in a lifetime the sensitivity, based on talent and experience, on how to make sails “breath”. Then, engineers (all of them yachtsmen but amateurs) asking yachtsmen to follow our graphs and tables, so contrary to intuition… it was not easy at the beginning, but sailors, after testing out in practice our idea, became its strongest supporters. Since they were asking to designers why one wing shape was better than another one, CFD visualization capabilities were really useful to support the engineers’ explanations to sailors.
I think that in a high-tech sports activity it is important to find a common language between engineers and “pilots” and in that sense, CFD has been a very good communication tool.

What is your America’s Cup forecast?

[Mario Caponnetto] It is difficult to say. Anything could happen due to meteorological conditions; also, boats are quite different from each other. Our competitors did a good job, with the advantage of designing their boat around rules they made themselves after seeing our boat. For instance they decided an engine could replace arms’ force and allowed movable ballast. We tracked the new rules and adapted our boat accordingly. Fortunately, there is still not a lot of time to wait. The America’s Cup match will take place in Valencia on February 2010.