« Back

Plotting a Course to America’s Cup victory


Contributed by Jonathan Binns
BMW Oracle Racing

Refining the Grid

The graphics of refined grids show the results of refining the grid for a boundary element code. The top image shows the dynamic pressure distribution for the older grid, the bottom shows the dynamic pressure for a refined grid.

The Researchers

The design team members of BMW Oracle Racing share a common, single goal: to design the International Americas Cup Class (IACC) yacht that will win the 32nd America’s Cup in 2007. The history of yacht design involves the use of leading-edge technology in all fields; much of which has traditionally been outsourced. BMW Oracle Racing has, however, taken a very different approach by doing as much of the research as possible in-house. One area which has remained almost entirely in-house is Computational Fluid Dynamics (CFD).

With viscous, free-surface and lifting surface effects all playing critical roles in sailing yacht design, CFD has always pushed existing technologies to the limit. The use of CFD in the design of sailing yachts is complicated by the existence of a large number of requirements placed upon the design.

For example, to win the America’s Cup a yacht is required to sail efficiently at all speeds up to 18 knots and at all heel angles up to 35 degrees. The yachts also must sail into the wind, requiring lift-to-drag ratios in excess of 5.0 for the hull, yet be able to sail away from the wind with near zero lift requirements.

Yacht Racing - boat 1

Photographer: Gilles Martin-Raget

Yacht Racing - boat 2

Photographer: Gilles Martin-Raget

To win, all of these conditions must be met more efficiently than their opponents. To analyze all these conditions for all possible design variations using the most complex analysis procedures available within the time frame of an IACC design cycle is a difficult engineering feat.

Covering the design space with the greatest efficiency requires investment of design resources into different CFD methods. In order to accomplish this BMW Oracle Racing has invested heavily in two CFD methods: finite volume prediction methods and boundary element methods. Calculation speed and physical detail variations in the methods mean that comparing the results of both bring greater insight than one alone.

Taking the Right Track to Better Hull Design

Jonathan Binns, PhD, a CFD and experimental operator with the BMW Oracle design team, says that the design team is currently performing CFD analyses using three finite volume codes and six boundary element codes. Each code carries its own advantages in terms of throughput, physical simulation detail and pre- and post-processing requirements. However, ultimately all of the results must be related to each other and then to the design of the yacht.

Refined GridsThis is where Tecplot has become important to our research effort in providing a universal format with a high degree of programmability,” says Jonathan. “Tecplot has provided the visualization link required for many situations.”

The graphics of refined grids in Figures 1 to 3 show the results of refining the grid for a boundary element code. The top image shows the dynamic pressure distribution for the older grid, the bottom shows the dynamic pressure for a refined grid. “These images provide information to the CFD researchers concerning the improvements made and visual proof to the designers that the results will provide a better answer,” says Jonathan.

The resulting plot shows the predicted variation in dynamic pressure coefficient for an IACC yacht sailing to windward. From the results Jonathan and the BMW Oracle design team learned that the refined grid and the newer program were producing a much smoother, more realistic pressure distribution. “Spikes in the pressure distribution have been dramatically reduced and so the results can be used with much greater confidence for subtle design geometry changes in these regions,” says Jonathan.

The graphics of the finite volume results (right image of Figure 4) against the boundary element results (left image of Figure 4) provide much greater understanding of the effects of the differences between the two CFD methods. “Armed with this information our CFD operators can better provide our designers with the information they need in the time frame they need it.”

Finite Volume Results

The plot was created using data generated by a boundary element time domain flow solver and an unsteady finite volume solver. To create the plot, Jonathan exported the final ten steps of a converged solution from the flow solver to an ASCII data file, and then imported those steps into Tecplot software using a customized add-on he wrote to average the final ten steps. The layouts were then linked to ensure that they could be examined from an identical angle and location.

Besides pressure distributions the design team also uses Tecplot 360 to visualize shear stress distributions for CFD data and for plotting large sets of XY data from experiments. “Tecplot also gives us the ability to easily plot and print multiple data sets of 4 million data points with overlays and graphics,” adds Jonathan.

The results of these visualizations are used to explore the largest possible design space and to gain a greater insight into flow topology. “In this way,” says Jonathan, “small incremental advancements can be made to produce greater lift for lower drag whilst sailing. Experimental data is used hand-in-hand with CFD results both as validation and as design insight.”

Visualizing their data with Tecplot not only provides the design team with a way to view physical representations of their CFD analyses, but also provides valuable design information on the effects of design variations.

Jonathan believes that Tecplot’s three greatest strengths are its Macro and ADK programming; its ability to handle very large data sets; and its common data format. Without Tecplot, Jonathan says that customized comparisons would not be possible, which might leave his dream of being a part of an America’s Cup winning yacht design team high and dry.