Are simulation tools used in sails design ? I do not talk about CAO but about means of numerical simulation of flows around sails ? If such tools are used, it seems to me that they have to take into account the dynamic behaviour of the sail, hence a very complicated and heavy code. How is it in reality ?

A problem similar to sails is parachute membrane fluid-structure interaction (FSI) -- see:

http://www.arc.umn.edu/

http://www.mems.rice.edu/TAFSM/PROJ/FSI/axi_para.html

http://www-sscom.army.mil/warrior/98/oct/airdrop.htm

"All airdrop systems (parachutes)encounter highly complex fluid structure interaction (FSI) phenomena as they deploy, inflate, reach steady state conditions, and ultimately provide a soft landing. Without an accurate representation of the inflated parachute shape, the FSI phenomena make it impossible to predict accurately the pressure distribution on a canopy surface. At the same time, the parachute's shape cannot be determined without an accurate representation of the pressure distribution (and other loadings) acting over its surface."

Good simulations of fluid-structure interactions such as sails and parachutes require high performance computing hardware (supercomputers).

I think you would have a look on aeroelasticity problems, such as studied by Mister Grisval from ONERA. You would certainly have to use 2 codes linked together. One for the deforming structure of the sail, and one for the air. Bon courage.

This is being worked on by the Dolan Sails subsidiary in Auckland, New Zealand. I know the woman who is leading the code development and she's one bright scientist and is doing a pretty fabulous job by all accounts. If you need a contact, I think I can track it down for you. The code itself is a full blown Navier-Stokes solver using overlapping meshes and membrane physics as far as I can see. If you're looking for code, you might have a hard time...

The sail and parachute problems are indeed quite complex. If you are thinking of developing a code (vs. just using one) then there are a few serious issues you need to think about before you decide what numerical methodology to implement.

I would recommend that you consider grid-free methods against traditional CFD for this highly complex problem.

The simplest grid-free method is the panel method with some model for the wake (and its interaction with the sail). The problem with this approach is that you need to know the exact location of flow separation!

A better approach is the vortex-boundary element method, which (coincidentally) our company http://www.Applied-Scientific.com is developing in conjunction with Sandia National Labs to simulate flow about parachutes! There are significant advantages with this approach:

1- The method is based on discretizing the vorticity transport eqations in the Lagrangian reference frame using vortex elements/particles. (It is similar but not exactly the same as SPH) The boundary element method is used to specify the sail geometry and to enforce flux-boundary conditions.

2- The method is completely "grid-free" within the fluid domain, and you need not waste your time thinking about gridding issues. All you do is generate vorticity on the surface (to satisfy the no-slip B.C.) and the rest is done by the code. This provides a tremendous simplification (and time reduction) of problem set up, especially since for time dependent problems such as this you'd need to mesh the domain at every timestep. Furthermore, no _ad-hoc_ "technologies" such as the moving-grid, overlapping mesh, etc. are necessary.

3- The method is dynamically adaptive and concentrates the computational elements in regions with significant vorticity. This is unlike traditional adaptive gridding techniques where within each timestep you'd need to go through the adaptation process: The Lagrangian motion of the elements ensures adaptation without any extra effort!

4- The Lagrangian computation makes the method virtually free of numerical diffusion!

5- Since the method is vorticity based, you only need to discretize the vortcity field which is concentrated in the wake and the boundary layers on the sail (you don't need to "mesh" outside these regions as you would with traditional methods). So you save significant amount of computational resources (less number of particles than you'd need finite volumes for, and there are no connectivity issues)

6- Far field boundary conditions and the continuity constraint are satisfied implicitly in the formulation,

7- The method is naturally parallelizable (with much less effort than traditional CFD), which is a must for this complex problem.

Adrin Gharakhani

The Dolan Sails http://www.sail-flow.co.nz/ researcher is Cheryl Fillekes. She is using the Overture framework http://www.llnl.gov/casc/Overture/ to grid the sails, flow region and masts.

See the Journal of wind Engineering and Industrial Aerodynamics 63 (1996)111-129. "Numerical simulation of the flow over sails en real sailing conditions" T. Carvet, F. Hauville, S. Huberson Godby