Optimisation for VPP's

[DSmith]

I was wondering what sort of optimisation algorithms are being used in modern velocity prediction programs(VPP's) and what their pros and cons are.

We are about to start coding the optimisation algorithm for a foiler velocity prediction program(VPP). With the large number of variables and non-linear nature of the VPP, gradient based algorithms are likely to be cumbersum and not necessarily arrive at the global optimum.

We are looking at using Simulated Annealing which was invented 20 years ago. The algorithm searches the design space in a semi-random manner for the local maxima. A more detailed description can be seen at... http://en.wikipedia.org/wiki/Simulated_annealing

Any advice/thoughts/opinions/guesses would be appreciate.

Dave

Dave, do you know Alan Smith? He's done a lot of foiler research and may be able to help you-he's down there too. E-mail me at: lorsail@webtv.net and I'll give you his e-mail. I've written him about your post and he may respond here. Looking forward to hearing the results from your project-good luck!

[alans]

David

Doug Lord is partially correct re some work I have been doing, mainly concerned with dynamics of foilers. Performance estimates are of course a by product of these simulation models. I do not normally use any optimization algorithms but simply "sail" the boat on fixed input parameters.

I have used one method for "quick look" performance results and that is to use an equation to set the sail to optimum thrust as a function of the apparent wind angle and then starting at a true wind angle of say 135 degrees allow the simulation to bring the boat to maximum speed. You can now increment the true wind angle by very small increments and achieve performance numbers for the full range af true wind angles.

Some of the maths on my web site may help

alans
http://www.highspeedsailing.com

[DSmith]

Originally Posted by alans

David

Doug Lord is partially correct re some work I have been doing, mainly concerned with dynamics of foilers. Performance estimates are of course a by product of these simulation models. I do not normally use any optimization algorithms but simply "sail" the boat on fixed input parameters.

I have used one method for "quick look" performance results and that is to use an equation to set the sail to optimum thrust as a function of the apparent wind angle and then starting at a true wind angle of say 135 degrees allow the simulation to bring the boat to maximum speed. You can now increment the true wind angle by very small increments and achieve performance numbers for the full range af true wind angles.

Some of the maths on my web site may help

alans
http://www.highspeedsailing.com

What I've written so far already plots the polar diagrams for a foiler with fixed parameters. I now want to optimise all of the parameters (CL of sail, foil sizes, foiling height, crew weight/position....) for particular racing courses and wind speed probability distributions.

Your work on the C Class foiler is very interesting and I will read through it in greater detail when I get a chance.

Cheers

Dave