america's cup yacht

[redcoopers]

Yes, you are exactly right.

Slender ship is used by Michell's eqn, but Michell's eqn is not the totality of slender ship. However, to make a velocity prediction on a sailboat, I would do the following steps:

1) Slender ship, Michell's eqn, or Delft parametric for initial design (I prefer Delft series because it may help judge what sort of lift we need. The downside of any code which uses slender ship is that lift cannot be predicted... if I'm wrong, please tell me how I can incorporate lift with slender ship theory, I'd like to know!)
2) SPLASH or another potential flow free surface code (boundary integral: for an example see Faltinsen) for wave-making resistance with the incorporation of the Kutta condition.
3) RANS (or another viscous code) for appendages or blunt body separation affects.

I tend to put all sorts of slender ship theories into the first step for yachts. Although many improvements can be made, I still think that these are only the first methods for sailboat design because of the absence of lift prediction on the keel and rudder. Michell's eqn can predict trim and heave moments and forces (see Yeung 197?). Moreover, I think that for most modern yachts, an improvement could be made to switch from slender ship to shallow ship theory. This would need a redesign of the calculations (anyone remember Green's functions... PDE's ?), and may be more applicable to yachts. However, I think that the basic necessity of yachts needing the Kutta condition prevents this technique from becoming predominate.

In the long run, what I think yacht design needs is a potential flow code which quickly calculates the wave-making resistance which incorporates the kutta condition. Boundary integral theories are quick, but an integral equation solver can be more accurate. Finally, having to make a wake-cut to satisfy Kelvin's theorem means that the solution is only as accurate as the cut which is made. I truly belive that this is the big problem to yacht hydrodynamics. Does anyone have suggestions?

[redcoopers]

Oh Leo, by the way,

Your SWPE5.0 code is very impressive. The wave prediction profiles on the Wigley hull are awesome (I would presonally almost kill for those results with my codes ). I'm just wondering, however, if you know what happens when the L/B ratio goes to ~ 4 or 3?

-Jon

[Leo Lazauskas]

I'm not sure what you mean by "lift" in some of your statements. Do you mean the lift due to leeway, or the dynamic lift (and trim), i.e. squat? Or both?

I'm also not quite sure what you mean by the Kutta condition. Do you mean it in the sense of a smooth detachment condition for a yawed hull? Or are you suggesting the Kutta condition is necessary for vessels with transom sterns?

Good luck with the RANS solvers. I hope you have a lot of available computer muscle and time!

You mention that lift on the keels and rufdders is an issue for you. I don't have a hydro code that includes lift and induced drag in ship calculations, but I can include them via other codes I have written. Unfortunately, the Vortex Lattice Methods (VLM) used by many workers are not accurate enough for my purposes. They work very well for lifting surfaces with planforms comprised of straight-line segments (e.g. rectangles, trapezoids, delta wings etc) but they are pretty poor and inconsistent for small aspect ratio curved keels, fins, and rudders.

I have put up some benchmark solutions for wings with circular planforms (for which there is an analytical solution) on my WWW pages. Lifting line theory is hopeless for this case, as you would expect. I'd be very interested to see how commercial codes perform on elliptical planforms (either flat or cambered), but I suspect that many commercial solvers can't handle infinitely thin surfaces.

I think that some aspects of "shallow ship theory" are relatively easy, but other are very tough. See, for example,
"Wave patterns and minimum wave resistance for high-speed vessels"
www.cyberiad.net/library/tsl02b.pdf

My first suggestion for improving resistance estimates has nothing to do with potential flow, but rather with the estimation of viscous drag. I think that one of the best improvements you can make is to dump the ITTC 1957 line. It is not physics-based and there are far better alternatives IMO, e.g. Grigson's algorithm. See:
www.cyberiad.net/bl/grigson/grigson1.htm

But a caveat - you still have to estimate the form drag which is not easy! There are many who still claim that the ITTC line is not a skin friction formula, but rather a correlation line and thus includes some allowance for 3D effects. I am guilty of thoughtlessly chanting that same mantra, but it is just not true. Since the 1978(?) ITTC, the formula is used as nothing more that a simple skin-friction estimator.

Regards,
Leo.

Originally Posted by redcoopers

Yes, you are exactly right.

Slender ship is used by Michell's eqn, but Michell's eqn is not the totality of slender ship. However, to make a velocity prediction on a sailboat, I would do the following steps:

1) Slender ship, Michell's eqn, or Delft parametric for initial design (I prefer Delft series because it may help judge what sort of lift we need. The downside of any code which uses slender ship is that lift cannot be predicted... if I'm wrong, please tell me how I can incorporate lift with slender ship theory, I'd like to know!)
2) SPLASH or another potential flow free surface code (boundary integral: for an example see Faltinsen) for wave-making resistance with the incorporation of the Kutta condition.
3) RANS (or another viscous code) for appendages or blunt body separation affects.

I tend to put all sorts of slender ship theories into the first step for yachts. Although many improvements can be made, I still think that these are only the first methods for sailboat design because of the absence of lift prediction on the keel and rudder. Michell's eqn can predict trim and heave moments and forces (see Yeung 197?). Moreover, I think that for most modern yachts, an improvement could be made to switch from slender ship to shallow ship theory. This would need a redesign of the calculations (anyone remember Green's functions... PDE's ?), and may be more applicable to yachts. However, I think that the basic necessity of yachts needing the Kutta condition prevents this technique from becoming predominate.

In the long run, what I think yacht design needs is a potential flow code which quickly calculates the wave-making resistance which incorporates the kutta condition. Boundary integral theories are quick, but an integral equation solver can be more accurate. Finally, having to make a wake-cut to satisfy Kelvin's theorem means that the solution is only as accurate as the cut which is made. I truly belive that this is the big problem to yacht hydrodynamics. Does anyone have suggestions?

[Leo Lazauskas]

Thanks for the kind comments, Jon, but I don't agree that the Wigley wave profiles in SWPE 5.0 are all that good. I have rewritten huge slabs of old code since that release and I can do much better now. I have also nearly finished a program called "Flotsm" that combines the methods of the U.S. Navy's standard ship motion program (SMP) with my total resistance and squat calculations. I'm trying to keep the input and output identical to SMP I/O so that the learning curve for users is
not too steep. I hope to release a demo version in the next few weeks, but chances are that it won't happen until mid-Olympics, when I can get some quiet productive hours in the middle of the night :-)

As regards accuracy for L/B ~ 3 or 4, all I can say is that it depends on the type of hull. Prof. L.J. Doctors uses Michell's methods for much of his work and he noticed that for some hulls with large transoms, predictions are sometimes better for L/B ~ 4 than for L/B > 10, which seems counter-intuitive.

Incidentally - while we are in mutual admiration mode - I thought your report (i.e. using SPLASH) was excellent. It certainly made me chuckle to see that SPLASH took several weeks to get up and running properly, even with assistance from the author, but that it was out-performed in some aspects by Noblessse's much simpler and faster slender body code. Your experiences seem to coincide with other criticisms I have heard of SPLASH (at least of
the pre-2000 versions) in that it sometimes needs to be tweaked differently for different hulls and for different sizes of the (wave-field) domain. On the other hand, the graphics are excellent despite its somewhat uncertain performance when used blindly.

On a final note, you should be a little wary of using the Wigley hull as an example for the performance of some codes. Despite its simplicity and smoothness, some derivatives are not smooth and this can cause numerical problems (and possible difficulties in producing acceptable experimental results). For example, SPLASH predictions (on the SPLASH WWW page) for the Wigley hull are no better or worse than simpler codes, as someone else on this forum pointed out a few months ago.

All the best,
Leo.

Originally Posted by redcoopers

Oh Leo, by the way,

Your SWPE5.0 code is very impressive. The wave prediction profiles on the Wigley hull are awesome (I would presonally almost kill for those results with my codes ). I'm just wondering, however, if you know what happens when the L/B ratio goes to ~ 4 or 3?

-Jon

[redcoopers]

Hi Leo. It's nice to find someone online who also speaks hydrodynamics! First, maybe, we should start a new thread? But on to the content:

I'm not sure what you mean by "lift" in some of your statements. Do you mean the lift due to leeway, or the dynamic lift (and trim), i.e. squat? Or both?

If I say lift, I mean hydrodynamic sideforce. Anything vertical, I refer to as sinkage or heave. This leaves trim as coinciding with a pitching moment.

I'm also not quite sure what you mean by the Kutta condition. Do you mean it in the sense of a smooth detachment condition for a yawed hull? Or are you suggesting the Kutta condition is necessary for vessels with transom sterns?

My main interest is in sailboats. To begin to predict lift, I need smooth detachment off the trailing edge of the foils (Kutta). But with transom sterns:
1) I don't care too much about boats with these sterns (my hobby is only sailboats... I don't want to do "ships" as a past-time)
2) there is so much turbulence / separation / eddymaking that it doesn't make any sense to enforce a potential flow condition here (but I know that there are a variety of corrections which can be made to "force" a better wavemaking solution).
3) I use the Kutta condition only to help predict lift. If we are not interested in net sideforce for a ship, then let's not complicate things further!

A smooth sailboat (let's say an IMS sled-type) does not have a transom stern, but it does have very smooth and wide aft sections. Moreover, the canoe body creates sideforce itself. While there is still a tremendous amount of turbulence and separation back here on these boats, I think potential flow could be at least more applicable. In theory, a wake cut could be made to satisfy Kelvin's theorem around the canoe body. The big question is: does anyone know how this cut should be made, where it should be placed, or how it interacts with the free surface?

One thing I've been toying around with is to use a simple boundary layer calculation to find out where flow separates. At this point, the outer separation layer is used as the "hull" for potential flow, while a simple friction and momentum calculation is used inside the layer. For slender ship theory, this method can work, since at some point, the longitudinal derivative of the separated surface will go to zero (hence resistance -> 0). However, if I use a boundary element code, there is not a real good way of doing this unless I take my separated surface and just carry it to the edge of my domain (but again, how do I treat the free surface?). Finally, the inner calculation with separated/turbulent flow remains very important, and a good method is needed to handle this.

But this brings me to boundary layers:

But a caveat - you still have to estimate the form drag which is not easy! There are many who still claim that the ITTC line is not a skin friction formula, but rather a correlation line and thus includes some allowance for 3D effects. I am guilty of thoughtlessly chanting that same mantra, but it is just not true. Since the 1978(?) ITTC, the formula is used as nothing more that a simple skin-friction estimator.

I remember that in my undergraduate resistance class (Prof Bhattacharya) the ITTC line is only one part of the model-ship correlation. What the 1978 ITTC conference said was that a method (such as Prohaska's) is necessary for calculating 'k'. Without including this form factor (hopefully from experimental data), the ITTC line is meaningless. Altogether, the semi-empirical method has had good results.

But I am very interested in Grigson's algorithm. I've been browsing your web-pages, and I was just wondering: what is your "present method" of your extension to Grigson's algorithm? The 2002-ITTC conference gave a "less-than favorable" view of Grigson's algorithm because of scaling effects. Myself, I have not worked with Grigson's method at all and do not know how it works. Because I am far too lazy to go browsing through RINA, do you think you could give a synopsis of the calculations?

Good luck with the RANS solvers. I hope you have a lot of available computer muscle and time!

Thanks! I'm working on a model right now which is viscous, but I haven't included turbulence yet. The best I can describe it right now is a particle-based lagrangian method. It handles vorticity inherently, and furthermore, because it is lagrangian, the convective acceleration terms vanish. But, because it is gridless and works based on time-stepping, it is very slow right now. My computer's been doing overtime...

-Jon

[Leo Lazauskas]

Jon:

Originally Posted by redcoopers

A smooth sailboat (let's say an IMS sled-type) does not have a transom stern, but it does have very smooth and wide aft sections. Moreover, the canoe body creates sideforce itself. While there is still a tremendous amount of turbulence and separation back here on these boats, I think potential flow could be at least more applicable. In theory, a wake cut could be made to satisfy Kelvin's theorem around the canoe body. The big question is: does anyone know how this cut should be made, where it should be placed, or how it interacts with the free surface?

This sounds a bit like the infamous "line integral". Is that what you are alluding to?

Originally Posted by redcoopers

...One thing I've been toying around with is to use a simple boundary layer calculation to find out where flow separates.

Are you using something akin to Stratford's separation criterion?

Originally Posted by redcoopers

But this brings me to boundary layers:
I remember that in my undergraduate resistance class (Prof Bhattacharya) the ITTC line is only one part of the model-ship correlation. What the 1978 ITTC conference said was that a method (such as Prohaska's) is necessary for calculating 'k'. Without including this form factor (hopefully from experimental data), the ITTC line is meaningless. Altogether, the semi-empirical method has had good results.

I agree that it gives reasonable results, but it could be better. I'm just uncomfortable using a formula that was originally based on an incorrect assumption and poor correlations, tweaked in order to keep a group of ship-builders happy in 1957, and then retained ever since because it is convenient for naval architects. I don't know what that is, but it isn't Science!

The ITTC criticised Grigson's 1993 recalculation of the Lucy Ashton data, but I don't think it had anything else to say about his algorithm. I do think that Grigson's algorithm needs further validation at very high Rn (> 10^9), though.

Originally Posted by redcoopers

But I am very interested in Grigson's algorithm. I've been browsing your web-pages, and I was just wondering: what is your "present method" of your extension to Grigson's algorithm? The 2002-ITTC conference gave a "less-than favorable" view of Grigson's algorithm because of scaling effects.

My "extension" is really only a hack to make the algorithm work for smaller Reynolds numbers than Grigson intended. The "Wake Parameter" graph shows the assumption I made to get the results.

Originally Posted by redcoopers

Myself, I have not worked with Grigson's method at all and do not know how it works. Because I am far too lazy to go browsing through RINA, do you think you could give a synopsis of the calculations?

Unfortunately it is too complicated to summarise here, and I'm going interstate for a few days. I'll expand the note on my WWW pages when I get back and find some spare time.

Cheers,
Leo.

[SailDesign]

Originally Posted by leo

I agree that it gives reasonable results, but it could be better. <snippage> I don't know what that is, but it isn't Science!

Speaking as a naval architect, Yes it is convenient, so "pppppppffffffffffffffffffttttttttttt"



Yeah, it ain't deadnuts accurate, but what is when you start adding waves and things? It is, as they say, good enough for ship work.

Steve

[Leo Lazauskas]

Originally Posted by SailDesign

Speaking as a naval architect, Yes it is convenient, so "pppppppffffffffffffffffffttttttttttt"



Yeah, it ain't deadnuts accurate, but what is when you start adding waves and things? It is, as they say, good enough for ship work.

Steve

No, it is good enough for *boat* work. For the Reynolds numbers of most sailboats the ITTC line is almost identical to more sophisticated methods.

BTW, it sounds like your inflatable friend has sprung a leak :-)

[Leo Lazauskas]

Originally Posted by redcoopers

I remember that in my undergraduate resistance class (Prof Bhattacharya) the ITTC line is only one part of the model-ship correlation. What the 1978 ITTC conference said was that a method (such as Prohaska's) is necessary for calculating 'k'. Without including this form factor (hopefully from experimental data), the ITTC line is meaningless. Altogether, the semi-empirical method has had good results.

But what this is really saying is that the ITTC line is being used as a skin-friction estimator.

See you in a few days,
Leo.

[SailDesign]

Originally Posted by leo

No, it is good enough for *boat* work.

You shippies.....

Originally Posted by leo

BTW, it sounds like your inflatable friend has sprung a leak :-)

Dang! Now, where did I put that patch kit?

Steve

[jonny_IRL]

sorry to butt in, but it looks like some of you might know some good sites on yacht design icould look at for an essay on resistance in racing yachts. i'm in the very early stages of marine engineering so something not too advanced would be good.
thanks