Hull Asymmetry and Minimum Wave Drag

Do you have any confidence in any of the various emperical models of resistance which are used everday to make design decisions? A real question, not a rhetorical question.

 

Gosh, do you guys ever sleep?

Back to the discussion - this post but DCockey:

has triggered the following question of mine:

"But how do you change LCB position alone on a transom-stern (immersed transom) hull. Would you be able to do it without altering, for example, Cp and at the same time without altering the nature and purpose of the transom-stern hull?"​

The replies were:

NoEyeDeer:

Obviously you would have to change the lines, but you would be able to do it without changing some of the common measurements. Within limits, you could do it without changing C_p. Nature and purpose? That's getting trickier. You'd have to be more specific I think.​

DCockey:

Probably by changing the forward half of the hull. Depends on what else you're willing to change and what you want to hold constant​

The point here is: you cannot get what DCockey wants (charts or investigations of minimum resistance vs. LCB only for transom-stern vessels) because it is generally not possible to modify only this single parameter without influencing the other important form parameters, or without modifying the hull lines up to creating a bad vessel for what it was intended - high displacement or planing speeds. This consideration is valid for canoe-stern hulls too, though to a lesser extent.
Once you have introduced such modification, you will not be able to say with confidence that the obtained change in resistance is due to LCB only, and not due to the new overall hull lines. This is what Ad Hoc has tried to explain, and what philSweet has also tried to explain. The LCB alone is nothing.

On a more general level, there is one little line written by philSweet at the page 2 of this thread, which has passed apparently unobserved but which tells an important true fact:

The propellers can significantly modify the wave near-field around the hull, particularily at low speeds. It has been documented for a transom-stern hull in (for example) this research: http://handle.dtic.mil/100.2/ADA386888
So there you have another variable: the propulsion. This parameter alone can cast doubts on many conclusions coming from bare-hull model tests, since it is not easy (or should I have said "impossible"?) to reliably reproduce in scale-model test facilities, and therefore is not easy (or should I have said "impossible"?) to include in general regression-analisys curves and formulae.

I will conclude this by saying that, like it or not, some questions are doomed to remain with no answer. Those who are in search of certainities or scientific truths may consider it as a disgrace. I personally consider it a big fortune, because it gives us a hope that the child in us will always have something new to discover and to stare at with wide-open amazed eyes.

Cheers!

 

I wish you well in trying to find a computer code (thin-ship, RANS,
or other) that will help you out. I also doubt you will even be able
to get a decent set of accurate experiments to help.

NA's might like experimental results, but I have no faith in them.
(The data, not the poor, over-worked paper-pushers. *)
The attached graph shows the spread of the wave resistance
coefficients for Wigley hulls which were free to sink and trim.
Note the 33% spread at F=0.4.

I don't believe that towing tank experiments will be any better
for hulls with more complicated geometries, like transom sterns.
Several sets of results from the world-wide CFD validation program
already show a 20% spread of total resistance for a destroyer hull,
and I bet there's more variation coming in the results from the
other 30 towing tanks.

Some might believe charts and data that show trends with
LCB or Cp etc, but if the data is based on dodgy experiments, they
are fooling themselves.

Nor do I have much faith in computer models.
I have yet to see computer predictions that fall between the
two lines for the full range of Froude numbers.
Sure, you will find some boffins touting their excellent
agreement for one or two points, but none are able to get it right
for the full range (which is not all that wide, to be honest).
I can, but only by choosing the optimal value of a particular
viscosity parameter, but that's not really fair. See:
http://www.boatdesign.net/forums/de...ave-resistance-integral-thin-ships-31481.html

That, I'm afraid to say, is the state of play in my beloved field.
Others may disagree, but I still maintain that hydro is at least
50 years behind aero. In short, you just can't expect the same
level of accuracy in hydro as in aero. (And that's pretty hairy
sometimes, too!)

All the best,
Leo.
* On 2nd thoughts, neither

 

How much confidence does anyone have or not have in the various charts about changes in resistances association with changes in hull volume distribution as characterized by LCB as a function of desired Froude number?

Why?

Several of the previous posts could be interpreted as saying they may be pretty meaningless.

 

I assumed that it's understood without saying that a good design is not optimized for a single parameter but is a set of tradeoffs and compromises among many. The "best" design of a complex device (almost) always is not optimum for any single parameter or other factor. But in making those tradeoffs it is useful to know what is being traded, particularly for something as fundamental as resistance.

Which raises the question of how much of the published design guidance is meaningful?

Understood. But in terms of improving the knowledge it may be worthwhile to start without propulsion effects, and then complicate the questions.

I'm looking for the something new to discover and stare at, even if it's small and not well defined.

 

Ok, that's perfect, I agree and I'm not going to debate that.

The only thing that I (and not only I) am pointing out is that this request:
"Does anyone know of any charts which are relevant to that question (note: LCB vs. Fv) as opposed to variation of resistance with cross-sectional shape, effects of slenderness ratio on resistance at various Fv, or design lanes for Cp and displacementlength ratio."
just cannot find a rigorous and definitive answer, since the above parameters are entwined together.

And also because there's another problem when it comes to testing of ship models and parent hull forms: They are costly. So when they are done, then it's usually with more materialistic than philosophical intentions - to transfer the lessons learned to real-life ships, so the investigations are limited to hull-forms which are known to be practically usable.

If a test for high-speed hulls requires that the parent hulls become so deformed in order to simulate just that LCB movement without modifying other form parameters, then it is not very likely that this test will actually be performed, becuse it would involve non-practical hull forms.

I believe that Ad Hoc's words in the post #12, though you understandably might not have liked the context around them, are the best answer:

In the beginning I was reasoning just like you, since I come from the aerospace world. Then with time I have understood that there are realistically some very particular difficulties when it comes to both the theory and the empirical testing of ship hulls, so now I indeed tend to see all these graphs and numbers as trends, not as absolutes. It adds that healthy dose of adrenaline to the science.

Cheers!

 

Unfortunately some assumptions have been made about my reasoning, my knowledge, objectives, and so on which haven't been accurate and in some cases have been completely wrong. Perhaps these assumptions are more reflective of the background and experiences of the person making them then of me.

Perhaps it would have been more productive if the discussion had focused more on the hydrodynamics.

 

At the very best.

I'm very suspicious of a lot of those graphs when they don't show the data that they have used in their derivation.

I've seen Saunders Design Lane used in very many papers and textbooks, and not once has anyone shown the ships that were used to derive the line.

We are supposed to take it as gospel that the data was accurate, the experimental methods and measurements were valid and not biased, and that the resulting line was a good fit to the data.

If that's how NAs work then IMO it is more of a faith-based system of accounting than science.

 

It also amazes me how much confidence is put into extrapolations of regressions of test data of models with shapes that have little similarity of the vessel of interest. But the general trends may be correct.

 

Yes, they may be correct, but who knows if the original data points are never shown?

And when you do see experimental data there's usually no error bars or discussion of the methods used by the towing tanks. Sometimes the scatter is tremendous. (As someone elegantly put it: "the data often looks more like the starry firmament than the Milky Way".)

You must have come across similar situations involving the performance of airfoils at low Rn. I've seen differences of 100% or more for the CL and CD of the same airfoil. Maybe some kind of trend can be derived from that rough data, but I wouldn't bet much on it, and less on any extrapolations to other airfoils. It's not much better for models in tanks.

The use of the ITTC line to estimate skin-friction is another joke. That line, to me, represents the current state of play. It is routinely used without acknowledging that it was basically drawn by hand to satisfy delegates at a conference in 1957.

So, yes, some of us are interested in the nitty-gritty of the hydrodynamics, but it's very tough when there is so little reliable data to compare to predictions.

In any case, you have available some free primitive tools, like thin-ship theory, to investigate the theoretical behaviour of wave resistance with LCB. But all you can reasonably expect is to come up with some insights into possible trends. Just like a real NA!

Including squat is obviously much tougher because there are so few codes around. I'm getting very good results for full-size rowing shells, but they are at the limit of slenderness and the results wouldn't be useful for other vessels. (The US Navy used a similar code I wrote and they got errors of about +/- 8% for the set of vessels they used in their independent tests, but I bet I could find as many that lie outside those bounds as inside).

And, yes, Daiquiri, I do sleep! It's just that I do so about 20-30 hours after I get up.

Good night/day to you all!
Leo.

 

D-

There are some things you can learn about a cow from a pound of hamburger and there are some things you can't. If you can't learn what you want from 1 pound of hamburger, There is only a very small statistical chance that looking at a second pound of hamburger is going to make any difference. I suggest that you are at a hamburger convention asking cow questions. If you had a physics explanation for how a single case operated, would that be satisfying? Keep working on understanding a single case. Repeat the same test over and over as your data gathering system improves. Don't change anything. Once you have the physics of one case, you can predict how the other cases will behave. No one here is against your inquiry. I think everyone participating is interested in it. It just appears to me that this is one of those cow questions that is best answered by killing 1 cow and dissecting it. Perfectly reasonable to try the hamburger approach first though. There was no way for you to know it was all just hamburger until you sampled some of it.


NoEyeDeer- regarding

I think almost all tank testing is constructed to answer the opposite question- "how much does it get reduced by". With only a small sample set, you can guess at one or the other, but not both. The test design is optimized for the second question so that a paper can get published. Sobol sampling was invented to address the issue of figuring out the causal relationships in a complicated system. It requires the smallest sample set for a given accuracy if you start out without any preconceived notions about what causes what. It is also very computer friendly. But I think it would be hard to recruit talent if a test facility manager told prospective employees that they would be spending five years doing random looking runs with no particular purpose in mind other than to build up a good reference set upon which to sample.

 

Phil, I'm interested in any data or other information which already exists and might be helpful, not in trying to get more information. There might be one or two folks at a hamburger convention who might have useful information about cows, in fact might even be cow experts who are attending the hamburger convention for assorted reasons. Unfortunately I don't have a cow available for disection and the facilities to do so I wouldn't be asking questions.

I expect you're correct that most tank testing is done with specific objectives and generally focused on quantities such as resistance rather than trying to understand the why. Certainly true of the wind tunnel testing of automobiles that I'm familar with. But occasionally some tests get run with the goal of trying to understand the why.

Lots of different methods have been developed for trying to determine causal relationships. From what I've seen most have been way over-sold or at least over-bought by folks who think that knowledge of what's being studied can be replaced with a certain method for a developing test matrix and then turning the crank.

The Delft Systematic Yacht series is an example of many tests made over several decades to produce a reference set to build emperical models upon. And DTMB used to run sytematic series. The information they provide is far better than nothing, though there is always the desire for more.

 

from D-

As far as generalized formulas obtained by recursive techniques whereby a whole bunch of coefficients and power factors are given and you are asked to apply them cook book fashion- no. Not one bloody shred of confidence in any of those that I have looked at (few in number). The reason is that they all search for the one best fit rather than looking at how a sample of subsets fit. Its like the prospectus says- Past performance is no guarantee of future results. If you test a hundred different formula against a data set for fit, one of them has to be best. There might be a reason or there might not be. Sampling is the usual way to distinguish between the two. Good old fashioned comparisons to long held beliefs is another (sanity checks, behavior at limits, inflection points, all that good stuff).

Starting from particular hull in hand, it shouldn't' be to hard to construct a decent custom model which includes a large portion of axiomatic data and a small amount of tank data. The point being that the axioms you use depend on the thing in your hand. Experience is inescapably a part of this. I happen to believe that this is actually a good thing, but I can't figure out how to explain that other than it is the normal progression of science to engineering.
Our obsession with parametric studies isn't just because they are more manageable, we actually like them. The trend is towards more axioms and less required data, or at least a more consistent set of axioms. So if I seem to have been at lock-horns with you here and elsewhere maybe that explains a bit of it. This has been great. The baiting has been polite and purposeful. And I've learned a heap of stuff. Thankyou.

Science is a cow.
hamburger is engineering.


That thanks goes out to everyone.

 

I'm not so sure about that, at least not some of it. But that's the way forums go. And the reason some forums acquire a less than positive reputation.

The highly amusing part to me of much of the "discussion" in this thread is that I'm in "violent agreement" with many of the sentiments expressed. I consider any theoretical model to be wrong at some level, with the important questions being how wrong and what is it good for (if anything). And I'm always skeptical of curves based on experiemental data. But that doesn't mean I don't look at them.

 

Yes I understand that. The thing is that this thread was going around in circles, as in:

"Hey why is this hull better with the LCB further aft?"

"You can't move just the LCB by itself."

"Yeah but these guys said the hulls they tested with the LCB further aft were better."

"You can't move just the LCB by itself."

"Yeah ok but these guys said....."

"You can't move just the LCB by itself."

"Arrggh."



So, to cut through all that and attempt to get a bit more clarity, I asked about a specific type of LCB shift that was produced by a specific change in a base hull, since this was the type of change that often seemed to be done.

I'm primarily interested in practical results myself, but was wondering if there had been any insights into what was causing the reduced resistance. If it's something nobody has bothered to research then that's just the way it is. Hey ho.