theoretical displacement hull shape for min drag

[Leo Lazauskas]

Originally Posted by Ad Hoc

Leo

".. As far as I am concerned, there's no real need to repeat the verification for every "suitably thin hull"..."

But for verification to show the solution converges consistently, a large sample is required. Just one or two samples statistically isn't enough validation.

I no longer see that as a great issue. Michlet allows the user to easily, for example, double the number of stations and waterlines so that surface area estimates are within 1% or less. Similarly, the number of "theta" intervals in the wave resistance calculations can be increased until the desired accuracy is required. As a further check, the spectral functions for a Wigley hull, although calculated numerically, are exact because of the transforms I use. This can be checked using as few as 3 stations and 3 waterlines.

I take your point about checking convergence. As I said to Don Mc, some of the high-powered CFD codes showcased at Gothenburg 2000 couldn't estimate surface area to within 5%.

Personally, I'd love to see more (for want of a better phrase) "blind competitions" where good experimental results are available for several vessels and competitors are given only the hull dimensions and shapes. The Wake-Off of a few years ago produced some interesting results, and had some code writers to quickly re-submit predictions after the experimental results were released.

So, got some good model and full-sized data that would be up to that sort of test?

[Guest625101138]

Originally Posted by Leo Lazauskas

Yes, I finished it and had it marked just before Ernie Tuck died in March. It was a very sad and quite sudden loss for his family and the hydro community. We still had a lot of interesting and useful stuff to do. Maybe, like him, I'll find a nerdy clot to do some of the work.

Leo.

Leo
My only knowledge of Ernie Tuck is through reading various papers he has authored or coauthored. I am personally very appreciative of the result you have achieved with Ernie and making a portion of those results so readily available through Michlet.

It is sad news. It is not an uncommon experience for me to lose relations and colleages to cancer.

I think I told you before that my eldest son is doing cancer research. Essentially looking for the magic bullit. At this stage they have now verified their process for drug evaluation. The real work will require more funding to get going by the end of this year. It is a long slow process.

Does having your thesis marked mean it is now Dr Leo Lazauskas or is it going through expert review?

Rick W

[Ad Hoc]

Leo

"...So, got some good model and full-sized data that would be up to that sort of test?.."

We've got around 20~30 years worth of tank test data and full scale sea trails. Some sensitive/confidential, some not. Some with loads of data some not so. Not sure if the data we have can assist, but no harm in asking/trying....just let me know what is the type of exact data that would really suit you.

"..I no longer see that as a great issue. Michlet allows the user to easily, for example, double the number of stations and waterlines so that surface area estimates are within 1% or less. Similarly, the number of "theta" intervals in the wave resistance calculations can be increased until the desired accuracy is required..."

Hmm...calculating surface area, is just down to the accuracy of the method employed, i don't see how this makes the program any better, simply because the WSA is now more accurate. I could manually measure every square millimeter or more, does that make 'me' suddenly able to say "I've cracked it".? It still relies on form function to obtain a decent frictional resistance. Or does the program say, ahh, very accurate WSA no need for form function, for example?

Narrowing down the WSA % errors,to within minimal values, of course helps, but it doesn't mean the results are more accurate, just means the input data is more accurate. Since whether the WSA is calculated to within an accuracy of 1% or 10%, the percentage error on the other variable (form factor) in the result remains. I've read many papers, as I'm sure you have too, where the form factor is different for different models and ratios and speeds etc, which is correct?. How would this aspect be improved simply by having a more "accurate" WSA, does the program say, ooo this hull has 99% accurate WSA i'll use a different formula, via a set of 'set' parameters?

Again, not sure i follow how increasing the number of 'theta' intervals increases the accuracy. Again, just reduces the % error, which does not equate to increasing the accuracy of the residuary resistance, which is based upon other variables. Since how does the program know what is accurate/correct and what is not, other than a series of algorithms and parameters?..or is this where the "accuracy" lies, in selecting a more appropriate set of algorithms.

Good luck with your PhD.

[Leo Lazauskas]

Originally Posted by Ad Hoc

Leo

"...So, got some good model and full-sized data that would be up to that sort of test?.."

We've got around 20~30 years worth of tank test data and full scale sea trails. Some sensitive/confidential, some not. Some with loads of data some not so. Not sure if the data we have can assist, but no harm in asking/trying....just let me know what is the type of exact data that would really suit you.

I was half joking. As I said, I wish that someone would organise more blind competitions to see how well CFD and other algorithms perform.

Originally Posted by Ad Hoc

Leo
I've read many papers, as I'm sure you have too, where the form factor is different for different models and ratios and speeds etc, which is correct?. How would this aspect be improved simply by having a more "accurate" WSA, does the program say, ooo this hull has 99% accurate WSA i'll use a different formula, via a set of 'set' parameters?

Sorry, I misunderstood what you were on about before. I didn't realise you were talking about code that somehow also estimates form factors.
I concede defeat - I can't predict form drag. When I feel the need to use form factors, I either use parameter versions, or I grudgingly use Prohaska's method. I guess that the only real concession I make is to use a more physics-based skin-friction line instead of the ITTC line.

Originally Posted by Ad Hoc

Good luck with your PhD.

Thanks. I'm over it now.

[Leo Lazauskas]

Originally Posted by Rick Willoughby

Leo
Does having your thesis marked mean it is now Dr Leo Lazauskas or is it going through expert review?
Rick W

The former, although I prefer "Your Serene Highness" in written correspondence.

[DMacPherson]

Ad Hoc:

I think Leo is talking about the calculation of the dynamic (speed-dependent) WSA. As has been mentioned, wave-making is just part of the puzzle. Viscous drag needs to be properly considered. The current practice of using static WSA is reasonable, but in the context of "academic" interest perhaps leading to better solutions, using the dynamic WSA makes sense.

Is it worth the effort? Depends on your acceptable error band. Our company worked on model expansion and VPP development for one of the AC syndicates back in the early 90's. We did a sensitivity study of the difference in expanded results using dynamic and static WSA, and given the large models and well-drafted test program, we had the data to determine that there was a measurable effect - one that we could use to better achieve full-scale results. As an aside, we also found negative K-values in the form factor (i.e., FFs less than 1), which of course is physically incorrect, but a result of using the ITTC CF line which contains a bit of form correction. This speaks to Leo's interest in the Grigson line. (We ultimately used the Hughes line for the AC work for better correlation.)

So, the fact that the CFD developers chose to try and calculate dynamic WSA - and then got it wrong by 5% - is an important part of any discussion regarding validation of codes. Validation to empirical results can be tricky. You first must validate the testing itself. I've seen CR vs FN curves for different tests of a standard Wigley hull that have a pretty broad scatter. So you have to take all of this with some reasonable attention to the ultimate objective.

I care about "what is", and much less about "what should be". I'm interested in validating the underlying code only in so far as it leads to a better treatment of the whole picture. In other words, the calculation of CW is only as good as the final calculation of CT, which is the real objective. I personally believe that final validation of hull codes should be to CT and nothing else, including calculation of all of the components that make up CT - CW, CF, WSA corrections (if any), and form factor. Arbitrary selection of any of these makes any conclusion about the "validity" of a particular code somewhat deceptive.

Regards,

Don MacPherson
HydroComp

[Ad Hoc]

Leo

I must confess, i ahve no idea about what your programs are about however, this

"...I concede defeat - I can't predict form drag. When I feel the need to use form factors, I either use parameter versions, or I grudgingly use Prohaska's method. I guess that the only real concession I make is to use a more physics-based skin-friction line instead of the ITTC line..."

does this mean Michelt, is just a more accurate way of calculating WSA (as also noted by Don re: dynamic) than existing hull modellers?..hence a more "accurate" result, since everything else is the same?

Don

"..I care about "what is", and much less about "what should be"..."

Fully concur there.

[Ad Hoc]

Don

We have had some interesting debates with Tank test houses regarding the way they calculate the Ct in recent years.

Our way is slightly different, (we do our own in-house tank testing) but yields the same results. In so far as we "assume" a position for the laminar flow portion and hence the laminar WSA and apply the studs in said region, and thus also obtain the turbulent WSA. It seems some modern test houses no longer take this approach.

But as you say:
"..I personally believe that final validation of hull codes should be to CT and nothing else, including calculation of all of the components that make up CT - CW, CF, WSA corrections (if any), and form factor..."

That is indeed the goal.

[Leo Lazauskas]

Don,
Some of the codes weren't able to esimate the static WSA accurately.
Some omitted the bulb from the destroyer hull, but that's not always clear from the published results. None of the results at Goth2000 showed dynamic WSA, IIRC.

I also found negative k_f for very fine hulls when using the ITTC line which prompted me to look at Grigson's work more closely.

IMO, the ITTC line doesn't contain "a bit of form correction".
I used to believe that, but:
1. The ITTC formula is not physics-based. It was proposed by R. Newton(?) at the 1957 conference and accepted by delegates who wanted something a little steeper than Schoenherr's line at lower Reynolds numbers. Hardly a scientific approach, but it meant the delegates got home earlier than if they were deadlocked.

2. As Grigson has noted, since 1978 the ITTC line is the recommended formula to be used in estimating the form factor: therefore,
"...there can be no form factor implicit in the ITTC57 2-D friction line, that would be a contradiction in concept".

I.e. the ITTC line is accepted as a "pure" skin-friction line.

What worries me is that non-physics based methods are sometimes used to pass judgement on prediction methods when they themselves are highly questionable.

The method of estimating form factors is also often really dodgy.
For example, as much as I love the work on the NPL series by Molland et al, I find their method of raising the transom out of the water at low speed in order to get consistent results very unusual. The form factor so derived is then applied to situations where the transom is immersed, and often considerably so when the hulls are free to squat. I'd hate to have my work judged without qualification against something that hairy.

Cheers,
Leo.

[Leo Lazauskas]

Originally Posted by Ad Hoc

Leo
does this mean Michelt, is just a more accurate way of calculating WSA (as also noted by Don re: dynamic) than existing hull modellers?..hence a more "accurate" result, since everything else is the same?

No. It is primarily a code that estimates wave resistance and far-field wave patterns of thin ships.

[Leo Lazauskas]

Originally Posted by Ad Hoc

Don

We have had some interesting debates with Tank test houses regarding the way they calculate the Ct in recent years.

Our way is slightly different, (we do our own in-house tank testing) but yields the same results. In so far as we "assume" a position for the laminar flow portion and hence the laminar WSA and apply the studs in said region, and thus also obtain the turbulent WSA. It seems some modern test houses no longer take this approach.

Do you use the ITTC line? Or something else?

[Leo Lazauskas]

Originally Posted by DMacPherson

Ad Hoc:

I think Leo is talking about the calculation of the dynamic (speed-dependent) WSA. As has been mentioned, wave-making is just part of the puzzle. Viscous drag needs to be properly considered. The current practice of using static WSA is reasonable, but in the context of "academic" interest perhaps leading to better solutions, using the dynamic WSA makes sense.

Is it worth the effort? Depends on your acceptable error band. Our company worked on model expansion and VPP development for one of the AC syndicates back in the early 90's. We did a sensitivity study of the difference in expanded results using dynamic and static WSA, and given the large models and well-drafted test program, we had the data to determine that there was a measurable effect - one that we could use to better achieve full-scale results.

When you (and AdHoc if you are reading this) measure the trim and sinkage of a vessel, do you also measure how much the water level around the vessels trims and sinks?

[DMacPherson]

Leo:

Can't calculate static WSA within 5%? That's really, really simple... T'would certainly make me wonder about the rest of the calcs.

Ad Hoc: FWIW, I've had the chance to look over a number of publications that Leo has authored and co-authored. His focus is, and always has been, on developing the wave side of the problem. (Leo, correct me if I'm wrong here.) The wave-making drag calculation is just part, but the work is thorough and well-developed. The Michell integral calculation is not my particular preferred approach for calculating CW for a variety of reasons, but given that as a starting point, I have no criticisms. My only comment is that when traveling from London to Rome, it nicely gets you to Paris. Then you need to get the rest of the way.

Don

[DMacPherson]

Leo:

Ah, the old "moving dish". Do we measure it? Well, the best answer is "somewhat". We measure sinkage and trim off the heave post, of course, but in the case of the AC work, we also photographed the hull and geometrically calculated the WSA from the photographed running WL on the model. Then we did a systematic series type of statistical regression for future prediction of dynamic WSA. We never correlated the sinkage with the mean WL from the photos to get the depression, however.

Don

[Leo Lazauskas]

Originally Posted by DMacPherson

The Michell integral calculation is not my particular preferred approach for calculating CW for a variety of reasons, but given that as a starting point, I have no criticisms. My only comment is that when traveling from London to Rome, it nicely gets you to Paris. Then you need to get the rest of the way.

And what mode of transport do you use to get the rest of the way? With CFD you might get a little closer to Rome, but only by digging downwards as you go until you hit magma and are forced to give up.