Michlet

What a beautiful paper Patrick. Seeing the cite for the Tuck et al paper was a bonus!

Are the offsets of the ship available? Or are they still considered top-secret by the Egyptian Navy? I'd love to try to predict the squat of the ship and it's drag in deep and shallow water.

Leo.

 

hi Leo -- Im pleased you liked the paper

can you read an IGEs NURBS file or do you prefer sections (if so how many is ideal)

 

Thanks, Rick.

I'm pretty sure it's the accidental inclusion of hydrostatic drag. No matter.

The good agreement you show in your Michlet run could be a fluke too.

I'm not sure which file you used as input to Michlet.
Were experimental values of squat specified in the input file?

Look at the size of the hydrostatic resistance.
It's comparable in magnitude to the wave resistance (Rwtrans + Rwdiv)
for nearly all speeds. You have to believe strongly in the hydrostatic drag and the way it is estimated to accept the good agreement!

And if we believe that we should account for the loss of pressure on the transom and include it in the resistance, shouldn't we also include the loss of pressure along the static waterline of hull to be consistent? It's a can of worms!

Inclusion or exclusion of squat will significantly change the value of the hydrostatic resistance so we also have to be careful not to cherry-pick. If we accept the good agreement here, we should probably use it the same way everywhere, i.e. with or without squat consistently.

Whether squat should be included, and how, is still not clear to me, or to some hydro gurus. There is a very interesting discussion of a paper by
Prof. L.J. Doctors and Dr. A.H. Day at:
http://www.iwwwfb.org/Abstracts/iwwwfb15/iwwwfb15_discussions.pdf
See page 18.

So is it better to use experimental or estimated squat in predictions of resistance? Confused? So am I.

For multihulls there is also the issue of the viscous interference mentioned by Patrick. For those interested in this effect I highly recommend searching for the work by Dr. N. A. Armstrong who, last I heard, was at Austal Ships.
That should confuse the issue regarding the use of form factors even more.

Leo.

 

Sections please.
Is it possible to get at least 41 stations and 41 waterlines? I need offsets above the static waterline as input when the hull squats.

Leo.

 

Just one more comment on that good agreement you showed...

The form factor for the 4a monohull is k=1.3, and for the cat with w/L-0.2 it's k=1.43.
If you add that to the predicted total resistance the agreement looks much worse.

Leo.

America's best-known and best-loved poet could have been referring to form factors when he wrote:

There are known knowns.
These are things we know that we know.
There are known unknowns.
That is to say,
there are things that we now know we don’t know.
But there are also unknown unknowns.
These are things we do not know we don’t know.

 

Michlet/GODZILLA gets me 95% there. I keep my world as simple as possible. I avoid transoms. My hulls are not prone to huge changes in trim at my speed of interest. This would be the same for anyone seeking an efficient hull. GODZILLA has taken me down a path where wave drag is very low. (As noted in another thread I get abused for lack of wake for drafting) It means I am not all that worried about the absolute value of wave drag. I am interested in any method of reducing viscous resistance that is non-polluting.

GODZILLA has guided me toward an easy-to-build hull using a hard chine that gives nothing away to a rounded chine hull. Flotilla shows it is a superior hull above my design speed. (I compared the results from V14 and V11 produced by Flotilla and there is more bow up with V14). Imagine what this could mean for the future of rowing sculls.

So it might be nice to keep chasing down all the little variations with the hull resistance but then there are other factors that become more significant.

Windage is currently a big ticket item for me - there are times when I am putting more power into the air than the water. Added resistance due to wind waves can be significant. Would certainly be nice to see how a hull might morph under GODZILLA if ambient waves were included in the resistance calculation.

Then there is the other side of the equation with the propulsion losses. Some 15% of the available power is lost here so a worthy contender for ongoing improvement.

Rick W

 

One of the winners of a Solar Boat race held in Canberra took a dinghy, changed the transom stern to a pointed one, and filled the thing with the allowable weight of batteries. A simple and effective strategy by an old naval commander!

I include air resistance in most of my own work, but the difficulty is in finding appropriate drag coeffcients.

For rowing I use a coefficient from wind-tunnel tests with standing people, but I'd like to get something more specific, eg. rowing shell + riggers + humans + hairdo. Hoerner's book also has some estimates for people in a variety of positions which should be updated one day.

L.J Doctors and A.H. Day tried this sort of optimisation many years ago.

Day, Alexander H. and Doctors, Lawrence J., Design of fast ships for minimal resistance and motions, Sixth Int. Marine Design Conf. IMDC-97, June 23-25, 1997.

They used Michell's theory and strip theory for the ship motions, but I can't remember if added resistance was considered.

I have code based on work by N. Salvesen and E.O. Tuck (yes, him again!) to estimate motions and added resistance, but the problem is getting accurate wave spectra that are applicable to very small boats. It's fine for big ships and yachts. A few years ago I cobbled together a hybrid SMP/Flotilla code I named Flotsm but I never got around to writing manuals for the beast. Maybe I'll finish it off in my dotage.

Leo.

 

just a bit of a note on the experiments we did almost 20 years ago!

Given the underlying assumptions (linear, potential flow, thin ship, resistance calculated from far field) you are never going to get good agreement in the hump region where there is local wave breaking whose energy is dissipated before it reaches the far field. Also for a transom stern running wet, (low Fns) there is a lot of separation and this is not picked up well. SO really I think that for vessels with a transom, you can expect to have reasonable agreement at higher Fns -- maybe 0.55 - 1.0 or there abouts, and for non-transom stern hulls you can also expect reasonably good agreement at the lower speeds -- but the wave resistance measurements are quite fraught with difficulty -- we used resistance wave probes as that time -- using 4 to measure the far field wave pattern. The analysis of the traces took into account the towing tank boundaries, but assumed perfect reflection, also at the higher speeds the wave trace was quite short. The models were quite small so their resistance was quite low -- hence a larger than desired signal to noise-ratio -- but then we had to make do with facilities available (20 years ago -- you forget how far computational power has come since those days; today you could probably get _much_ more accurate wave pattern measurement using photogrammetry) So I would probably estimate at least 10% measurement+analysis uncertainty in the measured wave pattern resistance.

cheers,
pat

 

It was a heroic effort, Pat, and nobody has, to my knowledge, produced such a comprehensive set of experiments.

Another phenomenon muddying the experimental waters are the boundary layers that build up on the sides of narrower-than-ideal tanks. This has not been investigated much.

I sometimes wonder whether that could cause some of the so-called viscous interference found in widely spaced catamarans. i.e. the interference is not between the hulls, as it is with closely-spaced demihulls, but rather between the outside of each demihull and the closest tank wall.

Leo.

 

possible, but the tank was 3.6m wide and at the wides spacing the catamaran hulls were CL / CL 0.8m
So dist from CL one demihull to tankside would be 3.6/2 - 0.8/2 = 1.4m
compared with the dist of cat CL to demihull CL was 0.4 so almost 4 times the distance.
Also with just a single demihull tested in isolation this "viscous" part was significantly less than for any of the cat configs and it was only 0.4m further from the tank wall ie 1.8m rather than 1.4m so I dont think it likely to be tank wall interference, but an interesting thought.
The thing that I always found interesting was that there appeared to be a big step up in "viscous" resistance going from mono to cat, (like maybe 10-15% if I recall correctly) but then the variation for different cat configs was negligible, even going from spacing of S/L=0.2 to 0.5 -- And it was consistent with all the hullforms tested (about 13) and different experimenters (Mustafa, myself, Dominic and Adrian)

Yeah Tony A. will give you some good discussion on form factor with data from 100m+ vessels -- but what effects are the waterjets having, and yes he is still at Austal Ships. We did a paper at the Sydney FAST conference a while back

BTW for the very original thin ship calcs I did for my thesis, I did modify the geometry for sinkage and trim that we measured at each speed and this did make the results fit the experiments much better.

Rick -- even very small trim will have a big effect on a hull that long. I found that the trim really switched from zero to about 1-2 deg (I think -- check the links to the reports I sent) at the hump speed (around Fn =0.5) I cannot remember if the effect was so pronounced for the non-transom stern Wigley hulls -- would have to try to dig out Mustafa's thesis from my library
cheers,
pat

 

Thanks for the clarification.

Just a comment on Table 2 of the Min paper. I think there are missing decimal points in the Displacement and Ballast. 29974 tonnes is probably a little high for a 14m long hull.

Leo.

 

oops -- yeah that would be kg

 

It seems I am being “bullied” into making a comment, despite my extreme reservations to post anymore, I have yielded, maybe owing to too much wine!

Firstly addressing the “thin” and “slender” issues.

Slender means a body/shape that has small transverse sections compared to its length, ie like a pole.
Thin, means that one transverse dimension is small with respect to its length.

The two are different as noted by Leo.

The ventilated transom is indeed an interesting one. There are 2 ways to approach this to ascertain what is going on to ‘describe it’, viz:
1, theory
2, model testing.

So, what is the theory based upon?. As already noted by Leo, has it limitations, therefore what can one do when analysing beyond the limitations? As for other theories, I have yet to see a theory that can mathematically model a hull surface to the point of being a true “friction line”. One hulls painting pattern and drying pattern is different from another, how is that modelled and taken into account?...or the scratches made by the fabricator when grinding out markings left by the tabs holding the hull plating together, or over zealous welders putting in too much heat creating a “staved horse” effect on the hull etc etc, what is the equation to describe all these variables? Is the density and temperature of the water established by theory?? Theory can only account for so much.

Well, that is where physical observations come into play, ie model/tank testing. Clearly when a model/ship is running at high Fn’s, if, like the series 64 in the papers attached kindly by Pat, has a ventilated transom. So, how does one produce a theory to account for this observation and produce consistent results?...since one cannot ignore quantitative real world events.

The notion of an extension to the stern as an appendage may appear, without logical deduction, but one cannot ignore the real world event. In the absence of being able to derive a mathematical theory that accounts for it (clearly not been done yet by anyone anywhere in the world), one must account for what is observed, despite flying in the face of contention.

Molland et al, recognised the fact that there is “some other mechanism” at play...what, as yet unknown. But, rather than just ignoring the results, owing to a lack of suitable theory to explain it, Molland et al, establishes a form factor. A fudge factor if you like. A fudge factor does not sit well with theorist nor hydrodynamists/mathematics because it is plucked from thin air, ergo has no logic. And one cannot argue against this logic. However, the observations are clear and consistently repeatable, ergo something is going on which has yet to be established.

Molland’s form factor is not dismissive of theory, nor confirmation of empirical measurements as the only way, it is just noting something odd is occurring that can yet be explained. What is required, is rather than a debate about whether one should use a form factor or that forms factors do not explain every ship shape and every Fn (which they can’t hence different form factors exists), one should be focusing upon breaking down the form factor that is used to correlate results; to find out what it is, does it really exist if not why not, if so, what is it. Focus on finding out what is the mechanism that is causing this phenomenon?

To suggest that test tank data is flawed, means that one is never trusting of observations and hence how can one establish a theory if one does not trust anything that one observes? It is then a theory based upon imagination, not observation. Since a “theory” by definition is a system of ideas that explains “something” that is measured or observed. In this case the “something” is the ‘form factor’, or rather, the consistent inconsistencies.

So, there shall always be systematic errors in any experimentation. The goal is to minimise them as much as possible. However, despite this, one needs to look at the implications of tanks and where errors may creep in. The sources are 1) systematic and 2) procedural.

1, Systematic errors, say reading the temperature of the water, has a certain range of accuracy, based upon the instrumentation and the observer taking the said reading. Reading the difference between say 1.0 degree and one hundredth of a degree is just a matter of what device is being used. It can affect the result. However, what are the implications on the final results of a difference of say 1degree. (Anyone who cannot establish an accuracy of less than one degree is not scientific in their approach). The difference in final results is relatively minor. Even a large variation in measuring temperature has minor effect on the overall result. Ergo systematic errors, if the experiment is done with due diligence with quantitative measuring devices, can be minimised to the point of being either negligible, or a known ‘constant’.

2, Procedural. This has a much greater degree of influence on the results. For example, where does one tow the model, how high above or below the VCG and LCG, what about at the bow, or stern, fixed or free to trim. The location of the studs to promote turbulent flow, what size tank, wide, shallow, with or without wave dampening etc. Is the model a fair representation of the ship, has it been scaled correctly, has it been set to the correct displacement, LCG/VCG etc etc?

Procedural errors account for a much higher percentage errors than systematic. Reading many papers as I also have, and peer reviewed some, this accounts of the main body of errors. Systematic is by comparison, small beer.

Many years ago we had a hull tank tested. The expected speed was 32knots, but 37knots was achieved on sea trials. Why so different? The exact same model was then tested in 2 other tanks, with differing results from the original. The procedure for tank testing highlighted errors in the methods used. There was nothing wrong with the measurements, just the procedure and hence understanding all the mechanisms involved when tank testing. Everything is a learning process and driven by observation. This addresses your concerns Leo of the distance from the demihull to the tank wall. I've not seen many papers addressing this aspect.

So, this brings us back to the “form factor”. Whether anyone likes it or not, there is an inconsistency in resistance measurements. But under certain conditions a constant can be added to account for this….yes. The constant may not be consistent over a wide range of Fn's nor hull forms, but can account for the uncertainty.

Many methods have been proposed, and some dismissed. But the reality is inconsistencies exist and as yet no theory can totally account for it under all conditions. The inconsistencies, once any systematic errors have been reduced or eliminated does not eliminate the inconsistency, ergo it exists.

There is an excellent paper by R.S. Guilloton, publish April 1960 RINA Trans. “The waves generated by moving bodies”. He uses Michell and Havelock’s work to establish “potential elementary wedges”. Using graphical means only, anyone can produce the results, given patience and time. Way ahead of its time.
The issue of transoms is taken differently in this paper too, in so much that it is assumed the hull lines, in plan view, eventually re-unite. This is then taken as the length of the vessel, but, with a section cut off.
The wave breaking issue, despite being ‘graphically produced’ is also addressed as also confirmed by Pat.

However one addresses the problem of ventilated transoms, it is clearly not a straight forward matter and what ever terminology is used and whether one should or should not account for it.

The series 64 hull used in the later 2 papers posted by Pat, is actually our hull. (or I should say the company we used to work for – we donated it during an EU lead project). We have also used 2 different scale models for this series 64 hull with the same results. For us, suggesting that if the procedure is quantitative and the measurements taken by calibrated devices, scaling small or large has minimal, if any, affect, other than systemic, owing to the scale and hence ability to measure something much smaller.

I would suggest it is worth focusing on why there are inconsistencies in so much as what is the mechanism driving this, rather than assuming errors in the measurements/readings are the problem. Systemic errors et al, can only account for so much; that just addresses the - which experiments do I believe - which everyone should ask when reviewing data anyway, no matter the source. This is just good engineering practice!

And yes we too have found going from a mono to a cat affects the results. (That is mono x2 is different to a cat of same demihulls). The same order of magnitude as you encountered too.

Therfore, do I believe that there are inconsistencies that are yet unexplained, if so, what is the mechanism? Avoiding a transom for example, because it cannot be modelled effectively nor quantitatively doesn’t push the boundaries of our knowledge or understanding. To take a page out of Prof Dawkins’ book, we need to fill in the “gaps”, to satisfy our curiosity rather than ignore it.

Sorry for the ramble….just a few scattered thoughts that come to mind before bed.

That’s me done now.

 

the models I tested in 91-96 were NPL-round bilge. I think some of the later work by Taunton may have been Series 64. But they are pretty similar.
Indeed --- filling the gaps and a better explanation of what is causing the discrepancy between measured Rt = measured Rw + ITTC Cf is the way forward

 

@Leo:
Leo, I am learning a lots about your software in this thread, thank you for having been so generous, objective and constructively critical towards both extremes (experimental and theoretical) of this discussion. And your work rocks.

@Pat:
Pat, thank you very much for having replied to the thread. Your testimony is valuable and has shed the light to many aspects of the research, both theoretical and experimental, laying the basis of the validation of your work done for Hullspeed. You have entered this forum through the main door, and I'm sure it's for good.

@Ad Hoc:
AH, I am glad that you are back (even if just temporarily). That was one beautiful post, intelligent and objective - a brief but densely written piece on science and philosophy of Ship Design. One of the best readings I've had on this forum. Deserves to be read by many and more than once.

I am learning loads of new stuff and am enjoying it. Thank you all. I just felt like saying that.

I'm hoping that at the end of the story it will be possible to draw some common guidelines about the correct way of using both Hullspeed and Michlet software, specifying once for all a checklist of things to be taken into account and with a due attention to the validation and cross-checking with other methods for resistance prediction. Putting it all together here (or, if you wish, in another appropriate thread) could be of great use for all new users of your software, imho.