Testing Boat Designs

[Jingo Lingo]

Hi all. I'd just like to ask how boat designs were tested back in the day before computers were around to do so. It's for a physics assignment so any feedback would me most appreciated.

[jehardiman]

Actually, nobody in their right mind uses only computers to design and test vessels. Computers, like any power tool, only let you work/make mistakes faster, they don't DO the job, or MAKE decisions.

BTW, the term "computer" is as old as "computor", dating from before Babbage's machine all the way back to the rennassiance at least. A compter is just a machine used to aid the computor in doing the drudge work with fewer (though faster and more consistant) mistakes. There are still some Computors employed in the engineering world, and thier insight should be sought when developing computer code

Back in ancient history, before personal computers, when I was in college, vessel perfomance was correlated (not tested as you say, because only full scale articles can actually be "tested", everthing else is a model) against tank testing. I would advise that you go down to your school's library and look in the card cataloge under hydrodynamics;tank testing. That should lead you the requsite texts. For what you want I would recommend Attwood.

[messabout]

Jehardiman has delivered the gospel. Yes, some computer programs are used to design boats, but serious design outfits still do tank testing before building a major project. Computers be damned. They are marvelous tools but they are even more subject than humans to the GIGO affect.

[FAST FRED]

40 Years ago the folks at the AYRS were using scale models and weights to test what became cat & tri hulls .

They would use a modified bathtub as water source for a sluce, and compare models on a sea saw type yoke.

FF

[Omeron]

Can somebody explain how tank testing process perfects a design?
You have this magnificant model, which must take hundreds of hours and thousand of dollars to make, test it, than decide you need to change this and that, and you go back and build another one,test it, and think it is not yet the perfect form, and go back and build another one, than another one.... than start switching keel, rudders...
To me, this seems like an endless process...both in terms of time and money.
Or is it that once you see how the first model behaves, you derive this data and feed it to the brain of the designer and/or computers and expect necessary modifications to be made based on that testing?

[MikeJohns]

Originally Posted by jehardiman

Back in ancient history, before personal computers, when I was in college

I did my first eng degree with slide rule and Etons tables in the 70's ( doesn't seem that long ago).
When I got the first programmable HP calculator I used it for all sorts of labour saving calculations. Then I went through a variety of early PC's. Now I have a state of the art multi core CPU Terrabytes of Hard disk space and more RAM than I can ever use doing large stress/strain FEA models and I still would never do more than a very rudimentary test a hull model on a PC. To replace tank testing, the CFD would be more expensive and considerably more unreliable than building a scale model and paying for the tank testing.

[MikeJohns]

Originally Posted by Omeron

Can somebody explain how tank testing process perfects a design?
You have this magnificant model, which must take hundreds of hours and thousand of dollars to make, test it, than decide you need to change this and that, and you go back and build another one,test it, and think it is not yet the perfect form, and go back and build another one, than another one.... than start switching keel, rudders...
To me, this seems like an endless process...both in terms of time and money.
Or is it that once you see how the first model behaves, you derive this data and feed it to the brain of the designer and/or computers and expect necessary modifications to be made based on that testing?

Looking at a video of the generated wave patterns often tells you as much as you want to know. Often it's just a case of making sure that your design has a reasonable resistance for the type, and no nasty characteristics. Sailing vessels are a pain, power vessels are a lot easier.

The more dilligent and well supported racing boat designers however really refine things as you describe.

[Omeron]

Thanks Mike.
Are you saying that for the size and displacement of your envisaged design,there is a target figure for resistance (form and friction i guess)at specific hull speeds, and if you are at or below this, you can consider your design succesfull, and mostly thats the end of the story?
If so, are these figures determined by the designers or are they universally known and accepted targets? Or are they calculated by design softwares?
Too many questions, but i believe important ones to understand where the art and science of boat design crosses over.

[jehardiman]

Originally Posted by Omeron

Can somebody explain how tank testing process perfects a design?
You have this magnificant model, which must take hundreds of hours and thousand of dollars to make, test it, than decide you need to change this and that, and you go back and build another one,test it, and think it is not yet the perfect form, and go back and build another one, than another one.... than start switching keel, rudders...
To me, this seems like an endless process...both in terms of time and money.
Or is it that once you see how the first model behaves, you derive this data and feed it to the brain of the designer and/or computers and expect necessary modifications to be made based on that testing?

First of all, models were cheap. They were made out of wax using a scaling pantograph mill straight from the lines plans. Normally, several models would be made, slightly different, and when you were finished you just melted them down. The exception to this were the standard series models. In these normally wooden models, all parametric varations of one hull form, they would be maintained as a library to test and calibrate the tank equipment and to compare to other standard series. FWIW, almost all modern computer programs to predict hull powering that do not use CFD directly, use the TANK data from the standard series as the basis of calculation, interporlating between the model parameter data points to predict your exact set of L, B, T, etc.

Second, do not think an computer "optimized" hull is ever the best. CFD is nowhere near ready for the average engineer to do speed in seaway or forefoot slamming calculations. And the smooth water "invisicid, irrotational" CFD suffers from parametric sensitivity. To prove this, take an "optimized" model and start varying G, rho, and nu within reason. The model will change, sometimes significantly. Since G, rho, and nu vary all over the world, have you really optimized the model. (This BTW is one of the major gremlins in things like an ACC design where the last 1% matters, you don't get the water you designed for.)

Yes, getting a real ship to sail in computer water is as difficult as getting a computer ship to sail in real water. That is why you do tank testing on a few of the final options, so that the real vessel will meet performance goals when tested in less that optimum conditions. As one of my professors advised, "Always pad your drag at service speed and add 10% for sea reserve to the engines, that way the owners will be buying you drinks in the bar, rather than sueing you in court".

[Omeron]

Dear Jehardiman,
Thanks for taking the trouble to explain it all.
My first comment is, people like us who are just intrigued and asking such
questions shuld be banned from this forum, so as to not take precious time of professionals in the business.
But then again, i am sure such questions trigger responses which may be more beneficial and intelligent to others in the profession.
I am curious why you say models were cheap, as if implying they are no longer.
What i understand from your response is that, unless you are in the business of perfecting that last one percent, which in itself may not be an exact science in its own right, it is not worth to go into tank testing.
To please your client... YES...
To put your mind at ease....YES
To avoid potential disputes... YES

But at the end of the day, it still boils down to what you say...
If in doubt, increase the safety factor, if you are still in doubt multipy by two..
Nobody will be able to know or even want to know the difference between one percent and two percent...

[jehardiman]

Omeron;

In todays economics, it is the ability to leverage sunk costs and the improvement in prime movers that has reduced tank testing. Why pay for a whole new design when just building a new hull with a more efficient prime mover is much cheaper. A normal bulk merchant of say 60,000 DWT is ~$25 million and in it's life will consume ~342240 T of fuel at a cost of $ 180/tn means a fuel cost of $61million, for a total cost of $86 million over the life. A $2 million design effort that saves 5% on fuel is bearly break even (i.e. 25+2+61-3 =85) and may not be if present value is included. The more efficent a ship becomes, the less advantage there is to investing money in efficency. As I have said about design before, the last 10% in performance will cost 90% of the money.

[Eric Sponberg]

In all my years of designing boats and yachts (going on 30 years), I have had only two clients request to go to a towing tank or a wind tunnel (one of each). It is extremely rare for run of the mill designs to go to a testing facility. Usually, the client does not have the budget to support the cost. Only really high profile racing sailboats or special mission or expensive powerboats go to the towing tank.

On the Moloka'i Strait 65, we went to the Institute for Marine Dynamics at the Memorial University of Newfoundland in St. Johns. The marketing/technical company for the tank, Oceanic Consulting, handled all of the details--arranging to build the model at MUN (1/5th scale, about 13' long), running the tests (took about 3 days), writing the report and supplying the photography. It was great value for money. Total cost of program was about US$31,000 (2002 dollars).

There was only one thing that we intended to get out of the testing, and that was the effective horsepower. We wanted an accurate prediction of horsepower going into the design so that we would be sure of our speed, engine power, and propeller requirements on the first shot. Engines and propellers are really expensive items, and you don't want to do trial and error at full scale with that sort of equipment. We did not use the tank to optimize the design. The design was what it was, we just needed to know the powering requirements. Could we have optimized the design further? Yes, but that would have set the design time back another couple of months and a few tens of thousands of dollars.

At the time we were doing our testing, the Swiss were testing their America's Cup designs in the same facility. This was for the effort in which they first won the America's Cup. They were testing 18 one-third scale models in 18 weeks. Each model had slight variations from the others in a predetermined program. The different variations probably had to do with nuances in hull shape and stability. Generally, the model testing at that level is looking for extremely small differences in performance. So for 18 huge models in 18 weeks--you can imagine the cost.

Eric

[RANCHI OTTO]

These are the pics of the model and the full scale craft:

Model length = 1.6 m

Craft length = 26.0 m
Speed = + 45 knots

Test carried out:
1) Bare hull resistance
2) Resistance with appendages
3) Selfpropulsion tests
4) test with propeller model in cavitation tunnel

Speed vs power forecast very closed to sea trial results

[Ike]

Many production boat builders did not go the model and test tank route. They built a full size prototype and put various engines in it and took it out and ran it. Then if they didn't like it they changed the hull and did it again until they got what they wanted. Then they made a mold and started building production boats. This was easier in the wood boats because you could tear the bottom out and change things easily. You could do much the same in fiberglass but it was a bit more difficult.

Mercury is a good example. They had a large test facility in Osh Kosh Wisconsin, and another one called Lake X, in Florida. They ran boats and engines to death at both of these places. They still have the Osh Kosh faciltity but he Lake X was sold and they have a new facility in Florida.

[chandler]

When did the swiss win the Americas Cup"? Sailing??