size doesn’t matter?
A counter rant to the does size matter reader rant from yesterday…
The use of 1/3 scale models restricted testing to a handful of facilities
in the world. In the US, the only tank capable of testing a 1/3 AC Class
model was the David Taylor Model Basin at Carderock. Anyone doing business
with Carderock knows 2 things: they produce excellent results and they
aren’t cheap (it could be that the AC guys got a deal, but I have no idea).
Depending on the size of the new AC class, a 1:5 scale model may open up
the use of other facilities around the world and significantly reduce the
costs of testing, not to mention the expense of building a model. In my
experience, a 30′ model could easily run $150-200K and the testing much more
than that. A 1:5 scale model will cost much less to build, will be easier
to handle, and will be less expensive to test. Other US facilities
potentially capable of handling a 1:5 scale model include the US Naval
Academy and Davidson Laboratory ("new" wider and deeper tank). RIP
Hydronautics – an excellent facility with many AC tests to its credit, now
the site a housing development.
Will the results be "worse?" This is a loaded question and it really
depends on what you seek to get achieve from testing. To say that small
scale tests (1:8 and smaller) did not produce useful results is incorrect.
Ken Davidson used small scale tests 1:12 I think) in the 30’s on Ranger and
the lab proceeded to test all te AC winners through 1980 I believe. While
its correct to say that small scale tests have larger errors associated
with reynolds scale effects, they do help resolve trends in resistance
useful in early design studies. In fact, smaller scale models are often
employed at this stage, and this has been true through modern AC campaigns
as well as Volvo. Limiting testing to 1:5 scale simply means that the
approach to design will be fundamentally altered. Any AC campaign can
contract their CFD out, and to my knowledge, most of the CFD gurus are not
full-time hires anyway. So, a new syndicate need not have embedded
experience with CFD. While CFD keeps getting better as turbulence models,
meshing techniques, and processing speed improve, don’t expect that CFD
alone will give you a better answer than even a 1:5 scale model test
because 1:5 aint so bad!
Back in 1989 when the new AC class was unveiled, the original PACT
(Partnership for America’s Cup Technology) was actually a group of
investors that sought to gain collective knowledge about the new rule. They
did so by pooling their resources and testing 1:8 scale models. 1:8! Good
God, the sky is falling! This may again be a smart decision, especially
among the smaller syndicates looking to get their foot in the door.
By the way, a little known fact about the tests performed on Mariner and
Courageous prior to the 1974 Cup was that Davidson Lab knew that Courageous
was the faster boat but was contractually prohibited from saying anything.
The scale effects skewing the results of these tests weren’t enough to
alter this conclusion. – Jesse Falsone.
Regarding Jesse Falsone’s counter rant, I would like to add to what Jesse provided.
As the person who oversaw the work on behalf of PACT, while the organization did indeed provide partial support for a 1:8 scale model tested at Davidson Laboratory, it would be a mistake to construe that as an endorsement testing at that scale, or any other. Most of PACTs calm-water resistance and side force tests were conducted using much larger models, and one objective of the Davidson tests was to see whether a smaller model could be scaled up to match those of a larger model.
Because of the subjective nature of unraveling the issues of scale effects based on final observed race performance of competing yachts such as the Courageous/Mariner matter, scientists prefer to resort to testing a series of models of various sizes up to and including full size; a so-called "geosim series" because all models are geometrically similar but different lengths. It is important to understand that an individual test result has no meaning until it is "expanded" to correspond to a full-size yacht result, and the challenge of small models is as much in getting valid expansion techniques as it is any feature of the results themselves. In fact, perhaps the most useful results of our SNAME paper was that it proposed improvements in how to scale up results from models of all sizes which have since been rather widely adopted for yacht tests: 3-D extrapolation, and stripping of appendages to name some examples.
In fact, small model results of 12-M hulls with full sterns had been under scrutiny since the 1970 match, and that controversy was what motivated the work we did at HYDRONAUTICS, Inc. on sailing yacht hydrodynamic scale effects.
As shown in the paper, models of various sizes of Valiant – perhaps the most suspect 1970 design – were not amenable to extrapolation between various sizes, and the existing methods worked with the least uncertainty for the larger Valiant model as compared to full size towing tests we conducted with Brit Chance.
It is far better to think of usefulness of test results in terms of the uncertainty level of the outcome rather than subjective characterizations about "usefulness" at various model scale ratios. Always ask the question: is the uncertainty level of a test result sufficiently smaller than the difference being evaluated that the results will provide a conclusive appraisal ?
Karl L. Kirkman