Smallest model scale for representative tank tests ?

[Boo2]

Hi,

Can anyone tell me the smallest scale model that will give representative behaviour in a tank test ? Full size yacht is a 33 foot "IOR style" monohull dating from the end of the '70s, and the behaviour I want representing is weather helm.

Thanks,

Boo

[luckystrike]

You don't need any tank tests ... ior hulls have always wether helm ... much of it!

Michel

[Paul B]

Originally Posted by luckystrike

You don't need any tank tests ... ior hulls have always wether helm ... much of it!

Michel

No, they don't.

Most IOR boats, as all decent sailing boats, were designed and set up to have 3 to 5 degrees of weather helm while sailing upwind.

This amount of weather helm allows the rudder to provide lift, working with the keel to prevent leeway.

[Perm Stress]

Kirkmann et all recomended at least 15feet ≈4,6m waterline length for hydrodynamic testing of sailboats models. [„Scale effects in sailing yacht hydrodynamic testing“ Transactions, SNAME New York 1974]

This is caused by the fact, that decreasing both size and speed of model foils, Reynolds number decrease dramatically, and water flow round keel and rudder became like oil or syrup flow -i.e. dominated by viscosity.

At present, tank testing of sailboat models is done at smaller scale also, BUT before using test results for any comparisons or calculations, they are "stripped": all the forces caused by keel and rudder subtracted. So it looks unlikely to be useful for weather helm research.

When I did small research for my sailing model, it appeared, that minimum Reynolds number of ~100000 (I calculate in Metric units) is about the minimum for sailing model foils, in order to have same relative size of foils and reasonable (qualitative only!) resemblance in behavior.

This mean, the model have to really quite big. And the size of model is not limited by scale -"1:5 is ok, 1:10 is too small", but it has absolute minimum size limits, due physical limitations.

P.S.
It should be remembered, that quality, cut and trim of sails have large effect on helm of sailboat. How will you account for this in the model?

[Paul B]

Originally Posted by Perm Stress

This mean, the model have to really quite big. And the size of model is not limited by scale -"1:5 is ok, 1:10 is too small", but it has absolute minimum size limits, due physical limitations.

Many serious projects are tank tested in scale of about 1:3.

[Perm Stress]

Originally Posted by Paul B

Many serious projects are tank tested in scale of about 1:3.

Agree 100% with this.

This also depend on the size of "original". For serious tank test work, "originals" tend to be several times bigger as 33feet. and at 1/3 scale they never come close to minimum model size limit.

What I was pointing to, is that there is an absolute limit to minimum physical size of the model itself, basically independent of scale factors.

[Paul B]

Originally Posted by Perm Stress

Agree 100% with this.

What I was pointing to, is that there is an absolute limit to minimum physical size of the model itself, basically independent of scale factors.

Agree.

For a 25 foot boat even 1:3 is not very useful, for the reason you mention.

Several years ago I was involved with a project where a customer was considering two design offices.

The first office told the client they would use their expertise in tank testing at a quite small scale to "scientifically" come up with the best shape. The other office tried to tell the client that testing with small models was not useful.

The client liked the "science" and went with the first design office. The result was a boat that was one of the slowest of it's class.

[LyndonJ]

Many professional model testing facilities around the world are testing models around 1.5 to 1.8 meters in length. Some good results have come from models as small as 900mm. But it depends on the hullform and the reason for testing.

Sea keeping for example is very well indicated with small models.

Hull and keel or rudder interaction with the hull is better on a larger scale although it can be simulated with a portion of the hull. The crux is the Reynolds number, you will get nice accurate results if the Rn is around 1e6 below this and the testing needs to be very precise and conducted by very skilled knowledgeable people.



There was a good thread on this recently about small model testing Leo Lazauskas is a resident guru on this.

[Perm Stress]

Originally Posted by LyndonJ

Hull and keel or rudder interaction with the hull is better on a larger scale although it can be simulated with a portion of the hull. The crux is the Reynolds number, you will get nice accurate results if the Rn is around 1e6 below this and the testing needs to be very precise and conducted by very skilled knowledgeable people.

.

The matter is, for displacement speeds and normal proportions of keels, ~1.5m LWL mean foils working at Re in the order of 1e5 at most ...

[LyndonJ]

That's why the foils are calculated and added later rather than run with the model. You can tether your result to real data for the foil and add it after testing the bare hull.

There are very few big tanks and they are very expensive they are feasible if you have millions of dollars to spend on development but for 99.99% of designers it's just not feasible to test 'big' models.

That's why all the other procedures have been developed or you use a wind tunnel for the foil interaction.

Curious What do you think is the big problem with Re around 1e5 ? I think tested with care and with good accurate load cells you can garner a huge amount of useful data and refine placement of appendages.

[Perm Stress]

Originally Posted by LyndonJ

That's why the foils are calculated and added later rather than run with the model. You can tether your result to real data for the foil and add it after testing the bare hull.

There are very few big tanks and they are very expensive they are feasible if you have millions of dollars to spend on development but for 99.99% of designers it's just not feasible to test 'big' models.

That's why all the other procedures have been developed or you use a wind tunnel for the foil interaction.

Curious What do you think is the big problem with Re around 1e5 ? I think tested with care and with good accurate load cells you can garner a huge amount of useful data and refine placement of appendages.

Here is some data I collected from various sources.

The lower the Re, the higher the drag and the lower the lift. L/D ratio decrease increasingly fast with decreasing Re number (in both tables CL-lift coefiicient, CD-drag coefficient and L/D-lift/drag ratio are quoted for 3 degrees angle of attack ):


Profile NACA 0012

Re Cdmin CL @3 CD @3 L/D @3
250000 0,0093 0,33 0,0107 30,8
500000 0,0083 0,33 0,0092 35,9
1000000 0,0075 0,33 0,0085 38,8
2000000 0,0072 0,33 0,0076 43,4

Note, that L/D for Re 2.5e5 is ~80% that for Re 1e6.
But for 1.5m LWL model at 30% hull speed re could be as low as 1.0e5.

Data in following table is compiled from 3 different sources, so some errors due to different procedures are possible; however,trend is more than obvious:


Profile NACA 0009

Re Cdmin CL @3 CD @3 L/D @3
50000 0,28 0,0190 14,7
60000 0,33 0,0156 21,2
2000000 0,0053 0,33 0,0076 43,4

From last table it could be seen, that at Re 60000, L/D is about 1/3 that at Re 2000000.

Essentially this mean NO SIMILARITY of forces so far, as modeling of foil behavior is concerned.

So, if our model has to model QUANTITATIVELY behavior of foils as well as behavior of hull, nothing less than ~4.7m LWL, as recommended by Kirkmann will do.

My guess would be for QUALITATIVE modeling of foils behavior ~3m LWL would suffice. But I would not expect too meaningful numbers from such an experiment.

As far as I know, small (~1m LWL) yacht models are tank tested with kind of standard keel and rudder, (to model influence of foils to flow round the hull) with low Re forces known well enough to be later subtracted from total forces (hull + foils), as measured. Of course, this process involve much guesswork and empirical data, when tank tested hulls are built full size and sailed for real. However, this kind of approach could became less accurate, when hull shape/speed or/and foils proportions wander far outside "beaten track". Because with no empirical data, linked to tank tests, amount of guesswork increase. So far this method looks to work well enough to be accepted by industry (although not totally 100% ). However, top designers for top-end projects use models over 6 m in length. And I think one of the reasons is low Re induced dissimilarities in flow round foils.

This problem with low Re numbers do not belong entirely to models.
For full size boats, efficiency of foils in terms of force coefficients also decrease at slow speed. If not to such dramatic extent as in tables above, because Re are still generally over 1e6 even at couple of knots.

Remember, how easy it is to stall the high-aspect keels and rudders, when sailing in near calm conditions? How careful and slow has to be every movement of tiller not to kill speed and maneuverability, when boat just slip along at 1-2 knots in light airs? How small deflection of tiller is necessary to stall the ruder? And how much more difficult it is to stall the rudder at 4 knots? How much more deflection of tiller stall the rudder, if at all?
Lower speed of flow is not all the story here. Every square centimeter of foils is less able to produce lifting force and create more drag for given lift too.

[LyndonJ]

Great effort Perm

And I agree, but that's not really a condemnation of small model testing. It illustrates very clearly the problems modelling keel-hull flow field predicition from small modsels. I still think a lot of very usefull design data comes from surprisingly small models, you just need to add the foil data separately becasue there is a compromise between 'ease of accuracy' and the limited bank account. You'll always be considerably better off testing a small hull than none at all , don't you think?

[booster]

Hi!
This is difficult. You have to get the medium scaled as well to get it perfect. Unforunatelly, several of such media are toxic. Don't do it.
Regards,
Booster

[Perm Stress]

By Lindon J:

"And I agree, but that's not really a condemnation of small model testing. It illustrates very clearly the problems modelling keel-hull flow field predicition from small modsels. I still think a lot of very usefull design data comes from surprisingly small models, you just need to add the foil data separately becasue there is a compromise between 'ease of accuracy' and the limited bank account. You'll always be considerably better off testing a small hull than none at all , don't you think?"

Originally Posted by Boo2

Hi,

Can anyone tell me the smallest scale model that will give representative behaviour in a tank test ? Full size yacht is a 33 foot "IOR style" monohull dating from the end of the '70s, and the behaviour I want representing is weather helm.

Thanks,

Boo

Keeping in mind the original intention -representing weather helm - adding foils later, in calculations, could be not accurate enough...