Const Fn for tank testing?

[curridronan]

I am confused with the principle of keeping the Froude number constant when tank testing a model in order to preserve dimensionality between the model and full scale.
In order to test the model at varying speeds whil keeping the Fn const. you have to build a different model of length associated to velocity to keep Fn const.?
Also to very the Reynolds number you therefore vary the Fn?
The only way to keep this Fn const at differing speeds is to very the viscosity which is not practicle!
Is it that when you test at so called 'constant Fn' at different speeds it is not really at const. Fn but since the Rn will change soo much it is regarded as accurate? Since America's cup tank test data is within 0.5% accurate?
Thanks!

[Olav]

I've never heard of keeping the Froude number constant. However, both model and full scale version have to sail at equal Fn of course, in order to get comparable results for certain speeds.

It's usually not possible to achieve Reynolds number congruency as the required speed to tow the model with would be too large. Apart from this it's absolutely impossible to have both matching Fn and Rn at the same time (unless model and full scale ship are of equal size). For this reason Rn is usually irgnored when carrying out model basin tests and some empirically gained Rn correction is applied later when calculating full scale values from model data. To make for the smallest possible error during the test, the basin water is often heated (around 40°C seems to be a common temperature) to reduce the kinematic viscosity (which in return increases the Rn of the model).

[dumbass]

"To make for the smallest possible error during the test, the basin water is often heated (around 40°C seems to be a common temperature) to reduce the kinematic viscosity (which in return increases the Rn of the model)."

I've worked at three different towing tank facilities, and none of them heated their water. Heating the water to 40C reduces the kinematic viscosity by 30%-35%, but the model and ship reynolds numbers usually differ by one to two orders of magnitude.

[Olav]

Admittedly my personal experience with model tests is limited to what we did at uni with the circulation tank. During our tests the water was heated to somewhat around said 40°C (38°C iirc) and we were told that it was common practice to do so.

The large Rn difference between model and full size ship still exists of course (which in our case was especially large due to the model size limitation around 2m at the test tank in Kiel), but every efford was done to keep it to the absolute minimum.

[tspeer]

Originally Posted by curridronan

I am confused with the principle of keeping the Froude number constant when tank testing a model in order to preserve dimensionality between the model and full scale.

The idea is not to keep Froude number constant, but to keep Froude number the same between model and full scale.

At a given Froude number, the ratio between gravitational and hydrodynamic forces is the same. As a result, the wave pattern is similar between the two scales, and thus the wave drag can be scaled from the tow tank to full scale.

The Froude number will change as the model changes speed if the model scale is kept constant. As it does in real life, too.

In order to test the model at varying speeds whil keeping the Fn const. you have to build a different model of length associated to velocity to keep Fn const.?

Sometimes a series of models of different scale are made and tested at the same range of Froude numbers. This helps to extrapolate the data from model to full scale because you have a trend versus model scale, not just a single point.

But as you say, this is expensive.
[/quote]Also to very the Reynolds number you therefore vary the Fn?
The only way to keep this Fn const at differing speeds is to very the viscosity which is not practicle![/quote]
Absolutely right. It's impossible to match both Froude number and Reynolds number with the same fluid and models of different scale. So the experimenter must pick which one is most important and find some other way to extrapolate the other.

The most common approach is to keep the Froude numbers the same, and thus the wave drag. The earlier boundary layer transition of the higher Reynolds number full-scale article is simulated by placing boundary layer trips just behind the bow and leading edges of the foils. This is intended to ensure that the boundary layer is fully turbulent, and there are no sudden jumps in characteristics due to laminar separation or rapid changes in the transition location.

The estimated skin friction drag of the model is then subtracted from the measurements, yielding the residuary resistance. The estimated skin friction of the full-scale model is then added back in to obtain the full-scale resistance.

Is it that when you test at so called 'constant Fn' at different speeds it is not really at const. Fn but since the Rn will change soo much it is regarded as accurate? Since America's cup tank test data is within 0.5% accurate?
Thanks!

You might look where you saw the term, "constant Fn," and take another look at the context. I suspect it means constant with regard to changing scale, not speed.

[Man Overboard]

Is it possible to do tank testing in say, a lake, and if so what type of gauge(s) would you need to record the data?

[melong]

I have heard of people towing both models and full size boats (NS14 racing dinghys) with a scale to measure resistance. Also John Wellsford, the New Zealand designer of small craft, has a weir near his house with a constant current running over it. The speed of the current varies from a maximum in the middle of the stream to very slow at the edges. He uses this for testing models of his boats at a variety of scale speeds.

[tspeer]

Originally Posted by Man Overboard

Is it possible to do tank testing in say, a lake, and if so what type of gauge(s) would you need to record the data?

Read, "High Performance Sailing," by Frank Bethwaite. He describes the test program he and his sons did to measure the drag of sailing dinghy hulls. They used a cross-arm on a motorboat to tow a hull on each side. One was a reference hull and the other a variation in design. This gave them a very sensitive measure of the difference in drag. They also measured the total drag of the two (IIRC), which gave them an absolute measure.

Others have measured resistance by towing, too. The instrumentation will depend on your test objectives and the resources available to you. It could be as simple as a mechanical spring scale attached to the tow line.

A whirling arm setup in a pond may be an affordable test facility that is in between open water towing and a tow tank.

[Man Overboard]

Tom,

I dug out my High Performance Sailing book; one thing that I noticed Frank says on page 250:

“First: we towed only real hulls which were ballasted to real sailing weights. We never trusted models because such unpredictable things happen in the scale change from model to full size. Even the most experienced workers still sometimes get the scale-factor wrong.”

I have considered towing two models on a T bar as described in the previously mentioned book to get comparisons of small changes to hull design. I think I would like to work out the math which means gathering as accurate data as possible.

You have stated:

earlier boundary layer transition of the higher Reynolds number full-scale article is simulated by placing boundary layer trips just behind the bow and leading edges of the foils. This is intended to ensure that the boundary layer is fully turbulent, and there are no sudden jumps in characteristics due to laminar separation or rapid changes in the transition location.

Is there a formula that is used to determine the size, and how far back from the leading edge to place the trip? I have read of two different techniques: one using a trip wire, and the other a trip strip. I am going to build models in the 6 foot to 8 foot range (about 1/10 scale) maybe you know of a discussion paper on this subject. I think I remember Marchaj touching on this subject; I have all of his books, maybe I can find something about it in one of them.

[tspeer]

You would need to estimate how much laminar flow might be present on the full-scale article, and place the trip on the model at the full-scale transition location.