Boundary-layer trip

My first thought was: what a great idea, it will save me a lot of tweeking with leak points in Rhino. Then I tried to use the coordinates-file from the DSYHS database instead of the IGES files. Result: they are at least as messy as the IGES files. It took ages till I had cleaned up the files manually. Excel helped a little. I have to think of an intelligent program code that lets the computer do the sorting and cleaning.

 

Thanks for the warning, Uli.
I'll have to think about what to do if I ever want all 70 or so models.

 

Thank you, Uli, for a very thorough paper (again). Interesting that the bumps in the experimental curve correspond to the roughness strip effects. Where do you get the velocity distribution for your boundary layer calculation?

 

The b.l.-calculation is my old workhorse that is described in http://www.remmlinger.com/Optimization.pdf
It is a simplification, becaus it approximates the hull with a body of revolution, but the results compared well with the Dehler 33 experiments. The velocity distribution is calculated based on Landweber's method: http://www.biodiversitylibrary.org/bibliography/48298#/summary
The velocity has only a minor influence. If the hull is well designed and there is only little wavemaking, then speed differences are small.

Concerning the paper about the b.l.-trip: I am still in contact with Jasper den Ouden from TU Deltft about the exact strip position. The version of the paper on my web-site will be updated as needed.

 

With no knowledge of how Michlet works, I would note that SYSSER72 appears to have a substancial, low aft overhang. I would imagine this would increase the WL at higher speeds (all the way to the stern), reducing wave making drag (while adding into wetted surface). If Michlet does not allow for the overhang, one would expect it to overpredict at higher speeds.

 

Yes, as you say, the overhang could lengthen the effective waterline, but
that will primarily affect the transverse wave resistance. Diverging wave
resistance depends more on the actual hull shape, and as you can see in
the graphs I attached, it is generally monotonic in Froude number, whereas
the transverse wave resistance oscillates a lot with Froude number.

Incidentally, large overhangs can play havoc with some non-linear codes that
use iteration.
For example, on the first iteration (usually a linear evaluation) the program
will find a stern wave that touches the underside of the overhang.
On the 2nd iteration, it includes the new wetted length and it finds that the
wave at the stern no longer touches the underside of the overhang, so
on the next iteration it doesn't include the overhang.

 

So perhaps the role of the overhang is to dampen out/prohibit the tranverse wave, not letting it grow under the stern... this would support/justify your Rtfunny approach, wouldn't it ;-)

Looks like that's exactly what is happening with my sims - I get this odd hobby horsing, with the stern wave appearing to push the stern up, resulting in an unrealistic pitching motion as the boat speeds up. Perhaps adjusting the time step to shorter will help, I'm usually trying to cheat by increasing the time step to make the simulation run faster.

You once asked if I get solitons, by the way - I do, if I make the basin very shallow (again to make the sim run faster, with less elements). I don't have an example at hand, but I will post it the next time it happens.

 

It would, but all I have found so far is an interesting correland for one hull.
I am still a long way from showing a physical connection. I'll get back to you
after I've examined all 70 or so Delft hulls and some others

That's a good result, Mikko. I'll be interested to see it when you time.

Another interesting exercise is to see if you can get a sudden drop in
resistance in a finite width, finite depth channel (i.e. a realistic towing tank).
Linear theory gives a very simple expression for the size of the drop in wave
resistance for that situation.

 

There is a s slight complication in trying to reconcile laminar flow,
transition to turbulence and then fully turbulent flow, and that is
that fully-turbulent flow might only be achieved at very high Rn,
somewhere around 10^9. Form factors can be quite different if derived
from small (< 2 metre) models. There is a very interesting paper that
examines form factors derived from 2, 4m, 8m and 10m models:
Keh-Sik Men and Seon-Hyung Kank,
"Study on the form factor and full-scale ship resistance prediction
method",
J. Marine Science and Technology,
Vol. 15, 2010, pp. 108-118.

Form factors have been shown to be Rn-dependent, which is a nuisance
for those NA's whose drag estimates assume it is independent of Rn.
In aircraft analysis, so-called "terminal" form factors are sometimes
used and these are usually constant as long as the flow is fully
turbulent, i.e. Rn > 10^9.

Thus, it might be imprudent to assume fully-turbulent on the Delft
Series hulls, even at the higher Froude numbers for those small
models. The Rn is still far too low, despite extensive BL trips being
used.

 

This is exactly the point that I was trying to make. Therefore I do not use form factors. I calculate the boundary layer in a similar way as XFOIL or JavaFoil do it. Attached is the variation of the friction coefficient along the hull for Sysser 1.

 

It did surprise me how large Rn must be to be sure of fully-developed turbulent flow.

I think I have mentioned the concept of "anaemic" BL before, but it might be
worth re-iterating the concept for the younger students who were playing
games on their iphones instead of listening to us old fogies.

Anaemic boundary layers can occur when trying to achieve high Rn using a
short fetch.
It seems reasonable that because Rn = UL/nu where nu is kinematic
viscosity, U is speed and L is length, that we can choose to increase either
U or L to achieve large values of Rn. (Varying nu by large values is
impractical).
However, anaemic BL arise when U is large, but L is not sufficiently long for
fully-developed turbulence to occur. This situation can also arise behind trips
if there is some re-laminarisation.

 

Try adding a small Angle of Attack (AoA)...then you know what the real world is about

 

It depends on how you get your jollies, I suppose.

I could also put it at an AoA in such a way as to align the flow to
be tangential to one side of the bow, and then treat that side as a
flat-plate flow and ignore the other side.

 

In this thread my intention was not the simulation of real world sail boats, instead I wanted to investigate possibilities how to distill the wave resistance out of the DSYHS towing tank results at zero AoA.

As you certainly know, there is a vast literature about the drag- and lift-forces on submarines under AoA. It is interesting to see, that there is a dramatic difference, if the "sail" and fins are added to the bare hull. This will be even more pronounced in sailing yacht hulls, because the keel-fin is large compared to the hull. The lift carry over from the keel to the hull changes the hull flow dramatically. The keel reduces the incidence of the flow along the hull and aligns it more with the centerline. This way the hull drag alone, without the induced drag from the keel, is not much different from the drag of the bare hull at zero AoA. So much to the complexity of the real world.
Uli

 

Uli,

I have been doing CFD runs on some SYSSER models to make comparisons. In your B-L trip paper Fig. 12 you have Cd starting up from 0,002 or so... from the Delft data for Sysser 72 I get Cd from 0,0033 and up (likewise in my CFD). The data you are showing is fullscale? In fullscale you wouldn't have the trips and no jumps in the viscous Cd curve?

Many of the SYSSERs have quite long and low aft overhangs (like the 72), so their actual wetted surface will be quite a bit larger than the given static area. To really get into the actual residuary resistance, you would also have to allow for this, would you?