Boundary-layer trip

Since the beginning of this year the raw data of the towing tank tests at Delft University are available online. To make good use of this invaluable data it is necessary to understand the influence of the 3 roughness strips that are used at Delft to trip the boundary layer.
I was hoping that someone out there in the large boatdesign.net-community was already confronted with such a problem. The literature about the properties of tripped boundary layers is sparse. I have compiled in the attached report everything that I could find in the literature. The data for the location of the transition and the length of the transition zone exhibit considerable scatter. Nevertheless the comparison of the experimental results with the b.l.-calculation is not so bad. Any comment would be helpful.
Uli

 

I would definitely have a read through of the paper:
Bertorello, C., Bruzzone, D., Caldarella, S., Cassella, S., Cassella, P. and Zotti, I.,
"From model size to full size. Investigation on turbulence stimulation in resistance model tests of high speed craft",
FAST 2003, Ischia, Italy, Oct. 2003, pp. A1-1, A1-8.

They found that the type of BL trip can cause differences of up to 20% (I seem to recall) in the total resistance for small (< 2.0m) models, especially for Froude numbers < 0.6

Bradshaw's report is also worth looking at:
Bradshaw, P.,
"Collaborative testing of turbulence models: Final report on AFOSR 90-0154",
Mechanical Engineering Dept, Stanford Uni., California, USA, Dec. 1992.

 

On a similar topic, can anyone point to data that compares transition in air vs the towing tank or open environment? The limited, and proprietary, data I've seen that was collected by AC teams on bulbs suggest transition occurs at much lower Reynolds numbers in real seas, but I've not seen anything definite on the subject.

It would be really great to have a correlation with a transition-related design parameter like Ncrit for water vs air.

 

I don't know of any data or comparison studies. My guess is you might want to look at boundary layer research including work done decades ago.

Two factors which should be kept separate in considering separation:
1) Air vs water
2) Ambient turbulence

Another guess is the ambient turbulence for tests depends on the research facility, and comparison data may not be readily available.

 

Thanks Leo,
I do not have the first paper, but Klebanoff demonstrated already in 1955 that the flow behind a trip wire and behind a roughness element are fundamentally different.
Bradshaw's paper is interesting and useful for the fully turbulent flow, after the transition zone.
If you ever come across measurements of the length of the transition zone for a tripped b.l., please remember me.
Uli

 

There is no physical difference between non-cavitating water flows and air flows at low Mach number. Both are Newtonian fluids. There might be a large difference in the turbulence level. The turbulence level in the ocean is discussed here: http://www.boatdesign.net/forums/hydrodynamics-aerodynamics/turbulence-level-ocean-47143.html
The turbulence level in the atmosphere at high airspeeds will be close to zero.

The parameter Ncrit describes the transition based on the linear stability theory (T.S.-waves). At the high turbulence levels in breaking water waves the linear instability is by-passed and nonlinear transition occurs. The exp(N)-method is normally not applicable for by-pass transition.

May be you refer to the paper by Binns, Albina & Burns "Looking for 'laminars': Measuring intermittency on the America's Cup race course". This paper is really disappointing, all diagrams are only qualitative, with the important numbers removed. There was no statement about early transition, so you must have some inside knowledge!
Uli

 

The location and type of BL trip can have a significant effect. Especially on semi-p and planing hulls at and around the hump. There are also 2 principal schools of thought on how to apply them, or rather note their effects when converting from model to full size.

1) The use of small studs at "some distance" from the FP. This has been a bit trial and error by many towing tanks to arrive at their norms. We have found that using a 10-15% laminar section yields consistent results. For non-planing vessels. The resistance calc's are then computed from the laminar and turbulent parts.

2) Using strips of sand paper of a certain grade, again at a "certain distance" from the FP. This method makes no allowance for laminar and turbulent. It is all assumed turbulent.

2 very different methods, yet the results have been consistent.

We have tried this ourselves, with the same results.

 

There is a good description of how to estimate drag due to boundary
layer tripping studs in Appendix A of the Uni of Southamptom Ship Science
Report 71 by Molland, Wellicome and Couser. (That's the one with the
enormous number of NPL experiments).
The same description was used in the latest ITTC Resistance Committee
reports.

 

Indeed. It was this that sparked or piqued our investigations further, at that time. Since results from another test house used a different method.

 

1. The FAST 2003 article is very good, but I only have a faint
photocopy, otherwise I'd show you some of the more important
graphs and comments.

The importance of the study is that they tested three geosims,
L=1.805m, 2.347m and 4.694m, without trips and with sand and with wire
trips at different locations. I know of no other comparable systematic
and comprehensive tests conducted by a group of very respected and
experienced ship hydro scientists. (Under-grad student projects are
always a little suspect, IMO).

The most important conclusion was:

"A small model (L < 2.0m) seems inadequate for the model-ship
correlation in the range Fr < 0.60 because of the appreciable
influence of laminar flow on the total resistance with the bare hull
and of the uncertainty about the value of the additional parasitic
drag due to the applied device in the tests with the turbulence
stimulators."


2. This report might be helpful to your research. If nothing more it
might give you a few more references to chase.

Jason Christopher Murphy
"A Novel Approach to Turbulence Stimulation for Ship-Model Testing"
http://www.dtic.mil/dtic/tr/fulltext/u2/a549033.pdf

 

I fully agree that models < 2m are inadequate.
Fact is, all tests at Delft were conducted with models of 1.6 or 2 m length. Because this is a phantastic database of more than 60 models that is now available for free, I am investigating methods to overcome this deficit of partially laminar flow. The last diagram in my above attached report shows, that my boundary-layer calculation can identify the laminar flow and the transition regions. If I use the calculated viscous resistance of this mixed flow and subtract it from the measured total resistance, I get a residual resistance that might be close to the true wave resistance. The diagram also shows that the ITTC-line dramatically fails for such small models.
At Delft they test the models always twice. The second time with a roughness strip of double width. The difference is assumed to be the additional parasitic drag of the sand grains.

My friend Google had found this already for me

 

The findings of Bertorello et al (and others) present the ITTC with an
interesting challenge. The last ITTC said that it was more important
to clarify the role of form factors on resistance than to find a
skin-friction line that is better than the ITTC 1957 line.
However, if measurements are unreliable at low Froude numbers
(especially for models with immersed transom sterns) then where does
that leave the Prohaska method which relies on extrapolating low Fr
tests back to Fr=0? Form factors would seem to be very unreliable if
derived from small models.
Transom sterns are a real problem because low Fr tests mean the transom
is usually completely wet. Bulbous bows also present some challenges,
but that is a slightly different issue.

Do you have a link where I can download the Delft tests? I tried a
couple of years ago but it wasn't as simple as just clicking and
downloading.

 

It is here, Leo: dsyhs.tudelft.nl - you just have to send an e-mail with a request for a password.

At least that's what they say in this article: http://www.misit.nl/docs/DO_2013_1.pdf.

Cheers

 

For all practical purposes in real world motions and turbulence, transition occurs at so low a Rn that flow is always turbulent for all real vessels (or so all my calculation to practice data shows and that's the way I code our programs). See Hoerner's discussion on boundary layer tripping as a function of velocity fluctuations in Fluid Dynamic Drag Chapter 2, Section 6 with specific notice to Figure 9 and the following thread.

http://www.boatdesign.net/forums/hy...in-friction-lines-some-comparisons-46272.html

Edit; Still reading the paper, but I notice that some of the scatter may be due to the pre-existing flow condition rather than the surface treatment.

 

I have downloaded the various files and imported the iges files into
Delftship Pro. I can get quite beautiful grids, drawings and graphics,
but there are a lot of leak points.

If I fix the leak points on the centreline, I can view hydrostatics
but I can't get the program to export a table of offsets or, more
importantly for me, I can't get it to create a Michlet file. If I
could do that I could check some of Uli's calculations and I could
also start setting up input files for the free version of Flotilla I
am working on. That would then allow me to see how closely I can
predict sinkage and trim.

So, has anyone managed to get Delftship (or some other program) to
produce a table of offsets from the DSYHS files that could be used as
input to Michlet?