laminar to turbulent

Since the initial flow over a surface (hull or foil) is laminar for more than the first foot or so, regardless of reynolds number ie. model or full size boat.

On a foil of short chord (less than a foot or so,) as the flow goes past the max thickness and the pressure drops, will the flow tend to stay laminar or will the change in pressure still cause it to go turbulent? (Even though its own viscosity has not tripped it up yet) And do programs like XFoil etc, correctly model this? Maybe the flow over the typical NACA 0012 is always laminar if the chord is short enough? (keel typical AOA)

thks, idkfa

 

One of the most difficult subjects to predict.. Just make an assumption all flow is turbulent can't go wrong.. IMHO

 

Very ineresting.....

Are you saying the flow won't transition to turbulent within a foot of the leading edge even if the the Reynold's number is otherwise large enough? Reference?

 

Except that due to surface roughness, you can almost guarantee that the flow will be turbulent. The only exception to this may be trailered racing boats and IACC class boats where there is extensive cleaning between races. For most other sailors this is prohibitively expensive.

You also find that you can only support laminar flow over a small range of Cl. This is unfortunate, as foils (at least keels and rudders) need to work at a very large range of Cls. Typically Clmax in both directions.

Camber changing devices will afford some benefit, but these are difficult to build and maintain, and are very difficult to design to retain laminar flow, even if you had it in the first place.

Consequently, "designing for laminar flow" which seems to have been very popular for some time is the wrong way to do it. You should be designing for the required performance, and then looking at whether a laminar-flow section is useful. Typically it isn't unless you intend to operate in a small Cl range.

As Teddy says, prediction of transition is phenomenally difficult, however, I think XFoil's transition output is reasonable, but you must remember that it is a 2D panel method, and therefore has limitations.

Tim B.

 

If you move the wing slowly enough through clean water, you may be correct.

-Tom

 

Indeed. That would be Reynolds-number dependent then.

 

When model testing you adjust speed to maintain reynolds and the turbulent flow behaviour is the same, but that initial portion of laminar flow needs further adjustment with a "comb". And that's my point, that initial behaviour? It's not (so) depend on reynolds number but some other intrinsic property, viscosity? Thus maybe the changes are not modelled correctly by software, and maybe for normal dinghy speeds a NACA 4 digit centreboard has similar drag of say the NACA 6 digit? ie laminar flow over most of it. Anyone with actual tank test results/pics. I would expect at larger AOA of a rudder, the drop in pressure would initiate turbulence.

 

When testing a scale model of a boat the speed is adjusted to maintain Froude number, not Reynolds number. The result is the scale model Reynolds number is lower than it would be in full size at the same Froude number. Assuming viscosity is the same the ratio of the scale model Reynolds number to the full size Reynolds number will be the model scale ratio to the 3/2 power. So a 1/4 scale model will have a Reynolds number 1/8 that of a full size boat at the same Froude number (Even with a variation in viscosity due to temperature or salinity of the water the scale model Reynolds number will still be smaller). This is the reason boundary layer trip devices are used on models in tow tank testing.

Well over a century of careful experimentation has demonstrated that Reynolds number is the appropriate scaling factor for boundary layer behavior considering size, speed and viscosity.

 

thanks DCockey, I have no excuse now not to give XFoil a try, have a mortal fear of garbage in garbage out.

 

As I understand it, there are several independent conditions that cause flow to transition from laminar to turbulent. The areas of interest are where the foil operates where you're not exactly sure which one is doing the deed. The drag bucket is an effect of transitioning between one cause and another as AOA changes at Re above 10^6 or so and can be predicted quite well. As the Reynolds number drops below that, more factors come into play, and prediction becomes more difficult. A good model for re=10^5 foil performace would be a boon for dingy designers and terrorists cobbling together UAVs in the 5 kilo class. I reckon I don't mind it not being out there for free.

 

as the fluid passes over the forward part of a foil, the pressure is dropping to the max camber point. This tends to help keep the flow laminar and attached. After the max camber point the pressure is raising to meet the free stream pressure at the TE, so it becomes very easy to transition to turbulent, and even loose attachment all together at high AOA.

And as pointed out, and surface roughness from dirt or growth or turbulence in the fluid, you loose the laminar flow. But to make matters worse, if you optimize the foil around the laminar flow conditions, and you are transitioning to turbulent, the foil properties change drastically and the performance is worst than if you had just had a good turbulent foil design . So you not only have the drag go up, but you loose even the performance a good turbulent foil will give you.

The last trend in foil designed for light aircraft was to try and design a laminar flow foil with reasonable performance in turbulent conditions. When it comes to aircraft, I just as soon design for turbulent flow, be safe and not worry about the mythical performance increase due to maintaining laminar flow. Depending on what you are trying to accomplish with your boat, I would also consider doing the same thing.

It is not very often you can maintain ideal conditions to maintain laminar flow in a practical craft of any type. If you are designing for very specialized purpose, like for short duration race, you might make it work, but otherwise expecting turbulent flow conditions would produce the most reliable results.

 

For 2D section design and analysis, XFOIL works well at Re<10^5.

In that Reynolds number range, transition occurs via laminar separation and turbulent reattachment, forming a laminar separation bubble. The SoarTech data are probably your best bet for validating XFOIL in this Reynolds number range.

XFOIL will not help you with hull design, however. The boundary layer development on a hull is probably in between axial flow and 2D planar flow in character. Some panel codes have the ability to do a quasi-2D boundary layer calculation along a surface streamline, ignoring cross-flow boundary layer effects. They typically don't handle free surface effects, either, but they might still be useful for understanding what is going on.

 

Thanks for the link. I recently searched the web for references, but all compiled data sources seemed to have been taken down. I'm slowly getting a bit more confident that the models will work ok for what I need.

 

Just some useless information about laminar flow wings. During WWII the designers where all going for laminar flow for the performance increase in their designs, which was achieved. But what they didn't know at the time until it became evident due to losses in combat, a laminar flow wing had a major drawback, hence the highlight part in Petros quote. If a laminar flow wing took damage from bullets or flak, it would tranistion from laminar to turbulent and the pilot would have trouble flying the plane. Which i guess was a good tatic if the enemy they where trying to shoot down wouldn't stay still for them, they would spray for the outer part of the wing, the targets pilot would lose the ability to dodge and become an easier target to finish off.

 

That's not necessarily so. It depends a lot on the design of the pressure recovery region. If you try to eek out the absolute maximum extent of the laminar flow, and as a result have a very short, steep concave pressure recovery that depends on the boundary layer not losing much energy before the start of the pressure recovery, then the section will probably experience flow separation and excessive drag if the laminar flow is not obtained.

However, one can design for generous amounts of laminar flow while still having acceptable characteristics if the boundary layer is fully turbulent. A balanced laminar flow design may only have a drag increase of 2 - 5 counts over a design that was intended for turbulent flow when both sections experience a fully turbulent boundary layer. But the profile drag may be 30% - 40% less than the turbulent case if the design amount of laminar flow is obtained. This is a very good risk vs reward payoff.

To design such a section, first fix transition near the leading edge to ensure a turbulent boundary layer. This does two things: it causes the analysis to assume worst case conditions, and it avoids having the transition point jump around when iterating to achieve the desired pressure distribution. Next, design for a rooftop pressure distribution at the design lift coefficient. Blunt the leading edge pressure peak when operating at angles of attack above the design lift coefficient - this will result in rounding off the leading edge pressure distribution at the design lift. Use a convex transition between the rooftop and the pressure recovery to ensure a short laminar separation bubble when operating with laminar flow. Finally, analyze the performance with natural transition to see the potential performance improvement compared to the fully turbulent case.

This design approach allows one to balance the performance improvement due to laminar flow vs the risk of premature transition. It comes down to knowing the application and what the acceptable level of risk is.

The maintenance of laminar flow also imposes requirements on other aspects of the design and construction. For example, waviness in the surface may need to be kept to under, say, 0.1 mm, or graphics applied to the surface may need to be kept under a maximimum thickness. The surface may have to be washed more often or kept covered to avoid contaminants.

But laminar flow isn't necessarily an all-or-nothing thing, either. Where there's a discrete bit of roughness, the turbulent zone will spread out in a 7 degree wedge downstream from the roughness. But outside of that zone, the flow will still be laminar. So there will be a probability distribution spanning the range of potential performance from fully turbulent to maximum laminar flow that represents the expected net performance in practice.