Aerofoil Selection

[Martin Harris - Moderator]

I've just been reading Tim Hooper's article in the current mag and it's prompted me to air something I've wondered about for a long time.

Tim selected an Eppler aerofoil, modified its trailing edge thickness and flattened the bottom to make it easier to build. No problem with that - it obviously worked well but how much would the modifications affect the very carefully designed output from Prof. Dr. Eppler's calculations?

My own rudimentary designs of a couple of 12th scale WW2 combat models used advanced aerodynamic research (well, I looked up aerofoils recommended for pylon racers based on rather speculatively calculated Reynold numbers and then stretched them chordwise on a drawing program to 10% thickness to comply with the rules and allow them to fly as fast as possible). Perhaps through luck rather than judgement, the models worked extremely well both in combat and when being flown for fun.

So the question for the proper designers and aerodynamicists on the forum is just how important is aerofoil selection at model sizes? For higher performance sections, is laminar flow as critical with our relatively thick air or even achievable by the average builder? Is it really a science or would we do just as well to randomly select a nicely curved symmetrical section for an aerobatic model, a semi-symmetrical one for sports models or a flat bottomed Clark Y-ish one for a trainer?

I can readily accept that for full sized aircraft and competition model gliders etc. there will be efficiency considerations that will favour particular aerofoils but really, do we need to spend very much time and thought on selecting an aerofoil for the average model? Have any of our prolific designers ever had to redesign a model's wing section to correct unwanted behaviour perhaps?

 

Edited By Martin Harris on 13/04/2013 22:47:35

[Pat (rActive) Harbord]

You make interesting points Martin. I would be very interested to know what clever people think about the subject.

[Biggles' Elder Brother - Moderator]

Well the Eppler sections were designed to be optimised on full size aircraft where the Reynolds numbers are very much higher than we see on a model.

To give an idea how much higher - full size aircraft operate at Reynolds numbers typically of a 1 million plus. Our models operate at around 40,000 - 20 times less.

Some Eppler aerofoil sections have been designed and tested down to Reynolds numbers of 30k - but they are very much the exception.

Given this you would have to conclude that the relationship between the viscous and inertial forces are very different indeed for the same aerofoil profile operating at a model scale and on the full size. One major difference would be that "plate flow" (the type of flow we have over a wing) is still well in the laminar region at the Reynolds numbers we are dealing with - so at no point over our wings does the flow transition to turbulent. In the full size the situation is of course very different with transition to turbulence happening quite earlier over the wing section. In a full size aircraft the flow is turbulent over maybe 70% plus of the profile length.

Another illustration of how different model aerofoils are full size is the comparative popularity of the Clark-Y section in modelling. For a sports model the Clark-Y is a tolerably good design - and of course its easy to make - which definitely adds to its popularity. But in full size practice it is a section that is hardly ever employed - very few full size aircraft have used a Clark-Y, the Piper Cub being the only notable example I am aware of. Why? Because at the sort of Reynolds numbers encountered in full size aircraft it is very inefficient lifting device. But in a model its actually quite an efficient aerofoil - in fact its limitation for us is that its almost too efficient and tends to produce "floaty" flight characteristics.

So, in answer Martin I'd say its almost meaningless to transfer highly optimised full-size profiles to models.

The other (positive!) way of looking at that statement of course is that if you do transfer a highly optimised full size profile you can feel free to "doctor" it! You are so far from its original design operating point it doesn't really matter! I'm not saying don't be influenced by such aerofoils - on the contary by all means copy them - just don't be too "precious" about it and if its convenient to alter it slightly (as Tim did) don't hesitate to do so as long of course as you don't do something really extreme and don't blame Eppler if it doesn't quite come out as you expected!

BEB

[Tim Hooper]

Thank You BEB!

I've often pondered the same question - especially given that many successful models simply use a flat depron plate, or a thin balsa sheet wing, and yet are great fliers.

tim

[TheFlyingCrust]

Some Eppler aerofoils are 'Low Reynolds Numbers' but, bearing in mind BEB's comments, I don't know if Low is low enough. Many moons ago I made a glider with E193 profile. Chord was, I think 7 inches. The model flew well but I have no idea how it would have compared to any other 'model' profile.

Have a look here for a pretty comprehensive list of profiles. Some are intended for model use - but I haven't tried them. Read this. It may be of interest.

Ian

Edited By Rentman on 14/04/2013 23:59:42

[Martin Harris - Moderator]

Thanks for the explanation BEB. Interesting that flow remains laminar on our models. At what sort of chord sizes / airspeeds does this situation change?

How much significance to flight characteristics is there in the flow remaining laminar - does it mean we get less sharply defined stalls than in typical full sized aircraft? I get the chance to do a lot of test flying and the majority of models do seem to have rather benign stalls unless prone to wing dropping. I'd say a large majority waffle around with full elevator although perhaps this is often due to conservative C of Gs?

The ultimate aerofoil abuse that I've perpetrated is a foamy profile Bearcat that I knocked up as an experiment

- 1/2 inch thick slab wings with square corners and it flies without any apparent vices - just don't ask it to go anywhere in a hurry in a strong wind!

[PatMc]

If a wing has an open structure the airfoil section isn't what it says on the tin.
That's probably the main reason moulded glider wings are so efficient.

[Martin Harris - Moderator]

That's certainly the case in full size gliders where rain drops can have a very significant effect on the laminar flow on the very accurately constructed and finished wings. I wonder how much difference it makes to a low speed glider model though?

It may well be that such models as F5b, pylon racers, dynamic soarers and large models with reynolds numbers in the high hundred thousands will be more critical - certainly they seem to approach the million figure quoted by BEB...

[Tom Wright 2]

Posted by PatMc on 15/04/2013 00:39:35:

If a wing has an open structure the airfoil section isn't what it says on the tin.
That's probably the main reason moulded glider wings are so efficient.

I wonder if the section thickness was increased by the mean average value of the sag , would the section then perform as said in the tin ?

Tom.

[Tom Wright 2]

Posted by Tim Hooper on 14/04/2013 20:57:38:

Thank You BEB!

I've often pondered the same question - especially given that many successful models simply use a flat depron plate, or a thin balsa sheet wing, and yet are great fliers.

tim

Hi Tim.

Symmetrical sections such as a flat plate , are only great at low wing loadings ,add weight ,and as the AOA is increased to get the extra lift required the drag goes ballistic , resulting in a sink rate comparable with a lead balloon. Quite sure you know this but worth a mention for others who may be looking in .

BTW the Blue Monster has 20 sq ft of flat wing and has an AUW of 12 lbs ,Just rounding the l/e and putting a modest chamfer on the t/e improved the LD very noticeably .

Tom.

[Peter Miller]

I use a variety of similar sections. Cannot tell the difference. If you use a very thick section with a blunt leading edge it will fly slower (Control line stunter) If you have a sharp leading edeg the stall will be sudden.

I used to use Clark Y set at 0 degrees incidence. Worked just like a symmetrical section.

Best example that I have quoted recently elsewhere.

I once read the write up on a plan in an American magazine. The authoir designed a very successful sailplane which won competitions. The opposition was using the fancy glider sections. He was using Clark Y. Nuff said!

[Biggles' Elder Brother - Moderator]

Posted by Martin Harris on 15/04/2013 00:36:23:

Thanks for the explanation BEB. Interesting that flow remains laminar on our models. At what sort of chord sizes / airspeeds does this situation change?

 

Reynolds number is given as:

Re = (density x free_stream_velocity x characteristic_length)/dynamic_viscosity

For us free_stream_velocity is airspeed, V

The characteristic_length is the wing chord, C

For air the density is about 1.2Kg/m3 and dynamic viscosity is about 1.8x10-5 Pa-s

The nearest Reynolds analogy we have for the flow is so-called "plate flow" and that tells us the transition to turbulence will occur when Re is greater than approximately 500,000.

Putting the numbers in then tells us for the onset of turbulence - right at the trailing edge of the wing.

V x C > 7.5

Bear in mind this is approximate and at best would only just trigger turbulence right at the end of the wing.

Looking at some typical model values then;

1. For a model travelling at 20mph (10m/s) airspeed the transition to turbulence would be at: 10 x C >7.5 so C> 1.3m - that's a very big wing chord for a model!
2. For a model travelling at 40mph - the wing chord would have to be at least 0.65m

And so on.

Now case two shows why big models flying at reasonable speed look more realistic in the air - they are starting to approach the flow conditions in a full size aircraft. Its worth looking at the conditions for a modest full size single engined aircraft. Take a Cessna 172 for example:

V= 63m/s - based on a cruise speed of 226 km/h

so V x C > 7.5

i.e 63 x C > 7.5

so C > 0.12m

The chord of the Cessna is about 1.5m - so the flow goes turbulent at less than 10% of the wing chord! In other words 90% of the wing chord is operating in turbulence. For our models, as shown above, we are struggling to get turbulence at all even at the very trailing edge!

BEB

Edited By Biggles' Elder Brother - Moderator on 15/04/2013 13:17:58

[Martin Harris - Moderator]

Posted by Biggles' Elder Brother - Moderator on 15/04/2013 13:15:45:

Now case two shows why big models flying at reasonable speed look more realistic in the air - they are starting to approach the flow conditions in a full size aircraft. Its worth looking at the conditions for a modest full size single engined aircraft.

I think I follow that reply OK but in what way does the onset of turbulent flow over some of the wing affect the realism of a model? Is it purely in terms of scale speed or something less definable such as the "sit" in the air?

I know that it was said to me that from a distance, my 1/3 scale Miles Atwood racer looked extremely realistic in flight. An experienced observer said that he thought it was a full sized aircraft when he was walking his dog near our field. That probably flew at a realistic 55 - 65 mph and had a chord of around .33 metres which puts it out of the fully laminar flow case - although only by a small amount...

[John Muir]

I looked this stuff up on line recently and it turns out that recent years have seen something of an upsurge in research into low reynolds number aerofoils because of the market in drones and wind turbines. Nobody seems to have bothered before, although I imagine propeller manufacturers must have been at least a bit interested. I think a lot of propeller blades over the years used the Clark Y section by the way. The research needs a 'clean' wind tunnel with minimal turbulence and that seems to be a stumbling block.

I read some research papers that had been published. In the end I learned that the Clark Y is still pretty good and that turbulator strips might help. I did read a lot more but completely failed to understand the rest. Then I fell asleep and droolled on my keyboard. Not the most exciting stuff I've ever read.

John.

[Biggles' Elder Brother - Moderator]

Posted by Martin Harris on 15/04/2013 14:49:43:

I think I follow that reply OK but in what way does the onset of turbulent flow over some of the wing affect the realism of a model? Is it purely in terms of scale speed or something less definable such as the "sit" in the air?

Reynolds number defines somthing called "dynamic similitude" (that's easy for you to say!). Basically this means that when we are considering a model if we want it to move the same way as the full size (ie react to lift, wind etc etc, the same way) then we need the aerodynamic forces to act on it in the same scale.

The problem with this is that aerodynamic forces are not a single entity - there are actually several different "flavours" - for the model to behave like the full size we need the mix of this cocktail of forces to be the same.

Two of these forces are the viscous forces - sort of like the "stickiness" of the fluid - and the inertial forces - related to the mass (inertia) of the fluid as it flows. In very slow flows (or with very "thick" fluids) the viscous forces dominate. In fast flows the inertial forces dominate. These two forces act on the object in very different ways and make it move in a different way.

Reynolds number is proportional to the ratio between the inertial forces and the viscous forces. So when Re is small the viscous forces dominate - the "stickiness" of the fliud makes it flow in nice neat streams and when have laminar flow. When Re is large the inertial forces dominate, they break up the flow streams and the flow becomes turbulent.

Now one of the conditions for a model (like ours or in wind tunnel) to behave and move like the full size is that the contribution of the forces on it from viscous and inertial effects must be the same as the full size - ie Re must be the same. Basically this is just saying that the balance between viscous effects and inertial effects must be the same. The magnitude of the combined forces (from all sources) has to be "in scale" with the models scale weight - but the relative balance between inertial and viscous forces must be the same.

So, in a wind tunnel test for example we might have a 1/6th model, but the flow speed we test it at might be much higher than 1/6 in order to bring the model's Re up to something more like the full size. Alternatively we might test the model in a different fluid - for example water. This has a much higher viscosity than air (about 5 times higher at room temperature) and so we can achieve similar Re values at lower speeds for the same scale factor.

In pratice to get true dynamic similitude is almost impossible as you need to match more than just the Re value and in a wind tunnel test you get as close as you can and then use all sorts of correction factors on the numerical results for the forces.

There is another factor to considered. Generally we want the flow to go turbulent as this results in considerably less drag. Our models operate in a high drag regime! On full size aircraft if there are problems in getting the flow to go turbulent early enough across the wing a device called "turbulator" can be used. This is basically a small (very small) step or fence. Its purpose is to unstick the laminar boundary layer and trigger the development of fully turbulant flow.

Now, here I have an idea! I've often wondered if this would help pylon racers? For a model a turbulator could be as small as the thickness of a layer of Solartrim or similar. So would strip of solartrim (or maybe a couple of strips one on top of the other) trigger turbulant flow across a pylon racer's wing - and thus lower drag? Any pylon racers out there willing to give it a go?

BEB

Edited By Biggles' Elder Brother - Moderator on 15/04/2013 22:28:04

[Martin Harris - Moderator]

I was following that until the bombshell at the end (shades of Top Gear?) where (if I understand correctly) you revealed that laminar flow is a bad thing.

Confusing, as I always thought laminar flow was the holy grail e.g. the laminar flow wing on the Mustang being key to its excellent high speed performance and, in particular, the fine surface finish required on gliders in order to keep flow laminar for as long as possible over an aerofoil designed to promote laminar flow. In fact I thought that turbulators actually moved the separation point rearwards by some black aerodynamic magic involving factors way beyond my understanding...

I have seen theories expounded that very fine and uniform roughness on a surface can reduce drag but this doesn't seem to have been adopted by any aeronautical application I'm aware of.

Of course, I suspect there's a massive difference between the Re of a P40 at altitude and of a competition glider operating mostly under 100 knots in the lower regions of the atmosphere but both seem to strive for the maximum laminar flow - if it's possible to explain in the relatively simple terms in which you excel, what have I missed BEB?

 

Edited By Martin Harris on 15/04/2013 23:23:38

[PatMc]

BEB, I'm sure you must be aware that there's nothing new about using turbulators on models. They have been used on some F/F comp models since the year dot, though AFAIK they have fallen out of favour in recent years.

Since models & full size aircraft operate in the same size air wouldn't the turbulators also need to be similar in size ?

[Biggles' Elder Brother - Moderator]

Its all to do with the behaviour of the boundary layer. Yes, in a perfect world you'd love to have completely laminar flow. But for ordinary full size wings this isn't going to happen, there is going to be a transition to turbulent flow somewhere across the wing. While you're right the turbulence itself is a problem (adding drag) the real issue is boundary layer separation at or about the transition point. This is the real drag issue for most wings - particularly those operating at lower Re values.

The reason is when the boundary layer unsticks from the wing it does two things. First out in the bulk flow around the wing it effectively changes the aerofoil profile - acting like an artificial surface. Second, and this is the really big problem, areas of stagnation (zero velocity) form in the low pressure recirculating air behind the separation. These create a lot of drag.

A turbulent boundary has more energy in it, this means it can adhere to the surface for longer as the pressure drops across the chord. So a turbulent boundary delays - or possibly even avoids - boundary layer separation. Which is a good thing.

It becomes a balancing act now. A turbulent boundary layer has more drag than a laminar boundary layer, but it sticks to the wing for longer thus delaying or avoiding, separation. Unless you are going very fast the gain in reduced drag by keeping the boundary layer in contact, outweighs the loss in slightly higher surface drag by the boundary layer being turbulent. So, on balance, on lower Re values, its better to settle for turbulent boundary layer and get the benefit of delayed, or even no, boundary layer separation.

In full size aircraft this is often employed in state-of-the-art gliders - where Re is relatively low. That's why I wonder if it might be transferable to model pylon racers - as although they go pretty fast are still probably mainly in the laminar regime but with possibly a late separation point leading to a smallish turbulent zone. I don't know if it would work - but it would be interesting to experiment - I need a tame pylon racer who can fly consistent times to see if there is a difference.

The early so-called laminar wings were an attempt to stop transition and possible separation by moving the maximum thickness point backwards - thus lessening the area of lower pressure in the back part of the profile. Once you try to do this you then want very smooth wing surfaces because the whole point is to not trip up the boundary layer and cause it to separate. More modern attempts at laminar wings do this a different (and very clever) way by forcing high speed oscillations in the air flow to sort of "lock it down".

If you are not "going for laminar" then there is actually some benefit in having a slightly rough texture to the wing surface as it promotes turbulence in the boundary layer. But there are other problems with that - especially at lower Re values, its OK for a 747 but not so good for glider - so hence use of the turbulator on an otherwise very smooth wing surface.

BEB

 

Edited By Biggles' Elder Brother - Moderator on 16/04/2013 00:27:13

[Biggles' Elder Brother - Moderator]

Posted by PatMc on 15/04/2013 23:28:36:

BEB, I'm sure you must be aware that there's nothing new about using turbulators on models. They have been used on some F/F comp models since the year dot, though AFAIK they have fallen out of favour in recent years.

Since models & full size aircraft operate in the same size air wouldn't the turbulators also need to be similar in size ?

Actually Pat I didn't know that - interesting. Do you know why they have fallen from favour?

BEB

[Martin Harris - Moderator]

Brilliant BEB - that seems a very clear explanation - even to my limited grey matter!

[John Muir]

Thanks BEB, very clear and easy to read.

I read about turbulators way back too, but I think the theory behind them has changed. The idea then was to turbulate the airflow earlier than would otherwise be the case, promoting the formation of a separation bubble nearer the front of the aerofoil thereby allowing the boundary layer to reattach further back. In other words, it was believed the turbulator promoted separation. Now it is understood that a turbulator actually keeps the flow attached, preventing separation until much further back on the surface. The effect though would be the same, to reduce drag, especially at low airspeeds and increase the stall angle.

I think turbulators on models were always very much experimental and usually took the form of a thin strip of tape spanwise along the wing. Don't know if anybody ever proved they worked or why they fell out of use.

By the way BEB, is it true that our models fall into a sort of 'transitional' area between truly low and high Reynolds number aerodynamics, which makes them really difficult to figure out?