LIFTROLL - what am I trying to achive

[atjurhs]

Hey guys,

I've decided to start designing my own thermal sailplane. Although I do intend to borrow from the success of other sailplane designers, I do want this to be a from the ground-up scratch design and build project. So I've started doing some homework, and downloaded some freeware and shareware, and have been playing around with the numbers a bit in these codes. Currently I am looking at various sets of inputs to a Excel spreadsheet called LIFTROLL.

There are three main output worksheets that result from the inputs you use. The first just shows the planform of your designed wing, the second shows a curve for values of Local CL, and the third output worksheet shows an elliptical lift distribution. The first and the last worksheets I pretty much understand what they say and know how to tweak the numbers to achive what I want. But the Local CL worksheet, I don't know what it is telling me other than the Coeffeicent of Lift. I guess what I'm asking is, what sort of a Local CL curve should I be working twoards?

In trying to answer my own question, I input the dimensions of 4 very successful thermal sailplanes and got back 4 very different Local CL curves. So this attempt to answer my own question only further confused me.

So now I figured I'd turn to the thermal gurus herein, and ask the question: what sort of a Local CL curve should I be working twoards in designing a thermal TD sailplane?

[redietz]

This is a wild guess, not an informed opinion, but it looks like the top plot may be less prone to tip stall than the bottom. I could have that completely backwards, though, since it looks like the tips on the bottom plot contribute very little toward lift anyway.

I also assume that the CL of the wing is the integral of local CL across the span, which would imply to me that the "deeper" plots have better total lift.

[OVSS Boss]

If you can find John Hazel in Michigan he can explain the whole deal, he wrote LiftRoll.

Marc

[cesarf3j2]

I think you want the local CL to diminish gradually towards the tip. In this sense the first and last diagrams should be better than the other two.

If the tip has higher local CL than the root I think the wing will be prone to tip stall.

[Bryan Quick]

A different perspective is:
- The coefficient of lift in a static environment (snapshot) over the span will show you which portions of the wing will contribute to the total and/or how it is distributed

- In order to understand what the behavior of the plane will be in any environment, you'd have to show the distribution of lift in that environment

e.g., to show the behavior of an aircraft close to the stall, you'd have to put it in that environment. High angle of attack, slow speed. To keep the tips from stalling, you'd want them to still be flying while the center of the wing is stalling.

It's impossible to know what the tip-stall behavior of those wings you graphed because we don't know under what conditions your calculations were made...if that was your question.

The questions you need to answer first in order to develop a design are in two categories, performance and structure. Once you have your performance requirements, you design the structure to meet those requirements. In one example, once you have decided on a wing planform, you can predict the loads on the wing generated in your predicted flight envelope. This allows you to estimate if the structure you can build using the materials and techniques at your disposal can withstand the loads (spanwise loading, torsional forces, etc.) you estimate will be generated in your desired flight envelope, fall within the weight requirments, and any other constraints you impose upon your design.

That's the short story for a wing.

Good luck-

Bryan

[schrederman]

As Dr. Schliecher said... If you achieve perfectly elliptical lift distribution, the entire wing will stall at the same time. Everything is a trade off.

Jack Womack

[Ollie]

Your design objectives are a meld of many things. For example: a range of attack angles, range of Cl efficiency, tip stall margin, wing loading, low induced drag, low profile drag, wing strength, a range of air speed, etc, etc.

You might better answers if you tell us what specific objectives for a context.

[nuevo]

What are you trying to achieve? For me, it's get that av/mx Cl as high as possible. That is try to get the most lift out of the entire wing.

It appears the vertical scales are not the same. Assuming they are the same, then ...

plot 3 is the "safest" as the center section will stall first with lots of margin of safety before the tips stall. Less chance of a tip stalling and nasty behavior.

plot 4 is very good. You can go quickly from #1 to #4 just by adding 1 degree of washout near the tips. I'd probably prefer #1 and wash the tips out to get something between 1 and 4.

As you can see from playing with liftroll, every adjustment affects every other adjustment. For example, changing a local chord affects the lift distribution of the adjacent chords.

OT: The relative vertical scales in the plots can vary your apparent results a lot. For example plot #1 and plot #2 (below) are identical, but the vertical scale is changed. Exaggerating the vertical scale can help you during the design to see where changes might need to be.

[nuevo]

The tradeoff for me is maximum lift of the entire wing to pull harder during launch, vs. less efficiency to give a larger tip-stall margin.

Oh, the wash out angles must be all relative to the zero-lift angles of each airfoil at the Re you expect to be flying at. For these, you need to start playing a program like Xfoil. Hint: study Mark Drela's designs closely.

[tonyestep]

First, note that in the Liftroll spreadsheet, the Y axis of the local CL graph is dynamically set. That is, the max and min values change depending on the data in the graph. This sometimes makes it hard to compare planforms; in particular, if you have a planform where the local CL is in just the right range, the scale can suddenly expand, making the planform look no good when in fact it may be fine.

Note that this happened in the graphs shown in post #1; the vertical scales are not the same in all graphs.

The fix is simply to go to Format Axis in the graph tool and make these values static. A good max/min is 0.80 max, 0.50 min. Then choose an angle of attack (AOA, cell B16), that will give an average CL between 0.65 and 0.7. This is in the vicinity of the CL that our planes achieve when thermalling.

The bottom graph in post #1 is the one least likely to tip stall, because the tips are farthest from the stall CL.

In general, what you're trying to achieve is a nice flat line across most of the span, with a rolloff at the tips. I compute the coefficient of variation of local CLs, and the RMS value of their differences from the mean, just to give myself a way to compare two planforms.

HTH

[atjurhs]

Hey Guys thanks so much for ALL your comments!

In order to further the discussion (and make a fair assessment of all four of the planforms. I did force them all to be on the same scale. Also, I should say (so that I'm not misleading) that plot#1 is the default wing planform that comes with LIFTROLL. It may or may not be an actual TD ship. And there was an input error on the #3 planform, so I corrected that.

- redietz: Initially, I think I had the same thoughts as you (before making the post) but I think what the others are saying is "schooling me".

- ceasarf3j: These are the kinds of thoughts I am after. That is to say, generally speaking what should I be trying to achieve in the wing's planform.

- Jack: One of these plots is one of your designs. Would you like to know which one?

- Bryan: thanks for your input, but I'm not sure how to make use of your "short story". It seems like it will lead me down a long (not wrong) road.

- nuevo: I'd also say that your "generally speaking" kinds of thoughts are exactly what I am after. BTW, I have input a few degrees of washout in #4. Also, one of these plots is Mark Drela's BD (as best as I could represent it), but it is not #4. Also #4 is very very close to the elliptical lift distribution curve shown in LIFTROLL.

- Tony HTH: I also appreciated your general obsrvations, directions, and thoughts. That's what I'm after, I think?

and I think a consenus is begining to emerge.....

So guys, based in all this, are there any new thoughts or ammendments to your original thoughts?

[tonyestep]

You're doing the analysis at an AOA that is somewhat higher than you'll be flying at. The CL is approaching 1.0, whereas in real life it'll average between 0.6 and 0.7. This may not make a lot of practical difference in figuring out your planform, but it's worth noting.

It will make a difference if you incorporate washout, as the difference in the effect on local CL will be affected.

Plots 2, 3 and 4 show a lot of CL reduction near the tips, more than is necessary to avoid tip stalls. Therefore, the wing that generated plot 1 will produce lower induced drag for a given CL. This translates into two benefits: a lower sink rate when circling, and less energy lost when pulling up elevator to start the zoom.

[atjurhs]

Thanks Tony! so if I understand correctly, what you're telling me is that I should be working twoards a LocalCL plot that looks like this "Ideal" one shown below, and at the same time keeping as close to an elliptical lift distribution as I can?

What I have noticed so far with the code is that for any particular planform as I tweak the amount of washout, I get closer and closer to a pure elliptical lift distribution, but the "fall-off" at the wing tips on the LocalCL plots get deeper.

edit: added AOA information

The AOA that I used for all of the wingplanforms was 10.3 That was the default value used in the original download, and I didn't know enough to change it.

Also, I'm not messing with any of the Control Surface Modeling stuff (all zeros). Should I be?

[atjurhs]

Well, maybe this only further confusses things, but in attempt to be truely fair and scientific, I plotted the Majestic wing planform and the Majestic wing planform with the Drela Mods in LIFTROLL.

If I understood correctly what you guys were saying then it appears that just the opposite of what I understood to be desired occurred when the Majestic wing planform got modified. In Drela's world of real Aero Engineer Gurus, the mods to that wing were made with much better analysis than done here, and pressumably gave better performance. So perhaps I have the wrong understanding of what you all have been trying to tell me??? Or perhaps none of this really makes a bit of difference, and I should stop analyzing so much and start drawing?

Having said that, the Majestic wing planform with the Drela Mods did show a large improvement over the baseline Majestic wing planform in the elliptical lift distribution plot (not displayed in this post). Maybe it is the elliptical lift distribution plot that is more iomportant to focus on???

[nuevo]

is your plot in post #13 hand drawn?

[atjurhs]

is your plot in post #13 hand drawn?

Yes

[markdrela]

Well, maybe this only further confusses things, but in attempt to be truely fair and scientific, I plotted the Majestic wing planform and the Majestic wing planform with the Drela Mods in LIFTROLL.
If I understood correctly what you guys were saying then it appears that just the opposite of what I understood to be desired occurred when the Majestic wing planform got modified.
I look at the spanwise distributions at two conditions:
1) High CL (low speed), while performing a sharp trimmed turn, typically at 30 or 40 degrees bank angle. On a poly glider, the trimmed turn requires holding some slip angle to the inside of the turn. On an aileron glider, the trimmed turn requires holding some outside aileron with zero sideslip angle. The wing design objective here is to not have a local Cl peak on the inside wing.
2) .Low CL (high speed), in level flight. The objective here is to keep the tips, which will have the lowest local Cl, inside the airfoil's drag bucket, with some margin.

One of the major Majestic wing mods was to eliminate the Majestic's tip stall tendency.
You can find the "before and after" Cl plots from AVL here:
http://groups.yahoo.com/group/Allegro-Lite/files/Vortex_Lattice/

I generally do not pay too much attention to the Cl distribution at high CL in level flight. Good behavior in a slow tight turn and high-speed penetration are much more important design points. The slow, level performance then is what it is, and it typically won't be compromised noticably anyway.

[Tony D.]

I use XFLR5 as a design tool, IMO it is far superior to LiftRoll as a design tool for wings and gliders. (Not putting LiftRoll down here just stating my opinion). Multiple airfoils can be used at different locations across the span and Polars generated for the wing or the whole glider.

I've attached some of the design output images from XFLR5 for my latest project 'Daimyo', this is what I shoot for in the handling of my thermal duration glider designs and for me these parameters work extremely well.

You can download the program here - http://sourceforge.net/projects/xflr5/



T.D.

[nuevo]

"Yes" thanks. The bumps didn't look quite right.

One thing to be aware of. Everything is a compromise. If you go for the "perfect" elliptical distribution, you will end up with very small tip chords. Those small tip chords mean lower Re and often narrower drag buckets. (i.e. might stall before the root).

[atjurhs]

Thanks Mark for the link, I'll see what I can learn from the pics.

Thanks Tony for the link, I'm betting after tinkering with XFLR5 I'll have more questions......

Thanks nuevo for the reality check.

In the end :rolleyes: all of this is probably WAY above my flying ability anyways as I am just an LSF2 pilot. So I'll probably utilize an already existing planform from a "well designed TD bird" and go with that as my starting point. But it's always more fun when you make your own mods :D get a little technical, and learn a few lessons....

[tonyestep]

Of all the posts so far, the topic of this thread, "what am I trying to achieve," is best addressed by Mark Drela's response, to the effect that the most important things you can learn about your wing are its characteristics in straight-line flight at low CL, and circling flight at high CL. However, if you can use the Liftroll spreadsheet to analyze model circling flight I haven't figured out how to do it.
XFLR5 has full wing modeling which does not use the AVL code but does similar stuff. If you want the full unvarnished AVL, you can download it free at:
http://web.mit.edu/drela/Public/web/avl/

[atjurhs]

AVL looks like pretty cool stuff, and actually with my line of work, I understand computational lattice-modeling fairly well, but that still would not necessarily indicate to me what to look for in the code's output as being desireable or not. I think I would need to take a class or two in aero-engineering to get anything really useful out of AVL (or any of the other codes previously mentioned for that matter).

That's kind of why I choose to "backwards engineer" this project and start analyzing the question by useing "inputs" of already existing and well designed TD ships, and seeing what the codes' outputs are. I expected to find outputs that were very much the same (at least for the well designed TD ships), and then I would know what I wanted to achieve - "a typical backwards engineering process". But what I found (at least for the TD birds I used in LIFTROLL) was that they had very different output curves (not what I was hoping for). And that's when I decided to post my question here, "what am I trying to achieve" still in the hopes of backwards engineering a wing planform. See how screwed-up I am :D So again, I'll probably utilize an already existing planform from a "well designed TD bird" and go with that as my starting point, and then make a few mods just to be happy (or is that just to be different? :D ).

[atjurhs]

After tinkering around somemore.....

I decided it would be best for comparitive purposes to put all the wings on the same set of plots. And then in order to get better resolution and differentiation between the curves, I set the axis to only display the right half of the wing (the left half is just a mirrior image anyways). The black dashed line in the Lift Distribution plots is the perfect ellipse.

So, now I guess I'll re-ask my question, but in a slightly different manner:

If they all had the same WS and fuse and tail feathers, which of these wings/planes would you like to fly and why?

Brown
Purple
Green
Yellow
Orange
Blue

[schrederman]

- Jack: One of these plots is one of your designs. Would you like to know which one?


OK... I'll bite... is it the third one? Which model?

Jack

[tonyestep]

Heck, I dunno. I'll say the purple (at least I think it's purple - on my screen it's sort of a reddish-brown). Why? At high AOA, it has a pretty even CL across the span but with plenty of rolloff at the tip so it probably won't tipstall. Of course, in a banked turn this distribution might be different, but anyway, given what we have to look at, I'll try that one.

[nuevo]

I like Brown best. I'm guessing it has the highest av/mx Cl

[atjurhs]

Nuevo, it doesn't surprise me too much that you would choose the Brown curve - you fly one (or at least it's cousin). The Brown curve is a BubbleDancer.

Jack, the prurple curve is your Bantam design which I believe to be a scaled down 100inch Yardbird? And I've had great success flying it! So Tony I can attest that it would be a good choice!

BUT actually, I was designing twoards the Orange one. It's a distant cousin to the Fusion (I made a few mods to the actual Fusion's planform). The Green curve is the actual Fusion.

So, I guess the next question would be, what would be the flight characteristics of the Orange wing??? Hopefully not tipstall tendencies???

[chuckand]

Jack, the prurple curve is your Bantam design which I believe to be a scaled down 100inch Yardbird? And I've had great success flying it! So Tony I can attest that it would be a good choice!


I suspect Jack will jump in when he gets to it... The airfoils are different between the YardBird and Bantam, though I do not know firsthand the significance or impact. Suffice to say the purple curve is the Bantam. Thanks for the inclusion. It's in very good company.

Chuck (in Aaaala Baaama)

[atjurhs]

Chuck, the LIFTROLL program does not take into account the choice of airfoil, just the wing's planform. So as long as the Bantam is a scaled down version of the Yardbird, LIFTROLL will yield the same purple line in either case. But thanks for the clarification between the two farm birds :D

[Hutch]

I use XFLR5 as a design tool, IMO it is far superior to LiftRoll as a design tool for wings and gliders. (Not putting LiftRoll down here just stating my opinion). Multiple airfoils can be used at different locations across the span and Polars generated for the wing or the whole glider.

I've attached some of the design output images from XFLR5 for my latest project 'Daimyo', this is what I shoot for in the handling of my thermal duration glider designs and for me these parameters work extremely well.

You can download the program here - http://sourceforge.net/projects/xflr5/



T.D.

I need to learn how to use that software. You really have to know what you are doing to get it to work. I haven't figured out how to get it to spit out the info I am looking for yet...suppose I should figure out what I am looking for first though. :)

-hutch

[ChuckA]

It takes a real expert to understand all the capabilities and limitations of modern computer programs for aerodynamic design. Back in the early 1970's, I was the indian among all the chief's in the CFC research section where I worked. It was my job to conduct wind tunnel tests to verify all the fancy programs the chiefs wrote. The wind tunnel data seldom agreed with their predictions and their first comment was "What's wrong with your wind tunnel data". I realize CFD has come a long way in the last 30 years but interpretation of the results is still an art. I have used a few programs in designing my models but still rely on flight testing and modifications to get the desired results.

[schrederman]

I read a lot. My biggest problem with engineering is that the imagination doesn't come into play quite enough. Interesting that the Yardbird and Bantam are mentioned here. Yes, the Bantam is sort of a scaled down Yardbird. That planform seems to work well. I'm going to build another Yardbird wing soon with my own secret airfoil. I guess I'll keep designing to the purple curve... after all, my wife dared me to make my bantam purple...

Jack

[bobby legue]

Go, Jack, Go!!!!

[schrederman]

Hi Bob!!!

[nuevo]

It takes a real expert to understand all the capabilities and limitations of modern computer programs for aerodynamic design.

True, very true. Anyone can generate plots. Understanding them is another matter.

[infopimp]

Great thread. Subscribed.

[Tony D.]

No computer program can give you a perfectly flying sailplane but they can certainly give you an excellent starting point for further development of your design. The tools available today combined with information freely available on the Internet let people who might not have tried rolling their own to have a good shot at designing a well behaved and decently performaning sailplane at low cost which will help keep them in the hobby and bring others in with them.

When you combine CFD with some experience building, designing, flying and thinking about gliders you have a very potent set of tools at your disposal which is part of the art you speak of Chuck.

In other words there is a learning curve and experience is required but it is not beyond just about anyone who is interested in taking it on.


T.D.

[atjurhs]

No computer program can give you a perfectly flying sailplane but they can certainly give you an excellent starting point for further development of your design. The tools available today combined with information freely available on the Internet let people who might not have tried rolling their own to have a good shot at designing a well behaved and decently performaning sailplane at low cost which will help keep them in the hobby and bring others in with them.

When you combine CFD with some experience building, designing, flying and thinking about gliders you have a very potent set of tools at your disposal which is part of the art you speak of Chuck.

In other words there is a learning curve and experience is required but it is not beyond just about anyone who is interested in taking it on.


T.D.

- Tony D - man you've not only hit the nail on the head but you've driven it all the way home. What you've written is EXACTLY what I'm doing/experiencing/learning.

- Jack - Your Bantam design is an EXCELLENT flyer! As an LSF2 pilot, I am now able to compete against the masters and sometimes win and sometimes come in 3rd. It's made me a much better pilot (and so has been the help of all those masters that I'm competing against). THANKS to all you all! So if the purple curve is not considered the "optimal curve" predicted by LIFTROLL, well then we just have another case of what Chuck has stated, and as Mark Drela pointed out LIFTROLL does not analyze the wing at a 30degree bank. So, I'd say yea man, keep desigining 'em to the purple curve.

Can anyone comment as to what they might think the flight characteristics (good or bad) of a wing built to the Orange curve would be?

[Fly2High]

I have been following this thread in the hopes to gain some insight into how designs are developed.

If a guy wanted to learn this, what courses would one take (at undergrad or grad level - if grad is needed) to be able to understand design theory better?

If specific books can be defined, that would be great too.

Thanks

Frank

OK, this sounds too simplistic for what I really am asking. I want to know what courses in engineering, math etc. I would need to take to be able to get a better understanding. I meant in no way that the ability could possibly be contained in a book or couple of courses. Believe me I know. I have bachelor's degrees in Biochemistry and computer science - done the hard way - 8 years of schooling as well as extensive grad work in Human cytogenetics.

I only do not know how to focus an aerospace education to be able to do these kinds of things

Thanks

[ciurpita]

if your interested in considering banked flight, along with the drag dependent on specific airfoils, you mey be interested in
http://www.amadistrictii.org/cjrcc/wing2/wing.html

[Tony D.]

Greg, is it possible to add new polars to WA?

T.D.

[ciurpita]

yes. the airfoil polars are in ascii format, i think the one mike garton uses. the catalog file indentifies the airfoil by name and filename

[jbeanelliott]

You might be interested in investigating Pennsylvania State University's class on sailplane design (atleast to get some recommendations on the background prerequisites). See their website at

http://www.psu.edu/dept/aerospace/sailplane/index.html

Except:
Since 1989, the Department of Aerospace Engineering of The Pennsylvania State University has offered a special undergraduate project course that has a strong emphasis on "hands on" design and fabrication. Specifically, a group of approximately twenty-five students, freshmen through seniors, is involved in the design and construction of a variety of different sailplanes. Students can and are expected to enroll in this course for every semester during their undergraduate study. During that time, the students experience the cooperative, multi-disciplinary team environment that is required for solving the problems related to the design of an aerospace vehicle. The basic course structure consists primarily of three components.

The first, lecture, provides the student with the necessary theoretical background of modern sailplanes and their design requirements.

The second component is concerned with design groups of four to six students, in which the students design and analyze sailplanes, such as their performance, structure, stability and control, and so forth.

The third component is the fabrication of parts that have been designed and analyzed theoretically, such as the current project of an Easy-to-Build sailplane made out of modern composite materials.

I have been following this thread in the hopes to gain some insight into how designs are developed.

If a guy wanted to learn this, what courses would one take (at undergrad or grad level - if grad is needed) to be able to understand design theory better?

If specific books can be defined, that would be great too.

Thanks

Frank

OK, this sounds too simplistic for what I really am asking. I want to know what courses in engineering, math etc. I would need to take to be able to get a better understanding. I meant in no way that the ability could possibly be contained in a book or couple of courses. Believe me I know. I have bachelor's degrees in Biochemistry and computer science - done the hard way - 8 years of schooling as well as extensive grad work in Human cytogenetics.

I only do not know how to focus an aerospace education to be able to do these kinds of things

Thanks