HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old May 26, 2010, 01:32 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
DOH! My bad. I'm getting my thread authors all confused. Sorry Montag!

If you're going to remove the windscreen for photo flights then you may want to look at making the rear of the "cockpit" into a rounded shape so it's more "airfoil like". Then just make the windscreen longer so it attaches back where the fuselage is back to full width. Some paint or other opaque portion of the screen would cover the fact that it's got a rounded area under it for normal duties. Given the size of the fuselage width needed to mount a hand operated camera as you're showing the reduction is drag would be well worth the mod.

I would also look at the idea of making the fuselage top view a more airfoil like shape. The way it is now I would not be surprised to see big separation bubble zones along the pinched area leading into the prop. And that isn't going to help with prop efficiency.

This last aspect is why I far prefer the idea of a twin tractor motor arrangement over a single centerline pusher. Well, that and I think it would be neat to have a twin motor setup.... Along this line I'm torn between a C119 style versus a more normal fuselage style like the DH Caribou.
BMatthews is online now Find More Posts by BMatthews
Last edited by BMatthews; May 26, 2010 at 01:42 PM.
Reply With Quote
Sign up now
to remove ads between posts
Old May 26, 2010, 02:12 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Quote:
Originally Posted by BMatthews View Post
DOH! My bad. I'm getting my thread authors all confused. Sorry Montag!

If you're going to remove the windscreen for photo flights then you may want to look at making the rear of the "cockpit" into a rounded shape so it's more "airfoil like". Then just make the windscreen longer so it attaches back where the fuselage is back to full width. Some paint or other opaque portion of the screen would cover the fact that it's got a rounded area under it for normal duties. Given the size of the fuselage width needed to mount a hand operated camera as you're showing the reduction is drag would be well worth the mod.

I would also look at the idea of making the fuselage top view a more airfoil like shape. The way it is now I would not be surprised to see big separation bubble zones along the pinched area leading into the prop. And that isn't going to help with prop efficiency.

This last aspect is why I far prefer the idea of a twin tractor motor arrangement over a single centerline pusher. Well, that and I think it would be neat to have a twin motor setup.... Along this line I'm torn between a C119 style versus a more normal fuselage style like the DH Caribou.
Thanks, Bruce. Those are some good ideas. I haven't tried to streamline the fuselage that much, but you actually just gave me a good idea - I can treat the fuselage top view and/or side view as an airfoil shape and use the optimizer to minimize the drag at 0 angle of attack, given the dimensional constraints I need. What fun! I'll then have two important parts of the entire plane (fuselage and airfoil) "optimized." In any case I'll give it a try and post the results.

Dan
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old May 26, 2010, 04:49 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
Ooooo..... that IS a good idea. And don't be afraid to look at tubulator trip strips at the widest portion of the fuselage to help the air "turn the corner" with less risk of separation.

For the top view I'd optimize for from 0 to maybe 2 degrees just to hedge the bet. For the side you will want to figure on anywhere from 0 to maybe +6 to give it a good range for the angle of attack increases from fast cruise to slowed for landing.

Now I'm keen to see what you end up with.
BMatthews is online now Find More Posts by BMatthews
Reply With Quote
Old May 26, 2010, 05:21 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Quote:
Originally Posted by BMatthews View Post
Ooooo..... that IS a good idea. And don't be afraid to look at tubulator trip strips at the widest portion of the fuselage to help the air "turn the corner" with less risk of separation.

For the top view I'd optimize for from 0 to maybe 2 degrees just to hedge the bet. For the side you will want to figure on anywhere from 0 to maybe +6 to give it a good range for the angle of attack increases from fast cruise to slowed for landing.

Now I'm keen to see what you end up with.
I'll probably keep it simple to start with and just do 0 AoA. I'll look at the drag polar from there and see if I actually should try re-optimizing for higher angles. For a symmetrical airfoil there should be a pretty flat drag slope at low angles of attack anyway. I'll have to modify my code a little bit to run it for symmetrical airfoils with a blunt TE. Hopefully I'll have time in the next couple days.
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old May 26, 2010, 09:14 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Preliminary work:

I was actually able to evaluate the current fuselage profile at cruising Re. Since the fuselage "chord" is much longer than the wing, the Reynolds number is about twice as high. I tested for cruise at about 460k. At this high Re, there was no BL separation until the trailing edge and a thin BL. The drag at 0 degrees was 0.0215. I'm not sure how valid this actually is.

Next step is to set up the optimizer with appropriate size constraints and see how much better it gets, then decide how to implement on a real airplane. XFLR5 might be of help there.
Montag DP is online now Find More Posts by Montag DP
Last edited by Montag DP; May 27, 2010 at 08:52 AM.
Reply With Quote
Old May 26, 2010, 09:17 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
And given that the fuselage pod qualifies for a thicker airfoil that should make it even more likely to stay within the low drag bucket around zero.
BMatthews is online now Find More Posts by BMatthews
Reply With Quote
Old May 26, 2010, 09:35 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Quote:
Originally Posted by BMatthews View Post
And given that the fuselage pod qualifies for a thicker airfoil that should make it even more likely to stay within the low drag bucket around zero.
True. One thing that I wonder about is how to do the "side view" shape. I do need to have the propeller high enough to ensure it won't scrape on the ground, and I don't want to have really long landing gear, so a symmetrical shape is pretty much out of the question. I think I may either leave it the same shape as it is or use the top half of the optimized airfoil and then modify the bottom half to suit my needs.
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old May 27, 2010, 01:21 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
I started to type another long explanation of how I'd likely do this myself but then I figured that it would be easier to just make a drawing up.

So here it is. A short vertical "wing" with two big wingtips. Drag is not so much the issue as providing a good flow of the least disturbed air to the pusher prop. And that's why the need for an idealized shape to provide this.

As I recall reading about the Cessna "pushme-pullyou" the plane would maintain and climb on the forward engine alone. But with the rear engine alone it was barely able to maintain altitude. This issue would be solely due to the poor airflow into the rear prop disc. And this trait is why I'm not a big fan of pusher props that have to be mounted behind a large fuselage.

Another option would be to mount the prop much the same as the Edgley Optica ( http://en.wikipedia.org/wiki/Edgley_Optica ) but without the big prop shroud. A huge bubble ball shaped pod for the gear and camera with a large custom made "dunce cap" spinner. to allow it to fair in the prop blast better. Again the idea isn't to turn the model into an efficient soaring device by making the huge pod slippery but to ensure that the air flows cleanly and closely around the pod to reach the propeller with the least disturbance.
BMatthews is online now Find More Posts by BMatthews
Reply With Quote
Old May 27, 2010, 09:06 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Fuselage drag minimization results

The optimizer did a very nice job decreasing the fuselage top profile drag. I set it so that the trailing edge gap was the same as for the original fuselage (for motor mounting purposes) and so that the max width was at least the old max width.

The drag at cruise went down from 0.0215 to 0.0144, and there is a much better "bucket" shape of the polar. The max thickness increased significantly and there is very little separation. This does very well for cruise Re, but I should probably check the estimated stall speed Re as well. I expect more separation there since it is so thick, but I don't think it will really be a problem. Next week I will work on implementing this shape into the fuselage design.
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old May 27, 2010, 09:34 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
The drag may be low but look at the streamlines off the end. That part between the line until they meet will be highly turbulent. And you're feeding that into your prop.

I may be making a mountain out of a molehill but I wonder how efficient the prop is as a pusher behind this compared to as a tractor. I know it's not in the cards to mount the motor up front but I suspect that you're paying a hefty efficiency price for having the prop in back. And putting it in the wake of the fuselage "Kamm" ending is just making it that much more costly.

Having said this I suspect your new shape will be lightyears ahead of the current one just by feeding it around the "nicer" shape even with the chopped off rear section.
BMatthews is online now Find More Posts by BMatthews
Reply With Quote
Old May 27, 2010, 09:44 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Quote:
Originally Posted by BMatthews View Post
The drag may be low but look at the streamlines off the end. That part between the line until they meet will be highly turbulent. And you're feeding that into your prop.

I may be making a mountain out of a molehill but I wonder how efficient the prop is as a pusher behind this compared to as a tractor. I know it's not in the cards to mount the motor up front but I suspect that you're paying a hefty efficiency price for having the prop in back. And putting it in the wake of the fuselage "Kamm" ending is just making it that much more costly.

Having said this I suspect your new shape will be lightyears ahead of the current one just by feeding it around the "nicer" shape even with the chopped off rear section.
Bruce,

Only a very small percentage of the total propeller sweep area will be inside that turbulent wake from the fuselage. Ignoring the wing, the vast majority of the propeller's intake will be the uniform flow outside the wake. Actually, a bigger concern is going to be the wing wake and downwash. I'm kind of set on the pusher configuration, though, due to the camera and the fact that I don't want to spend the extra money for a twin setup.

This was a good exercise for the fuselage, however, and I'm sure it will be much better than the current design, as you say. I'll post the new fuselage design once it is finished.

Thanks for your inputs. I definitely wouldn't have thought to do this otherwise.

Dan
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old May 27, 2010, 11:09 PM
B for Bruce
BMatthews's Avatar
The 'Wack, BC, Canada
Joined Oct 2002
11,164 Posts
Glad to help. I look forward to seeing the results.
BMatthews is online now Find More Posts by BMatthews
Reply With Quote
Old May 28, 2010, 12:43 AM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
One interesting note: the new fuselage shape roughly doubles the destabilizing effect it has on pitch, which basically just means the CG will need to be a little farther forward.
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Old Jun 02, 2010, 06:38 AM
Registered User
United States, KS, Andover
Joined Oct 2005
539 Posts
Quote:
Originally Posted by Montag DP View Post
I had to do a group project for an engineering optimization class this semester. We were allowed to choose our own, so I chose airfoil optimization. It was a group project, but I pretty much did the whole thing. But that's beside the point. I thought the results would be of interest to this forum.


--- snip ---


EDIT: By the way, each of the six cases took about 1-1.5 hours to reach a solution on my laptop. I don't count the XFoil function evaluations but I'd guess it's around 1000-2000 each time. I also ran into difficulties with XFoil not converging in some cases, so it's no simple task. I imagine adding 2 more design variables will increase the computational time to over 2 hours.
This is really cool stuff. Keep at it. I'm doing something similar using Xfoil and the Python based Genetic Algorithms package called PyEvolve, which is free. I'm striving for functionality at this point at the expense of speed, but at this point it is fairly quick. I typically use a population size of 20-30, but usually I'm in the ball park by 50 generations or so. I guess that translates to anywhere from a few minutes to an hour or so. I'm just playing with it now. The only mod I made to XFoil was to comment out the calls to the plotting routines. This cut the run time by about 50% on my old tired PC. My airfoils are parameterized differently from yours and I end up with more design variables (generally around a dozen), but with GA that doesn't cost any more. The expense comes in generating a lot of solutions. Going to a more generalized parameterization method would be my only recommendation to you to open up your design space a little. I handled poorly converging airfoils in two ways. First, I reset the default number of iterations to 200 ("iter" command under "oper") and secondly, if the airfoil still won't converge, I kick it out of the gene pool (at least for that generation).

Very neat stuff. Thanks for sharing.
Mark
markschaffin is offline Find More Posts by markschaffin
Reply With Quote
Old Jun 02, 2010, 02:54 PM
Sink stinks
Montag DP's Avatar
United States, GA, Atlanta
Joined Apr 2005
4,513 Posts
Quote:
Originally Posted by markschaffin View Post
This is really cool stuff. Keep at it. I'm doing something similar using Xfoil and the Python based Genetic Algorithms package called PyEvolve, which is free. I'm striving for functionality at this point at the expense of speed, but at this point it is fairly quick. I typically use a population size of 20-30, but usually I'm in the ball park by 50 generations or so. I guess that translates to anywhere from a few minutes to an hour or so. I'm just playing with it now. The only mod I made to XFoil was to comment out the calls to the plotting routines. This cut the run time by about 50% on my old tired PC. My airfoils are parameterized differently from yours and I end up with more design variables (generally around a dozen), but with GA that doesn't cost any more. The expense comes in generating a lot of solutions. Going to a more generalized parameterization method would be my only recommendation to you to open up your design space a little. I handled poorly converging airfoils in two ways. First, I reset the default number of iterations to 200 ("iter" command under "oper") and secondly, if the airfoil still won't converge, I kick it out of the gene pool (at least for that generation).

Very neat stuff. Thanks for sharing.
Mark
From what I've read, genetic algorithms are very commonly used in aerodynamic optimization because the design space is so multimodal (lots of local optimum points). We never really covered them in our class though. I expect also that the more design variables you add the more multimodal it becomes. When I was doing the project I originally had a parameterization that used 11 design variables, but I soon found it would be impossible to estimate gradients using this formulation because changing one variable slightly could screw up the entire shape. The GA would probably be able to work around that as well.

I'd like to hear more about this Python script. I'm not too well-versed in any programming language besides Matlab and VBA for Excel, so I don't think I'll try it myself, but it does sound pretty cool. Feel free to post results in this thread want.

Dan
Montag DP is online now Find More Posts by Montag DP
Reply With Quote
Reply


Thread Tools