Carbon Fiber Tube Fuselage

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
Make just a 2 ply skin, and the max panel size is pretty small before stresses go too high. So you can add plies or make the skin a sandwich panel. You get strength and stiffness at much less weight with sandwich panels. Well, it does not take much of a sandwich panel to allow you to skip the internal structure entirely.
I did run some numbers for flat panel, for my material available in my shed: On two rigth most columns - sandwich panel. Far similar reserve factor or deflection - you are forced to space support 3-4 denser..

1605098083309.png
 

BJC

Well-Known Member
HBA Supporter
Joined
Oct 7, 2013
Messages
12,715
Location
97FL, Florida, USA

Is this an example of a Warren Truss
Design/Build ultralight ??

Neat Web site anyway.
That is an impressive project, with lots of neat features and on-going development. If anyone knows the designer, builder, or pilot, please encourage them to post information here.

Anyone have information on how well the tube bonding and fitting attach bonding is holding up? Any drawings of the wing swivel arrangement?


BJC
 

Tiger Tim

Well-Known Member
Joined
Apr 26, 2013
Messages
3,628
Location
Thunder Bay
Discussion on how to make a carbon tube truss fuselage comes up fairly regularly on the forum and the the rough approximation of what gets proposed (and picked apart) is that you basically replace steel tube with carbon. The thing is, steel can be welded at each tube cluster while the closest equivalent for a composite tube is some sort of bonding where none of us can seem to ever reach a consensus on how best to bond the stuff.

What if we’re using the wrong example? Instead of looking at welded steel trusses what if we looked to early wood trusses. Early enough that good adhesives hadn’t come around yet. What you’ll find are a great many fuselages where (nearly) all of the wood members were in compression while a maze of crisscrossing wires in tension held the whole works together. I’m not claiming it would have fewer pieces or go together faster; it wouldn’t. What it would accomplish is to take bonding out of the equation for the DIY-er. And maybe be a tad lighter if you were really careful designing it.
 

Geraldc

Well-Known Member
Joined
Nov 12, 2011
Messages
525
Location
nz
Could be done by replacing single diagonal tubes that work in tension and compression with 2 diagonals in tension.(strut and wire)
You could use carbon tow and wrap each strut to locate it.
The section can be smaller because all loads are taken in tension.
Going back 100 years with new materials.
1605150645716.png
 

poormansairforce

Well-Known Member
Joined
Mar 28, 2017
Messages
1,106
Location
Just an Ohioan
Could be done by replacing single diagonal tubes that work in tension and compression with 2 diagonals in tension.(strut and wire)
You could use carbon tow and wrap each strut to locate it.
The section can be smaller because all loads are taken in tension.
Going back 100 years with new materials.
View attachment 104019
That picture and this thread remind me of when I had to explain to my kids why we don't use rotary phones, typewriters, 8 tracks, cassettes, LPs, answering machines, and outhouses anymore. There are easier ways to do things now.
 
Last edited:

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
Could be done by replacing single diagonal tubes that work in tension and compression with 2 diagonals in tension.(strut and wire)
You could use carbon tow and wrap each strut to locate it.
The section can be smaller because all loads are taken in tension.
Going back 100 years with new materials.
View attachment 104019
1605175530275.png

Find 10 diffencies in tail section.
 

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
And some number for panel/tube with R500mm radius.

1605181750782.png

Conclusion - you really want to avoid flat panels in composite at any cost..
 

Vigilant1

Well-Known Member
Lifetime Supporter
Joined
Jan 24, 2011
Messages
5,882
Location
US
And some number for panel/tube with R500mm radius.

View attachment 104025

Conclusion - you really want to avoid flat panels in composite at any cost..
Interesting. Some questions, if I may:
1) Does your method consider only the stiffness of the panel (due to compression of one skin and tension on the other) or does it also account for the tension from rib to rib (two fixed ends of the panel)
2) Have you been able to check/validate the results of the formulii/spreadsheet (e,g. do the deflections match a known actual layup/panel, etc)?

The idea of a composite flexible (in one direction) sheet ("black plywood") with CF stiffening riblets in one direction is still rattling around in my head. Would your method be able to calculate deflections for a CF sheet of, say, .25mm constant thickness and varying spacing (e.g. 25mm, 50mm, 100mm) of riblets of varying height (5mm, 10mm, etc) and constant . 25 mm thickness?
 

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
Interesting. Some questions, if I may:
1) Does your method consider only the stiffness of the panel (due to compression of one skin and tension on the other) or does it also account for the tension from rib to rib (two fixed ends of the panel)
2) Have you been able to check/validate the results of the formulii/spreadsheet (e,g. do the deflections match a known actual layup/panel, etc)?

The idea of a composite flexible (in one direction) sheet ("black plywood") with CF stiffening riblets in one direction is still rattling around in my head. Would your method be able to calculate deflections for a CF sheet of, say, .25mm constant thickness and varying spacing (e.g. 25mm, 50mm, 100mm) of riblets of varying height (5mm, 10mm, etc) and constant . 25 mm thickness?
1) Plate of infinite length, fixed on both side lines, lines are spaced at determined distance. Load is uniform, via plate pressure. All fabrics are 45/-45 degree. Reason - worst case scenario, wing skin designed for torsion load, transfer load from lifting force further down to other elements - ribs.
2) Will do it for some interested to me cases.

On third one - no it is not possible to run such a simulation on macro level. That i was able to do - just add some pre-defined sections (with know inertia/modulus) to panel.

I did this for myself to have a way to simplified optimization - you have known number of fabric, and you can combine 1-2-3 layers in skin, and support at different spacing to achieve same deflection/RF. And choose lightest one / easiest to make. Or on tapered section - you have to space ribs according to locla radius of panel to achieve uniform panel rigidity.

All for preliminary results, but i like its better than black metal or wood thinking approach.
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,796
Location
Saline Michigan
I did run some numbers for flat panel, for my material available in my shed: On two rigth most columns - sandwich panel. Far similar reserve factor or deflection - you are forced to space support 3-4 denser..

View attachment 103982
I like that this give you a reasonable comparison between known and understood skins and the new ones to you, and at a variety of rib spacings so you can make some solidly educated choices on what you need.

Is there a text I can refer to for the method used here? I am a mechanical engineer, so I start with Timoshenko and then go to Roark's for special stuff. For skin panels that see "inflation" loads, I go to Roark's chapter 11, then superimpose that upon loads due to bending and torsion of wings and fusealges.

I am trying to understand the inputs and outputs of this:

Rib spacing (assumed to be in mm) only sets the width of a rectangular panel. What are the lengths of the panels? The infinitely long case puts the least constraint on the panel, while real panels are constrained on all four edges. A look at Roark's table 11.4 will give some idea of how much deflection and stresses occurs in changing the length of the long sides;
Loading is uniform over these panels? And is it 20, 30, 40... or 120, 180, 240 kg/m^2? Fuselage panels may be close to uniform, but wings at high AOA tend to be forward loaded, meaning you usually have to superimpose results from triangular load and uniform load cases to get to a reasonable approximation of reality;
Deflection is mid-panel deflection out of unloaded case, yes?
Reserve factor is what? It appears to be a ratio but of what we can not tell. Is it FOS on the material strength or on some maximum deflection?

Billski
 

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
Thank you for reply

Is there a text I can refer to for the method used here? I am a mechanical engineer, so I start with Timoshenko and then go to Roark's for special stuff. For skin panels that see "inflation" loads, I go to Roark's chapter 11, then superimpose that upon loads due to bending and torsion of wings and fusealges.
Here is some information on used methodology: ScienceDirect.com | Science, health and medical journals, full text articles and books.

In contrary to you - I have graduated computer science, software modelling. I know that simulation is not always 100% proof - but it may give you good data for comparision. And as i wrote - i will make some samples this weekend, and will test them. If it is in +- 30% deviation ratio - it is ok.

On my input date height is 500mm - 20", it is separated by lines with fixed boundaries from displacement, not rotation. And lines itself are space by 20mm "empty spacers". Just to make calculations faster.

1605276184620.png

Loading is 20 (4.1 psf), 30 (6.14), 40 (8.2 psf), 50(10.24) and 100(20.48) kg per m2 x 6.

On wing loading - yes you are 100% correct, and i did this before on full skin calculation - but in normal case, you only have to check it at first 1/4 of chord to handle wing load * 2 (triangle area) * (you reserve factor) * 2 (reserve factor for composite).

Deflection is max at center of panel. On reserve factor - as far as i was able to understood - additional overload to be taken before failure at given deflection. I did some comparison on 20 or 40 loads, and i am not getting twice RF on 20 kg load.

And here are some screnshots. I am taking data from the middle.

1605276996058.png
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,796
Location
Saline Michigan
I get FEA of panels (FEA works...) , two parallel edges simply supported, two alternate edges not restrained (infinite length case). Reporting deflection at center of panel and a reserve factor based on ? I still can not tell. Tsai and Wu failure criteria cited has safe designs when criteria is less than unity. The makeup of the Reserve Factor is still not clear. Is it the inverse of a failure criteria?

The lack of reserve strength in some of the cases indicates that inflation alone will cause disqualification of some schemes.

I also still do not get the notation of loading in "kg per m^2 x 6" what is the "x 6" about? kg/m^2 seems quite complete to me.

Comments on the modeling and its impact:

FEA works when it is applied like the real system it is modeling;

The skin loadings as represented here are due to moving air outside and stationary air inside, not wing loading. Wing loading is the difference between the skin loading on the top vs the on the bottom. In fast airplanes, you can have very high skin loadings even at low lift, while in low speed airplanes, skin loadings are small.

Leaving two edges free is the case of the panel being infinitely long. According to Roark's Table 11.4 Case 1a (simply supported on all four edges), the infinitely long panel has 2.6 times the max stress of the square panel and the deflection is 3.2 times the deflection of a square panel. Stresses and deflections of real panels will be substantially smaller than for the illustrated panels.

Simply supported edges - where they are supported against translation, but not against rotation - might be appropriate in a case where torsionally soft members support a panel that ends at the supports. In composite wings this case will be present at the ends of each wing panel or where an access panel is attached. Where the skin is continuous across the supporting members, the fixed edges (Table 11.4 Case 8a) peak stresses and deflections similarly improve with transition from infinitely long panels to more nominal proportions, with max stresses being only slightly higher...

Then using curved panels shows further reductions in panel deflections and stresses.

In total, it appears that Stanislavz has picked the most severe cases to study, and should be secure, if somewhat overbuilt, using this method for sizing. Being as we have other loading issues to deal with besides inflation, it is good to have significant reserve strength under these issues.

Billski
 

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
I also still do not get the notation of loading in "kg per m^2 x 6" what is the "x 6" about? kg/m^2 seems quite complete to me.
Example for 50 kg per m2 ~10 psf. Pulling 6 g at the nose we 10 * 6 * 2 = 120psf of skin loading. At trailing edge - we will have 0. At mid chord - we will have 60 psf. And you want to multiply it by 2 for ultimate RF. Based on Jim Marske book, similar simplified solution found in other book. :
1605335133064.png

The makeup of the Reserve Factor is still not clear. Is it the inverse of a failure criteria?
As i said before - it is not very clear. Logically thinking - panel loaded at 10psf and having reserve factor of 5, will have reserve factor of 1 loaded at 10psf. But it is smaller according to FEA analysis.

Leaving two edges free is the case of the panel being infinitely long. According to Roark's Table 11.4 Case 1a (simply supported on all four edges), the infinitely long panel has 2.6 times the max stress of the square panel and the deflection is 3.2 times the deflection of a square panel. Stresses and deflections of real panels will be substantially smaller than for the illustrated panels.
Thank you for pointing this. Yes, i will redo it, on real life it will be easier to check it too - ie. opening in two layouts of plywood with some air at the pressure blown in for testing.

Being as we have other loading issues to deal with besides inflation, it is good to have significant reserve strength under these issues.
As i mentioned before - On single skin you are limited on rigidity/max allowable deflection. Ayes opener for me was Cri-cri - ribs was so dense (45/90mm first half of them wing/last half of the wing) - but it was needed to achieve minimal skin deflections for laminar flow. On minimal weight - being done from 6mm 100 kg/m3 klegcell, they are lighter than full cored wing. 6/45*100 = 13.333 kg/m3 at root portion of the wing, to the half of it at tip half of the wing. Or 0.83 / 0.415 lb/ft³. DOW flotation billets are at 2 lb/ft³ at least to be used for full cored wing..
 

wanttobuild

Well-Known Member
Joined
Jun 13, 2015
Messages
744
Location
kuttawa, ky
A cluster or node is easily constructed out of carbon fiber with the introduction of the simple PIN JOINT. Totally changes the node into a much simpler fiber path.
With carbon fiber you gotta think outside the box, and this method requires the introduction of alum or 4130.
you could disassemble the fuselage and make changes as each tube is connected with the pins
 

Bill-Higdon

Well-Known Member
Joined
Feb 6, 2011
Messages
1,091
Location
Salem, Oregon, USA
A cluster or node is easily constructed out of carbon fiber with the introduction of the simple PIN JOINT. Totally changes the node into a much simpler fiber path.
With carbon fiber you gotta think outside the box, and this method requires the introduction of alum or 4130.
you could disassemble the fuselage and make changes as each tube is connected with the pins
Shades of Great Britain's Hawker Hurricane & Wellington
 

Vigilant1

Well-Known Member
Lifetime Supporter
Joined
Jan 24, 2011
Messages
5,882
Location
US
Obviously, need to electrically isolate the metals from the CF.
 

Victor Bravo

Well-Known Member
HBA Supporter
Joined
Jul 30, 2014
Messages
8,622
Location
KWHP, Los Angeles CA, USA
A cluster or node is easily constructed out of carbon fiber with the introduction of the simple PIN JOINT. Totally changes the node into a much simpler fiber path.
How do you get one pin axis going in three or four different directions at once, where you have several tubes coming together at different angles at one intersection on the fuselage?

The weight of the metal connector has to be added to both ends of every tube, so that takes back a little of the weight advantage.
 

stanislavz

Well-Known Member
HBA Supporter
Joined
Aug 21, 2016
Messages
933
Location
Lt
A cluster or node is easily constructed out of carbon fiber with the introduction of the simple PIN JOINT. Totally changes the node into a much simpler fiber path.
With carbon fiber you gotta think outside the box, and this method requires the introduction of alum or 4130.
you could disassemble the fuselage and make changes as each tube is connected with the pins
I could think of it for tubes in soft metals - easily deformed to flat shape. And some of that shape may be connected using one pin. But on composites - it is highly recommended to avoid any stress concentration. Which Pin is..
 
Top