Quantcast

Composite wing spar web design - composite webs, foam webs, etc.

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

UAVGuy

Active Member
Joined
May 1, 2020
Messages
29
I need to design a wing spar for a large RC UAV.

I understand how conventional (ie steel) beams work, can calculate stresses, etc.

I'm mystified at how to design a fully composite spar because I've heard that composites don't like compression loads, leading to buckling on the compressed flange (cap) and web compression between the caps.

If you give me the strength (compression and tensile) of a set of caps and a distance between them, I can calculate ultimate loads, etc, *assuming* the web doesn't fail. But how do I design a composite web ? And how do I prevent the spar cap in compression from buckling ?

I'm mystified at how some designs use foam as the web ! For example, in a LongEze wing there are upper and lower "spar caps" of considerable size, which I understand. But the web between them is insulating foam ! I'm guessing in that design that the spar caps are tied into the wing skins which spread the compressive/shear load between the spars over the entire wing area, thus making the entire wing surface the web. Am I right ?

If so... how far can this be taken ? How narrow of a piece of foam can I use to support the spar caps in a design ? How do I calculate this ?

What if I want to make a more complicated web... say a piece of foam with a composite on each side. How do I calculate the strength of that ?

Any and all hints and advice on composite spar design will be appreciated.

Thanks
 

Marc Zeitlin

Exalted Grand Poobah
Joined
Dec 11, 2015
Messages
734
Location
Tehachapi, CA
And how do I prevent the spar cap in compression from buckling?
Same way you do it with metal - you make the compression cap thicker than the tension cap.

For example, in a LongEze wing there are upper and lower "spar caps" of considerable size, which I understand. But the web between them is insulating foam!
No it is not. In all the Rutan derivative canard aircraft, the spars are solid glass unidirectional caps, with a solid glass shear web of +/-45 degree glass. The shear web glass is supported in buckling by the interior wing foam (in the wing) or a slab of foam with glass on the opposite side to help stiffen it inside the main spar. I suggest you look at Chapter 15 of the Long-EZ plans (Main Spar) and Chapter 19 (Main Wing) as well as Chapter 10 (canard - essentially similar construction methodology for caps/shear web) for a clearer view of what's actually inside the spars.

I'm guessing in that design that the spar caps are tied into the wing skins which spread the compressive/shear load between the spars over the entire wing area, thus making the entire wing surface the web. Am I right?
Nope. In the EZ family of aircraft, the caps take the bending loads, the shear web(s) transmits the shear from top to bottom, and the wing skins take the torsional loads, pretty much like almost all heavily loaded wings of whatever material and construction.

On the VE, the bending loads are low enough that the spar caps and shear web stop about 2/3 of the way to the tip, and the wing skins take all the bending loads from there outboard, with the wing foam taking the shear loads. The VE is a small, relatively light plane.
 

UAVGuy

Active Member
Joined
May 1, 2020
Messages
29
I agree that the LongEze main strut has a solid glass web.

I took a closer look at Chapter 19, the wings. It appears that a sheer web is built up in step #4, when the leading edge cores (FC4 and FC5) are removed. Thank you for pointing this out.
 

lr27

Well-Known Member
Joined
Nov 3, 2007
Messages
3,822
You might look at some of Mark Drela's designs at charlesriverrc.org
He shows the spar construction and the estimated strength. Somewhere. I've seen some discussion by him about physically testing spars. I think he, or at least someone, tested wings that used the foam cores for the shear webs. This was probably on the defunct Allegro Lite Yahoo group, though. As I recall, tbe top spar had to be thicker than is usually seen.
 

UAVGuy

Active Member
Joined
May 1, 2020
Messages
29
Same way you do it with metal - you make the compression cap thicker than the tension cap.
So if I'm using a unidirectional material and it is rated for X PSI in compression, I just calculate the compression stress in the cap and size it appropriately ? That is it ?

How about the web, if I'm sandwiching a piece of foam with a composite material ? How do the properties of the foam play into it ? What is the difference between a web of thickness T, versus a web with 2 thicknesses of 1/2T, separated by a bit of foam ? Seems to me it is way stronger, but I know the foam is compressible in X, Y and Z and will sheer between the layers easily compared to if the web was just thickness T.
 

UAVGuy

Active Member
Joined
May 1, 2020
Messages
29
This was very helpful !
 

RogFlyer

Member
HBA Supporter
Joined
Jun 27, 2017
Messages
19
Location
Lewisham
UAVGuy - if you are used to conventional (ie civil engineering) design, a reasonable way to think about this is as if you were designing a concrete beam - which is, after all, a composite structure. For bending, assume linear strain through the thickness, with zero strain at the calculated position of the neutral axis.
Then the steel plus concrete take the compressive forces, at the same strain. For the tension side, the steel takes the tension forces, ignoring the tension capacity of the concrete (it is assumed to be cracked, thus zero - high strength concretes and prestressing are different models).

For "glass" (fibreglass, Kevlar, carbon, basalt, etc) and foam composites, the load carrying contribution of the epoxy resin is ignored. It is there to transfer the strains between the glass fibres. Only the glass reinforcing dimensions are used to calculate the moment of inertia. Wrapping plies are sometimes added to try to ensure everything works as a unit - sometimes these are included in the stress calculations, depends on how conservative you wish to be.

As in steel and concrete beams, there are limits on the flange widths before one needs to consider flange buckling. The design shop I sometimes work with say they have not found anything they consider to be reliable on flange widths for composite beams, so they stick to flange overhang from the web is less than depth/8. So for a 160mm deep beam, the flange width is limited to 2*20mm overhang plus the web thickness. And they prefer depth/10, ie 32mm plus web thickness. As they point out, though, they are not specialists in composite design.

A web of thickness T and 2 webs of thickness T/2 with a piece of foam between them will have the same raw shear strength - in the absence of web buckling. The foam stops the web from buckling, increasing the available web load from the Euler buckling limit to closer to the limit stress of the glass. For glass reinforced webs, it is usual to ignore the shear stress capacity of the foam.

If the web is foam only, then one runs the shear stress capacity based on the foam shear stress. An important consideration here is that joint between the shear web and flanges can carry the co-planar shear stresses at the nominal strain, to transfer the bending shear.

Hope this helps, and that it is not too inaccurate.
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,144
Location
Saline Michigan
Listen to Marc Zeitlin...

Composites are more complicated than metal, in that the various lamina can have greatly different stiffnesses in bending, shear, and torsion, but they must all strain together under these loads. You may find these threads to be useful.



Billski
 

Vigilant1

Well-Known Member
Lifetime Supporter
Joined
Jan 24, 2011
Messages
5,179
Location
US
In those cases where a composite spar web is supported only on one side by foam (e.g. a foam D-cell with spar caps at the aft top and bottom with a CF spar web laid up on the aft face of the foam), do designers typically find that the web gets sufficient support from buckling from that foam on one side? Or are other measures often required to prevent buckling (spanwise stiffeners in the web, a sandwich panel trapping the web on the back side, etc).

Thanks.
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,144
Location
Saline Michigan
You might look at some of Mark Drela's designs at charlesriverrc.org
He shows the spar construction and the estimated strength. Somewhere. I've seen some discussion by him about physically testing spars. I think he, or at least someone, tested wings that used the foam cores for the shear webs. This was probably on the defunct Allegro Lite Yahoo group, though. As I recall, tbe top spar had to be thicker than is usually seen.
While Mark Drela's work is interesting, please remember that he is a world class fluid mechanics guy. When he talks aerodynamics, we should all listen.

That being said, Mark Drela is NOT a structures guy, and it shows when he talks structures. As a genuine structures guy, I will tell you to please seek guidance on aero structures someplace else. Basics are in Bruhn or Peery and Azar. For composites, you need to add Jones or Tsai and Hahn. Yeah, I am recommending whole textbooks. Let's get to where I see shortfalls.

The first fundamental error is designing the web for only shear. The web deforms with the spar caps, so the web MUST be analyzed as carrying shear PLUS the extension/compression of the adjacent caps. This applies to metal as well as composite beams. Using the simplification results in shear webs failing early. The other thing that comes out of this is that usually the way to make strength at low weight is to beef both the caps and web. Beefing the web only is the heavy way to make strength as it takes a lot more web when it is the only thing changed to make strength.

The second fundamental error is thinking that composites can be safely designed using allowable stresses. The combined strain state of each lamina must be checked against a full failure criteria, much as we would check von Mises stress in metals. Surprises of the bad kind await those who design composite structures without doing full-up analysis. This is not to say that simple rules have not been used, but many of them have succeeded by use of large FOS (far short of optimal designs) and conservative use more than by wise design.

If you are willing to use bridge-builder levels of FOS, you may get away with simplifications, and in the RC world, the weight penalties may still be trivial. If you are designing to reduce deflections (and strains are then modest), you will likely be OK using Jim Markse's methods. The consequences of structural failure are low in the RC world as well, so perhaps simplified design rules are OK there as well.

Billski
 
Last edited:

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,144
Location
Saline Michigan
In those cases where a composite spar web is supported only on one side by foam (e.g. a foam D-cell with spar caps at the aft top and bottom with a CF spar web laid up on the aft face of the foam), do designers typically find that the web gets sufficient support from buckling from that foam on one side?
In massive foam core wings, the web can work fine with core on only one side. This is because the foam is large and the web is thus well constrained, providing the needed support against buckling that would otherwise destroy the spar's load carrying capability.

The alternative is a free standing spar and a web core of an inch or less. You had better put some lamina on both sides of that web core. Here the stiffness and support against the web buckling is almost entirely the the thickness (or lack thereof) of the web. To achieve enough support against buckling, you need much more bending stiffness of the shear web panel. Split the web with some on one side and some on the other side of the core, and it is way stiffer and works fine.

Keep in mind that the web does have to support not just the shear load at each station, but that the top and bottom of the web extend and contract with the caps too. As I pointed out above, the light way to strength is usually achieved with some beefing of both caps and web rather than just beefing the web.

Billski
 
Last edited:

Jay Kempf

Curmudgeon in Training (CIT)
Lifetime Supporter
Joined
Apr 13, 2009
Messages
4,088
Location
Warren, VT USA
How large is the UAV? Will it fit under the 55lb rule? 12' span-ish...? What is Vmax? Slow fat wings are pretty easy. Thin fast wings are tougher but more suited to just foam core. A shear webs and caps can be inserted in the core before skinning. Hollow, female molded wings are more steps and yes you can build a spar in a separate operation. Depending on size and speed it can be simple or complicated. Working on the same problem right now at about 12' wingspan with a laminar airfoil and hollow construction female molded, one off CNC foam tooling. I am using 3D printed parts to make simple forms to mold shear web like parts that have bonding flanges that bond to the UNI spar caps laminated with the skins. You can cut forms out of foam as well like Boku just did on his Carbonmax Puddle jumper. Bond area is key. When the parts are small it is harder to control all the little cloth bits but doable with some cleverness.

The below pic shows the 3D printer parts in place in a wing skin just as an example. Those get skinned over and more parts get added. Two bagging operations for the whole wing and some Hysol bonding. This is one outer wing section. Both skins in one mold.
 

Attachments

Lendo

Well-Known Member
Joined
Feb 6, 2013
Messages
522
Location
Brisbane
Jim Marske told me he didn't allow for Buckling, as Buckling starts with the Web there are formula for this on the Internet look for Cantilever type formula. The rest of his approach is straight forward and the Buckling formula aren't so difficult.
The way Billski does his analysis is a all together, at a another level completely.
George
 

kent Ashton

Well-Known Member
Joined
Aug 15, 2008
Messages
756
Location
Concord, NC
This guy has a good site showing how the shear web that connects the spar caps are built in a Long-ez. Chapter 19 is the wing
These structures are so quick and easy to build you could make a load-test item pretty quick
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
98
Location
Altus SW Oklahoma
/snip/
A web of thickness T and 2 webs of thickness T/2 with a piece of foam between them will have the same raw shear strength - in the absence of web buckling. The foam stops the web from buckling, increasing the available web load from the Euler buckling limit to closer to the limit stress of the glass. For glass reinforced webs, it is usual to ignore the shear stress capacity of the foam.

If the web is foam only, then one runs the shear stress capacity based on the foam shear stress. An important consideration here is that joint between the shear web and flanges can carry the co-planar shear stresses at the nominal strain, to transfer the bending shear.

Hope this helps, and that it is not too inaccurate.
I am an innocent bystander - but I am interested in the difficulty of adhering a foam anti-buckle fill to the web walls in a box spar. I seem to recall people pasting foam wings with glass bead resin mix prior to laying a glass skin - is this usual technique?
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,144
Location
Saline Michigan
I seem to recall people pasting foam wings with glass bead resin mix prior to laying a glass skin - is this usual technique?
I believe you are talking about slurrying the foam. It is customary on open wet layups of foam to fill the surface with a "slurry" - equal volumes of resin and microballoons - and squeegee off as much as you can before applying glass. This has been done to make the layer of resin in the broken cells of the foam lighter than if it were straight resin.

As to paste, the terminology is usually "dry micro" - aprox 4 parts micro per one part resin. This used to fill nail holes, dents, etc prior to laying glass.

Billski
 

User27

Well-Known Member
Joined
Apr 5, 2015
Messages
52
Location
England
Go find your local glider repair shop and look at some of their broken spars. I cannot tell you how to design the spar (aero & systems guy) but I do know the spars have up to 15 wraps of carbon cloth at the root as a shear web. The spar is often formed first and dropped into the wing mould with the wing core chamferred down over the spar to allow the spar to be full height of the wing section. Bear in mind the wing section thickness of glider wings has been reducing to reduce drag, which results in heavier spars to react the loads. Sometimes a relatively dense foam is used in the centre of the spar (compared to the wing skin foam core), but I don't know the relative numbers. perhaps a few test pieces are in order?
 

tdfsks

Well-Known Member
Joined
Aug 29, 2005
Messages
93
Any and all hints and advice on composite spar design will be appreciated.
With all respect, you have a big knowledge gap and I am not sure that a forum like this can teach you to design a composite spar. You need to hit the books. Unfortunately this is one of those questions that has no simple answer.

As an experienced structures engineer, with years of experience designing and stressing composite structures, including spars, I can offer a few comments to help guide your learning.

First, I think you need to understand how composites fail and what allowable stresses / strains you can use. This is the first area where composites differ greatly from metal structures and you need to develop a good understanding of the materials, their failure modes and failure criteria (Tsai, Hill etc). As a general rule composite structures are designed to strain limits and 4000 micro strain would be about right for the tension cap of a uni glass epoxy spar at limit loads. The compression allowable would be lower but I don't have an exact number handy (maybe 3000 microstrain from memory but I am reluctant to provide an exact number for reasons I will discuss below). These numbers will be different for other materials such as carbon. Whilst I have a lot of allowable data, much of it is proprietary and so cannot be shared. Unfortunately this is the reality of designing with composites. There is very little data in the public domain as much of it has been generated by companies developing and certifying their own aircraft. Also one other observation on allowables. Contrary to the urban myth that composites do not suffer from fatigue, they can fatigue and the use of these low allowables is important to ensuring a long and trouble free life (lot of papers in the sailplane community on this).

Why do we design to strain limits ? In simplistic terms it is because the resin matrix is more brittle (lower strain to failure) and will crack long before the ultimate strength of the fibres. When doing coupon testing of composite materials in an lab, it is amazing how much audible cracking there is at relatively low stresses way before the coupon ruptures. So we need to set conservative strain limits to preclude this sort of cracking (which is why composite structures are never as light as the ultimate strength to weight ratio of the basic materials would suggest). Of course E = stress / strain so you may see some people using conservative stress limits (JAR VLA for example does provide a stress allowable for spars than can be used without further justification) but generally the convention is to use strain.

Strain allowables for shear webs will be lower and I am not going to provide any numbers here because of the multitude of factors that influence them such as the laminate thickness and core density etc. There are analytical approaches for predicting the buckling stress of thin webs on foam cores but honestly the theory can only go so far before you need to test (or design conservatively). Also bear in mind that in some composite structures the caps will carry a good portion of the shear loads .... I will leave that for you to think about (shear stress = VQ/It). One thing I would caution about is interpreting shear allowables that you might find in books or papers. Generally shear allowables for composite materials are obtained using a rail test or similar in which a flat plate of composite material is clamped between two pairs of steel rails with a small gap between them. This test produces a perfect shear failure in the laminate whilst completely excluding buckling. You will never achieve these allowables in a spar web that is laminated over foam core because the failure modes are different.

The other factor that needs serious thought with composites, when setting the allowable stresses, is the operating environment and how this affects the material properties. Elevated temperatures will reduce the strength and modulus of composite materials and cores. Also long term moisture absorption will affect strength. Read up on the so called glass transition temperature (Tg) of your resin matrix material ... you should quickly realize why most resins used in wet layup structures will require a white paint to be used to keep temperatures down in the structure when exposed to sunlight. Anyway, you need to knock down your allowables to account for these effects. All major certification standards for airframe structures require the consideration of environmental effects. At least one of the European standards (JAR VLA) provides some knock down factors that may be used without additional justification. It is the difficulty of establishing these environmental effects that was one of the reasons that composites were so slow in coming to certified Part 23 aircraft. The test program is daunting. As an aside, the simplified approach used in Europe for gliders and JAR VLA aircraft compared to what is required for Part 23 certification is what allowed the Europeans to get ahead and produce certified composite aircraft many years before the first Part 23 composite aircraft. So what does a homebuilder do ? One approach would be to research and adopt some of the standard knockdowns or allowables that are used in sailplane design etc. If you are using prepreg materials then your material supplier may have pre-qualified the materials you are using through the AGATE program and be able to provide allowables that include the effects on elevated temperature and moisture exposure. You can find a lot of info on the AGATE program on the web. Also one of Martin Hollman's books (Composite Aircraft Design) discusses environmental effects (temperature anyway) and presents some data that he derived from his own testing. I am not sure if these books are still available (no longer on the Aircraft Designs website) as he passed away a few years back.

As you have pointed out, the compression performance of composites is not as good as for tension. In simplistic terms, the reason for this is that the fibres are like small columns that are stabilized by the matrix (i.e. the resin). These columns buckle before they get anywhere near the ultimate strength of the fibre. So we can say that ultimate tensile performance is fibre dominated and ultimate compressive performance is matrix dominated. This is very simplistic and I would suggest you read up on this. Anyway, as in a metallic structure, we need to check multiple failure modes in compression. The compression allowable of the cap material itself and then local or column buckling of the cap would also need checking. Compression allowables for materials are derived using one of a number of laboratory tests that preclude any type of buckling so be careful and make sure you understand where your allowables come from and do not use them out of context.

Detailed design in composite structures is vitally important. For bonded joints look at the work of Hart Smith (numerous papers). You need to learn how to design structures to eliminate inter-laminar stresses (tension normal to the plies such as would occur in a spar cap with a kink in it for example) because there is no reinforcement in this direction and composite structures do not perform well at all when these types of loads are present. Also you need to make sure you check inter-laminar shear stresses where the web joins the cap to make sure you have enough area in the joint. Also you will need to think very carefully about how to design wing attachments. As someone else suggested above, have a look at how sailplane designers have done this. Bearing stresses and shear stresses will be important considerations there.

I would be happy to comment on a specific design concept if you post it and provide some more guidance on the analysis approach.
 
Last edited:

Tommy222

Member
Joined
Mar 23, 2019
Messages
11
Don’t know if this would add much here but, back in the 90’ a company in FL sold kit planes made of fiberglass and foam. AeroComp is the company and the planes were CompMonster, CompAir 4, 6, 7, 10. All their wings were made with 1/2” foam board wrapped in cloth and bonded with Vinyl-ester resin. They made a spar about an inch thick. They had two wings, a tapered and a symmetrical. They ran up to 650hp turbines on those planes and never had a spar failure. Too bad the rest of the plane worked as well.
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
98
Location
Altus SW Oklahoma
With all respect, you have a big knowledge gap /.../no simple answer.
As an experienced structures engineer, with years of experience designing and stressing composite structures, including spars,/.../you need to understand how composites fail /.../you need to develop a good understanding of the materials,/.../composite structures are designed to strain limits and 4000 micro strain would be about right for the tension cap of a uni glass epoxy spar at limit /.../ I have a lot of allowable data, much of it is proprietary and so cannot be shared. /.../. Contrary to the urban myth that composites do not suffer from fatigue/../Why do we design to strain limits ? /.../ I will leave that for you to think about (shear stress = VQ/It). /.../ You need to learn how to design structures /.../ you will need to think very carefully about how to design wing attachments.
Not sure why, but this note left me longing for anosmia. Couple or four thoughts:
1) Beware of the note that starts, " With all respect...".
2) because the linear elastic limit is not well defined for aluminum alloys, the material is often specified at 0.1% strain. Oh look: 1000 microstrain! [instead of a conservative glass 4000 microstrain limit?)
3) Imagine a regulatory body so ill-conceived as to offer working rules of thumb which allowed Europe to get WAY ahead of the US on composite airframes?
4) I suspect anyone who can offer usable guides to composite structure needs all due respect! <g>
/end of unwarranted rant/
 
Top