# Composite spar design & composite wing design?

### Help Support HomeBuiltAirplanes.com:

#### karoliina.t.salminen

##### Well-Known Member
Hello,

I am thinking of a concept which is targeting at both high AR and very good empty to gross weight ratio at the same time. The weight of the resulting wing will pretty much dictate the feasibility of the concept. I want to calculate the weight based on actual materials that are going to be used, instead of using the rule of thumb estimation equations from Daniel Raymer's book I have. If I know how much fabric I am going to use, I can estimate how much resin is going to be used and resin + fabric + foam equals the weight of the wing without the control mechanisms (and weight estimation of these will be next step after this basic wing estimation).

I think there are several people designing their own planes, some have already designed and built even many of them, and they have already calculated their spars and wing skins. So looks like there are some of the best people in the world to know this topic on this forum. So could you please help me to get started with this? I had some stress analysis long time ago at university but there was absolutely nothing about sizing of composite structures where the stress direction is to be known and the fiber direction will be aligned with the stressed direction. I have attended several composite courses but these have been targetting for composite aircraft repair, not actually designing one from scratch. I know there is a lot to study, but everybody has to start someday. I would like to avoid "metal thinking" as much as possible and be able to use the benefits of the fiber direction as much as possible.

The application is: molded composite (carbon) wing (fabricated on female mold) of large span and relatively
large AR and I need to know how much the spar and skin might weight to be able to conclude the feasibility of the concept as the wing will determine if the plane will be too heavy or good enough.

1. Which books should I get, which publications should I read, are there any excel sheets made for it by someone - to learn the sizing of composite spars (carbon spar).
I found some spar calculator xls for other materials from this forum, but have not found for carbon spar anything except
for the basic program from Martin Hollman's book. I have not yet had the efforts to porting it to C++ unlike some other programs from his book that I have already a C++ version done, so I can't use it for preliminary sizing of the spar (yet) and I would like to know if someone else has excel sheet for that already so I wouldn't need to port the program. To understand what the program does would be helpful too: The book does not explain how the sizing method works, it only tells about basics of calculating matrices and then it shows the Basic code and then it expects that you get it what was the methodology used. There is a information gap in between.

2. Same question for the wing skin and other structure. The basic structure will be sailplane-like, so there is skin on both outside and inside of the wing panel and there is foam or honeycomb between these and then minimum amount of ribs inside the wing (so that the skin is the main load bearing component on twist and the spar takes the bending loads). Some kind of preliminary sizing would be needed for that too to end up with the wing weight estimate. Basically I need to know answer for the question: how many layers of the fabric I need per square meter of wing skin for each wing station.

I know that for optimized results FEM analysis is needed but I think it can not be made if the starting point is not known, so I would need to end up with this starting point.

Any advice on this would be great. If you have a good excel sheet for the purpose, if you could share it with me, I would be very happy as well. Thanks in advance.

Best Regards,
Karoliina

##### Well-Known Member
Here, Orion has summed up all the "major" composite books:
https://www.homebuiltairplanes.com/forums/aircraft-design-aerodynamics-new-technology/8950-technical-references-books-technical-papers-software-etc.html#post89319

As for weight, the majority, even for larger AR's is usually in the skin. Some sailplane wing weights:
Duo Discus Turbo: 90+10 kg, fuselage+payload around 600 kg
15-18 meter racer: 45-55 kg, fuselage+payload around 300 kg

Some random remarks:
*One should always have multiple layers of cloth.
*Ribs aren't a necessity. No ribs is sometimes lighter and easier to manufacture. In fact I went through some trouble to have a very weak rib (fuel slosh), because a load-bearing rib would increase wing weight and complexity.
*Start with a given geometry (taper, spar location, spar height/width ratio) and calculate the thicknesses and check against the final stresses and displacements. Then start varying taper, sweep, spar location and so on, this is far easier than writing an optimization loop.

#### Jay Kempf

##### Curmudgeon in Training (CIT)
Hello,

I am thinking of a concept which is targeting at both high AR and very good empty to gross weight ratio at the same time. The weight of the resulting wing will pretty much dictate the feasibility of the concept. I want to calculate the weight based on actual materials that are going to be used, instead of using the rule of thumb estimation equations from Daniel Raymer's book I have. If I know how much fabric I am going to use, I can estimate how much resin is going to be used and resin + fabric + foam equals the weight of the wing without the control mechanisms (and weight estimation of these will be next step after this basic wing estimation).

I think there are several people designing their own planes, some have already designed and built even many of them, and they have already calculated their spars and wing skins. So looks like there are some of the best people in the world to know this topic on this forum. So could you please help me to get started with this? I had some stress analysis long time ago at university but there was absolutely nothing about sizing of composite structures where the stress direction is to be known and the fiber direction will be aligned with the stressed direction. I have attended several composite courses but these have been targetting for composite aircraft repair, not actually designing one from scratch. I know there is a lot to study, but everybody has to start someday. I would like to avoid "metal thinking" as much as possible and be able to use the benefits of the fiber direction as much as possible.

The application is: molded composite (carbon) wing (fabricated on female mold) of large span and relatively
large AR and I need to know how much the spar and skin might weight to be able to conclude the feasibility of the concept as the wing will determine if the plane will be too heavy or good enough.

1. Which books should I get, which publications should I read, are there any excel sheets made for it by someone - to learn the sizing of composite spars (carbon spar).
I found some spar calculator xls for other materials from this forum, but have not found for carbon spar anything except
for the basic program from Martin Hollman's book. I have not yet had the efforts to porting it to C++ unlike some other programs from his book that I have already a C++ version done, so I can't use it for preliminary sizing of the spar (yet) and I would like to know if someone else has excel sheet for that already so I wouldn't need to port the program. To understand what the program does would be helpful too: The book does not explain how the sizing method works, it only tells about basics of calculating matrices and then it shows the Basic code and then it expects that you get it what was the methodology used. There is a information gap in between.

2. Same question for the wing skin and other structure. The basic structure will be sailplane-like, so there is skin on both outside and inside of the wing panel and there is foam or honeycomb between these and then minimum amount of ribs inside the wing (so that the skin is the main load bearing component on twist and the spar takes the bending loads). Some kind of preliminary sizing would be needed for that too to end up with the wing weight estimate. Basically I need to know answer for the question: how many layers of the fabric I need per square meter of wing skin for each wing station.

I know that for optimized results FEM analysis is needed but I think it can not be made if the starting point is not known, so I would need to end up with this starting point.

Any advice on this would be great. If you have a good excel sheet for the purpose, if you could share it with me, I would be very happy as well. Thanks in advance.

Best Regards,
Karoliina
Karolina,

Great question. From what I have been involved in there is no generic spreadsheet that can estimate and analyze any wing config. But it isn't all that hard to build a spreadsheet for your intended design if you already have a configuration in mind. The number of dimensions required for a first pass formula calculation of max stresses in the key spots and weight can be done in one spreadsheet. This can be used to do "what-ifs" until you have a more sane candidate for full analysis. The projects I have been involved in where weight was the most critical key variable involved making coupons of the intended layups so that weight per square surface area could be accurately assessed. Weight ratios of resin do not really characterize scrap and joinery or differing methods like prepreg vs. infusion vs. wet layup with vacuum vs. wet layup without vacuum. We used a glass top table to make coupons. Cheap, quick and easy. Once you have this you can then use these numbers in your final design. I fake out Solidworks with actual layup numbers for instance with a general layup of .010 inch of carbon layup each side of a .250 inch foam core, I then calculate a surface square inch of this materials weight and then I model all of those parts .250 inch thick and apply that density to that shell model. That way I get an accurate model that I can estimate CG of the component and the component in the assembly. Then that model can feed directly into composite layer analysis. But a first isotropic pass will give you general flexural characteristic of the shape to tell you what to look out for in the overall assembly.

Management of wet laid up joinery much like calculating interference drag is the hardest thing to really estimate and the easiest thing to get wrong. When you have 5 people around a large wing trying to get all that adhesive in the right places at the right amount. MacReady had a really hard time dolling out epoxy on the Gossamer Condor wing (20 college students each with a tiny cup of epoxy and responsible for one rib joint at time).

The other thing that has caught me by surprise each time I analyze a new configuration is how much beef has to be added to the wing for attachment especially for individually removable wing panels and especially if you want the conservative safety factor of aerobatic level of strength in the wing for emergency maneuvering. What I have done is modeled and done basic FEA on the spar and attachments alone by doing a distribution per linear length of spar of the entire weight of the airframe just as a first approximation. This significantly simplifies the amount of work required to be done to optimize the spar design and what I have found is that the results scale very well to the final analysis.

Just some approaches that have worked for me to get from theory to the real world quickly. Looking at other airframes is a good way to start but the devil and the possibility of innovation is in the details in the end. The range of possibilities is from the Global Hawk and the Rutan Voyager for largest payload to airframe and range capability to say the F104 on full tilt for a lack thereof and everything else in between is possible.

#### Aircar

##### Banned
I take issue with the statement that 'the majority, even for larger aspect ratios is in the skin' (weight) -- in practice the wing skin on typical gliders is not much more than minimum balanced layup and handling strength gauge -- with relatively thick wing sections and well stabilized skin -on the other hand the spar cap material goes up as the exponent of span ratio and the distance it must traverse goes up directly -- thus the spar bending material is the most significant fraction and most sensitive to AR . I once graphed the weight of Glasflugel aircraft from the 13.6 M Salto to the 22 Metre Glasflugel 604 ( a valid test of span given the same manufacturer,material,payload and technology ) -the 15M libelle and 17M kestrel plus 19M slinsgby Kestrel give a smooth graph that shows the exponential increase in wing mass --principally spar cap material.

Carbon lowers the weight but shouldn't invalidate the general principle -- back breaking rigging of 19m Kestrels (the largest two piece wing ) rams home the point. The SB 10 needed a trailer mounted winch for the centre section back in 74 --and lots of hands .

Getting some measured wing and other weights from something similar is the best reality check for initial design use.

#### wsimpso1

##### Super Moderator
Staff member
Log Member
Karoliina,

First things first - put aside Hollmann. His books are scary dangerous...

I use Timoshenko and Gere for mechanics of beams and such, and either Tsai and Hahn or Jones for composites. Abbott and von Doenhoff is the bible for developing the aero loads into the wing. The big thing I found out in grinding through my own work is that basic beam sizing (Timoshenko and Gere cover it as does Hollmann) results in beams that are undersized in both the caps and in the web. This is because they do not carry load independantly - the shear web carries the shear deformation and it moves with the caps, so its strain state at the caps must take that into account.

I set up a series of spreadsheets that are more than a little user hostile. It allows me to first determine shear, bending and torsion for the wing at a bunch of stations, then estimate the skin thickness, size the caps and shear web, for each of the stations. Enter a dose of reality - minimum external skin for surviving building, handling, airshow morons and bird strikes is probably 3 UNI or the equivalent, and that is usually way more than basic sizing indicates you need. 2 BID on the inside surface is probably a minimum there too. Once you set the skins at a minimum of these numbers, you may find your design work simplified.

I also set up a spreadsheet that gave me the Ixx, Iyy, and J of the skin with a standard skin, then varied it over the range of chord size in my bird, then fitted a curve to it using LINEST() in Excel. That allowed me to automate the contribution of the skins.

Then I set up the equations to calculate the Ixx, Iyy, and J for spar and skin as a unit and by the specific plies the structure. Yeah, Outer skin laminate, Inner skin laminate, Outer wrap top, bottom, and sides, top cap, bottom cap, shear web sides, shear web wrap onto caps.... With some careful layout of the system, and starting at the known skin outers, you can build up the structure, determine its centroid, and the other beam characteristics. Set up the ABBD matrix for the wing, and put in the loads, and solve it using Gaussian reduction. Yeah, I am a bit of a geek...

When you are done getting the solution strains and decompose them to lamina strain, then check the failure criteria, you can start beefing where necessary. If you beef the web (it is what will likely be showing failure), you will eventually get a strong enough spar. If instead you add some beef to the caps as well as to the web, you will get to strength at lower weight. Really.

If all of this sounds like a lot, well, it is a fair part of one undergraduate ME class and an entire semester of a graduate level class. Welcome to composites.

Some other hints. The skin shear due to torsion of the wing is significant, but so is the effect of moving air outside the wing and stationary air inside. It tires to pull the skin off the wing. I actually made use of Roark's to figure out how big the rib spacing could be, and decomposed it to make it applicable to sandwich composites. The key to reducing the number of ribs is using enough core thickness. In my bird, I used 3/8" PVC cores and a few ribs, but I might use 1/2" in the next design, if there ever is one. Realistically, you have to close the ends of the panels, the ends of the fuel tanks, mount the aileron bellcranks to something, and come up with some way to keep some fuel at the pick-up during a slip.That pretty well means four ribs each forward and five aft. Adding more core thickness than what is required to get to that is just added weight. Yeah, mine has more...

Enough for now.

Billski

##### Well-Known Member
I take issue with the statement that 'the majority, even for larger aspect ratios is in the skin' (weight) -- in practice the wing skin on typical gliders is not much more than minimum balanced layup and handling strength gauge -- with relatively thick wing sections and well stabilized skin -on the other hand the spar cap material goes up as the exponent of span ratio and the distance it must traverse goes up directly -- thus the spar bending material is the most significant fraction and most sensitive to AR . I once graphed the weight of Glasflugel aircraft from the 13.6 M Salto to the 22 Metre Glasflugel 604 ( a valid test of span given the same manufacturer,material,payload and technology ) -the 15M libelle and 17M kestrel plus 19M slinsgby Kestrel give a smooth graph that shows the exponential increase in wing mass --principally spar cap material.

Carbon lowers the weight but shouldn't invalidate the general principle -- back breaking rigging of 19m Kestrels (the largest two piece wing ) rams home the point. The SB 10 needed a trailer mounted winch for the centre section back in 74 --and lots of hands .

Getting some measured wing and other weights from something similar is the best reality check for initial design use.
Yes it does, completely.

I've found (and understood from those who do it for a living) that the things that define skin weight are:
*Stability/buckling, both from flight loads and for the "moron factor".
*Skin deformation in flight (aero loads, no longer laminar flow)
*Fuel/ballast deformation in flight. Even with sandwich skins this can be enough to deform your profile and ruin laminar flow (the Nimbus 2 is notorious for it).

Almost all need a "minimum thickness" of the skin, both the foam and the composite. Changing from glass to carbon will only allow a very small reduction in thickness, since buckling is often the driver. The spar on the other hand goes almost proportionally.

On most modern 18M (60 ft) gliders, typical spar weight is around 20% of the wing panel weight and that includes the overlapping spars in the fuselage, despite their aspect ratio of around 30.
So cutting the aspect ratio in half would only yield a 15% lighter wing panel weight one would think.

The results I got in fact showed that, when reducing aspect ratio within certain ranges the wing weight went up. Not what you'd expect, but my results were confirmed by other people. Since you increase the chord, the small weight reduction in spar weight is more than offset by the larger weight of the thicker wing skin, since your chord gets longer, you need more resistance to (elastic) buckling of the skin. Most certainly not a general outcome, but worth a thought or two.

Just some food for thought, a typical plastic glider with the same "total fuselage mass" and wing area as a piper cub has roughly the same wing weight, despite it's aspect ratio being 4 to 5 times higher...

This also shows the great benefit of computer code (being either an Excel spreadsheet or advanced programs), you get results you didn't expect (nobody would) and thus discover new things.

#### Rick McWilliams

##### Well-Known Member
What stress allowable do you choose for unidirectional graphite epoxy spar caps? 100Ksi tension? 50Ksi compression?
What shear allowable stress for 45degree glass epoxy shear web? 7Ksi?
Do the plies of the web need to be distributed between the plies of the caps? Can they be just half below the cap and half wrapping on top of the cap?
What bearing stress alowable glass epoxy alternating 45 & 90 BID pad for a bolted attachment? 32Ksi?

I have seen a composite LSA airplane with outer wing skins of just 1 layer of 10 oz BID glass inside and outside of 1/4" 4 lb/ft3 PVC core. The skins looked smooth after a couple of years of use. The skins weighed about 0.3 lb/ft2. Is there any lighter "minimum gauge" composite skin.

#### orion

##### R.I.P.
What stress allowable do you choose for unidirectional graphite epoxy spar caps? 100Ksi tension? 50Ksi compression?
What shear allowable stress for 45degree glass epoxy shear web? 7Ksi?
Do the plies of the web need to be distributed between the plies of the caps? Can they be just half below the cap and half wrapping on top of the cap?
What bearing stress alowable glass epoxy alternating 45 & 90 BID pad for a bolted attachment? 32Ksi?

I have seen a composite LSA airplane with outer wing skins of just 1 layer of 10 oz BID glass inside and outside of 1/4" 4 lb/ft3 PVC core. The skins looked smooth after a couple of years of use. The skins weighed about 0.3 lb/ft2. Is there any lighter "minimum gauge" composite skin.
Like alloys of metal, composite physical property values will depend on a number of variables which unfortunately prevents any attempts at generalization. Values will depend on the fiber, the epoxy, the method of application, the quality of application, the cure and environmental issues. Given the number of variables, the property values can represent a significant range, which leaves the designer with only two options: First, use only those materials with a reputable database and second, do your own testing with the materials and techniques that you plan on using. The only reputable database for graphite and glass materials is that which was developed under the NASA AGATE program but in using those values it is important to select only those materials that the tests represent.

Personally though, I think that in using composites the designer should not only use only verifiable property databases but also get away from metal type structures such as those that are represented by the use of concentrated spar configurations and use structural configurations that can not only take advantage of the materials in question but also work at developing simpler, more redundant structures.

##### Well-Known Member
I have seen a composite LSA airplane with outer wing skins of just 1 layer of 10 oz BID glass inside and outside of 1/4" 4 lb/ft3 PVC core. The skins looked smooth after a couple of years of use. The skins weighed about 0.3 lb/ft2. Is there any lighter "minimum gauge" composite skin.
Judging against the Nimbus 2 (or similar aircraft), that skin thickness is far into the "ruin laminar flow" territory. No worries if you're flying very slow (low aero loads) and don't carry ballast/fuel in the wing. No worries at all if you don't care about laminar flow, such a very thin laminate can be extremely thin if you have the aerodynamics of a Piper Cub. If I recall correctly, a monocoque skin of around 10 oz BID is used in several composite ultralights. It warps and deforms a lot, but it flies just fine.

#### Rick McWilliams

##### Well-Known Member
I missed the memo: Spar caps and shear webs are out of style to carry wing bending loads? Bolted or pinned connection through a shear web is passe? If a structure is redundant, how can it be minimum weight?

The composite LSA had performance as if it maintained laminar flow. There was no visible tension field buckling at any load condition that I flew.

#### Rom

##### Well-Known Member
I missed the memo: Spar caps and shear webs are out of style to carry wing bending loads? Bolted or pinned connection through a shear web is passe? If a structure is redundant, how can it be minimum weight?...
I don't think Orion was implying wing spars are obsolete. With metal construction the load is transferred from the skin to the ribs to the spar. With composite the wing skin and spar become a monolithic structure. The whole structure is supporting the load.

#### Voyeurger

##### Well-Known Member
Billski said,"Yeah, I am a bit of a geek..."

A bit of a geek? You sir, are a total geek (from my outlook). And sir, I admire you greatly.

You, Orion and Auto have certainly filled Karolina's plate for the near future.
Best,
Gary

##### Well-Known Member
The composite LSA had performance as if it maintained laminar flow. There was no visible tension field buckling at any load condition that I flew.
I meant the "inflation" of the D-nose since you have a huge pressure difference between the inside and outside of the skin. I for example am looking at roughly a tonne per square yard which is plenty to inflate even a very thin sandwich, let alone a single layer. Dick Johnson concluded that the few hundred liters in the N2 wings were enough for a sag that changed airflow enough to significantly reduce L/D. We're not talking huge movements, but even less than a millimeter is enough to significantly reduce laminar flow.

To be honest I first only designed my D-nose for the fuel loads (during a few G's) and the total (lift+drag) loads, limiting their deflection. Only after Billski's remarks I checked for the Cp distribution, with pretty worrying outcomes. Structurally it wasn't a problem, but those loads created half a promille of "skin warp", enough to trigger turbulent flow.

#### Rick McWilliams

##### Well-Known Member
Inflation of the D section? A Tonne per squre yard? you must be going very fast. Where does the internal section vent to the outside? I would choose a location on the bottom surface about 75%c. The aileron pushrod hole would seem like a fine vent. At high CL the suction peak near the nose would apply high forces, but the same suction peak would be the end of laminar flow.

I would expect rib spacing to be important to maintain local airfoil geometry. I am sure that there is a rib spacing and wing skin core tradeoff. The skins were 1 layer of 10 oz glass epoxy on both sides of a 1/4 inch 3.7 lb/ft2 PVC foam. Not just 0.010" of glass epoxy.

#### wsimpso1

##### Super Moderator
Staff member
Log Member
Allowable stresses? This is not metal.

In composites, any given structure sees the deformation of the whole structure, and when you get done decomposing to the local strains and checking the strain based failure criteria, which boils down all four strains through five (or six) failure points for your unique situation, you finally know if that ply failed or not. IF you force yourself into using "allowables" you either will expose yourself to surprises in the form of failing plies or you will have to set the allowables so low to always protect your design that your design will be overweight, sometimes by large percentages. Since these are airplanes, you can not afford that luxury.

Now, that being said, I do use book values of strength of my materials for starting points in my design work. 70 ksi for tension and compression of UNI Tape in spar caps, and 24 ksi for BID used in skins. But as soon as I have an initial size, I can dispose of those numbers and rely upon the failure criteria as I iterate the number of plies (and core thicknesses where applicable) in each laminate.

And then, I never did get into buckling... A whole 'nother discussion.

Billski

#### Rick McWilliams

##### Well-Known Member
It would seem to me that a spar in bending made with caps and shear web loads the composite materials in a simple manner. The caps are primarily in tension and compression with modest interlaminar shear. The web is primarily in shear. Spar caps almost always use unidirectional graphite in epoxy. Shear webs often use fiberglass epoxy with a 45 degree fiber direction. The applicable failure criteria are thus reduced. So a conceptual isotropic stress allowable might be useful.

Bolted attachments are always troublesome when joining to composite materials. How do you make a load distributing pad for these loads? The wing attach bolts that attach a one piece wing to the fuselage comes to mind. I would locally replace the foam shear web core with G10 fiberglass, and build up a pad area with additional 90 & 45 plies co-cured with the web. I would use a bearing stress allowable of 32Ksi. I hope for a laminating resin bond strength to the G10 of 2Ksi. How would you design a bolted attachment to a shear web?

#### Tom Nalevanko

##### Well-Known Member
Rick,
Your approach sounds reasonable to me. A lot of current designs use a phenolic insert co-cured into the web to take the loads of the attaching bolt. It seems that in these types of designs, less bigger fasteners are better than more smaller fasteners because of tear-out stresses, etc..
Blue skies,
Tom

#### Rom

##### Well-Known Member
I was also wondering about the common approach in applying hard points in the spar. If the phenolics typicaly run from cap to cap or or are they just at the bolt connections with a large enough bond area with the shear web. If sufficient bolt depth and size through the hard point, I would think a lighter weight phenolic other than G10 could be used. I plan on using a metal sleeve through the hard point since composites typically don not like point loads.

#### orion

##### R.I.P.
Allowable stresses? This is not metal....
No, it's not however given appropriate limitations and a proper database, the use of strength properties is as valid as the strain criteria. In our work I start with a first cut using conventional methods, the results of which are compared to previous work or other examples. A second analysis is then done using the Scan-n-Solve analysis software that resides within Rhino. This however has significant limitations due to its assumption of isotropic materials but it can identify potential areas of stress concentrations or other problems. And this is then followed by the FEA modeling, analysis and optimization. Since the latter analyzes failure as a function of strain, given proper input of properties, interpretation of results as a function of stress or strain can be interchangeable. But again, this requires a proper database of the particular material one is using - guessing or eyeballing these inputs can result in substantially under-designed structures.