Best Stragegy for modeling composites

[phoenix221]

Hello,

I have recently switched to Solidworks+SurfaceWorks so please bear with me...

I am looking at a project for modelling an airframe made of glass fiber composite materials. I was wondering what the best strategy is for this project?

If you have experience modelling composite shell structures in SolidWorks, please pass on some of your tips and best practices. If you know of a good tutorial that focuses on composite/shell structure modelling in Solidworks, please point the way.

Thanks in advance

 

[G1DESIGN]

do you want to get preform and pattern information from this model

 

[phoenix221]

G1DESIGN,

The project will use moldless construction techniques. I will have to use the model for FEA.

 

[CorBlimeyLimey]

Excuse my ignorance, I am not familiar with your field, but what difference would the material make in creating a model? I assume you would use thickened surfaces & if need be, create multiple layers. Are you laying up the composites over a framework?
Sorry for the questions, but this piqued my curiosity.

from (the City of) Barrie, Ontario.

What happens if you get scared half to death twice?

 

[phoenix221]

CorBlimeyLimey,

There is no framework... the entire thing is one giant shell structure with local reinforcements .

I think one of the issues will be the way the joints are modeled... The "sheet metal" feature allows for quick modeling of panels, joining them properly is a whole other matter...

Firberglass panels when joined cannot be welded... they in effect are "glued" together by applying additional plies of fiberglass over the joint.

Since I will be using this model for FEA, it would be important for my SolidWorks model to be "FEA friendly", i.e. to be as accurate as possible, otherwise my FEA analysis will be flawed.

I am sure my current perception of the issues is incomplete however that is the reason why I am posting

 

[macPT]

phoenix221

I'm also curious about your problem, but lacking experience in this field.

Using SW, I think the best you can get is as Cor said: use thiken surfaces and ofset surfaces to create layers. But I don't know if the generated geometry is acurate enough for FEA.

What FEA are you using? Some FEA programs have special solvers for these cases. If you are using one of these, try to adjust your modeling technics to the input requirements of the FEA.

We can say that SW is FEA friendly if you create a part made of steel, aluminium, plastic,..., that is materials that have uniform mechanical properties (at least in each direction).

Regards

Regards

 

[G1DESIGN]

most fea programs will import solidworks data

as stated above a lot of what you export will be dictated by how you wish to analyse the structure
model the fuselage in sw but ensure you have the scope to model each laminate layer as an offset surface to represent each ply of material you will lay up

when you export this model you will then have to decide how to analyse the model in shells as per each individual ply or as a single structure a lot depends on which program you are using some will allow you to combine each layer to make a hole others have ways of defining the properties of the laminate as a complete structure if it was me i would decide how you want to do your analysis at the modeling stage

as for joining and the lap joins you mention this can be an entire topic of analysis on its own

i will try and get some info from a company whodo our fea they recently did a monocoque car and showed me how they did each laminate ply

ill find out more

 

[cpretty]

phoenix221,

I think the hardest part of this will be getting some of the effects of anisotropic behaviour of the material. Unless it is made of chopstrand, then your layup will affect the results greatly.

Possibly you will need to create each layer of material as a separate part, assemble it, then assign materials to each part to get the axes for the material aligned correctly. You will probably also need to define the interface between the 2 parts to simulate the matrix material.

I would start from simple models first to check that you can create lay-ups that perform as theory predicts. Examples of these would be multilayer systems that have assymetrical layups. When loaded in a beam type setup, you should be able to create twist even though you have an on centre load.

Work up from there. Your model is going to be entirely dictated by the constraints needed for the FEA to work properly.

Remember that FEA is a tool, not an answer. If you put rubbish in, you will get extremely pretty and convincing rubbish out. It is quite difficult not to put rubbish in.

Craig

 

[cpretty]

Another issue with the FEA. Almost all FEA does not work for brittle materials (I would think glass fibre is brittle). If you are looking at ultimate strength, then be very wary of results.

 

[tstanley]

To expand on what I said in the Computer aided Engineering thread, I used Solidworks to model a composite part that was something like a fuselage shell. I didn't have surface works so I modelled it with thickness exported it as a parasolid I imported it into Nastran 4W and tried to use the mid-surface command to create the surface, but it was to complex with corners etc. so just the outer surface was used. Probably you will to better with surfaces.

Once you get the surfaces, Nastran's laminate elements can take care of the various plies and thicknesses etc. No need to make multiple surfaces. On the down side, Nastran doesn't have the draping capability that was mentioned in the other thread, at least not yet.

Tom S.

 

[phoenix221]

>What FEA are you using?

I will be using Lusas... however I am still learning Lusas so I need a little help in determining the modeling strategy.

>Possibly you will need to create each layer of material as a
>separate part, assemble it, then assign materials to each part
>to get the axes for the material aligned correctly.

This sound like a lot of work. There has to be a better way

>a lot depends on which program you are using some will allow
>you to combine each layer to make a hole others have ways of
>defining the properties of the laminate as a complete
>structure

So far I tried the approach where each larger composite laminate is modeled as a solid with a certain thickness. My thinking is that as long as I can specify the anisotropic material properties, based on # of plies, etc., the FEA software should correctly handle the panels. Whith this approach the joints become a challenge! I am not sure that the joints will not be a challenge regardless of the approach ... considering that the structure is most likely to fail around joints, due to delamination, etc., I would like to get that part right!!!

What say you?...

 

[G1DESIGN]

have you checked lusas website

i used to work for a company called europa aviation who did exactly what you are thinking of doing

i do recall the fueselage was split and they had similar concerns on the lap join perhaps you can drop them a mail as to how they solved this problem they are pretty freindly and support most home builders of kit planes(not that im saying this is what you are doing)i also know they brought in one of the bigwigs from BAE

>Possibly you will need to create each layer of material as a
>separate part, assemble it, then assign materials to each part
>to get the axes for the material aligned correctly

enginuity fea uk used exactly this approach on the monocoque car they recently showed us dunno why, they are good at what they do so this may be the best approach

dunno if this helps but in my experience of carbon composites its more a mother of invention approach like get the basics right and you will have no problems

 

[phoenix221]

>>Possibly you will need to create each layer of material as a
>>separate part, assemble it, then assign materials to each part
>>to get the axes for the material aligned correctly
>
>enginuity fea uk used exactly this approach on the monocoque
>car they recently showed us dunno why, they are good at what
>they do so this may be the best approach

This approach is a HUGE amount of work! Some actual parts have 50 to 60 layers ... you still have to align each part (now layer) perfectly!!! There is no way I can follow this approach... I don't have an unlimited reasearch budget :-(

>...in my experience of carbon composites its more a mother
>of invention approach like get the basics right and you will
>have no problems

That sounds promising... what are the basics that must be right?

 

[phoenix221]

>have you checked lusas website

I actually did again the claim is made that "...lay-ups independent of the component to be analyzed...". I'll have to look into this further to find out just exactly what this means

The software seems to be quite complex and it seems that you need to define each layup individually. While this allows non-linear analysis, it also involves a large amount of work.

I am wondering whether I need that amount of precision? Linear analysis would allow me to treat laminate properties integrated together as if forming an homogeneous material matrix. This seems to me less work, since depending on the number of plies, thickness of core, etc, the composite component is treated as a solid otherwise... i think this is much easier to model...

If anyone out there has a different take on this, please chime in