Monocoque Chassis FEA

[BenB]

To make my monocoque model I made a solid body that is shape I want the car to be. I then used the shell feature with the thickness of my carbon layers + the honeycomb. This is an easy way to represent the monocoque for modeling, but it doesn't work so well for FEA.

How do you do FEA analysis on a monocoque? I couldn't imagine its efficient to model the inside layers, outside layers, and core separately and then do the analsis on an assembly. I am using mechanica. Maybe Ansys has some tools that make this pretty easy, but I'm pretty sure mechanica does not.

In general I was wondering if somebody could give an overview of how they do composites analysis? This would be a big help.

[BenB]

To make my monocoque model I made a solid body that is shape I want the car to be. I then used the shell feature with the thickness of my carbon layers + the honeycomb. This is an easy way to represent the monocoque for modeling, but it doesn't work so well for FEA.

How do you do FEA analysis on a monocoque? I couldn't imagine its efficient to model the inside layers, outside layers, and core separately and then do the analsis on an assembly. I am using mechanica. Maybe Ansys has some tools that make this pretty easy, but I'm pretty sure mechanica does not.

In general I was wondering if somebody could give an overview of how they do composites analysis? This would be a big help.

[ad]

Ben,

What FEA package are you using?

In ANSYS (Workbench or Classic) you can model composites using a thin shell model (ie constant thickness) and layered section properties. You can also apply ply drop off if desired.

However this is not trivial to implement, especially in Workbench as the ANSYS Classic commands (or APDL) need to be imported and placed within the correct processor using the 'command object'

Heres some hints;
- Using layered shell elements you can build layers on the modeled shell body and specify the offset criterion (ie. inside, outside, midplane etc) the fibre orientation with respect to principle orientation (ie layup) what matid (ie which type of fibre) and thickness and the number of integration points for the through thickness of each layer.

-The next part is to define a local coordinate system for each 'face' to represent the principle axis of the fibre, and then align the material.

-Define orthotropic material data, linear or nonlinear with a specified failure theory to represent failure between layers.


This is a very time consuming, iterative process especially in Workbench as it is not officially supported until later releases. ANSYS Classic has a help section on modeling composites and layered sections.

Hope this helps.

[BenB]

I'm using pro-mechanica...it is the FEA package that is integrated into pro-e. I don't think its has the type of analysis tools that you are referring to in ANSYS, but thanks for the information. I will look in the help for the topics you listed.

Has anybody else done this type of analysis? It must be pretty common for all the teams that do monocoques.

Since the outter plies are the most stressed in tension and compression, and provide the most stiffness, how good (or bad) of an approximation would it be to just forget about the core and just run the FEA on the model that I described above?

[Jersey Tom]

Getting composite layup FEA to match up with reality is very, very difficult. The teams that do it well have people who do masters and phd work on this specifically. If you don't get it dialed spot on you can be off by an order of magnitude.

[TG]

To get a stiffness driven composites design that has in plane loadings is not that difficult. Where you will run into difficulties is with strength driven designs (predicting failure is very, very hard as there are many different ways for composites to fail) and out of plane loadings (esp. with tri-axially woven composites, like non-crimp fabric, plus there aren't really any good standardized tests for out of plane loading). So as long as a design for a monocoque doesn't require too much out of plane loadings (the areas of suspension attachment come to mind here) then you should be ok with getting a decent prediction of the stiffness of your structure.

Andrew, are there any good tutorials for these functionalities? I took one a while back, but it just was basically a plate with no ply drop-offs and no curvature and I don't recall having to select principle axes at all. Basically it just had you select an orthagonal structure and input the ply orientation, nothing that wasn't trivial. Also, which shell element do you use? I know there are a number of different shells that support layered and orthagonal properties with various limitations to each.

[ad]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TG:
To get a stiffness driven composites design that has in plane loadings is not that difficult. Where you will run into difficulties is with strength driven designs (predicting failure is very, very hard as there are many different ways for composites to fail) and out of plane loadings (esp. with tri-axially woven composites, like non-crimp fabric, plus there aren't really any good standardized tests for out of plane loading). So as long as a design for a monocoque doesn't require too much out of plane loadings (the areas of suspension attachment come to mind here) then you should be ok with getting a decent prediction of the stiffness of your structure.

Andrew, are there any good tutorials for these functionalities? I took one a while back, but it just was basically a plate with no ply drop-offs and no curvature and I don't recall having to select principle axes at all. Basically it just had you select an orthagonal structure and input the ply orientation, nothing that wasn't trivial. Also, which shell element do you use? I know there are a number of different shells that support layered and orthagonal properties with various limitations to each. </div></BLOCKQUOTE>

Hi TG,

I only know of some internal training material that ANSYS circulates to their customers, but there may be tutorials on their website which you are free to paruse.

Actually saying that it is a trivial act to implement is somewhat inadequate. It is if you are manually specifying the composite data within ANSYS. There are a few third party programs like LaminateTOOLS which make this very simple. And include manufacturing methods such as draping effects on fibre warping and fibre orientation etc.

Its been a while since ive used composites in ANSYS but from memory it was SHELL180 or 181. But it ultimatly depends on which element matches your problem.

The reason for the local coordinate system adpation was that i made some effort to adjust fibre principle axis orientation with that of the actual layup. Referencing the global coordinate system for fibre orientation is fine for planar surfaces. Such as the example you mentioned.

We used actual material test data using the resin, layup and manufacturing process to represent the failure theory and orthogonal properties of the material.

[Rachel]

I used NEIWorks which is an add-in for SolidWorks and runs off of NEI Nastran. Just as others have said, I selected the outer surface of the monocoque and then defined a laminate cross section. Defining the laminate was fairly easy as it pops up a window where you add plys and honeycomb all at once, select ply orientations, and failure theory. Understanding the analysis was a whole other story.

After knowing more on the subject, I wish I could do it all over again. I believe the proper way to start is to get some hand calculations going on a sandwich panel in bending. Hopefully these will match closely with your FEA and give you some assurance. The ideal thing to do is perform some actual tests on your laminate and see if these results match your FEA. If the results do not match, changes should be made to the defined properties in the program so they match the physical testing. Most of the data you find on your materials will be ideal values. Composites are finicky in that their quality depends on many things such as lay-up technique and the expertise of the person doing the lay-up. Another example is that our prepreg was very expired. It worked great but the physical properties of the material no longer matched the ideal ones. It would have been nice to have some physical testing done to fine tune the analysis.

After you have your FEA and testing and/or calculations matching...good luck applying that to a more complicated surface.

I just recently came across a program (CompositePro) that basically does the hand calculations for you, using Classical Laminate Theory. It seems like a simple program but has a lot of function to it. The handy use I found for it is to define a laminate and have it pop out the 2D and 3D properties. I then took this into CosmosWorks and defined an orthotropic material (I'm not sure if the student version of Cosmos will do orthotropics). Again using a simple panel in bending, the numbers from Cosmos and Compro matched within 4%.

Ben, you should definitely include your honeycomb core in your analysis. Your chassis stiffness will be all about honeycomb...or more specifically, the thickness of the honeycomb.

As far as which program to work with... Altair Engineering provides a free copy of their Hyperworks program to all FSAE schools. I haven't used this program before but "free" sounds like it's worth checking out.
http://www.altair.com/corporat/press/20060711.htm

[Steve Yao]

As Rachel pointed out. There is basically zero chance of having accurate analysis without having done your own testing to find the material values. There are simply too many variables in fiber manufacturing, matrix materials, layup processes, individual technique, curing processes etc. Then you get into the multitude of core materials available and how they are bonded within the laminate...

[BenB]

Great information thanks a lot.

[TG]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by SEY:
As Rachel pointed out. There is basically zero chance of having accurate analysis without having done your own testing to find the material values. There are simply too many variables in fiber manufacturing, matrix materials, layup processes, individual technique, curing processes etc. Then you get into the multitude of core materials available and how they are bonded within the laminate... </div></BLOCKQUOTE>

Yes, but you can still design to B-basis, 95th percentile of the 90th percentile, (or A-basis, 95th percentile of the 99th percentile, if you wanted to be overkill in the design) from the Mil handbook and you probably should come up with actual values that exceed these. The whole idea of the A- and B-basis values are to account for variability in the manufacturing of the parts and provide a minimum design value. If you wanted to make the most absolute light parts (we're talking factions of ounces here) then yes, you would go take sample coupons of which you would test. But then you must also account for loss of stiffness and strength due to moisture absorption and how are you going to do that? A boil test? Boil tests are only partially accurate but not completely accurate as they will not give you the same ammount of moisture absorbed in a material in 24 hours as a part can have in a real world environment for even a month.

So the fact of the matter is you will NEVER have entirely accurate data to work from, but there is data that will give you a very good statistical probability that the properties will be at least equal to, if not exceeding, of the ones listed.

[Steve Yao]

TG,

Could you give me specific references. I had not heard of this "A-Basis" and "B-Basis" composite design specs?

I have a hard time believing any handbook spec is keeping up with the constant development of every composite fiber, matrix, and processing method, but I have not seen it so I will refrain from commenting further.

[TG]

Mil Handbook is great! There are several different handbooks covering metals, plastics, ceramics, composites, etc. with all of the tested properties listed. Sure Matweb is great for convenience, but Mil Handbook is more complete with material descriptions, etc. I know its public domain info but I access it online through my schools library.

The guy who taught my composites manufacturing courses last semester works at the Boeing plant here in Mesa and is also a contributor to Mil Handbook 17 (the composites one). This is how he explains the A- and B-basis to me (it's all design of experiments, which I haven't taken yet):

There are many contributors to the handbook and they will each take some material and cure it. From the material, they will take a number of coupons to test and these will all be part of a single "batch". Each person doing this is in a different environment than another (elevation, temperature, humidity, etc.). When all of the batches are combined, they will have statistical significance and form a bell curve in which each property will fall in. The value reported in the A-basis will be the 95th percentile mark of the batch (95% of the coupons tested in the batch will be above this mark) of the 99th percentile mark for all of the batches (99% of the batches will fall above this batch). The B-basis is the same except it is the 90th percentile instead of the 99th percentile.

You will see ratings various temperatures and moisture absorption, like cold temperature dry, room temperature dry, elevated temperature dry, and elevated temperature wet. Elevated temperature wet is generally, but not necessarily, the worst. Also, properties trend towards being the same at colder temperatures, regardless of moisture levels... again, this is in general but not necessarily always true. When designing you should always pick the worst as you will most probably encounter all of them. If you do know the operating environment and it excludes the worst condition, you can disregard those values, but I'm in Arizona and I expect I will encounter all of them. My professor told me the only time he has seen someone allowed by the government to omit values like this is with the B2 bomber where in some areas of the fuselage, the composite skin is around 1 foot thick and it was proven that it was impossible for the laminates in it to become saturated with moisture. The thickness that the FSAE monocoques are are probably at most 1/1000th of that thickness.

The FAA will only allow for structural items in an airframe to be analyzed using A-basis properties. They will allow for B-basis to be used in non-structural parts to reduce weight. I really don't see any reason for a FSAE car to use A-basis as it will just be overweight in a car that has a stiffness driven design and has more than enough strength already.

One thing I will note is that Boeing does its own material characterizations. They will thoroughly test materials from manufacturers and create certain parameters that they may test coupons from incoming material shipments and hold the manufacturer liable through contract to maintain these characteristics. My prof said the last he heard of the price to characterize a material was a million bucks. With that being the case, Boeing limits its designers to certain materials that have be characterized and it must be shown economically that a new material will yield the cost to characterize it in cost savings in design or manufacturing. FSAE teams obviously don't have a million dollars to throw at characterizing a material and should be able to get away with using the values in the Mil Handbook.

[TG]

I forgot to mention that the tests are all ASTM standardized tests to assure consistency in testing procedures, which are extremely important in testing composites as the slightest variance can give dramatically different results.