Composite Plate Design

[CoyoteBros]

I own a very small sporting goods company that makes small part for elite athletes. I can honestly say that the product is a hit but I am doing a lot of my R&D blindly. The parts are cut from plate is made to my spec. I need to find a way to more intelligently adjust my lay-ups to achieve the required flex characteristics. I know what I need but I just do not know how to get there. With my current development method I can spend thousands and lose 2 months with every experiment.


A brief history:
The original product was made from a 17 layer prepreg lay-up. 9 layers at 0 degrees and 8 at 90 degrees. This lay-up came in at about 0.102” thick. The whose who in the sport loved it and but demanded that I lose about .005” in thickness. Since then I have been rearranging the plies on various 16 layer constructions trying to get the stiffness back to where it belongs. The only obvious thing I have not tried is to go to a higher modulus fiber.

I am looking any of the following.
A) a high quality vender who will make me a few very small batches of plate with various fibers and lay-ups.
B) Computer modeling for predicting stiffness of a specific lay-up vs. another
C) Any good advise from someone who understands my problem.


I know this forum is not really intended for non-engineers but none of the engineers I know seem to be able to help.

Thanks a ton


John
Coyote Brothers Research

 

[gwolf]

The tool which you need to predict stiffness of laminates is called "Finite Element Analysis". This type of software is quite expensive (many $thousands/year) and is hard to use.

This looks like a job for a small engineering consultancy, they will be able to quickly simulate different layups. There are a few who post on this site, just wait and watch this post fill up with replies.

 

[RPstress]

You don't necessarily need FE for relatively simple stuff, such as stresses and deflections of beams and plates, possibly like basic analysis of skis or snowboards.

If you've got a simple criterion to meet for a laminated plate then an engineer could probably easily calculate the relevant properties using some laminate analysis software for equivalent plate properties and a copy of Roark. This should get you in the right street, from which you could do some practical stuff to refine the laminate.

For cheap/free laminate analysis software try

http://www.thelaminator.net/

(30 bucks)

http://composite.about.com/od/swlamanalysis/

Google laminate analysis for more.

Are you able to tell us what your criterion is, i.e., how you're testing the laminates that you make? No need for actual numbers if it's commercially sensitive.

What fibers are you using from what supplier? It might be as simple as switching to intermediate modulus.

-R.

 

[CoyoteBros]

I have looked at some of that software and to be honest I do not know enough to use it.

My manufacture uses Toray T700G. I assume that is a commodity fiber. It is all unidirectional and the math tells me that I am at .006 per layer. I am using MQC and feel like what I get from them is pretty consistent. It is really hard to find a supplier who will work in my small volumes and still return a good product.

If I knew where I could order a few square feet of test material at a time I would go crazy with exploring the options. On the other hand I would rather be able to predict the material property and get back to the tasks at hand.

The original 17 ply material was layered up Zero / Ninety – alternating layers (Is Isotropic the right term for that? ) When I originally went from the 17 ply to the 16 ply material I found that I could regain some of the needed stiffness by reordering the plies. Since the final part is clamped along one edge in it’s final application I was able to take away plies that run parallel to the clamp and add them back perpendicular to the clamp to a net increase in stiffness. I understand that middle plies do very little so I have moved the additional perpendicular (to the clamp) layers out to the skin of the plate.

More recently I thought that since the flex happens at a 45 from the perpendicular (to the clamp) layers I would cut the parts so the most fiber would be angled out to the greatest flex point. This seems to have been a failed idea. I had greater flex. I assume that even though the stress was directed in the primary direction of the fiber the fact that the length of the fiber was longer resulted in greater flex. Below is a link to a picture of a first generation part. The clamp attaches all the way across the top and the greatest point of flex is at the second big hole down on the right

Here is a photo of the part coyote_brothers_research.com/photopage/blackfin.gif (remove the “_” from the URL)


Examples of my lay-ups can be found at coyote_brothers_research.com/ply.xls (remove the “_” from the URL)

As for testing, I only have a method for testing the final part. The part flexes at a 45 from the primary bias. My testing simulates the flex as it would be used in the real world but does not test the plate in an unfinished form. I know that my fins have been tested by others who found surprising results. They tested on the Zero and the Ninety and were baffled by the results Since the product flexes at a 45 in use that is the only way to rate it.

I am currently thinking about upgrading to intermediate modulus fiber for some or all of the plies but feel like this is blind experimentation. If it does not yield the needed flex I am out a ton of $ and will have waited 6 to 8 weeks for not improvement.

Note I added a “_” to all the web links so Google would not send my customers here. I hope this forum is obscure enough so …. Heck if my industry wanted to know my lay-up they would have done so by now….

 

[RPstress]

Because you're testing the final part it makes it hard to do simple analysis which would give realistic deflections for different layups. However, some general trends should still be true: putting fibers in line with the biggest stretching and compression should reduce bending deflection and making them at 90 degrees to that should increase deflection. (As a quick note, the 0/90 layup is a cross-ply, and is orthotropic; the stiffness at 45 degrees is very much less than the stiffness at 0 and 90 degrees. To approximate isotropic (more or less as a metal behaves) you would need to use 0/90/+45/-45.)

You did what should have been a sensible thing in putting in fibers at about 45 degrees (I assume the fibers at the top and bottom of the laminate ran approximately parallel to a line from the square hole in the clamped area to the second big round hole down). Did the max deflection still occur at the same place? How are you loading the fin? Do you just hang a weight from one of the holes? If so, which one?

Generally, the maximum curvature of the laminate should be occurring towards the clamped edge, where it should also be being stretched and compressed at the surface the most when bending; this means that thickening the laminate towards the clamp should decrease deflection most efficiently, pretty much like a shark's fin.

T700 should be a fairly premium grade of fiber from Toray, though I'm not sure what the 'G' means (doesn't appear on their website). It is high strength, but standard stiffness. Intermediate modulus (Toray T800, Hexcel IM7 (6000), Toho-Tenax IMS) should increase stiffness by about fifteen percent in bending with more-or-less the same high strength. This ought to equate to a decrease in deflection of very nearly the same percentage. That might be enough to replace the loss of bending stiffness. [(.105/.1)^3 ~= 1.16 factor increase in bending deflection for a thickness reduction from .105" to .1", though it's not quite thickness cubed for a laminate like yours.]

To get a feel for structures and materials, see professor Gordon's books, "The New Science of Strong Materials, or Why You Don't Fall Through the Floor" and "Structures, or Why Things Don't Fall Down".

NB:

http://www.coyotebrothersresearch.com/ply.xls

couldn't be found.

 

[opse0]

If I were you I wouldn't go thru the finite analysis method.
You probobly have a lot of experience from the tests you've done, as well the feel of what you need.
What process are you using to lay up & cure the laminate ?
what type of tooling ?
Would it be possible to give a "shut height" to your tooing?
Are you counting on bag pressure only ?
Are you using autoclave pressure?

You may be able to get the required thickness with a 17 ply thickness by applying more pressure on the laminate.
Example, if you were using a 2 part tooling process where the male tool would be aluminum, & the female tool steel, the CTE of aluminum when properly calculated would give you the thickness you want.

 

[CoyoteBros]

RPstress , Opse0 & all

Hey thanks for the replies.

I pulled the web page with the lay-ups down in the last week. I have gotten some very informed help since I first posted here and did not want to expose more of my semi-confidential details to the world longer then I had to. I was hoping that my actual URL would not end up here….

As for the layers Rstress’s assumptions about lay-up, you are correct. The revelation that my plate is base is only stiff in the exact direction that the fibers run should have been obvious but was not clear until recently. My new adviser has opened my eyes to what you guys take for granted. I always thought that the 0 and the 90 supported each other pretty well and the 45 would just fall in line….. according to you guys I was way off base.

I am looking at the T800 option but am nervous about the cost of these experiments. I am not sure if my plate manufacturer is using a hydraulic press or a vacuum but the consistence is great. Since I am on a shoe string budget consistency means as much as a theoretical target. I believe that they are using a “Shut Height” of some sort. The whole plate is with in a thou or two across a 4’ x 5’ plate. To me that is pretty dare good.

On your opse0’s last comment it is my understanding that thickness is a huge % of total stiffness so the same amount of fiber with more resin squeezed out would not replicate the 17 layer construction. Machineability is also an issue. My experience with low resin plate is that my machinist had a lot of trouble with it.

It is my hope that with everything being the same my new adviser (new best friend) can offer enough calculated guesses on my needs to solve the issue.

Thanks to all you guys for giving this some thought. If anyone wants less vague information about what I am doing drop me from the Coyote_Brothers_Research.com web site. (Remove the "_")