What am I doing wrong???

General Composites Discussions Composites Talk
#1

As of about 2 months ago I had never attempted to make any composite pieces. Recently my job has had me making samples for destructive testing. I am having a really hard time making consistent samples, or even controlling the variability within a given sample.

Let me step you through my process, so you can point out where I may be causing the problems.

Starting with 2 waxed steel plates. I am using 8.7 oz plain weave 6" wide e-glass, cut to 8" long per sample.

Weighed out 20 gr of resin w/ catalyst, and poured onto the glass. Worked the resin around and in with the tongue depressor.

Placed second plate on top and secured with 4 c-clamps. This squeezes out the excess resin. I typically end up with fairly consistent thickness using this method. (.0115"-.0125"). The plates are put into an oven for 30 min. at 100C to cure. Pulled out and left to cool to room temp before breaking the plates apart.

The 6"x8" sample is cut into 1/2" wide pieces and then broke on the tensile tester. A breaking force, elongation, width, and thickness is taken from each sample.

My problem is that I am getting a very wide range of numbers for the breaking strength. The average is around 220 lbs, but the standard deviation is anywhere from 7 to 21, which is not enough resolution to detect any changes we make to the fibers with our coatings and additives.

Things we have tried that did not result in better/worse spread in data.
*de-aeration of resin after mixing
*Vacuum bagging, and use of roller to get rid of air bubbles/voids
*Mixing methods for resin
*Application method of resin onto the fiber
*Wet out time (time before I put it in the oven to cure)

Let me know if anything needs more explanation or detail.

#2

You have no control over vf in relation to pressure.

For simplicity you may try to use instead of clamps…bolts and apply different torque values to the sandwich plates. You should start to see some degree of consistency. You should also be able to convert applied torque to a surface pressure the laminate will see with a little math. as well as test laminate characteristics at different applied torque values.

You can also calculate the laminate vf by weight to aid in your tests. I think a more consistent vf and a more consistent thickness will yield better results.

Hope this helps.

#3

Using bolt torque may get you ±25% because friction bolt thread pitch and other things come into play… Depending on how precise your tests need to be… However there is another way to get a precise pressure force on the laminate and it be simple.

You can use springs and bolts and compress the springs to a certain length for their designed load. Springs and bolts are cheap and very easy to get.
For simplicity’s sake you could order a 2.5, 5, and 10 lb engineered springs to get a precise control on force.

Or approximate clamp force with bolt torque as mentioned above.

Hope this helps.

#4

I’m also assuming fiber orientation is consistent. Other than that… I’m out of Ideas.

#5

How accurately are you laying and cutting the fabric? Are the fibers skewed, wavy, etc? If the fibers are misaligned you could get some significant deviations.

I would also calculate the Vf using the weight of the fiber before and after saturation. Using the specific gravity of the fiber and resin you can calculate the theoretical Vf. You can also calculate the theoretical thickness for the calculated Vf. If you are seen differences between the thorectical thickness at the calculated Vf and actual thickness then you know that you voids within the laminate.

Adam

#6

VF has been calculated on several samples. Raw is 8.7 oz/yd2 which I verified. I end up with a .59-.63 VF consistently. Is +/- 2% consistent enough?

The idea for using die springs to provide a consistent repeatable load is one I may have to try soon.

I cut the fabric as close to perpendicular as I can, but it is never perfect. The fabric does tend to move around once wet out, and we never have perfect alignment. We have attempted to normalize the data based on the number of stressed ends that extended from one end of the sample to the other, but that was inconclusive. I think my biggest problem is that I’m working with only 1 ply, but our load frame for breaking the samples is only rated at 1000lbs, so I can’t just add a bunch of layers to normalize the variation.

I’ll grab some pictures with the microscope to show the types of bubbles/voids.

#7

Is it one layer of 8.7 oz/yard E-glass?

If so a Vf of 60% would yield a thickness of .0076". 2 layers would be .0152" with a Vf of 60%

You must be thinking of Wf (fiber fraction by weight). 60% by weight with e-glass would be 40% by volume. 1 layer of 8.7 oz/yd2 with a Vf of 40% would yield a laminate that is .0114" thick.

Your thickness is varying .001" Which would be a Vf variance of 3-4%.

#8

Thanks for the correction. Yeah we have been doing everything to a percentage by weight. Yes I am only doing a single layer. I have tried doing a 2 ply, but the jaws for the instron don’t have enough clamping force to keep the sample from slipping before it breaks.

Based on your numbers, it would appear that the theoretical and the actual are very close. Since I’m new to this composites thing, what would be a good Vf/Wf to shoot for? Is 3-4% to broad a range?

#9

1 layer is hard to get good results, the more the merrier. How wide are your samples, and are you using end tabs to control the breakage in the gauge length? Also, is 30min at 100c enough time for the resin? If you do that for all of them, however, you test one set that day, the next set 2 days later…if the resin wasn’t fully cured out of the oven, it might have cure a bit more in the set that had 2 more days to sit.

#10

The samples sheets are 6"x8", but that is cut down to 1/2" x 8" strips that are tested. I don’t seem to have a problem with breakage right at the jaw interface, it seems to be a fairly random assortment of where the samples break. I have tried tabs on several samples, and did not see a change in results.

We did a DOE, to determine time/temp for curing(strength), and that was the optimal result. I have had to leave samples sit over the weekend before I got around to testing them, and there was no appreciable difference in results.

I’m thinking the problem lies in only working with a single ply. I’ll see about designing a different set of jaws for the tensile tester that will hold a multi layer sample.

#11

normal wedge grips would be the way to go. Not sure of cost, but there are various sizes and wedges. we use them up to 30klbf