"Remolding" Thermoset CFRP Sandwich Structures

[MartinShane]

Short form version of my question: Has anyone heard of returning an Epoxy/CF structure (That is fully cured) to its mold and heating it above the glass transition temperature of the matrix as a means of improving accuracy of the molded surface when compared to the ideal (CAD) surface?

Long form version:
After finishing school I began working for a company that specializes in composite part manufacture.
A major customer requires curved sandwich structures with tight tolerances on the inside (convex) face of the parts. Various sizes are produced, all are surface mapped using a laser tracker for comparison to customer CAD data before shipping.

(Parts are generally produced in a multi-stage assembly using single sided molds with vacuum bags which are oven or autoclave cured depending on size. Carbon fiber prepregs with nomex honeycomb core and matched core bonding adhesive is the norm.)

Because parts are fairly large, labor intensive, and relatively low volume it is critical to produce parts that pass inspection. As a result, a failing part may be reworked to try and bring it into tolerance. One particular means of correcting a part that does not meet tolerance is to return the finished part to its original mold, reapply vacuum, and heat the part above Tg for the material, and hold for an hour or more.

When I learned that this was done I was skeptical about its benefit.
When I learned that it was both endorsed by the customer and actually seems to work in some cases I was very surprised.
More than two years on, I still have not come up with an explanation for why this works that I can point to with any confidence.

I do not know if this indicates that -
-Parts in question are not fully cured to begin with due to some process variation
-Residual stresses in the laminate / sandwich structure generated during manufacturing are being released
-I understand the "Set" in thermoset less than I think I do

The couple of people in the company who I have tried to pose similar questions to either attribute it to the particular resin we use, or wonder why popcorn costs so much at baseball games.

So... has anyone seen something similar? Am I missing something obvious?
Let me have it. I can take a shaming.

 

[Compositepro]

There is nothing surprising about what you describe. All polymers are visco-elastic materials that creep under load even below Tg. Fiber reinforcement helps a lot to reduce creep. The chemical cross-links in thermoset polymers prevent remelting but they can be stretched and deformed, given time and temperature. Study-up on shape memory polymers.

 

[berkshire]

Adding to Composite Pros remarks.
Remember that most thermoset epoxy resins follow a logarithmic cure pattern, and continue to cure long after you think they are "Done"
Sometimes a freshly cured part under steady pressure will move even without heat.
B.E.

You are judged not by what you know, but by what you can do.

 

[MikeHalloran]

A yacht of my acquaintance has FRP mufflers. Think Cherry Bombs, but 2 feet in diameter and 8 feet long, made of fiberglass reinforced thermoset about 1/2" thick.
The plastic is not ordinarily a concern, since the seawater that's used to cool the engines is injected into the exhaust gas before it reaches the muffler.
If, however, an idiot butchers the injectors, the water flows only in the bottom of the exhaust tubing.

The boat was several hundred miles into the Atlantic when the crew smelled hot plastic and started investigating.

After the mufflers had cooled, i.e. by the time the case got to litigation, the mufflers were still intact, but the shells had collapsed, and the resin had re-hardened into the new, unsatisfactory shape. It was still hard and tough and ably resisted the caress of a big hammer, as it did when in its intended shape.




Mike Halloran
Pembroke Pines, FL, USA

 

[berkshire]

This phenomena catches you by surprise when you least expect it. My first practical experience with it was in 1979, I was constructing fiberglass sailplanes. We were busy making sub components, in this case glass fiber spar caps. We had built a storage rack from tubes bracketed to the wall.
The spar caps were stored there for later use, until somebody noticed that after about a month, they were sagging between the supports. We had to take them off the rack, put them back into the mold, and gently heat them until they came back, we were sweating because, if they had not , we would have had to scrap the whole batch.
B.E.


You are judged not by what you know, but by what you can do.

 

[MartinShane]

I'll chalk creep up as "Something Obvious."

It's probably of note that in my case, and it sounds like the other cases described, the deformations are largely taking place out of the plane of the laminate such that the reinforcement fibers are un/under-engaged in supporting the load that causes the deformation, whether those deformations are desired or not.

It came to me that I already knew of another example of permanent deformations occurring in a thermoset CFRP part - namely carbon fiber bicycle wheels - where permanent deformations to the brake track can result under extended heavy braking (ie. descending a mountain). Why this didn't register until reading about your examples is just proof I had blinders on.

Thanks all. Often need a reality check working in an environment without some more experienced engineers on hand to ask stupid questions of in person.

 

[RPstress]

In theory deforming a cured thermoset laminate at elevated temperature doesn't break and remake any fiber-matrix bonding; it only causes creep of the matrix. If someone knows a way in which the fiber-matrix bond itself can deform by one sliding over the other without the bond breaking please let me know.

After cure the thermal strain in the fibers may deform the resin permanently a bit, so putting the part back in the same mold may undo some of that deformation. Putting it in a different tool with more curvature would change the shape a bit more. Berkshire's experience of the parts being significantly deformed with no applied temperature and load shows what may be possible (being flexible and a bit heavier, glass fibre will have allowed this creep to be more pronounced under self weight cf. carbon).

Usually a part will undergo spring-in during cure and cool down, which slightly unloads the fibres from the as-cured state they're in while the part is still on the tool. Deformation at elevated temperature after cure to creep the matrix may increase the load in the fibers a bit. However this is likely to be a small fraction of the fiber strains from the cool down. Any laminate analysis done to predict fibre strains as-cured is usually conservative and 'remolding' the part is unlikely to increase the chances of fibre failure.

For resins which like a post-cure the instructions often say that it's ok to post cure unsupported (at say, 180°C after an initial cure of 140°C), but it's usually best to use the original tool or at least provide some support to parts which overhang and might droop under self weight.

This emphasises the possible effects of subsequent post-cure processing, e.g. a secondary bond. You can rarely be sure of the exact final shape of a part. Process modeling can help, but only so much. Experience tells. MartinShane, is the origin of the practice lost in the mists of time? Or is there some residual rumour about where it came from?

 

[MartinShane]

In this case the length of time is not very long - 4 years maybe - but how/why the process was adopted has been lost.