Thursday, November 29, 2018

Rear Bumper Fabrication Project – Part 3: Process Development & Test Parts!


It’s been a long time between blog updates again, but I will blame social media (and getting married!) as a major contributor to this condition as I have posted numerous quickie updates to our Instagram and Facebook project pages as progress has been made on the rear bumper.  The lure of easy, on-the-fly updates has really put a damper on conventional blogging and somewhere, I need to find a better balance.  Also, I have been getting more and more video requests and I continue to contemplate that with more seriousness.  In the meantime, let’s get caught up!

In Part 2, we left off with a polished mold that was ready to begin the process of molding our first test parts.  I was under no allusions whatsoever that I would sacrifice a few test pieces to learn the intricacies of molding such a complex part.  However, a few of the more important lessons learned along the way had more to do with choices in materials options than technique.  More on that in a bit.

For the first test parts, it was decided to target a finished part thickness of 0.125” using a composite laminate matrix of fiberglass cloth, Soric infusion core material, and infusion grade epoxy resin.  This combination laminate would be produced using the vacuum resin transfer method of molding and results in a compression-molded part of near perfect resin to laminate ratio with very little waste and almost no odor or mess.  Under a near perfect vacuum, the laminate stack is under about 7 TONS(!) of compression force over the surface of the mold before the infusion resin was introduced!

Like anything scratch built, there are numerous ways to approach the creation of a composite part like this.  Rules are rarely “hard-and-fast” and it is likely that several attempts and techniques will be needed to arrive at the best overall manufacturing solution.  Our little bumper project has certainly been no different.  Along the way, we have made some rather unexpected materials discoveries along with a number of small (and some not-so-small) setup and process details that make huge differences in the quality of the finished part.  And of course, we have had a few failures along the way as evidence of our steep learning curve.

Molding the First Part – Success in Failure


Someone once said: “Success is a terrible teacher.”  The longer I live; I have come to understand this concept more and more.  If we were immediately successful at everything we attempt, there would be no growth.  No improvement.  No evolution of a craft.  We would be effectively stagnant.

With this idea firmly in mind, the transition into our first molding trials was much easier as we were predisposed to accept the inevitable failures as our most valuable teachers.  In this, the evolution of the molding process and techniques has been quick to develop and our part quality has improved very quickly as a result.  But as I have committed in the past, I will share the “ugly” so anyone who follows this work will know just what was involved should they ever endeavor to do something like this themselves.

Like most endeavors, research into the processes and materials involved is a place that demands discipline and time.  Around our shop, I am most often accused of “overthinking” most things.  However, I am reasonably confident that do so has saved untold thousands of dollars in wasted materials on many aspects of our projects.  However, it may be true as well, that I have cost considerable time in my rather slow evaluation process.  In any case, I am sure the argument will rage long after I’ve been sent to the shed and handed the “noodling” off to someone else.

As it goes, our research into our desired molding process had many aspects pinned down with certainty.  For example, we knew exactly what laminate components we would use (e.g., epoxy resin, reinforcements, layup order, vacuum pump and resin trap setup, general vacuum bagging configuration, etc.).  These details are relatively easy to confirm as there is a large database of experience out there to tap for this type of thing.  However, we discovered that there are equally numerous “intricacies” in this realm that seem to be left to your own discovery.  This became disturbingly obvious on our first attempt to mold our first bumper sample.

From the beginning, we committed to using the best available materials we could get our hands on to avoid having issues as much as possible.  In that, we chose the top-of-the-line vacuum bagging materials based on recommendations from respected suppliers and seemingly objective reviews by users.  This ended up being our first and most critical mistake.

Our first mold preparation and composite stack layup went particularly well and we were able to get into the vacuum bag assembly phase with everything looking textbook perfect.  The vacuum bag assembly went well from all aspects until we reached the point where the actual vacuum bag material came into play.  We chose a bag material with the highest available stretch properties (Stretchlon 200) and was advertised as being compatible with all typical infusion resins available (particularly epoxy resins).  From the off, we had difficulty chasing down vacuum leaks that were extremely small but had to be fixed to allow a proper cure.  This involved many hours of careful leak detection effort before we could achieve full vacuum integrity and could move on to the infusion process.  This should have been our first warning with this particular bagging film as later investigation showed numerous “micro-perforations” in the film as it stretched to conform to the part. 

However, there was one unexpected reaction that doomed our first try in a major way.  Specifically, when the moment came to introduce epoxy resin into the mold under vacuum, the bagging film adversely reacted with the epoxy infusion resin, turning into a rather greasy sheet of cling wrap that lost all physical integrity at every single location where an even moderate feature was positioned.  In the course of about 20-30 seconds, about a dozen massive holes opened up in the vacuum bagging film and the entire mold was fouled after being roughly 75% through the infusion.  At this point, there was really no choice but to let it cure off and hope there was no damage to the mold. 

Fortunately, when the resin had fully cured, the failed part pulled from the mold (albeit with fairly considerable effort) and gave us an opportunity to inspect the surfaces that did see full saturation and we were thrilled with the results.  With this result in hand and the experience well documented, more “noodling” ensued in the quest for root cause of the bagging film failure.  It was discovered that others had experienced similar issues with resin compatibility with this particular bagging film despite of the manufacture’s insistence that resin intolerance was not an issue.  After considerable research, we found a different bagging film (Stretchlon 800, same manufacturer) that ended up working well with our resin choice and was much easier to seal.  MAJOR lesson learned.

Mold preparation is the first step in any composite layup.  Since we had a very nice, Class "A" mold surface, we were able to apply chemical mold release to the mold and begin our layup for our first part trial.
In our first layup stack, we used three different types of fiberglass cloth.  here, two identical patterns are being cut from a roll of fiberglass fabric.



We included a hexcel core material called Lator Soric in the layup.  This material build thickness into the part while allowing fast resin infusion through the honeycomb channels.  When the part is fully cured, the remaining hexcel structure is an excellent strength enhancement.

Here is a closeup of the Soric core material.  The honeycomb channels made into the material allow resin to flow more freely through the material, providing faster and more complete wet-out of the entire reinforcement stack.
While not as "drapable" as fiberglass fabric, the Soric is quite conformable and worked well in the bumper mold.  Here, we have trial fit the Soric over the bumper buck to see how well it would conform to the mold cavity.



In this shot, all of the fiberglass fabric and Soric core material is in place and the mold is ready for the polyester fabric peel ply to be added.
here, the peel ply layer has been applied over the laminate stack.  Peel ply is a fabric barrier between the laminate stack and bagging consumables.  When peeled away from the cured part, it leaves a textured surface that requires no sanding before bonding additional elements (like mounting hardware, etc.

The red screen in this shot is the plastic infusion mesh that is placed on top of the peel ply layer.  This mesh allows even distribution of vacuum across the entire mold surface and allows the resin to flow through the media evenly across the entire part during the resin infusion process.
This is the completed first mold trial with the vacuum bag pulled down under full vacuum and ready for leak check.  It tool far too many hours to chase down every micro-leak we encountered using this particular bagging film and that was our first indicator that things would soon unravel on this attempt.
After all vacuum leaks were addressed, we were able to pull the vacuum bag down to full vacuum in preparation for the resin infusion.
We used an excellent epoxy infusion resin formulation from Composite Envisions.  This resin is extremely easy to work with and has a viscosity that is perfect for the resin infusion process we use.  Unfortunately, our first trial was spoiled by a bagging film choice that did not live up to its advertised capability on a number of critical levels.
In all its ugliness, you can see in this shot how the first trial was on it's way to success before the bagging film failure.  In only about 35 seconds, almost 75% of the laminate was infused before the resin flow stopped.

A closer inspection showing the transparency of the laminate stack and where the infusion line stopped.  For reference, there are three full layers of fiberglass fabric between the camera and the Lantor Soric hexcell core material.  Top shelf materials are key!
 
Up close inspection shows massive bubbles in several different layers of the laminate.  This occurred immediately when vacuum was lost due to the failure of the bagging film.  However, it is interesting to note how the facing layers of fiberglass fabric can be seen within the laminate ahead of the Soric core material.

Even as a failed, incomplete part, the strength of the cured composite matrix cannot be denied.  This is my fat, well-over-200-pound ass standing on the failed first trial part with it straight on the shop floor and the deflection in the part still keeps the edges comfortably off the floor at the lower edges by a solid 5/8" to 3/4".  This was eye-opening to say the least!

Molding the Second Part – Success Can be Bittersweet


In preparation for our second molding trial, we collected all of our notes and decided to apply all of what we learned (a good thing) and a few theoretical remedies we thought might help (not always a good thing).

One very helpful change was to modify the vacuum infusion fittings that allow resin and vacuum to be applied to the laminate through the vacuum bag.  This allowed better vacuum distribution prior to introducing resin into the bag and similarly, it allowed for better resin transfer into the mold during the infusion process.  Go team!

Next, we reversed the pathway the resin would flow across the mold to allow the mold to saturate more evenly.  This was another great move as the improvement in resin flow was obvious.  Score!
Our third improvement involved the method of laying up the fiberglass fabric into the mold.  We used smaller, more tailored pieces to assemble the fabric stack and this allowed for easier conformability to the more intricate mold surfaces and made for an overall easier and quicker layup.  We’re on fire now!

The final (and nearly fatal) idea was to not rely entirely on the chemical release agent to prevent the part from sticking to the mold.  Based on our first de-molding attempt, it was surmised that the release agent wasn’t quite up to the task and perhaps the application of an extra release mechanism might be a worthwhile step.  So, we waxed the mold using the same release wax we used to pull the mold off the car based on its demonstrated success.  Bad idea!

As it turned out, the mold infused beautifully and we were quite ecstatic that out new bagging film was on point and the resin flow through the laminate stack was much improved.  For that moment, we were ROLLING!

The next day would prove to be extraordinarily bittersweet.  The mold was stripped of the bagging materials rather quickly and we were rewarded with the early indications of an absolutely bubble-free and uniform composite thickness of about 0.137”.  However, when it came time to pull the part from the mold, the bottom of our program dropped right out. 

The “sweet” part was that the bumper we ultimately produced was near perfect after trimming.  Even though it is technically “useable”, its total thickness is a bit too thin for our liking and we will need a thicker laminate in the final part.  But the endgame here was that the mold and part it produced, along with the bulk of the vacuum resin infusion process was an overall success.

The “bitter” came about as a result of some simple errors during our mold surface preparation.  As it turns out, in preparing the mold surfaces for the second trial, we cleaned the mold a bit too well and essentially removed the “base” of the chemical release agent from the flanges of the mold.  Then, it was discovered that the vacuum bagging sealant tape was placed about ½” further outboard from the original molding position leaving a thin strip of exposed mold surface without enough mold release.  This was compounded by the discovery that the petroleum solvents in the release wax we chose to add had reacted negatively with the water-based chemical mold release, rendering both agents much less effective than either would have been by itself. 

The end result was a part that was even MORE difficult to remove from the mold and due to the adhesions that occurred around the outer mold flange where no mold release was present, the gel coat was damaged as the part was pried from the mold.  Fortunately, however, the polished mold detail surface was not damaged other than to have a very light etching, similar to water spots, permanently affixed to the mold surface.

This simple shot is very telling relative to a failure in our second molding trial.  The mold was cleaned very thoroughly just prior to this picture being taken and this effectively removed most of the chemical release agent base we had established from the first trial.  The masking tape indicated that we moved the bagging tape about 1/2" further out on the flange edges.  The areas under the masking tape would eventually have too little release agent applied and this allowed resin to adhere to the mold surface during curing.
Our second trial layup was improved by using smaller pieces of fiberglass fabric that allowed better conformability to the mold details.

Here, the peel ply, infusion mesh, and resin feed tube and fittings are in place for the second trial.

The new vacuum bag film worked very well on the second trial and was much quicker to verify perfect vacuum than the first bag film selection that plagued our first molding trial attempt.  The photo above shows the second mold setup ready to be infused with epoxy resin.

This picture shows the mold about ten seconds after the infusion was started.  The resin front is just visible as a darker, edge across the bottom edge of the mold and as a larger dark spot in the center around the resin inlet point.  The resin supply cup can be seen just at the bottom of this image.
A close-up of the mold after being completely infused with resin.  The composite materials are fully saturated with absolutely no bubbles.  This method of construction achieve a near-perfect laminate to resin ratio and ensures the lightest and strongest part possible.
This shot shows the second mold trial completely infused with resin with both the feed line and vacuum line clamped off for curing.  After 24 hours, the mold was opened and the part removed (albeit with a few new issues we discovered as covered in the text above).


Trail 1 & 2 together for comparison.  The lower sample (Trial 2) is in rough trim and looks very much like the bumper we want!  While technically two failed attempts, the evolution is undeniably positive.

What Now?


With a host of additional notes collected and many, many problems solved, the next steps are rather clear and somewhat laborious.  First, the mold flanges are being repaired to restore the smooth molding surfaces.  This requires careful sanding and preparation of the flanges to allow additional gel coat to be applied to restore the surface and allow it to be sanded and re-polished. 

Second, the mold detail surfaces will be stripped and the entire mold wet sanded and polished, starting with 800 grit and progressing all the way out to 2000 grit and then polished to a fresh “Class A” surface.  This will restore the entire mold to “as new” condition and we will get on with our third molding attempt while incorporating all of our lessons learned.

The biggest differences we will make are a move from epoxy resin to vinyl ester resin.  This will allow the use of a much wider variety of fiberglass materials (like CSM) and is much cheaper and more impact resistant than epoxy.  It is also less “reactive” than epoxy which should help in overall materials compatibility as well.  Also, we will be modifying our laminate stack to produce a thicker overall part that will be easier to work with in terms of body fit and it will be stiffer and more temperature stable in the elements.  And finally, we will stick to the chemical release agent exclusively and quit trying to outsmart ourselves.  Fingers crossed! 

Until next time!

Here is the bumper pulled from the second trial in rough trim.  Even after weeks of trial fitting to the body, the part fits the mold perfectly!
We have been using the second trial bumper sample as a crude masking device to protect the mold surfaces while we begin repairs to the mold flanges that resulted from our mold release faux pa during the second trial.

Here is where we are today:  The damaged mold flanges have been sanded smooth and the voids that were created in the gel coat have been filled with fresh tooling gel coat and allowed to cure.  We will finish these flanges with 800 grit paper and them proceed to wet sand and polish the entire mold to return it to Class "A" finish specs before we attempt another pull.