A set of coupons for destructive testing to determine optimum core treatment in sandwich panels. Coupons were laid up and vacuum bagged directly on a sheet of -inch tempered glass, a great method for making flat panels.
In the first two articles of this series, we discussed making blanks for CNC molds. Once you’ve CNC’d your molds, you’re ready to start making parts. Or are you? In the next several articles, we’ll discuss how doing a series of coupon tests before you start making actual parts can help you gain a feel for the nature of composites as well as dial in your methods to attain optimum strength-to-weight ratio parts. In this article we’ll tear up some coupons to investigate and characterize sandwich core treatments. (As discussed previously, these articles deal with composite sandwich structures, the norm for modern aerospace components). We’ll continue our investigation into testing in the next several articles, looking at both qualitative and quantitative testing of coupons as well as scale-model testing. Let’s begin by explaining how we started clipping coupons in the first place.
A Recipe for Success
After completing the molds for my SR-1 Project (a speedplane for setting records in the c-1a/0 category), I enlisted the help of friends and laid up the first part, a control surface (a good first part due to the relatively small size and simple geometry). Although the resulting part was satisfactory, there had been some confusion about methods during the layup, and the part seemed unnecessarily heavy.
Over the past several years, I have worked in a number of shops to learn composite construction techniques. I knew from these experiences how to do a layup, but after doing several more parts I realized that the lack of a set of well-defined procedures was problematic. There was too much variability in parts quality (primarily as measured by the fiber/resin ratio—more on that later), and I needed better instructions for friends who had little experience working with composites but wanted to participate. In my mind, TLAR (That Looks About Right) just isn’t an acceptable method when building parts for a high-performance airplane that has strict weight restrictions, strength requirements, and is meant to set speed records.
As an example, one volunteer was tasked with making a slurry of epoxy/microballoons to squeegee onto the foam core before insertion into the layup. The volunteer had no idea how thick the slurry should be, or how much to apply/squeegee off the foam core. (As an aside, there seems to be some agreement that the term “slurry” refers to a mixture by volume of 1:1 micro and epoxy. “Thin micro” is a 2:1 ratio, and “thick micro” is a 5:1 ratio). Afterwards, as I reflected on the layup, I realized I myself didn’t have a recipe for micro (other than TLAR), or even why we paint micro on the core in the first place. That’s the way I’d done it in other shops, but how did I know that was any better than painting the core with straight epoxy—or nothing at all, for that matter? I decided before doing any more layups that I needed to dial in my procedure so that we could obtain consistent, repeatable results with empirical evidence to back up our methods. I wanted a recipe—a recipe for success.
Glass microsphere “floating” (i.e., not bonded) within the epoxy matrix. The author found microspheres to cause potential bonding issues in composite sandwich structures. (Image from Effects of Surface Preparation on Long-Term Durability of Composite Adhesive Bonds [DOT/FAA/AR-01/8])
Defining your Goals
Composites are all about minimizing weight and maximizing strength. In contrast to the Rutan method, which skins a hand-shaped or hot-wired male foam core with glass or carbon fiber (see sidebar below), most modern composite ships make extensive use of vacuum-bagged, molded parts of sandwich construction (i.e., outer and inner layers of carbon fiber or fiberglass separated by a thin foam or honeycomb core). The strength of the part is primarily dictated by the number of layers of carbon (or glass) used, and so the weight from the laminate is basically a fixed quantity that depends on how strong you need the part to be. (The core on the other hand provides stiffness, not strength per se, and typically accounts for only 5-10% of total part weight.) However, the resin binder (i.e., epoxy or vinylester) is a variable: in a resin-rich layup, the resin can easily weigh more than the fabric itself, while on a resin-lean part, the resin content may be less than half the weight of the fabric. Over the weight of the plane, differences in fiber/resin ratio (FRR) can make a huge difference, as shown in the table (right). This was the primary parameter I wanted to dial in.
For carbon/epoxy laminates, a 50/50 (by weight) FRR is roughly considered the low end of the acceptable FRR range, while 70/30 is considered the high end (based on discussions with composites engineers, articles in various industry journals such as Composites Today, and resources such as AGATE [Advanced General Aviation Transport Experiments]). Lower than 50/50 means the part will be unnecessarily heavy, and in fact excess resin will hurt the structural properties of the part. On the other hand, a 70/30 ratio results in a very light part, but unless there is very even resin distribution (possible with prepregs but difficult with hand layups), this risks dry spots that will also severely compromise part strength. My FRR goal for SR-1 parts was to achieve a 60/40 ratio, which ensures we will attain relatively lightweight parts while maintaining a margin of safety below the 70/30 threshold (60/40 is a pretty common standard for FRR). By and large we have been able to achieve this using the techniques that will be described in a subsequent column—techniques influenced by the coupon tests discussed here.
To give you an idea of how these ratios would look in a real part, consider a wing that for strength purposes requires 30 pounds of carbon fiber (remember, the fiber provides the structural strength; the epoxy simply holds the fibers in place). The table above shows the final wing weight with different fiber/resin ratios.
Clearly, controlling fiber/resin ratio can have a huge impact on part weight. Ignoring this parameter is to ignore the strength-to-weight advantages offered by composites.
The second parameter I wanted to determine was core treatment. Although peel resistance (between the face sheet and core) can be quantitatively measured using a test called the climbing drum peel test, I was more interested in a qualitative analysis of the bond failure.
Adhesive bond failure can be classified into three categories: substrate failure, adhesive failure, and cohesive failure, as shown in the illustrations (left). Because the foam core has far lower structural properties than either carbon fiber or epoxy itself, we would like to see substrate failure when composite sandwich panels are subject to destructive testing.
Illustrations courtesy of 475 High Performance Building Supply (https://foursevenfive.com adhesive-bonds-and-failures-whats-going-on)
Coupon Test 1: Core Treatment Methods
Before attempting to dial in methods for controlling fiber/resin ratios, we decided to focus on core treatment methods. The following six methods were tested, based on practices I have seen used in other shops:
- No treatment; core placed into layup stack dry.
- Neat treatment: core squeegeed with a layer of neat (straight) epoxy.
- Thin micro: core squeegeed with thin micro (two parts micro to one part epoxy, by volume).
- Slurry micro: core squeegeed with slurry micro (one part micro to one part epoxy, by volume).
- Seal 1: core painted with epoxy thinned 2:1 with alcohol and allowed to cure before placing in stack.
- Seal 2: core painted with epoxy thinned 2:1 with acetone and allowed to cure before placing in stack.
Two coupons were made for each treatment method. One coupon was rolled with a porcupine roller, while the other coupon was not. All 12 coupons had cores of identical size (6×6 inches) and weight, and the finished dimensions were identical (7.25×7.25 inches). Cores utilized a typical 2-core/1-laminate schedule, which is to say two layers of 6-ounce plain-weave 3K Hexcel carbon fiber on the mold (i.e., exterior) side, 1/4-inch foam core, followed by a single layer of 6-ounce carbon on the interior face.
After curing, each coupon was subject to a drop test in which a pair of pliers was dropped from a height of 30 inches to simulate dropping a tool onto a wing panel, for example. The pliers were dropped four times (on each quadrant of the coupon) on both the single- and double-ply sides. A damage score was assessed according to the following scale.
- 0 No visible damage
- 1 Visible dent, or fracture without penetration
- 2 Penetration <1mm
- 3 Penetration 1-2mm
- 4 Penetration >2 mm
Foam residue is present on the skin when peeled from the core. This is evidence of substrate failure—that’s good.
Adhesive failure. The microballoon slurry interface has lifted clean from the carbon skin. This indicates that the bond between the micro and skin is even weaker than the foam itself—that’s pretty weak!
Damage tests are often (but not always) performed with a hemispherical indenter/impactor (i.e., a steel ball) so that the indenter always presents the same shape and therefore an identical impact to each coupon (and is thus a standardized and repeatable test among labs). That said, I feel the plier drop has its own advantages. First, you are more likely to drop a pair of pliers on a wing than a ball bearing. Second, the fact that slight variations in the impact angle of the pliers result in widely varying damage scores (from no visible damage through relatively deep penetration of the ply) tells me what range of damage I should expect if indeed I do drop a tool on a composite part—and that it won’t necessarily be visible on the surface.
Pliers were dropped 30 inches onto panels to simulate a medium-sized tool dropped onto an airplane skin.
Impact to the core from the plier drop is seen here in a well-defined circle of foam torn from the core when the skin is peeled off. Pliers were dropped onto each quadrant of the coupon, thus four damage sites.
After the drop test, each coupon was broken and the upper and lower skins peeled away. The type of bond failure (substrate or cohesive; adhesive failure was inapplicable) was then noted. The results of this first coupon test can be seen in the coupon test graph below.
Coupon test comparing various core treatments. Bars represent coupon weight, while the black diamonds show damage score.
As can be seen, treating cores with micro or sealant resulted in undesirable bond failures. In contrast, either inserting cores dry or squeegeed with neat epoxy yielded 100% substrate failure (remember, that’s good, because it means the foam failed, not the bond).
Also as would be expected, no core treatment yielded the lightest coupons, and the pierced (porcupine rolled cores—the right bar of each pair) cores are slightly heavier than the non-pierced cores (pierced cores absorb more epoxy into the piercings left by the roller and are about 2 to 3% heavier on average). Piercing had no effect on coupons exhibiting substrate failure, and little effect overall.
Typical damage to double-ply face. The plier drop rarely penetrated the double ply, but usually penetrated the single ply.
The damage score, as shown by the black diamonds, shows no particular trend other than that sealed core coupons are more resistant to puncture damage. This makes sense, since the epoxy sealant significantly toughens the core, although at the expense of increased weight and poor bond characteristics.
The main notable observation from the drop scores is that single-ply skins (at least regarding typical 6-ounce plain-weave carbon fiber) are extremely fragile and easily punctured, while double-ply skins are significantly tougher. That is one reason why composite aircraft use at least double skin thickness for external facings, even if a single ply is sufficient from a strength point of view.
That wraps up our investigation into core treatment. In next month’s article we’ll continue our investigation into optimizing FRR by experimenting with a variety of core types and perforated plies.
Eric Stewart is designing and building the SR-1, a speed plane for setting records in the FAI c-1a/0 category (takeoff weight less than 661 pounds, including pilot and fuel). You can see more at facebook.com/TheSR1Project, including additional photos and videos of the subjects in this series of articles.