Rapid Prototyping and Experimental Design

Coupon testing, part 2.


Rohacell and Divinycell foam cores inserted into the layup. Cores are staggered to minimize the effects of local variations in resin saturation, which could affect coupon weight.

Last month we looked at different ways of treating foam cores before inserting them into the layup stack. This month we’ll continue our foam core tests, looking at different varieties of foam. Next we’ll look at attempting to control fiber/resin ratios using perforated plastic in the laminate stack. Let’s start with the foam core tests.

The smaller cell size of Rohacell IG-F is apparent when comparing Rohacell 31 IG-F (left) and 31 IG (right).

Core Types

A large variety of cores are available to homebuilders. These include foam, honeycomb, and end-grain balsa, among others. (For an in-depth look at core types, check out “Getting to the Core of Composite Laminates” in the October 2003 issue of Composites Technology.) The most salient variables to the homebuilder are mechanical properties, density, and price/availability. Since the SR-1 Project is very weight sensitive, we looked at various core types to determine which would offer the best combination of structural properties and low weight.

Rohacell IG/IG-F is Evonik’s industrial-grade foam core. The F designation indicates the fine-cell version. Likewise, Divinycell H is Diab’s industrial-grade foam, and F is their aerospace offering with finer cell size. Both Rohacell and Divinycell are closed-cell foams and only absorb resin into open surface cells, so finer cell size means less resin uptake and thus a lighter layup. The numbers following the grade indicate the density in kilograms per cubic meter. For example, Rohacell 51 IG-F is industrial grade, fine cell, 51 kilograms per cubic meter density (about 3.2 pounds per cubic foot).

Airshow coverage sponsor:

Rohacell is a polymethacrylimide (PMI) foam, in comparison to the polyvinylchloride (PVC) used in Divinycell. From a practical point of view, what this means is that for the same density, Rohacell’s nominal allowables (i.e., mechanical properties) are approximately 30-60% higher than equal density Divinycell. Divinycell’s continuous operating temperature is nominally 160 F, while Rohacell’s is 275 F. This higher performance is, of course, reflected in Rohacell’s higher price. Data sheets for both these foams can be found on the manufacturers’ websites.

Coupon weights and average drop test scores for various foam core types.

Coupons for Test 2 were identical in size/laminate schedule to coupons for Test 1 (see Part 1 of this article in last month’s KITPLANES). All cores were squeegeed with neat epoxy prior to inserting into the layup and weighed for resin uptake. As can be seen, there is little difference in weight, since the core makes up at most about 15% of the total coupon weight. Drop test scores also show no particular trend. These tests do not, and are not meant to, reflect the different structural properties of the foams, which are higher for the Rohacell PMI foams. All coupons showed substrate-type bond failure, as indicated by the green color of the bars. Right-hand bars had cores pierced with the porcupine roller and are characteristically slightly heavier.

In addition to the coupon tests, samples of Rohacell 51 IG-F and Divinycell H45 were both subject to environmental tests with acetone and avgas. After 24 hours submerged in acetone, Divinycell felt slightly rubbery and also warped slightly when dried. After 24 hours in acetone, Rohacell was unaffected. After one week submerged in 100LL avgas, Divinycell was again slightly rubbery and warped when dried, although to a lesser extent than when exposed to acetone. Rohacell appeared unaffected by avgas.

Perforated film sizes are easily distinguished by color.

Coupons trimmed, labeled, and weighed.

Coupon Test 3: Perforation Film

As mentioned in Part 1 of this article, the major goal of the SR-1 Project’s coupon testing was to dial in the layup process, in particular to be able to control the fiber/resin ratio (FRR) with standardized procedures. Since each coupon has only three constituent materials (carbon fiber, epoxy, and core), by weighing the dry carbon fiber, core, and final coupon, we are able to determine the weight of epoxy in each coupon and thus the fiber/resin ratio. No attempt was made to account for the small quantity of resin that is absorbed into the core and which therefore, strictly speaking, does not figure into the FRR ratio. Accounting for this quantity would give somewhat higher values of the “true” FRR.

Fiber/resin ratio and film open area of various perforated films.

Fiber/resin ratio can be affected by many variables, including, but not limited to: type and viscosity of the resin system used; type of carbon fiber (or glass, Kevlar, etc.); shop temperature and humidity during layup; atmospheric pressure and available vacuum pump pressure, core type, and density; and the types of peel ply, perforated film, and breather/bleeder used to absorb excess epoxy from the layup. It should also be noted that smaller parts tend to have higher FRRs than larger layups, so it’s not unusual to see FRRs drop a few points when moving from small coupons to large wing skins, for example.

Because several early practice parts for the SR-1 had FRRs near or exceeding 70/30 (remember, our goal was 60/40), it was decided to experiment with several of these variables to see if FRR could be more precisely controlled. A variety of coupon tests were conducted that looked at type of epoxy (Rhino [aka Jeffco] 1307 resin/3176 hardener vs MGS 285 resin/285 hardener), vacuum pressure (28 inches Hg vs 26 inches Hg), bleeder/breather weight (4.5 ounces. vs. 10 ounces), peel ply (1.8 ounces vs. 3.0 ounces Dacron fabric), and perforated films of varying permeability. Not all of the experiments yielded meaningful or useful results, and discussing what didn’t work—while interesting—is beyond the scope of this article. That said, the graph for Test 3 shows how reducing the open area of the perforated films, for example, had somewhat unexpected results.

Comparison of Airtech P (left) and Airtech P2 (right). Failure of P2 to bleed epoxy properly lead to uneven saturation of the coupon. An even pattern of dots like that seen on P indicates a good layup. No dots suggest the layup is too dry. In contrast, if the bleeder/breather becomes totally saturated, the layup is too rich. To make matters worse, saturated bleeder loses its ability to function as a breather.

The horizontal axis shows perforated films of decreasing permeability, or open area (i.e., the area of the holes as a percent of the total film area, as indicated by the black diamonds). Even when reducing the open area by a magnitude of 100 (from 2.0% to 0.02%), the FRR was only reduced by about 10%. In fact, Airtech P31 was the only film that made an appreciable difference in throttling resin flow into the bleeder. Although P2 looks effective, the holes in P2 are not clean perforations. Small flaps left from the manufacturing process had a tendency to block the holes, resulting in very uneven bleeding (see photo) and artificially lowered the FRR.

I currently focus on three variables to control FRR: temperature, part size, and bleeder cloth. Elevated shop temps will lead to over-bleeding (i.e., FRR too high) since the resin is made less viscous by high temps, so I try to avoid doing layups when it is above 80 F. Also, small parts have a tendency to over-bleed. In this case I either put a small pinhole in the bag to reduce vacuum pressure until the resin has begun to gel, then plug the pinhole, or wait until the resin has begun to gel slightly before pulling vacuum. Finally, I prefer to use heavy (10 oz.) bleeder only for large parts, since on small parts it can lead to over-bleeding. I use light (4.5 oz.) bleeder for small parts and layer it if necessary to avoid saturation.

How About Honeycomb?

Some readers may wonder why honeycomb was not included in the tests described in these coupon articles. The main advantage of honeycomb core for homebuilders is honeycomb’s ability to conform to compound curves (although foam does fine with shallow simple curves, and can be thermo-formed with a hair dryer if desired) and its high service temps (60% room temperature strength at 400 F for Nomex). The lightest density honeycomb readily available to homebuilders is 3 pounds per cubic foot, which is identical to the 3 pounds per cubic foot foams tested here; so although it has better mechanical properties, there is no weight advantage (1.8 pounds per cubic foot honeycomb is available, but can be difficult to source). Industrially manufactured honeycomb cores are typically bonded to the face sheet with an intermediate heat-activated bond film and caul plate. This film is just sufficient to create a bond fillet between the honeycomb and face sheet without filling the honeycomb cell with excess resin.

This photo illustrates the challenge of bagging with honeycomb cores. This cutout from a molded cowl shows the exposed upper surface (left) to be resin starved—there is little or no resin on the edges of the honeycomb. In contrast, the bottom surface of the same cutout (right) shows excess resin, with some honeycomb cavities completely filled. The upper surface had very little peel strength, while the lower surface required brute force just to remove a small corner.

Homebuilders can certainly make honeycomb-cored panels (hopefully we’ll do just that in a future installment); it just takes a certain amount of experience to reliably manufacture panels with a good upper skin to core bond and decent weight specifics. The reason this is challenging is that when a wet laminate is placed on top of the honeycomb core, gravity tends to cause resin to pool at the bottom of each cell. As a result, the top facing may be poorly bonded if resin has drained away from the upper face sheet, and the pooled resin results in unnecessary weight. The alternative is to make honeycomb panels in a two-step process, bonding the honeycomb to a bottom face sheet in the first step, allowing this to cure, then flipping the panel and again bonding to a face sheet. This avoids the gravity-induced problems, but is obviously more work.

For these reasons, we chose not to include honeycomb in these tests (although they feature in the next article of this series). That said, if you wish to utilize honeycomb panels for your project, but do not want to invest in the time and materials to make them, pre-manufactured panels (in any combination of foam, honeycomb, or balsa core and carbon or fiberglass face sheet) are available from companies like Aerospace Composite Products. We encourage curious readers to conduct their own tests on these materials and share their results here in KITPLANES.

Brian Paris, senior in mechanical engineering at CalPoly and summer intern on the SR-1 Project, overlooks a set of coupons testing various perf films on non-cored laminates.


I want to make clear that the presentation of information in this article is meant to be illustrative rather than prescriptive. There are a large number of variables that affect the quality of composite assemblies, and it will benefit builders to refine their methods with coupons before embarking on structural components. Most of us do not have access to climate-controlled clean rooms, and differences in epoxy systems, environmental effects, shop equipment, and bagging processes will yield different results for different builders. I’m merely trying to illustrate one method by which to bring some science to what is definitely still very much an art.

In the next installment of this series, we’ll look at machine testing of coupons to derive ultimate strength values. It’s pretty darn cool!


Many thanks to Evonik Industries (maker of Rohacell) and DIAB Group (maker of Divinycell) for supplying foam core samples for coupon testing. Evonik is also a sponsor of Rohacell foam core for the SR-1 Project. Thank you to Airtech, which provided samples of perforated release films for the perf test. Thanks also to the composites engineers who provided advice and feedback on these tests, especially Peter Schwarzel, P.E., of CarbonWorks, Australia.

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.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.