Wing skin vacuum bagged from a mold of CNC’d tooling foam. It’s really not all that different from the flat panel we bagged in the previous column.
In the previous article of this series, we looked at vacuum bagging flat sandwich panels using a glass table as a release surface. In the next two articles we are going to go a step further and take a detailed look at bagging panels in molds. The process is similar, but using molds requires a few extra steps. Let’s start off by talking a bit about the molds themselves. (Note: Although I refer to carbon fiber throughout the article, the methods could apply to fiberglass as well.)
(3) Wing root mold CNC’d from HDF for Paulo Iscold’s ASH-30 glider experimental wing.
(4) Composite female mold made from male master mold. Mold prep is just a few coats of wax and PVA (optional). Mastic bagging tape can be applied directly to the perimeter of the mold outside the scribe line for vacuum bagging. Molds such as (1) and (3) typically require bagging of the entire tool as the mold is fairly porous compared to (4).
Molds for our types of projects will almost certainly be one of two types: either tooling foam/HDF (high density fiberboard) or composite. (We are going to ignore metal molds, which are really beyond the scope of what we are discussing here, though they are common at the industrial manufacturing level.) If you’ve designed your part on a CAD program like SolidWorks, creating a female mold of that part is relatively straightforward. This female mold model is then CNC’d into tooling foam or MDF. On the plus side, this is the easiest and most accurate way to make a mold (assuming you have the CAD skills and access to a CNC). On the downside, tooling foam and CNC time can be expensive.
Low-density foam is less expensive but more prone to dings, and the surface quality of the bagged part will require a little more primer to bring to paint than a part that comes out of a glossy-surfaced composite or metal mold. Higher density tooling foam can give you essentially the same surface quality as a composite or metal mold, but will be expensive; tooling foam cost is proportional to weight. HDF molds give a surface quality between low-density foam and composite, but require sealing (or double bagging) for mold release. They are also extremely heavy—you will need several people or a forklift to move an HDF mold of any appreciable size.
Part on the left was made directly against waxed/PVA’d epoxy-densified 10-pound tooling foam and shows characteristic orange-peel texture due to relatively large cell size of foam. Part on the right was made using 30-pound foam and shows the smoother surface provided by higher density foams.
Part on the left was made in a composite mold, as reflected in the high gloss of the finished part. Part on the right was bagged against 12-pound tooling foam using double-bag method. Orange-peel texture is minimized by use of peel ply between the mold surface and part; it imparts a dull finish.
Obviously, the double-bag method used with foam or HDF molds to achieve part release adds to the complexity of the layup process, and you will need at least two vacuum pumps: one for the first (release) bag and a second for the layup itself. I personally have a third pump ready to go in case either of the two main pumps have problems. (Having a backup $100 Harbor Freight pump paid for itself on the one occasion when one of my pumps overtemped, and I was able to save an expensive part.)
Pre-CNC, the traditional method of mold making was to make a male mold and then make a splash mold of fiberglass or carbon off of this. The benefits of this method are lower cost, a very nice mold surface that requires minimal body-working of the part before paint, and the ability to see and manipulate the male mold, thus knowing what the final part will look like. The main disadvantage with this method is that it requires making the male mold, which can be quite time consuming.
There are hybrid approaches to each of the above methods. For example, a tooling foam mold can be machined .25-1.0 inch smaller than the OML (outside mold line, i.e., surface or edge of the part). A screed of epoxy tooling dough can then be added to the mold, building it back up larger than the OML. After cure, the screed is CNC machined to the exact OML. This achieves a very nice mold surface equivalent to a splash mold while avoiding the work associated with making the male mold. On the downside, the tooling dough adds a certain amount of cost and increases complexity at the machining stage. However, it decreases complexity at the molding stage by eliminating the need to double bag, since the smooth surface can be waxed for release.
Male mold machined from tooling dough. Mold basic structure is constructed of plywood templates covered with expandable steel lath.
Another novel hybrid approach for quick and cheap rapid prototyping that can be used with simply curved (i.e., flat wrap) parts is to hotwire blue or pink EPS foam insulation board and then bond a sheet of thin (.016-inch) aluminum to the foam. The aluminum gives a glossy surface finish and good part release with just a simple waxing. Be aware though that insulation foam can crush under strong vacuum, especially around unsupported edges. This can be alleviated by pulling less than full vacuum or supporting the edges of the mold with some kind of framework. This method is documented in post #22 and 23 of the following VansAirForce.net thread, which discusses a project to mold composite wings for an RV-6 Reno racer.
Splash molds can also be pulled from CNC’d male molds. As with the above approach, the male can be machined smaller than the OML, built up with a tooling-dough screed, and re-machined to the OML, providing a perfect surface from which to pull the splash mold. The male mold can also be machined from a dense material such as wood, which is then primed and painted to a high gloss.
Locator pyramids allow these two molds to mate perfectly when joining a multipart piece. These locators were simply hand formed from clay, but premade pyramid or hemisphere locators are also available.
Scribe lines on the mold transfer to the part showing where to trim the part. This part came from a glossy surfaced mold; the very slight texture on the surface is due to the use of PVA. For a perfectly glossy surface, forgo the PVA—but be sure the mold is very well waxed!
A few final notes on the mold design itself. First, when using multi-part molds (such as an upper and lower wing skin set, for example), make sure you include good locators so that the molds can mate up nicely. Ideally, the molds should self-center on tapered dowel pins, or pyramid or hemisphere locators as they are brought together. It’s also helpful to include datums, trim lines, or other reference marks either on the mold itself or out on the flanges to help locate ribs, bulkheads, cutouts, etc. This is one nice aspect of composite molds, since it’s possible to scratch out very fine scribe lines on the gel-coated mold surface that transfer nicely to the finished part. Low-density tooling foams lack the surface definition necessary for this.
More references: Past KITPLANES articles from this “Rapid Prototyping/Experimental Design” series have discussed some of the above topics. In addition, the two-part article from April 2015, “Black Beauty,” also provides an overview of a wing made using the double-bag method. Two excellent online sources of information for making splash molds and molded parts are Mike Arnold’s excellent AR-5 and AR-6 videos (which he kindly made available for free on YouTube before passing away; see at The Arnold Company channel) and Bob Kuykendall’s HP-24 website updates from 2001-2007).
Foam board can be pre-curved somewhat with the application of heat. Here a Rohacell wing core is being formed to better fit the mold. Honeycomb is better for compound or deeply curved applications.
Rohacell is used in the flat wrap section of the wing skin while honeycomb is used in the compound curve wingtip.
If you haven’t read Part 1 of this article on bagging flat panels, now would be a good time to do so. The material preparation steps are similar, but I’ll point out some important differences here.
For molds with complex curves, you need to consider the curves when choosing fabrics and cores. Foam cores are generally not used with compound curves. A very small amount of curvature may be doable, but the issue is that you may have trouble with bridging when you pull vacuum, or the core may push the fabric out of place as it sucks down to the mold. In this case it’s better to use honeycomb (which comes in regular or overexpanded varieties, with the latter better able to comply to curves), or simply eliminate the core altogether and increase the layup schedule with a few extra plies of carbon. When using cores, don’t forget to scarf the edges.
The above consideration of curvature also pertains to fabrics. Drapability and stability (i.e., the warp and weft staying straight and perpendicular) are inversely proportional. Twill drapes nicely but keeping fibers aligned during handling can be difficult. Plain weave seems to possess a good balance of conformability and stability. Stitched fabrics or unidirectionals with a support scrim have the greatest stability and thus are the most difficult to work into curved molds. On the other end of the spectrum, the ultimate in conformability is unwoven tow, which can be laid anywhere you want some extra unidirectional stiffness.
abrics in counterclockwise order of decreasing stability/increasing conformability: unidirectional with fibril netting, stitched biaxial, harness weave, plain weave, twill weave, and tow. Far left yellow/black swatch is a carbon/Kevlar hybrid cloth.
Cloth should be weighed before beginning your layup so you know approximately how much resin you’ll be using.
Templates are great when cutting material. They minimize waste, and you can write notes regarding orientation, number of plies, etc. on them. Also during the layup, time is usually at a premium, so not having to trim oversize pieces of fabric saves time. For large templates, I usually place some transparent 4-mil plastic on the mold and mark the trim lines I want, then use this plastic as a cutting template. For smaller or especially curvy parts, I cut a piece of carbon oversize and dry fit it in the mold. I then mark the trim line (usually the edge of the mold plus 2-3 inches) with a Sharpie (silver shows up best on carbon, black for fiberglass). I remove the fabric, flatten it back out, trim, and transfer the shape to a piece of dollar-store foam board. I then cut out another piece from this template and do another dry fit to make sure the template shape works OK and modify as necessary.
Template includes notes for orientation on mold, fabric alignment, and number of plies. Foam-board templates are easier to trace out on fabric, but plastic templates fold up nicely for storage.
For the vacuum bag, plastic sheeting from the hardware store (3.5- to 6-mil polyethylene) is cheap and works fine but has limited conformability; we typically only use it for flat panels or very shallow curvatures such as wing skins. For deeply curved parts such as a fuselage tub or a rib with flanges, bagging film like Stretchlon works great and is more puncture resistant than polyethylene.
For tape, I like ACP’s black sealant tape, which comes in 50-foot rolls for not much more than the 25-foot gray tape. However, it is about half as thick as the gray mastic, and I find it difficult to make bags with. So if you are bagging directly to a glass table or glossy composite mold, the black tape works fine. If you are making a bag by sealing sheets of plastic, use the gray tape.
Peel ply and breather/bleeder are the same as used for fabricating flat panels. To make sure pressure is distributed evenly around the part, use one vacuum port for every 6-12 square feet of part.
That’s all we have space for this issue. We’ll continue next time by looking at prepping your materials and then the layup process itself. See you then!
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.