Worth the effort

Finally! The shell of the center section is joined. It was a long and tedious way, but worth it:

We had to solve many problems to get there! It is astonishing how well it worked out considering that a couple of years ago we had not much more besides an idea.

Joining the shell was more or less straightforward. We mixed an adhesive based on expoy-resin (130 g), cotton flakes (15 g) and thixotropic agent (3 g). The adhesive was spread on the contact surfaces and the shells were pressed against each other with the mold. While the aft joint is wide, and hence, strong enough, the contact surface in the front is small. Notice also, that the aerodynamic pressure is highest here, as the stagnation point moves around this region. A bursting shell would we simply catastophal. Reinforcement was a must. We laminated and vacuum bagged a layer of biaxial non-crimp carbon fabrics (200 g) from the inside and outside. This should be enough to keep it together in all circumstances.

Here’s a video of how we joined the shells:

 

The next steps? Well, we  have enough to do in the center section: reinforcements of the ribs, canopy, fitting of spars, seat, controls, etc. We hope to start soon with the pattern of the cockpit and canopy. If everything works out as expectect, we should have at the end of the year a cockpit and a clear canopy (produced by Plexiweiss GmbH).

Reinforcement of ribs

Yesterday and today we laminated the reinforcement of the ribs. We used 200 g/m² non-crimpt biaxial carbon fabrics and a covering layer of 100 g/m² flax fabrics. Biaxial carbon fabrics are great when you have corners. The stiff carbon fibers have then an orientation of ±45°, so that the fibers are bent less and wrapping them around the corner is much easier. Also, this orientation is optimal for shear loads.

P1080400_mP1080401_m

You might ask why we used flax fabrics. Well, they protect the load-bearing carbon fibers from external abrasion! Flax behaves much like aramid (Nomex® and co.): It takes some energy until it is damaged. Such a covering protects not only the load-bearing fibers, but also the pilot from sharp edges.

The reinforcements are curing now in vacuum at roughly 200 mbar residual pressure. The surplus resin is taken up by breather felt. We used also a covering layer of peel ply to obtain again a rough surface.

P1080402_m

P1080403_m

Two are better than one

The main ribs and those beside them have to take-up the weight of the pilot while entering the airplane: Climb from behind and walk on top of the wing. These ribs have to be doubled near the joint to the shell to distribute the pressure on a larger surface:

P1080390_m

P1080391_m

After bonding, one layer of 200 g/m² biaxial non-crimp carbon and 100 g/m² flax fabrics will further reinforce the joint.

We also made a couple of small ribs to stabilize the tail:

Diese Diashow benötigt JavaScript.

Wedding preparation

The upper and lower shells are almost ready to marry. We are making the final adjustments:

Diese Diashow benötigt JavaScript.

It’s a long time since both molds were together. The next couple of weeks we’ll install some more ribs…

We also tested if the aft spar bridge fits well:

This bridge is only a piece for testing and fitting. The test piece for the main brige is beeing build by Siggi as of right now. The final bridges will be built by Eichelsdörfer GmbH using Siggi’s jigs. The material will be 1.7734.5 steel (better than SAE 4130).

CNC milling of ribs

Last time we wrote about how verly light AIREX® panels can be produced. Now it’s time to use that panels. Obviously one can use a simple saw and cut ribs off the panels. However, the ribs are large and need to be somehow transferred. A piece can be drawn in a CAD program and printed on several pieces of paper or tranferred by mere measuring and painting on the panel. Both approaches are tedious and precision is an issue. We are lucky to know somebody nearby with a self built CNC machine. With a good drawing and some experience it works like a charme. Thanks again for the help, Marcel!

Everything starts with making a good drawing a the piece:

rippe

I use a simple but good 2D-CAD program (QCad for Linux). If you look carefully at the drawing, you’ll see a chamfer at the left edge, which makes the piece more complicated (3D instead of only 2D). This drawing was imported in SolidWorks—a great but expensive 3D-CAD program—and converted via a STL file into G-Code by another tool. G-Code is what the milling machine understands. So much for the theoretical part…

Theory ends and practice starts after having a milling program. The panels have face sheets of carbon and flax fabrics. Carbon creates very stiff but brittle laminates. Flax keeps the carbon together and substantially increases robustness—similar to aramid. The AIREX® core is light but also soft. This mixture is difficult to mill properly, when the wrong router is used. Probably milling aluminum is easier than these panels. Anyway, the first time we used a router for metal. The result was pretty good, but the edges were quite fluffy. Nothing that cannot be corrected with some 400 grit sand paper. For the following pieces we bought a composite router (Karnasch 29.1783). It was not cheap, but the result was great.

Here’s a video of the machining:

I think the video speaks for itself!

We started to fix the ribs on the upper sandwich of the center section. Soon we’ll be able to „marry“ the to halves.

 

Production of AIREX® panels

We need several ribs to support the center section. These have to be installed before the upper and lower halves are joined. The main ribs conduct the wing’s shear to the spar bridges (shear load of about 5000 N  or 1100 lbf). Other ribs are needed to support the shape of the center section, e.g. to be able to walk on top to enter the plane. The ribs are subjected mostly to shear, which means that the fibers should be oriented ±45° in respect to the chord line. Thin plates tend to buckle under shear. Hence, we decided to use a 6 mm sandwich with biaxial non-crimp carbon fabrics face sheets. To prevent splintering in case of an accident, the carbon is covered by a layer of flax fabrics.

Though a honeycomb core is well suited to create ribs, the much cheaper and easier to process AIREX® PVC-foam cores are a good alternative. Their weight is somewhat higher, but considering the small area of the ribs, their contribution to the final weight is small.

 

AIREX® provides a good and long processing guide (see for example copy at R&G). Essentially these foam cores are suited for almost any available processing method. This does not mean that the different methods do not result in different qualities. We compared two processing methods: vacuum resin infusion and usual vacuum bagging.

Vacuum resin infusion

We created small resin channels with a screwdriver. This way we hoped to prevent having to use a flow media.

p1070807_m

Trimming the face sheet material (100 g/m² biaxial non-crimp carbon fabric and 100 g/m² flx fabric):

p1070808_mp1070809_m

 

Vacuum resin infusion with HP-E300GL epoxy resin and MTI® hose:

p1070813_m

Result:

p1070828_m

The face sheets look perfect. The surface was covered with peel ply to obtain a rough surface for further processing. The flax fabrics are almost not recognizable besides at the edges. It would be perfect if there wasn’t a small „but“: low fiber volume ratio (fvr). The final panel weights 1550 g (55.6 oz) for 0.66 m² (7.1 ft²), which means that it weights 2.3 kg/m² (7.8 oz/ft²). Too much! I expected roughly half that weight, which means that the fvr is catastrophic. But wait, vacuum resin infusion is known to produce high volume ratios! So how is that the result is overwetted with resin? Where’s the resin gone? It was sucked by the porous surface of the foam core! This is the crux when vacuum infusing foam cores. Some people fill the surface of the foam before infusing, but filler weights also and makes processing more elaborate.

In total: it worked alright without flow media, but keep an eye on temperature (not too cool). It is a good alternative for applications were weight is less critical, but it is certainly not suitable for airplanes.

 

Vacuum bagging

The face sheet material was separately wetted with HP-E200GL epoxy resin and vacuum pressed to the core. Processing is more elaborate than in infusion, but this approach prevents the porous surface from filling with resin.

 

Peel-ply, perforated release film and absorber material were used on top of the face sheets. The panel was then pressed between two standard shelf boards. A strong vacuum with a pressure lower than 20 mbar (0.3 psi) was applied, which is rather unusual in vacuum bagging.

p1080047_m

The face sheets were almost as nice as in vacuum infusion. More important: much less resin consumption and a higher fvr. A fvr of over 55% was achieved with 200 g/m² carbon face sheets. This is an excellent value! For 100 g/m² carbon we achieved over 45%, which is still a very good value for vacuum bagging. A panel with 200 g/m² carbon and 100 g/m² flax weights about 1.3 kg/m² (4.3 oz/ft²), which is not that bad even compared to a honeycomb core.

Conclusion

Use vacuum bagging with very low pressures if weight is critial. It is more elaborate than vacuum resin infusion, but much higher volume fraction ratios can be achieved.

Update

I uploaded a video of the vacuum infusion. Enjoy!

Almost there

The top side of the center section is ready! It was a long and difficult path to get there…

We tried to prevent some problems we had last time with the lower side. Most prominent is that we produced a left and right side for the inner face sheet. Last time we had problems with wrinkles while draping the geled sheet. Using two halves reduces the risk substantially and is much easier in handling.

Production of the inner face sheet started with a dry layup on top of the pattern. Last time we used a drawed tamplate while laminating the sheet, which was difficult to set correctly up. It is much easier to make a dry layup on the pattern.

Flax fabrics:

p1070732_m

 

p1070730_m

Originally we wanted to vacuum infuse the face sheet and join it with the honeycomb core when geled.  However, we had massive problems to get a sealed bag and the infusion of one half failed. I think the following picture shows my mood that night:

p1070740_m

Already here it showed that having two halves is better than one, as we lost half of the sheet instead of all. The problem was that the bag was made of the wrong material: Never use a single layered PA bag for infusion, use a three layered PE/PA/PE bag. No matter if the PA bag is slightly cheaper. Usually a small leak is not a problem, as long as the part produced is not something like a visible carbon surface. But, this time the leak was directly at the suction line, so that the resin was not sucked properly and the laminate was not wetted. Anyway, we made the dry layup of one side again, we wetted the old way with a roller and vacuum pressed them for about twice the resin’s pot time.

After having glued the geled face sheet with 3M 9323 epoxy-resin on the honeycomb:

p1070794_m

The whole thing was cured in moderate vacuum for 24 hours and then heat treated for 15 hours at 60 °C (140 °F):

p1070800_m

 

We are proud of the result: very light and extremely nice

p1070805_m

 

We have to cut and level the joining edges of both the upper and lower sides. Also some ribs are needed to support the shape, in particular where the pilot will walk on top to enter the plane. Having done this, we’ll be finally able to join these into one piece. I guess the picure above gives anyway a good impression of how it will look like.

Merry Christmas

p1070780_s

Glass bubbles

Not many recipies are found when it comes to filling honeycomb cores with glass bubbles thickened resin. R&G provides here a list of reciepies for usual additives. They advice to use 30 g of 3M 0.2 g/cm³ glass bubbles for 100 g of resin. We will denominate this as 30% fraction from here on, though officially the mass fraction is only 23%.

The recipie of R&G is too thick to fill the small channels of a 3.2 mm honeycomb core. As a result, air is trapped in the core weakening it. Take, for example, this carbon sandwich:

P1070241_m

Using less bubbles provides a better mixture, which fills easily the channels. However, less glass bubbles means also more weight. So what is the best mixture?

To answer this question, we made some systematic tests with a low-viscosity resin from HP-Textiles designed for vacuum resin infusion (HP-E300RI). Four different fractions were tested:

p1070713_m

Cross sections show a separation of the mixture into phases of pure resin and glass bubbles:

wabenschnitt_beschriftet

This means that the mixture was oversaturated. Pure resin does not only add unnecessary extra weight, but also tends to have additional shrinkage leading to marks on the finished surface.

An estimation of the volume fractions and extrapolation delivers the optimal amount of glass bubbles (about 18.5%):

microballon

Having a figure for the optimal recipie, we made a further test in which the effect of adding a thixotropic agent (1.5% fraction) was investigated:

wabenschnitt2_beschriftet

It looks that we found the best recipie!

With a good feeling, we filled the core of Schneewittchen:

p1070720_m

Getting the sandwich ready

This weekend we started to organize everything needed to get the missing face sheet finished. Many things have to be done before we start with it, for example filling the core. The honeycomb core has to be filled with a glass bubbles-resin mixture where it is subjected to high localized loads. This prevents the core from failing and is, among other places, needed on the left side of the cabin where the pilot will walk on top of the shell to get into the airplane. Though the glass bubbles reduce much of the weight of the mixture, it is still much heavier than an air filled honeycomb core. Thus, the places to fill should be considered very thoroughly.

Last time we filled the core and glued the face sheet in one shot, which was extremely stressfull. It takes ages to mix enough filling material, and having a wating face sheet that is slowly but surely curing does not make it easier. This time we will do it in two steps and started today to mark the places to fill:

p1070708_m