Nose Cone Mold

We manufactured the nose cone of Kılavuz Kaizen, our TEKNOFEST 2026 rocket, in our own workshop. The 3D-printed mold set we designed for it consists of twelve parts in total: a six-piece female mold, a three-piece male mold and a three-piece core that fills the male mold from the inside. Below we explain both why we structured the mold this way and how the manufacturing went.

The part we made

The nose cone is a tangent ogive, 370 mm long with a Ø129 mm base. Our speed range was decisive in choosing the shape: the rocket flies at around Mach 0.9, just below the transonic band, and an ogive keeps drag low in this range. Conical or Haack-series shapes were also on the table, but the ogive is noticeably more tolerant of manufacturing errors. The mold work was hard enough already; we did not want the geometry to fight us too.

The shoulder dimension comes from the body. Our body tubes are Ø129 outer and Ø125 inner, so the shoulder sits directly inside the body. This joint is also our primary separation point: pressure from the hot gas generators shears the shear pins and starts the separation. So we could not use an interference fit here and designed the joint as a clearance fit.

The wall thickness was a safety margin decision. The theoretical calculation found 1.7 mm sufficient, but at the CDR stage we increased it to 2.0 mm along with all the body tubes. Our workshop capabilities and molding precision were limited; we did not want to leave room for a weakness caused by manufacturing. The 2 mm gap in the mold is the direct result of this decision.

How the mold works

Three sets work together. The body on which we hand-wrap the glass fiber fabrics is the male mold; its outer surface corresponds to the inner surface of the part we produce. The cavity of the female mold, which we close over the wrap and bolt down once wrapping is done, forms the outer surface of the nose cone. The 2 mm left between them is the laminate itself. The third set, the cores, goes inside the male mold with a gap-free fit; their job is to keep the 4 mm male shell from collapsing inward when we tighten the female mold with bolts.

We split all three sets into three along the axis: tip, body and shoulder. The reason was build volume. The cone is 370 mm, 570 mm with the shoulder; printing it in one piece was impossible. We split the parts to fit the printer's 256 mm build volume, and none of the twelve parts exceeds 255 mm on its longest edge — the largest, the female body half, sits just under the limit at exactly 255.00 mm. We also split the female mold into two symmetrical halves at the parting plane, because we are making a closed shell and the mold had to open.

We did not want to print the male mold solid; the body section alone would have taken 2.5 liters of material. Instead we split it in two. The outer 4 mm shell is printed at 100% infill, because its surface will be sanded and must be void-free. The inner core makes do with 30% infill and eight walls; all we need from it is to withstand the clamping load. The shell provides the precision, the core carries the force, and material drops to a third. We modeled the female mold with similar logic, as a shell wrapping the cavity with a constant 40 mm thickness, and printed it at 40% infill — it must not bend under bolt clamping load, but printing a solid block made no sense either.

There are two separate systems at the parting plane, because alignment and clamping are different jobs. The alignment pins that hold position are 2 mm high and arranged in two rows on each side of the cavity: 52 on the body half, 33 on the tip half and 14 on the shoulder. The bolts that provide the force pass through Ø4.3 mm holes along the flange; about thirty-three M4 bolts and nuts in total. Since warping is inevitable in a printed plastic mold, we did not want to rely on a single point for alignment and spread it over as many points as possible.

The whole set takes up 8.9 liters in total; with this recipe that means about four kilos of filament. The mold itself was a manufacturing job in its own right. All per-part print settings are in the attached “Print Details” document.

Printing and manufacturing

We printed the mold on a Bambu Lab printer with a 0.2 mm nozzle and ABS filament. The fine nozzle was chosen for surface quality: since the mold surface transfers directly to the part, we wanted to keep layer lines as small as possible. In return, print time increased considerably.

In the CDR report we had planned vacuum infusion for the nose cone, but in practice we did not use that method; we made the part by hand lay-up and compression in the mold. First we smoothed the mold surfaces by progressive sanding and applied mold release. Then we cut the glass fiber fabrics into pieces, hand-wrapped them onto the male mold, closed the two halves of the female mold over the wrap and tightened them with M4 bolts. The compression seats the laminate in the 2 mm gap between male and female: excess resin is squeezed out and the mold itself sets the wall thickness. We left the part to cure in this state.

Demolding was planned from the start. The nose cone is a closed shell whose only opening is the Ø125 mm shoulder mouth; the body section of the male mold is also Ø125 mm. So pulling the male mold back out whole was geometrically impossible. The solution was to sacrifice both: we broke the female mold off the part, and softened the male mold left inside with hot air to remove it. This was also one reason we insisted on ABS for the male mold — it softens in a controlled way under hot air instead of collapsing suddenly. As a final step we sanded the seam lines left at the parting plane.

The biggest gain from switching from infusion to compression was less preparation: no vacuum pump, flow mesh, sealant tape or bagging film, and no leak-check step. In return, control over the fiber-to-resin ratio decreases. In infusion the vacuum sets the ratio; here it comes down to how much resin we apply during wrapping and how evenly we tighten the bolts.

Outcome

We produced the nose cone in a usable state on the first attempt. There was no scrap and no second attempt was needed. The part was integrated into the flight configuration and we completed the reporting stages. The launch has not taken place yet; we will evaluate the nose cone's behavior under flight loads and landing shock after the launch.

We also made one revision to the nose cone along the way. In the CDR we had designed the internal plastic part that keeps the parachutes from jamming in the tapering tip with a Ø20 mm hole in the middle; when we saw how hard it was to pass the shock cord, carabiner and swivel through it, we enlarged the hole to Ø34 mm. We also changed its material from PLA to ABS, because at launch-site temperatures there was a risk that PLA would soften and fail to do its job. Where we should have avoided ABS in the mold, here we had to switch to ABS — a good example of how the material decision changes with the function.

Looking back

Our most obvious mistake was material choice. The female mold and the cores did not need heat resistance, since curing happens at room temperature; in these two sets ABS only brought us warping risk and a laborious print process. Instead of writing “doesn't matter, ABS preferred” in the print instructions, we should have simply said PLA. The only place where ABS really paid off was the male mold we removed with hot air. Likewise, instead of using the 0.2 mm nozzle for everything, we should have chosen per set: the male mold is the only set whose surface transfers to the part, and printing the other two with a larger nozzle would have saved a lot of print time.

That the mold ended up single-use is also worth thinking about. Because we broke the female mold and removed the male mold with heat, no part can be reused; four kilos of filament and a very long print went into a single nose cone. For the male mold this was unavoidable, the geometry left no other way. But not for the female mold — already split into halves, it should not have had to be broken; the release agent and draft were not sufficient.

On the surface side, the seam lines are a clear cost. Because we split the female mold in two, two opposite lines remain along the axis on the outer surface of the part; together with the tip–body and body–shoulder joints there are four lines in total, two longitudinal and two circumferential. We sanded them all smooth, but the 20 µm surface roughness we targeted depends on how well these lines are leveled — and we had also decided not to apply a PVC film wrap to the nose cone.

Finally, we made two documentation errors. The print instructions say “print 1 of each of the 9 parts listed below”, while the list has 12 parts; we wrote the document when there were six female and three core parts and did not update the number when we later added the male mold parts. An error that leaves room for misunderstanding by whoever prints the mold. Similarly, in the OpenRocket file the nose cone is still set as ogive, shape parameter 0.8, i.e. a secant ogive, while the mold CAD and the CDR text use a tangent ogive. Even if the effect on flight performance is small, the simulation model should have matched the part we produced.