The jet of water exiting the bottle is constrained only by the bottle's internal neck diameter. No special adapters or reducers are required, which lowers the barrier to entry because any standard soda/seltzer bottle may be launched as-is. It is the simplest configuration possible.

Several excellent existing designs using the Gardena garden hose adapter. In many parts of the world, including the US where I live, these are not readily available. Some launchers use custom 3D printed nozzle/base interfaces, but this was insufficiently reliable due to the porosity of 3D prints and weakness to shear forces.

While a full-bore nozzle approach has inefficiencies related to drag from higher peak velocities, it also allows for use of a “launch tube” where a PVC pipe is inserted into the bottle neck. This guides the rocket during liftoff and reduces the chance of things going (literally) sideways. 

Picture: Freeze-frame of rocket launch showing the full-bore diameter column of water (upper frame, above lights). Version 1.2, Core B Universal. 

Axially-directed (i.e. upward) launch forces are contained by triggered-release clamps that hold the bottle to the launcher which are functionally distinct from parts that contain pressure in the bottle. This decoupling allowed latitude for separately optimizing these two key functions so each was maximally robust. This was critical to the iterative design process especially in early phases when I didn't understand basic aspects of global soda bottle neck geometries and what pressurized air system parts were easily available.

This image of the orange Alpha-stage prototype launcher demonstrates early Clamp/Collar/Pin development focusing on axial forces only. Note the lack of an O-ring or any pressure containment.

An example of a coupled configuration would be a classic friction-fit wine cork. The cork both holds the rocket down by friction and provides the sealed pressure envelope. This configuration does not allow for a triggered release, which is important for safety and the visceral joy of counting down to a launch.

Simplicity suggested clamping directly to the flange found on the neck of soda bottles. These exist specifically for clamping to pressurizing equipment for injection blow-molding and bottle-filling.

You may recognize inspiration from the ingenious Clark cable-tie retention mechanism, which uses nylon cable tie heads to interface to the bottle neck along with a pipe collar that holds them in place. This design uses six 3D printed Clamps that pivot from the Base and hold onto this flange. (n.b.: capitalization indicates a specific 3D model by the same name.) Initial versions prior to v1.4 required three separate Clamp designs to fit different bottles found worldwide. 

With coordinated Core and Clamp redesigns in 2023, a universal Clamp design (pictured) now securely holds all known (to me) soda/seltzer bottles in modern production.

Expanded discussion about Clamp design can be found in a further discussion about the Base subassembly.

A circular-cross-section O-ring forms the interface between the Core and the bottle and fits inside the bottle neck, as per the image (green O-ring). This allows for over 15mm of axial positioning margin, which is critical when harmonizing with the range of standard soda bottles.

O-rings are robust to countless cycles of use and easily sourceable worldwide. In this regard, they are superior bottle interfaces compared to expanded PVC pipe sections, wrapped tape, and custom-cut rubber gaskets.

The challenge was to finding the right O-ring out of the thousands of options.

The promise of 3D printing is that a part can be printed anywhere in the world. However, these parts are limited by their material properties, and the intrinsic porosity of fused-deposition modeling (FDM) plastic 3D prints mean that holding pressure with exclusively 3D-printed parts is an exercise in futility. This requires use of specialized non-printed parts.  These parts must be easy to source globally.

Early Alpha and Beta designs required use of ASTM Schedule 40 PVC pipes and fittings (common in the US) and brass adapters between NPT pipe threads and Schrader valves (challenging to find even in the US).

A great deal of effort was expended on finding a region-independent solution. The search led to the four non-printed parts pictured: the green O-ring, black ice-cream cone shaped TR414 tire valve stem, black air hose extension, and red plastic straw. They are pictured disassembled with the red 3D-printed universal Core B and an orange 3D-printed Schrader valve extractor, then pictured assembled into Core B and inserted into a bottle. No adhesives or additional tools are required.

It happens that one specific standard O-ring – the nitrile rubber 15 ID x 22 OD x 3.5mm CS O-ring – perfectly interfaces between the bore of a standard soda bottle (21.74mm) and one particular dimension of a TR414 tire valve (15.2mm). This same O-ring can be used to interface between the bottle and 3D-printed parts for the advanced Core A and C designs.

These are parts that the global supply chain has made widely accessible. The O-ring is a standard size. The TR414 tire valve is a standard automobile/light truck tire valve made by the millions each year. The air hose extension is commonly attached to doubled tires on light trucks/RVs/trailers but similar can also be found as extensions for bicycle pumps. The straw can be any similarly-sized tube. 

These parts enable a remarkably robust pressure-containment system anywhere in the world there are cars, for less than $5 USD.

In initial Alpha and Beta designs, the Core and Base were physically inseparable. A failure in one or the other meant throwing the whole thing out, such as the first pictured launcher with fractured Core after a bolt worked its way loose (n.b.: always use a lock-nut). Failures were common in early stages.

Thankfully, the decision to decouple rocket retention from pressure containment allowed separating the Core and Base into separate subassemblies. Once separated, testing cycle time dropped from weeks (to build full Base/Core systems) to days (rapid iteration of Core designs).

In the current design, Cores are removable (second picture) and interchangeable between launchers. Three separate Core types are available. Further discussion of each design is in [*** Design Notes: Core].

Additional usability benefits emerge: water-filed rockets can be loaded with upside-down Cores away from the launcher and set up for launch rapidly. First. this avoids the usual water spillage when rockets are loaded onto vertical launchers. Second, multiple Cores can be set up with a single launcher to shorten time between launches. When dozens of rockets are launched (e.g. Scout activities) this is especially helpful.

3D-printed parts are subject to slicer software/printer/material variability. Additionally, 3D-printed parts are anisotropic, meaning that they have different material properties in different directions. This is demonstrated with the pictured Beta launcher fractured along layer lines, which are weak to shear and tension forces. 

To some extent, these vulnerabilities can be mitigated with geometry. In practice, this means:

• Overhangs are generally limited to 45 degrees (beta version had many 60 degree overhangs)

• Bridges are short

• Print bed interfaces are wide for good bed adherence with some chamfer to allow for first layer squash

• Low sensitivity to print bed flatness

• Serpentine walls and enclosed cavities to increase layer bond strength (most strength is in the walls, not the infill)

• Part interfaces allow for generous tolerances without jamming

• Layer direction is optimized for forces (e.g. Clamps printed sideways lying on their edges)

Some of these features can be seen in the pictured Version 1.2 CAD cutaway view. 

For the universal launcher, all parts can be printed without supports and are very strong with 3 walls and 20% infill. The launcher has evolved from failure under light launch conditions to robust to an 80kg dynamic load (me jumping on it), even with standard PLA filament.