Lemon Launcher water rocket

The brief

Our product design specification framed the rocket as an educational kit for A-level and first-year university students: a standard 2 litre PET bottle powered by pressurised water, with a 3D-printed nose cone and fins, and a BBC micro:bit on board to measure flight data. It had to be easy to assemble, safe to launch, reusable and able to show measurable performance.

The team and my role

Ten of us worked in three sub-teams. The design team did the concepts, CAD, flow simulation and printing. The research team covered materials, costs and test support. The communications team built the website and ran the presentations. I was in the research team, and my part was researching candidate materials for the printed parts, with sources, which went into the materials comparison on our project website. Teammates did the CAD, flow simulation, thrust tests and telemetry.

How it was designed

Concepts

The design team modelled three fin shapes (classic, delta and swept-back) and seven nose-cone variants in SolidWorks, and ran Flow Simulation on the nose cones.

Final design

The bottle (36 cm long, 10.2 cm in diameter) carries a nose cone with a 114 mm maximum diameter, a 117 mm fin sleeve, and four swept fins with a 100 mm root, a 30 mm tip and a 75 mm span, each located by two pins.

Manufacture

The printed parts took about 7 hours on an FDM printer.

Telemetry

One micro:bit on board measured vertical acceleration on its 8 g range and radioed it to a second micro:bit plugged into a laptop.

Safety

Our risk assessment covered bottle burst, sharp edges and impact on people, with deburring and a cleared launch area. The launch was planned for 20 May 2025 at Falmer.

(opens full size)

(opens full size)

(opens full size)

Materials research

Our criteria were high stiffness, low density, good dimensional accuracy and reliable 3D printability. We compared six candidates: three printable plastics (PET, HDPE and LDPE) and three fibreboards (LDF, MDF and HDF), which can be machined but not printed. We chose PET for its strength, printability and low cost. We rejected HDPE and LDPE for their lower tensile strength, and the fibreboards for low strength and harder manufacture.

Results

The research team ran seven static thrust tests, logging force at 10 kHz.

Material for the printed parts came to about £8.43 (0.56 kg of PET at £15/kg).

Looking back

(opens full size)

We built and tested a working rocket as a team, but our evidence was thinner than our website claimed.

• We measured the wrong thing. For an ideal jet, peak thrust is about 2 × pressure × nozzle area, so at a fixed pressure it shouldn’t change with fill volume. What fill volume changes is how long the thrust lasts. Total impulse (thrust integrated over time) was the number to compare, not peak force.
• The data were too noisy to support conclusions. Repeating 910 ml at 4 bar gave 360 N and 249 N, 37% apart, and the only 5 bar run came out lower than most 4 bar runs. Our highest reading is above the ideal limit of about 304 N at 4 bar, so it was probably a transient spike or rig ringing rather than real thrust. We didn’t tare or filter the 10 kHz signal, and we ran too few repeats to tell.
• The telemetry couldn’t measure the boost. The micro:bit’s accelerometer tops out at 8 g, but the boost phase is several times that, so the readings would have clipped exactly when they mattered.
• Our targets were never tested. The website promised 200+ ft and 15–20 s flights, but no flight was measured. Even with no drag, a simple estimate gives under 200 ft, and a 15–20 s flight would need a parachute, which we didn’t have.
• Our specifications contradicted each other. The website, poster and risk assessment gave maximum pressures of 5, 8 and 10 bar, and the material appears as PET, PLA and ABS in different places.
• Fit. The nose cone and sleeve bores (100.27 mm) were smaller than the 10.2 cm bottle diameter we’d recorded, and the nose cone had to be resized two days before launch.

My own part, the materials research, had weaknesses too:

• The property values didn’t match our parts. We credited PET with a 9 GPa modulus and 150 MPa tensile strength. Typical unfilled PET is nearer 3 GPa and 55–75 MPa, and printed parts are weaker again between layers. Our figures were closer to oriented film or glass-filled grades.
• The reasoning was muddled. We listed PET’s high density as a benefit when our own criterion asked for low density. We rejected HDPE and LDPE as unsuitable for high pressure, but the printed parts are never pressurised: the bottle is the pressure vessel. The real reasons to avoid them are low stiffness and how hard they are to print, as both warp and grip the print bed poorly. The table also had placeholder zeros for some fibreboard properties, and priced PET as bulk resin ($1.52/kg) while our cost analysis used £15/kg.
• We never recorded what we printed in. After selecting PET, our documents list PLA and ABS for the actual parts.

What I’d do differently

• Write the specification first, with one set of numbers and a test method for each target.
• Rank materials with a weighted matrix on specific stiffness and strength (E/ρ, σ/ρ), impact toughness, printability and cost, using data for printed filament rather than bulk resin, and record the material actually used.
• Run at least three repeats per test condition, tare and filter the load cell, report total impulse alongside peak thrust, and compare each result with the 2·Δp·A estimate.
• Measure apogee directly with a barometric altimeter, or use an accelerometer rated well above 8 g.
• Measure the real bottle and print test rings before printing full parts.