A grade 11 physics build: a bottle rocket powered by pressurized air and water, with a parachute recovery system to bring it down safely. We weren't allowed to test before the graded launches, so every design decision rode on judgment alone.
System 1 — pendulum trigger. A string and an internal weight were meant to detect when the rocket flipped at apogee and release the cap. The airflow during flight held the cap on, so it never released and the rocket crashed, destroying the mechanism.
System 2 — pressure-balance nose cone. A cleverer design relying on the rocket being near-stationary at apogee with no airflow on the nose cone. An internal balloon and external rubber band pulling on a side flap would overcome the (now-absent) air force and pop the nose cone, deploying the chute. A hairpin flap held everything in place until launch, when it was pushed down — leaving only flight airflow holding the cone on.
It also failed: I placed the parachute below the balloon, which blocked deployment. Reversing the order would have worked. Static and friction on the large garbage-bag chute likely didn't help either — a different material would have been better. In grade 11 I didn't think material choice mattered much; looking back, it clearly does.
Both recovery systems failed to deploy, but the post-mortem was the real value: a clean root-cause chain pointing at deployment sequencing and material choice — lessons that have stuck ever since.