Projects

Five projects, from sketch to hardware.

Different domains, same approach: strip the problem to physics, explore real options, then build and test what matters.

01 · Consumer product · Team of five

Oyster microfibre filter

A washing-machine filter that captures clothing microfibres before they reach the ocean. From an open brief to a costed concept to a tested prototype.

Exploded CAD assembly of the Oyster filter — 01 · Exploded assemblyMotor, housing, bearing cap, mesh cylinder and propeller insert on one shaft axis.

Final CAD render of the Oyster filter in its housing — 02 · Final CADPacks into the dead space behind the machine's service hatch.

Fatigue FEA result of the filter cage with a colour scale — 03 · Fatigue FEAThe analysis flagged the shaft crossing as the low-life region.

Four photos of the 3D-printed Oyster prototype and its test rig — 04 · Prototype & testAssembled unit, mesh cartridge, propeller insert, and the test rig.

1 / 4

The idea

Spin a cylindrical filter so the flow is driven outward through its whole wall. The filter area becomes the cylinder's surface, not the pipe's cross-section: far more area in the same space, and longer before it clogs.

Alejandro's part

Torque calculations, stress and fatigue FEA, the final mesh selection, prototype manufacturing and part of the concept sketching. Pump blades and the CFD study were led by teammates.

Result

On a repeat test run the printed shaft snapped exactly where the FEA had flagged. Cause: weak inter-layer shear from the print orientation. Fix: steel, a higher-modulus polymer, or an injection-moulded production part.

75.7 %fibre capture in the first test, with real microfibre-laden water

25 µmnylon mesh; beat stainless steel on mass and recyclability

12–18 Vlow-cost DC drive, sized from worst-case start-up torque

02 · Flight hardware · EuRoC 2021 · Technical lead

Darwin II airbrake

Active apogee control for a supersonic rocket: built to overshoot on purpose, then brake back to a 9 km target.

CAD render of the three-flap airbrake with central gear-rack and servo — Three flaps, one central gear-rack, one servo: the brake moves as one body.

Schematic of the control idea, not flight data.

Two constraints

Symmetry. If one flap opens before another, the asymmetric force yaws the rocket at supersonic speed. Three identical flaps on one central gear-rack and one servo make that mechanically impossible.

Fail-safe. No supplier had a torsion spring of the right size, force and travel, so retraction moved to a redundant control board on its own power supply that drives the servo shut if the primary fails.

What happened

On the flight, the airbrake never deployed, and the mechanism wasn't the reason. The integrated control system was tested for the first time at the competition itself. The lesson we carry into every project: test the integrated system before it matters.

Alejandro's role

Technical lead of Endeavour, the University of Edinburgh student rocketry team (about 60 people). The airbrake was a team build.

×23safety factor on the aluminium, from ANSYS FEA with a deliberately pessimistic drag coefficient

11.3 MPapeak stress, against 250 MPa yield

133 Nload per flap, fully deployed

0.1 mmworst-case flap deflection

Mach 1.7maximum speed of the rocket

03 · Test engineering · Solo project

Five test rigs for chalk

Measuring the failure strength of ordinary chalk in bending, tension, shear, torsion and compression, with no lab and no testing machine. Every rig was designed and built from workshop scrap.

Chalk sample clamped as a cantilever with a water bottle hanging as the load — BendingCantilevered sample, loaded by a water bottle filled until fracture and weighed on a kitchen scale.

Chalk sample hanging from a hook with wire knots, a water bottle hanging below — TensionTwo opposed constrictor knots keep the load axial.

Twin-plate guillotine shear rig clamped to a workbench — ShearTwin-plate guillotine, version 2: four guide rods stop the plate rotating and binding.

Wooden torsion rig holding a chalk sample with a lever clamp — TorsionA grease-packed homemade bearing isolates torsion; residual friction is measured and subtracted.

Compression rig: a cylinder held vertical by four guide wires on a wooden board — CompressionA piston in a slotted cylinder, kept axial by four guide wires.

Chalk sample with a helicoidal fracture surface next to the torsion rig — EvidenceHelicoidal fracture from the torsion rig: proof the load was isolated.

1 / 6

The constraint

Each load had to be isolated cleanly, or the number would be meaningless. The first shear rig let the plate rotate and added friction to the reading; version 2 removed the parasitic moment.

The result

Consistent, physically sensible values on all five rigs, with the spread growing as the apparatus got more complex, exactly as expected. The real deliverable is the method: a defensible answer from improvised tools and clear reasoning.

Mean failure force per rig

Newtons · 3–4 tests per rig · hover or tab to a bar for the individual tests

Compression249.9 NTests: 254.8 · 235.2 · 259.7 N
Shear55.5 NTests: 56.8 · 54.9 · 54.9 N
Tension28.6 NTests: 29.4 · 30.0 · 27.1 · 27.6 N
Bending4.9 N4 tests, 4.7 to 5.3 N
Torsion2.9 NTests: 2.8 · 2.5 · 3.4 N

Compression and tension load the same cross-section, so the almost 9× gap shows how much stronger chalk is in compression. Bending and torsion forces act through a lever arm, so those bars compare rigs, not materials.

04 · Concept & embodiment · Solo project

Seabed drilling rover

A tethered, remotely operated vehicle that drills the ocean floor: a novel machine with no off-the-shelf precedent to copy.

Architecture

A sealed chassis deployed by cable and winch, with the drill on silent-blocks like a car engine, isolating vibration and keeping the design modular.

Mobility

Four independent in-wheel motors and Mars-rover-style swing arms that adapt to uneven seabed and lock flat while drilling.

Materials

Ashby-chart selection against explicit objectives: leak-before-break for the chassis, hardness and corrosion resistance for the wheels, flexibility without failure for the arms.

Cost modelDe-risking: Bottom-up estimate by subsystem, with adaptation factors for bespoke development.
FMEADe-risking: Every subsystem rated severity × probability. The winch cable and the chassis seal came out as the risks to control first.

05 · Thermal concept · Biomimicry · Solo project

Passive cooling from termite mounds

Termite-inspired cooling had only been used in large commercial buildings. The challenge: apply the same principles to an ordinary family house.

Hand sketch of a house cross-section showing the convective flow circuit — Convective loop: outer walls drive a stack-effect flow; air is cooled against high-thermal-mass walls and by evaporation.

1 · ExtractDistil the mound to its mechanisms: convection between low- and high-thermal-mass structures, and induced flow.

2 · TranslateOuter air gap, high-mass inner walls and conduits, breathing outer walls, a chimney network, optional evaporative cooling.

3 · CheckA first-order stack-effect calculation before any claims, with a full thermo-fluids study as the next step.

≈0.46 m/sstack-effect airflow, first-order estimate for a 100 m² house

Start a project

Have a product to build?

Tell us what it is and where it stands today. We'll tell you honestly how we can help.

alejandro@vgengineering.ch Ways of working