CAD modelling, manufacturing, analysis and testing. Select a project to see the full write-up.
A lightweight sign that spells “hoco?” in glowing cursive, built on a black 3D-printed frame and flown beneath a Ruko F11 Gim2 drone at dusk as an invitation to homecoming.
A six-fingered, radially symmetric claw driven by a hydraulic syringe through a rack, bevel and spur gear train. The fingers scoop a table-tennis ball up into a chamber, then an iris diaphragm closes to hold it in place.
Stress data from a static simulation controls how a lattice is built inside the connector that joins the claw to its arm: material is concentrated where the part carries load and left open where it does not.
A 3D-printed guard modeled over a 3D scan of Nick’s own hand, built from photos in Meshroom, made to slip inside his goalkeeper gloves and protect an injured hand without getting in the way.
Designed and built for a toy company brief: a spring-loaded board throws a marshmallow into the path of a rubber-band paddle, all on a base that swings the shot with a bevel gear.
Six ways of taking mass out of the same 150 × 40 × 10 mm printed part, designed in nTop, built to about the same weight and broken three times each on a home-built bend rig to see which keeps the most strength.
A science fair project asking whether a wheel with its weight off-centre speeds up and slows down differently from a balanced one. A 3D-printed wheel spun on bearings, washers set the imbalance, and a light sensor timed every rotation.
A science project that flew the same rocket on different sized motors and measured how acceleration and final altitude changed with each one.
I am a high school senior applying for undergraduate mechanical engineering programs. My passion lies in solving practical problems with my CAD, additive manufacturing and maker skills.
Whether I am hanging a glowing sign from a drone, fitting a guard to my own hand, or building an Arduino rig to measure the friction of a spinning wheel, I approach the problem pragmatically and within the real world constraints. In the end, the drone still has to fly, the guard has to fit inside a goalie glove and the Arduino has to run fast enough to read the speed sensor’s data.
My collegiate goals center on enhancing my building skills by developing the mechanical engineering rigor that will allow me to solve bigger problems than will fit on my desk and help more people than just me.
A lightweight sign that spells “hoco?” in glowing cursive, built on a black 3D-printed frame and flown beneath a Ruko F11 Gim2 drone at dusk as an invitation to homecoming.
The word was drawn as one flowing cursive line so flexible LED strips could trace it, with loops for each letter and a question mark at the end.
A black 3D-printed spine with thin bracing rods holds the strip in the letter shapes and gives the drone one rigid place to attach, without adding much weight.
The strips run through a switch and a 5 V regulator from a 2S LiPo battery pack. The sign is hung beneath the drone and flown at dusk, when the glow stands out against the sky.
To find how much the drone could carry, a basket was hung beneath it and nails were added a few at a time until it reached its limit.
The sign is only a few millimetres wide, so it is easiest to read against a darkening sky. The black frame disappears into the dark, and the letters seem to float.
Modelled, not measured: a straight line between Adafruit’s two published points for a 1 m strip (3 V at 200 mA, 5 V at 1.2 A), with both strips sharing one 10 Ω resistor. Battery life assumes 80% of the 1400 mAh pack is usable and a 85% efficient regulator.
A six-fingered, radially symmetric claw driven by a hydraulic syringe through a rack, bevel and spur gear train. The fingers scoop a table-tennis ball up into a chamber, then an iris diaphragm closes to hold it in place.
Each finger is a multi-link arm on a ring around the body. The same finger repeats every 60°, so the six fingers close evenly and scoop the ball up into the chamber.
Once the ball is inside the chamber, a door gear turns three door arms on a ring and closes the iris across the opening, so the ball stays in without the fingers staying tight.
A rack and bevel gear turn the syringe’s straight-line push into rotation. The mechanism was modeled and animated in CAD, drawn up with numbered parts, then printed in PLA and assembled with pins.
The gear shaft carries the motion down the chassis to the door gear, which closes the iris.
Stress data from a static simulation controls how a lattice is built inside the connector that joins the claw to its arm: material is concentrated where the part carries load and left open where it does not.
A static simulation of the connector shows where stress concentrates, here as the bright ridge between its two raised bosses.
The stress data drives the lattice directly: the size of each cell and the thickness of its walls change with the stress at that point.
The part keeps a solid shell and fills the inside with a lattice that is denser in the load-bearing areas, so material is prioritised there.
In the stress plots, lighter colour means greater stress. Compare each stress plot with the lattice cross section beside it to see the cells follow the load.
A 3D-printed guard modeled over a 3D scan of Nick’s own hand, built from photos in Meshroom, made to slip inside his goalkeeper gloves and protect an injured hand without getting in the way.
The hand was photographed from many angles and the photos were turned into a 3D mesh in Meshroom, free photogrammetry software that works out the shape from where features appear across the photos. The mesh records the hand’s true contours, such as the creases, bony points and curves that a standard shape would miss.
The guard was built directly over the scan mesh in CAD, so its inner cup follows the shape of the hand and its outside is smooth and chamfered to sit inside a glove.
The guard was printed in black filament and checked in the hand, holding it in a closed grip to see how it sits and where it presses.
Designed and built for a toy company brief: a spring-loaded board throws a marshmallow into the path of a rubber-band paddle, all on a base that swings the shot with a bevel gear.
The brief was a marshmallow launcher for young kids that reaches different distances and has something exciting about it, with the launching system held in tension. Limits: a 10 × 10 × 10 in workspace, a trigger, an unmodified marshmallow and no more than 12 in of tape.
After sketching several designs, I settled on a springboard that throws the marshmallow into the path of a swinging paddle, then added a rotating base so the shot could be aimed with a bevel gear turned by a crank.
Almost every part was printed so it could be remade quickly; the bases and dowels were cut from wood by hand. Fit problems were found in assembly and fixed in a second round of printed parts.
Four test shots with the rubber bands on the top hooks, measured to where the marshmallow first landed and to where it stopped rolling.
Pulling the handle slides the draw back, which swings the release arms off the springboard, so the spring throws the marshmallow upward. The same motion frees the retainer, and rubber bands whip the paddle round to hit the marshmallow in the air.
Six ways of taking mass out of the same 150 × 40 × 10 mm printed part, designed in nTop, built to about the same weight and broken three times each on a home-built bend rig to see which keeps the most strength.
Consistent infill (the control), variable infill thickness, variable cell size, a plain gyroid lattice, ribs, and a topology-optimized shape, all made in nTop at the same designed mass. Three of them (both variable infills and the topology-optimized shape) were driven by simulation results.
Each part sat on two supports with a rod across its centre. A lever pulled the cord down slowly until the part broke, and the peak force on a 500 N sensor was recorded. Every design was broken three times.
The simulation predicted that topology optimization would carry the most. In the lab it carried the least, so the prediction and the physical result were compared to see why.
Load at failure, with the average mass of the printed parts:
Print time, filament and filament cost to make three of each design:
It had the lowest peak stress in the simulations, so it was predicted to carry the largest load.
The consistent infill carried the most (about 497 N, with one trial past the sensor’s limit) and topology optimization the least (about 51 N).
A science fair project asking whether a wheel with its weight off-centre speeds up and slows down differently from a balanced one. A 3D-printed wheel spun on bearings, washers set the imbalance, and a light sensor timed every rotation.
The idea started with bike lights whose battery pack sat on one side of the hub. If the weight is not over the axle, does it affect how the wheel performs? The hypothesis was that it would not, because the average friction on the bearing is the same over a full turn.
The wheel was designed and printed with a lug for the pulling weight, a stopper so every test starts from the same point, a visible mark for tracking, and holes for washers on each side. Two rigs held it: one with the wheel upright and one with it flat.
Reflective tape covers a quarter of the wheel. A light sensor on an Arduino-based board calibrates a threshold from its own highest and lowest readings, then logs the time of every rotation and the time since the start.
Average of four trials. The same total mass, split differently between the two sides of the wheel:
Average of four trials. Rotations made before the rotation time passed 200 ms, and in total:
A wheel with its centre of mass off the axis should speed up and slow down at the same rate as a balanced one.
They accelerated in less time and also slowed in less time. Slight imbalance did not change slowing much, but the more uneven the weight, the fewer rotations the wheel made before stopping.
A science project that flew the same rocket on different sized motors and measured how acceleration and final altitude changed with each one.
Used each motor’s published thrust curve in a simple simulation with drag to predict acceleration and apogee before launch day.
Packed a small barometric altimeter and accelerometer into the nose bay and padded it so it survived the launch and landing.
Flew each motor three times, averaged the results, and checked the altimeter against angle measurements from two ground observers.
Altitude at the top of the flight, averaged over three launches per motor:
Measured by the onboard accelerometer during the motor burn:
Peak acceleration rose only slightly with motor size. The big gain in altitude came from the longer burn time, which kept pushing the rocket up while it was already moving fast. Drag made the real apogees lower than the no-drag prediction by about a quarter.