Home/Mechanical/CAD, drawings and coursework

Side view photograph of a yellow Dolphin torch on a carbon-fibre patterned surface — One of the reference photos the Dolphin torch was rebuilt from.

CAD, drawings and coursework

WSU coursework · SolidWorks, ANSYS, MATLAB · 2023–25

The rest of the degree worth showing, from 2023 to 2025: drawing sets, mechanisms, structural and fluid analysis, robotics, and a design for the circular economy.

Dolphin torch drawing set, 2023

This was my first full drawing set, for ENGR2024. I photographed a Dolphin torch from every side, rebuilt it in SolidWorks and drew it up as 13 A3 sheets: an isometric, an exploded view with an 11-item bill of materials, then one sheet per part. Each part carries its manufacturing process and material, mostly injection-moulded HIPS and ABS PC, with rubber, copper and aluminium for the button and contacts. Moulded products like this are mostly draft angles, blends and datum choices you can't see from outside, which made it good practice for design intent.

Shaded SolidWorks isometric of the yellow and black Dolphin torch — The rebuilt torch in SolidWorks.

Reference photo looking into the open torch body at the switch carrier and contacts — Inside the real torch: the switch carrier and contacts, drawn on sheets 6 to 8.

Drawing sheet: exploded view of the Dolphin torch with an 11-item bill of materials — Exploded view with the bill of materials: part, quantity, manufacturing process, material and drawing number for all 11 items. Sheet 2 of 13.

A3 part sheet for the torch's switch link, fully dimensioned with a render — The switch link, ABS PC and aluminium. Sheet 6 of 13.

E-bike drawing package, 2023

The e-bike was a group project, with a rear hub motor and the battery on the down tube. I drew the full package in SolidWorks: 55 A4 sheets, starting with four assembly sheets for the bike, both wheels and the disc brake, then one sheet for every part, from the aluminium frame to the brake caliper, pads and shaft collars. Every sheet carries its material and finish, with general tolerances of ±0.1 mm and ±1°.

Shaded render of the e-bike: red frame, battery on the down tube, rear hub motor and disc brake — The e-bike, from the main assembly sheet.

E-bike main assembly drawing with top, side and front views — Main assembly, sheet 1 of 4.

Tube frame part drawing with tube angles, lengths and diameters — The aluminium tube frame, one of 51 part sheets.

Brake caliper sub-assembly drawing with a red render — The disc brake assembly.

Mechanisms in motion, 2023

MECH3005 had two individual labs in SolidWorks Motion, run alongside the unit's group gearbox project. The first was a four-bar linkage: model the base and links, draw them up, drive one link with a motor, then plot the angular velocity and acceleration of every link and the path a point on it traces. The second did the same for a belt-driven fan on two pulleys, with pulley sizes set for each student.

SolidWorks render of the assembled four-bar linkage on its base — The four-bar linkage.

The four-bar linkage with its trace-path curves overlaid — Trace paths of points on the linkage.

Eight SolidWorks Motion plots of angular velocity and acceleration for links 1 to 4 — Angular velocity and acceleration of each link over five seconds.

Belt-driven fan and pulley assembly with the blade-tip trace path — The belt-driven fan, with the blade tip's trace path.

Fluid dynamics in MATLAB, 2024

Here I wrote my own solvers in MATLAB. The first half modelled a cylinder oscillating in waves, with drag, inertia, a spring and a damper, stepped with fourth-order Runge-Kutta to find how much power the damper could extract: about 43 W in the base case, and up to about 110 W at the best damping. The second half solved flow past square and rectangular prisms, found a stable time step by testing Courant numbers, and measured drag and Strouhal number from the vortex street behind each one.

Plot of instantaneous damper power over 40 seconds with its time average — Instantaneous damper power, averaging 42.8 W.

Plot of time-averaged power against damping coefficient, peaking near 700 — Power against damping: the best damping extracts about 110 W.

Velocity vectors behind a square prism with vortices numbered 1 to 4 in red — The vortex street behind a 10 mm square prism, vortices numbered.

Plot of Strouhal number rising with Reynolds number for square prisms — Strouhal number rising with Reynolds number.

Heat exchanger CFD, 2024

My first Fluent project, for MECH4001, compared three shell-and-tube designs: a multi-bend tube in counter-flow, the same tube in parallel flow, and a straight tube in counter-flow. They moved 313, 303 and 156 W. Counter-flow beat parallel flow by about 3%, and the straight tube did best per unit area but, with about a third of the area, moved half the heat. The energy balance closed within 0.03 W on the straight design but was out by 22 to 28 W on the bent ones, because I'd coarsened their meshes to finish in time.

Half-section SolidWorks model of the multi-bend heat exchanger: blue shell, copper tube looping the length of it — The multi-bend design, which moved the most heat, modelled as a half section in SolidWorks and exported to Fluent.

Half-section SolidWorks model of the straight-tube heat exchanger: blue shell, one copper tube running diagonally through it — The straight tube: the best per unit area, but with about a third of the area.

Temperature volume rendering of the hot fluid through the multi-bend tube in counter-flow, 80 degrees Celsius falling to about 61 — Hot-side temperature through the multi-bend tube in counter-flow: 80 °C in, down to about 61 °C.

Temperature volume rendering of the hot fluid through the same tube in parallel flow, 80 degrees Celsius falling to about 64 — The same tube in parallel flow only gets down to about 64 °C.

Temperature volume rendering of a multi-bend tube heat exchanger — Temperature through a multi-bend design.

Cold-side velocity streamlines fanning around the shell — Cold-side streamlines through the straight-tube design.

Robot arm control, 2024

I built independent joint control and trajectory planning for the first three joints of a PUMA robot arm in Simulink, for MECH4004, with payloads up to 1 kg. The joint inertias came from the parallel axis theorem, the controller gains from the arm's 10 Hz structural resonance, and the joint speeds from the motor limits. The report is honest about its limits: large, fast moves could still send the model unstable, and the trajectory planning needed more debugging. A second assignment in the same unit trained an image classifier with MATLAB's Deep Learning Toolbox.

Simulink model with position and velocity gain loops for three joints feeding a PUMA model block — The Simulink model: a control loop per joint around the PUMA model.

Simulink scope of the three joint velocities over the 20 second test run — Joint velocities over the test run.

Mobile robotics, 2024

MECH4003 set two MATLAB assignments. The first modelled a tricycle robot and compared front-wheel drive with rear-wheel drive through the same manoeuvres. On the full speed profile, the front-drive version understeered and couldn't make the turn. It then localised the robot with a 1,000-particle filter on noisy range and bearing readings to landmarks. The second found lane lines in road photos, using thresholding, thinning and a Hough transform to fit them and draw a driving line down the middle. Its main lesson was how much the threshold tuning mattered.

XY paths of front and rear-wheel drive through a U-turn; the front-drive path runs wide — Rear-wheel drive makes the turn; front-wheel drive runs wide.

Hough transform space with the two lane-line peaks marked — Hough transform space, with the two lane lines as peaks.

Road image with detected left and right lane lines and a red driving line between them — Detected lane lines, with the driving line down the middle.

Modular climbing dome, 2024

For ENGR3025 I designed a playground climbing dome with a slide, for children aged about 3 to 12, around a circular economy. The frame is modular 3003 aluminium beams, so a broken beam can be swapped instead of the whole dome, with recycled LLDPE panels, stainless fasteners and nylon washers. Materials were chosen in Granta EduPack: aluminium takes more energy to recycle than steel, but won on weight, corrosion and service life. Each material has its own end-of-life route, and a maintenance plan sets when each part is inspected or replaced.

Hand sketch of the climbing dome and slide with frame joint details — The dome's structure, sketched by hand.

Hand-drawn life-cycle loop from virgin and recycled inputs through use to recycling — The product's life cycle, material by material.

Granta EduPack chart of yield strength against density for steels and aluminium alloys — Material selection in Granta EduPack: yield strength against density.

FEA fundamentals, 2025

The individual half of MECH4002 ran alongside the suspension project, and I wrote it up in LaTeX. It started with springs, bars, trusses and beams solved by the direct stiffness method in Excel and checked against ANSYS, then element formulations by hand: four-node quads, constant-strain triangles and bricks. The clearest lesson was a plate modelled with just two constant-strain triangles, which read peak stress about 33 times lower than a fine mesh, because a constant-strain element can't capture bending. The last assignment covered shell elements, cyclic symmetry on a pulley and a pre-stressed modal analysis.

ANSYS von Mises stress on a plate meshed with two constant-strain triangles — Two constant-strain triangles: peak stress 0.31 MPa.

ANSYS von Mises stress on the same plate with a fine mesh, showing bending — The same plate on a fine mesh: 10.3 MPa, with the bending the triangles missed.

ANSYS von Mises stress on a hollow cylinder modelled with shell elements — Shell elements on a hollow cylinder, half symmetry.

One 45 degree cyclic-symmetry sector of a pulley with von Mises stress — Cyclic symmetry: one 45° sector of an eight-hole pulley.

Ten pre-stressed mode shapes of a post on its base — The first ten pre-stressed mode shapes.

What I took from it

Across these units I learned most of the tools I use now: SolidWorks for drawings, motion and simulation, ANSYS Workbench and Fluent, MATLAB and Simulink, Excel for matrix methods, Granta for materials and LaTeX for reports. Some of it was solo and some was group work, and all of it ran to deadlines, which taught me to scope an analysis to the time available and say plainly in the report what that cost, like the coarser heat exchanger meshes.

Message: Send me a message
LinkedIn: in/jeremy-alec-osborne
Based in: Sydney, NSW