Robert Pemberton · London, UK
Engineering Design Portfolio
This portfolio is a collection of my mechanical and aerospace design work. It covers academic research and industry projects, including manufacturing and test equipment.
Some of my best design work is directly on products released to the market; however, I am unable to share any pictures. The Product Design section describes the key points of each project.
Fixture design
Five-axis manufacture
Parametric geometry
Flow simulation
Printed & assembled
Handheld tooling
Selected Work
Aeroacoustic Research
Dyson-funded · Institute of Sound and Vibration Research
An experimental investigation into applying noise-reduction methods developed for flat plates to outlet guide vanes (OGVs) and a high-speed rotating fan, while measuring the resulting acoustic and aerodynamic tradeoffs.
1 · Research question
The research investigated whether leading-edge serrations and slits, or porosity applied across a surface, could reduce noise when transferred from flat-plate research to curved OGV airfoils while minimizing aerodynamic loss.
2 · Geometry challenge
Dyson supplied solid-body geometry to protect intellectual property. The OGV curvature and local angle change continuously along the span, so every cut had to follow the local surface. After SolidWorks proved too slow for the required iteration, I moved the work to Siemens NX and produced nine printable designs.
OGV design development
Two porous, three slitted and four serrated OGV concepts varied the geometry of each mechanism so its acoustic effect and aerodynamic penalty could be tested.
Local reference planesCylindrical cutting bodiesPorosity · 2 concepts
Equally spaced cylinders were constructed normal to local vane planes, patterned radially around all 17 vanes and subtracted from the baseline body. The fully porous design used 1 mm holes spaced 10 mm apart; the partial design used two rows of 2.5 mm holes. The openings allow flow from the high-pressure lower surface to the low-pressure upper surface.
Slits · 3 concepts
Curves from a radial plane were projected onto the vane and then onto a local tangent plane. The resulting profiles were thickened, patterned around the OGV and subtracted. The three designs varied slit width from 1 to 1.5 mm and slit length from 5 to 10 mm.
Serrations · 4 concepts
A sinusoidal curve was drawn on a plane tangent to the vane, thickened into a cutting body and patterned around all vanes. The set compared 3, 5 and 10 mm wavelengths at a 10 mm peak-to-peak amplitude, plus a half-amplitude variant.
Manufacture & rig testing
Serra-B and Slit-A were produced by fused-filament fabrication (FFF). I moved to stereolithography (SLA) for Porous-A, Serra-A and Serra-C because its smoother surface finish would help limit skin-friction drag. The FFF-printed OGVs were not reprinted in SLA after their tests showed no noise reduction.
Acoustic and aerodynamic data were collected from 2,000 to 7,000 rpm and at several pressure differentials around the 4,400 rpm operating point. None of the five OGVs produced a meaningful sound-power reduction against the baseline, demonstrating that OGV interaction was not the dominant broadband noise source.
Evidence changed the project
Rather than ending with an unsuccessful OGV campaign, I extended the project to the fan itself, which was a known dominant noise source. Limited time meant I could manufacture only two fans, so I designed one with leading-edge serrations and one with trailing-edge serrations.
Structural validation
Each fan required approximately one week of machining from a single block of 6082-T6 aluminum. Structural analysis therefore verified the most aggressive serration geometry before material and machine time were committed.
Design case
The 10 mm peak-to-peak leading-edge serration was selected as the highest-stress geometry.
Load case
Analysis represented 150% of the centrifugal load at the 7,000 rpm test maximum.
Mesh convergence
1.5, 1.0 and 0.6 mm meshes converged to within 8 and 6 microns of maximum displacement.
Material
6082-T6 aluminum: 70 GPa Young's modulus, 0.33 Poisson's ratio and 2.7 g/cm³ density.
Five-axis manufacture & result
The serrated geometry required five-axis CNC machining. Cutter failures and machine errors forced several restarts because even small asymmetry could create unpredictable vibration at speed.
The tested trailing-edge fan produced a 3 dB reduction in a frequency band, with peak reductions up to 8 dB. The acoustic benefit came with a 2.98% to 3.62% reduction in axial flow.
Research evidence
The dissertation documents the CAD development, manufacture, structural analysis, acoustic testing and aerodynamic comparison.
Production Equipment · Industry
Racal Acoustics (INVISIO Communications) · Production engineering
A four-part robotic gluing and pneumatic membrane-placement process developed to waterproof microphone housings while protecting acoustic performance and eliminating the variation of the original manual operation.
The problem
An existing microphone needed thin membranes bonded directly to the front and rear housings. Each seal required a circular adhesive bead followed by accurate placement and controlled pressure around the complete membrane lip.
Manual gluing produced variable bead width and occasional blobs. Excess adhesive could move away from the circular lip and across the membrane, causing inconsistent frequency-response results during acoustic testing. Rejected microphones then had to be broken apart so the internal assembly could be salvaged while the housings had to be discarded, wasting labor and production time.
The solution
I replaced the hand-gluing operation with a robot program and designed removable four-part trays that locate the housings beneath the dispensing needle. I then replaced the upward-facing vacuum jig with hand-held vacuum/pressure stamps, a four-way pneumatic valve, UV curing and final pressure testing.
The same tray remains with all four housings after gluing. The operator no longer removes and flips each housing individually, reducing handling and keeping every part in a known position throughout the process. The aim was maximum consistency between operators, with every assembly remaining within the required acceptance band.
Robotic adhesive dispensing
The robot controlled the dispensing needle in three axes. It applied adhesive along the housing path, stopped dispensing and retraced the path to spread the glue evenly, then lifted clear of the housing and moved to the next one to repeat the process.
Original membrane process
The original fixture faced upward. An operator positioned a membrane with tweezers, slowly flattened it against six vacuum holes, then pressed a glued housing down onto it. Turning off the suction was expected to leave the membrane attached to the housing.
Pressure was difficult to apply evenly around the seal. Adhesive was often squeezed onto the fixture, allowing the membrane to remain stuck to the jig instead of the housing. The six vacuum points also allowed wrinkles that could alter the later acoustic response.
Redesigned membrane process
The new hand-held stamp uses vacuum to collect the membrane, neutral pressure while it is brought onto the adhesive, and positive pressure to release it deliberately onto the housing. This third state prevents the stamp from depending on passive release.
A four-way valve selects the state. In its center position both supplies are blocked, while the Dymax 5 relief valve protects the pneumatic source. In the air and vacuum positions, the unused supply is redirected through the hidden fourth port under the valve to atmosphere.
Stamp geometry & internal channels
Both stamps combine a secure grip, alignment guides and a mechanical depth stop. Internal channels divide one pneumatic connection into 25 outlets in three rings, supporting the membrane more evenly than the original six-hole jig and reducing wrinkles.
Knurling across the body let the operator hold the tool securely with a light grip, preventing unwanted shaking and making membrane placement more consistent.
The surface touching the membrane is printed black. The contrast makes it immediately obvious whether a thin membrane is present and correctly located.
Twenty-five outlets replace the original six vacuum holes, spreading support across the membrane and sharply reducing wrinkles.
External legs keep the contact surface parallel to the adhesive ring so one side cannot receive more pressure than the other.
Gaps between the guide legs let the operator see the stamp approach and learn the correct placement motion without obscuring the joint.
Guide height creates a mechanical stop, limiting the stamp to the experimentally established placement depth without squeezing adhesive across the membrane. Changing the offset by 0.1 mm could change yield by 15%, so the front and rear tools were tuned independently.
Built & implemented
The mechanical-stop height was established separately for the front and rear housings through physical trials. Each tool had to press firmly enough to close the complete adhesive ring without pushing material into the acoustic area.
The finished stamps were connected to the Dymax 5 through the four-way valve and introduced with the robot trays, curing process and pressure verification. The process was then adapted for additional microphone variants.
Outcome
Production yield increased from 40% to 98%.Automated gluing was four times faster than manual operator gluing. The complete station made adhesive coverage, membrane pressure and release repeatable while reducing acoustic rejects, material waste and rework.
Production Equipment · Industry
Fixtures shaped around the operator and the process
3A
Microphone processing
The requirement
Several microphone variants had to be held at a repeatable height above an electromagnet. Each microphone sits in the black spring-contact tray shown above the blue fixture in the production photograph. Spring force changes the closed tray height between variants and also tries to push the tray back out of the fixture.
Problem with the original jig
The earlier station used a latch to resist the spring force. Inserting or removing a tray required one hand to operate the latch and a second to hold the tray, slowing a highly repetitive production task.
The mechanism
I replaced the latch with an adjustable friction bridge. The operator pushes the tray past light friction with one hand, and that same friction holds the tray at a controlled field position. An indexed screw is adjusted once for the microphone variant at the start of a production run and remains fixed for the whole batch.
The outcome
The additively manufactured fixture became a faster, universal station for multiple tray heights and was introduced at the supplier for production use.
3B
Roll-pin insertion
Requirement
Assemble two components with two roll pins simultaneously using an existing hand press. The pins had to remain aligned with their holes throughout the stroke. The supplier variant bolts onto its press; the in-house variant clips onto ours.
Design response
The fixture separates the functions into a fixed base, a replaceable alignment bridge and an upper press attachment. Steel guide rods pass through the bridge to drive both pins into the suspension assembly.
Maintainability
The bridge contains the guide holes and is intentionally replaceable. If those holes wear, alignment can be restored without remaking the entire jig.
Implementation
Most components were 3D printed, with steel used in the pin-driving load path. The supplier fixture was bolted to its hand press, while the in-house fixture clipped onto our press and located through the bed.
Test Equipment · Industry
Adaptable long-duration durability testing
A maintainable, adjustable test platform for repeatedly flexing different product components through programmable distances when existing equipment could not reproduce the required long-duration loading.
The requirement
Products qualified for demanding service environments also require life testing to demonstrate long-term mechanical durability. The rig had to accommodate different components, mounting positions and flex distances while remaining easy to repair after a very high number of cycles.
Adjustable actuation
The linear actuator mount can move laterally and vertically to align with different test articles. Stroke distance is programmed electronically; when a required position lies outside the actuator's available travel, its mounting block can be relocated along the repeated hole pattern.
Maintainable structure
Sheet-metal plates are joined with replaceable corner brackets, through-bolts and nuts rather than permanent joints. Black printed covers protect the exposed bracket geometry. An electronics enclosure on the reverse side contains the actuator power supply and programming connection.
Built hardware
The manufactured fixture uses the same adjustable actuator layout and bolted panel construction shown in CAD. The selected photograph omits sensitive product information while showing the production-scale structure and actuator.
Custom test development
Designed cyclic-compression testing for two helmet softliners, using a mechanical roller to apply thousands of repeatable load cycles where standard equipment did not reproduce the required condition.
Thermodynamic Research
BEng Aerospace Engineering · University of Surrey
An independent academic study combining CAD, thermal analysis and aerodynamic CFD to balance motor cooling, drag and weight.
Electric propulsion · 2018
The E-Fan X demonstrator, developed by Airbus, Rolls-Royce and Siemens, proposed replacing one of a BAe 146 aircraft's four jet engines with a 2 MW electric motor. This university project used that concept as the context for a simplified cooling study.
Without the actual duct geometry, I simplified the working region to an annular tube and investigated motor-support struts used simultaneously as cooling fins. My model assumed a 1 MW motor and 30 kW of waste heat, distinct from the demonstrator specification. The design problem was to conduct heat into the fins, then transfer it to the passing air by convection without excessive flow resistance.
Thermodynamic trial
Heat conducts through the solid motor and fins, then passes into the moving air by convection. Early SimScale trials established the mesh, thermal interfaces and boundary conditions before refining the geometry.
Thermodynamic optimization
I increased material near the motor and compared repeated fin arrangements. Symmetry sectors reduced computation while retaining the inlet velocity and pressure-outlet conditions; heat input scaled with the sector’s share of the full motor.
Large-model finding
Even the 32-fin configuration did not reach thermal equilibrium at nominal heating within the modeled 2,000 s. Fin-only cooling was not demonstrated for this large-motor case; another approach, such as liquid cooling, would need investigation.
A smaller reference model
I redirected the study to a 195 mm Schübeler EDF. With an assumed 500 W heat load, the report records equilibrium at 412 K for the baseline model.
Aerodynamic optimization
I replaced the rectangular baseline with three streamlined airfoil-section variants of similar average thickness. Four design variants were compared for viscous and pressure drag, with further thermal checks around the thinner leading and trailing edges.
| Design | Viscous (N) | Pressure (N) | Reported total (N) |
|---|---|---|---|
| D5.1 | 8.5 | 17.2 | 25.7 |
| D5.2 Lowest reported | 6.2 | 9.3 | 15.8 |
| D5.3 | 8.3 | 11.7 | 20.0 |
| D5.4 | 7.9 | 8.3 | 16.2 |
Original report values. D5.2’s components sum to 15.5 N; its reported total is 15.8 N.
The study established a comparative CFD workflow for balancing conduction, convection and drag. It did not provide an experimentally validated production cooling system.
Formula Student
Aerodynamic design · BEng Aerospace Engineering · University of Surrey
A three-element rear wing developed through 28 CAD and CFD iterations for the University of Surrey Formula Student car. The team placed fifth overall in the UK competition.
Design requirements
The rear wing had to exceed 130 N downforce at 12 m/s, cost no more than £660, remain below 8 kg excluding DRS and swan-neck mounts, and achieve a high-quality surface finish while satisfying the rulebook's working-volume, edge-radius, strength and stability requirements.
Why three elements
A comparison with Formula 1 showed that the Formula Student working volume was 205% larger while its typical cornering speed was far lower. The wing therefore had to be developed for this operating point instead of copying a two-element Formula 1 arrangement.
Flow separated at 16° incidence and the concept remained far below target.
Interference between the two layers increased the drag penalty and still missed the requirement.
Attached flow, 97 N drag and aerodynamic efficiency of 1.8 at the 12 m/s design condition.
Iterative workflow
Airfoil coordinates were transformed in Excel using scale, rotation and translation, then used to build the three-dimensional wing geometry in CATIA. Each iteration received a fine SimScale mesh and was evaluated at 12 m/s.
Pressure and viscous forces were resolved into downforce and drag, then compared with the previous version. Autodesk Flow Design visualizations showed separation, attachment and upwash, guiding position, chord and incidence changes of only a few millimeters or degrees at a time.
Final aerodynamic selection
B2 was selected because it combined approximately 180 N of downforce with 97 N of drag and visibly strong flow attachment. Designs producing more force carried much larger drag penalties and lower aerodynamic efficiency.
Moving from the earlier 80 N configuration to B2 increased downforce by about 100 N and exceeded the 130 N requirement by 50 N. Sharing its main airfoil section with the front wing also reduced the number of manufacturing molds.
Team integration
The team manufacturing concept separated each airfoil into upper, lower and corrugated internal sections for prepreg carbon-fiber construction. Structural detailing and swan-neck mounting were completed by other team contributors; these images show how my aerodynamic surfaces entered the wider assembly.
Project evidence
The linked SimScale project shows the digital iteration history; the team CDR records the requirements, CFD method, results and manufacturing proposal.
Product Design
Selected product development and supporting manufacturing experience
Key design contributions to released hardware and other products. Proprietary customer-product imagery is not shown.
Product redesign · Industry
Led the headset redesign (through SolidWorks) addressing seven failure modes behind a 25% reject rate. The work combined root-cause investigation, DFM and production changes across 35 plastic and rubber injection-molded components, saving tens of thousands of pounds and improving production-tool life.
Structural redesign · Validation
Redesigned a headset boom-arm system to increase stiffness fourfold and remove unwanted movement in a severe vibration environment. The revised hardware was qualified through MIL-STD-810G vibration and ingress testing.
Product development · Creo/Windchill
Developed four headset-to-respirator cable assemblies and two throat microphones from CAD through manufacture and mass-production qualification, including salt- fog, thermal and cable-bend testing.
Sustainable products · Internship
At Ecoinno, designed meal trays, coffee capsules and bottles in Siemens NX, then designed the wet molds used to collect recycled composite material and the heat- press tools used to form and finish each product.
Manufacturing & quality
Planned end-to-end manufacturing routes for aerospace and defense assemblies, generated FAIR documentation, verified tolerances with CMM and manual metrology, and led NCR/CAR root-cause containment. Nesting and routing improvements reduced waste, cost and energy use.