Inclining Transparent Walkway (Vibram Senior Design)

Vibram is one of the largest shoe sole manufacturers in the world, and their product testing facility — the Boston Connection Lab — wanted to evaluate shoe performance at various inclines by watching the sole flex through the underside of a transparent walkway. You cannot buy such a thing — Vibram’s own inclined walkways are neither transparent nor electricity-free — so the ask was a transparent, primarily mechanical system with easily interchangeable parts. This was my senior design capstone at Boston University, Team 173 — about eight months, with the finished device delivered in May 2025. Five of us built it: our project manager led the team, one teammate led CAD and the Vibram communications, another led simulation in the fall, a fourth led the transparent-surface materials research, and everyone machined and assembled. I was the camera research lead and did force and deflection calculations, the scissor jack concept, and the House of Quality and Pugh chart organization; on the build I was the lead for manufacturing and the second lead for manufacturing planning — a teammate led the planning — and my subsystems were the gym-bench integration, the steel tube supports, the gas struts, and the camera system.

Getting to delivery meant killing our leading mechanism two-thirds through and verifying its replacement — the work of four rounds of FEA across six months.

What Vibram asked the walkway to do

The load requirement carried a safety factor of two on the person’s weight, increasing the maximum weight for the walkway from 250 to 500 pounds. The lift had to be mechanical — the $2,500 budget pretty much ruled out electrical solutions anyway, and the customer did not want electricity. The finished unit also had to fit through a door into the facility, and that requirement is what set the walkway’s width.

Requirement	Target	Achieved	Verified by
Incline in fixed steps (relaxed from 0–45° in 5° steps)	0–40°, 10° steps	met — bench pins natively index 0–40° in 10° increments, one bolt pair per angle	delivered device, demonstration
Support a walking person with SF 2	500 lb	by analysis — FEA passed at 500 lb; the walk-on peaked at about 450 lb	SolidWorks FEA + hand calcs + walk-on test
Fully mechanical lift, no electricity	—	met — bench pins + gas struts + hand-cranked camera carriage	inspection of delivered design
Transparent 66 × 24 in walkway	66 × 24 in	met — 0.75 in cast acrylic, 24 × 66 in	BOM + delivered device
Operable by one person (two acceptable)	1 person	met — one pull releases both bench pins; gas struts nearly self-lift the ramp	use during validation; no formal test documented
Railings on top of the walkway	—	met — parallelogram handlebars, back bar vertical at any angle	delivered design
Removable acrylic, modular, stock parts	—	met — toggle clamps release the panel; all-bolted 8020; nearly every part off the shelf	delivered design
Easy to move on wheels	—	met — retractable casters	delivered design; no mobility test documented
Camera perpendicular to the walkway at all times	90°	partial — rail rides parallel to the acrylic frame at every angle; image capture validated at 0° only	presentation evaluation slide
Maximum cost	$2,500	not met — 2025-03-10 BOM totals $2,926.21 with buffer	bill of materials
Fit through a door into the facility	—	met — delivered into the Boston Connection Lab	delivery

Concepts for the lift mechanism

The lifting and indexing concepts on the table: hydraulic, jack, scissor jack, two pulley variants, winch, crank, linear slide, lead screws, a simple human lift, and a beach-chair locking mechanism. The scissor jack was mine.

I organized the House of Quality that scored all of it: fail-safe design ranked most important, and the specification weights put safety factor first at 16% (design goal 2) with maximum cost second at 14% ($2,500). The Pugh analysis reused those HoQ weights, and the manual human lift scored highest on simplicity and low cost. The candidates also got deflection numbers: we modeled the jack design as a cantilever with a composite acrylic-and-aluminum modulus of about 2.19 × 10⁶ psi, where a 400 lb load at the end of the walkway deflected it 0.743 inches with the jack 2 feet along — moving the jack to 3 feet cut that to 0.27 inches. The full matrix and weighting method are in the House of Quality report.

Rather than describe 8-foot mechanisms on slides, I 3D-printed hand-sized models of the standout concepts so Vibram could hold them at the November 5 review. The feedback from our Vibram contacts settled several concepts: maintaining a pulley system’s wires and counterweights “would be a nightmare,” linear slides were too costly, and their strongest response went to the compact model with a rack-and-pinion front and a beach-chair locking back strut.

That’s the design we carried out of the fall: a manual lift with lawn-chair indexing — machined grooves in steel tubing as index points, a steel bar slotting in, clevis-ended aluminum struts up to the acrylic cage. The fall down-select and design are documented in the fall halfway report.

Why we dropped the beach-chair mechanism

The beach-chair mechanism was our leading design almost halfway through the project. The FEA that killed it split cleanly by angle: at high angles, with the support arm near perpendicular to the ground, the results were great; at low angles, as the arm neared parallel, the supports were insufficient.

There was simply no way to support the weight we wanted with that mechanism.

The strikes had been accumulating anyway: the hinge point was costly to manufacture, operating it was strenuous, the lift-lower-move-the-strut sequence was too complicated, the machined index teeth came out costly and imprecise, and we could not verify a high enough factor of safety. About two-thirds of the way through, we killed it completely and pivoted. The pivot cost less than it sounds: the frame was highly integrable, so only the bench-to-base and bench-to-frame interfaces changed — not a lot of trouble at all to move pieces of 8020 around.

Two 1,200 lb gym benches as the lift

What replaced it was prefabricated: two home gym benches, modified and bolted into the base frame. Each bench is rated 1,200 lb by its manufacturer, its alloy-steel pin mechanism natively indexes to 40 degrees in 10-degree increments, it’s a stock product Vibram can repurchase, and it hits the top two House of Quality criteria — fail-safe design and 300+ lb. The lift that replaced months of our own mechanism design cost $69.99 a bench. We trusted the 1,200 lb rating as a starting point, then verified it — FEA before assembly, physical testing after.

Integrating them was one of my subsystems, and it reused the benches’ own hardware: the pads people lean on came off, and their existing screw holes were designed to line up with the 8020 — each bench’s inclinable portion bolts to the acrylic frame, its base into the base frame. The subsystem down-selection is in the spring 2025 presentation.

A beam ties both bench pins together through 3D-printed connectors, so pulling one handle pulls both pins at once. The two gas struts — IAQWE 28-inch, 120 lb heavy-duty — are sized to almost entirely lift the ramp by themselves from the compressed state; with the pins out, the user adds only a few pounds of force, and the pins hold everything the rest of the time. The load path at incline was the late fix: the original plan had the benches cantilevering the walkway’s weight, I calculated that was not feasible, and removable steel tube supports were added late. Each tube bolts to the sides of both ramp frame and base, and I calculated the bolt positions so each pair yields exactly 10, 20, 30, or 40 degrees — on the BOM they’re 4 ft and 6 ft carbon steel rods with ball-joint rod ends, the short and long struts the user installs for the height they want.

Why the plate is 0.75-inch cast acrylic

Four transparent surfaces made the material table: polycarbonate, acrylic, tempered glass, and structural laminated glass.

Property	Polycarbonate	Acrylic	Tempered / laminated glass
Yield strength	~60 MPa	~70 MPa	120–200 MPa
Elastic modulus	2.3 GPa	3.2 GPa	70 GPa
Impact resistance	250× glass	17× glass	baseline
Scratch resistance	Mohs 3	Mohs 3.5	Mohs 7
Refractive index	~1.58	~1.49	~1.52

Glass carried the best load-bearing numbers and lost anyway: too expensive for the budget and much heavier than either plastic, adding strain to the whole system. Between the plastics, acrylic won on the two properties that matter for this machine — a lower refractive index, the least image distortion for the camera below, and a higher yield strength; weight and cost were close enough not to drive the decision. Acrylic’s scratch softness we accepted — the user just keeps rocks out of their shoes — and UV yellowing can be stopped with a coating. Cast acrylic beat extruded because the spec sheet detailed a higher resistance to fracturing. For thickness, hand calculations came first — simple statics plus Roark’s Formulas for Stress and Strain, 8th edition, on our advisor’s recommendation — with the stated caveat that flat plates are highly sensitive to aspect ratio and edge conditions, so FEA followed. The October pass treated the plate as a beam supported at two points, and it showed where the real constraint lives: strength alone needed just 0.161 inches of acrylic, but holding deflection to the 0.1 inches we set as the maximum required at least 0.87 — deflection drives the thickness, and at the time that meant 1-inch stock. The fall FEA round then ran four flat-plate scenarios at 500 lb point loads: minimum panel thickness of 1, 0.75, or 0.5 inches for one, two, or three panels. The spring load case was deliberately unkind: 500 lb on a 3 × 9 inch patch at the exact center of the acrylic — a foot planted on the weakest spot — plus gravity, cradle bars fixed, frictional rather than bonded contact between acrylic and aluminum for the most life-like scenario, and a fine mesh down to 0.1 in. The answer: a minimum 0.38 inches for the plate not to collapse at a factor of safety of 2, with deflection analyzed alongside stress to prevent distorting the camera image.

We shipped 0.75 in. Half-inch acrylic deflected an undesired amount; 1-inch was deemed too heavy for the average user to lift; 0.75 in withstands 300+ lb with minimal deflection and stays light enough to lift with the gas strut assist. The fall analysis figures and the fall BOM are in the fall final report.

Finding the first part to fail

This was being delivered to a customer for their use, and I did not feel comfortable accepting risks and just letting them slide — every identified risk got a hand calculation plus FEA verification. Four rounds of FEA ran in SolidWorks Simulation across six months — too many complex geometries to stress-test accurately by hand. The failure points, analyzed one by one: the acrylic breaking, the frame holding it, the steel supports connecting frame to base, and the base itself. The gas strut load came from regular cantilever equations, the bolted supports got simple hand calculations first and FEA after, and the acrylic-and-frame FEA covered stress and deflection plus fatigue.

The full-system FEA put the first point of failure at the gym-bench pin, specifically where the metal connects to the plastic grip. The mitigation is the same gas struts that assist the lift: if the pin fails at 40 degrees, maximum inclination, the struts greatly ease the walkway’s decline and the user remains unharmed. One structural decision was made on ethics grounds before any analysis: no welding. None of us had the training or certification to make safe, consistent welds — NSPE rule 2, practice only in your field of competence — so the structure is entirely bolted 8020, a material all five of us knew well. That vote was unanimous.

Rails, clamps, and casters around the acrylic

The stack, end to end: the 8020 base, the benches bolted into it, the acrylic ramp frame bolted to the bench tops, and the handrails, gas struts, end supports, and camera rail all hanging off the ramp frame. The acrylic rests on 1 × 1 in internal beams inside a 2 × 2 in 8020 perimeter, where rubber tape and toggle clamps keep it from moving while still letting a user pull the panel out in seconds — Vibram’s removability requirement.

Railings were a hard safety requirement: at 40 degrees the user at the far end of the walkway stands about 5.5 feet in the air. Two handlebar options died on geometry: bars fixed to the base frame would need a roughly 8-foot cage — bad for stability and for machining time and cost — and bars rigidly perpendicular to the acrylic frame point backwards at high incline while a human leans forward, risking head strikes. We paid for the third option: a parallelogram design with 10 pivots at $30 each, tied to both frames so the back bar stays vertical and the rails stay parallel to the walkway at any angle — inspired by Vibram’s existing inclined walkway. Retractable casters at the ends of the base frame handle mobility, and everything is aluminum except the acrylic and the steel — the benches, the support tubes — because the ramp is lifted by hand and the whole unit has to roll.

The camera system under the walkway

I led the camera research, and the requirement was strict: the camera below must stay perpendicular to the walkway at all times. In the spring the camera design questions went to me and a teammate, and I ran the subgroup that answered them — three of us, meeting notes mine. The recording decisions came first: a phone rather than a dedicated camera, because capture and export are easier, there’s no camera to buy, and it’s simpler for the user; continuous video through the phone’s native camera app instead of triggered stills, so Vibram can extract whatever they need from the footage; no data cable — the videos stay on the phone and export digitally; and the camera stays unmoved while filming so the picture is stable, mounted as far below the acrylic as the geometry allows. The phone mount we bought instead of printing: friction-fit prints aren’t universal and crack, tensioned universal prints rely on rubber bands, and a purchased universal mount cost about $16 — I ordered it. For driving the carriage we weighed a belt drive, very smooth but needing at least 10 feet of belt; a pulley-and-spool, very simple but easy to break; and a motorized drive, scrapped outright because this machine was to have no electromechanical components. What shipped is a phone holder on a linear rail under the acrylic, connected to a gear system: the user cranks the holder out from under the frame — the rail extends past the end of the ramp — loads their phone, cranks it back under, and controls it by remote while it’s underneath. The crank-out exists because I did not want the user to have to reach under and potentially put themselves in a pinch point or any danger.

Camera capture was validated at 0 degrees; there is no documented validation at the 10-to-40-degree inclines. The side-view camera we dropped — a nice-to-have, not a must-have, per the customer.

Machining and assembly

We had pretty much all of McMaster-Carr available as well as the machine shop at the school, and that led to a ramp consisting almost entirely of 8020. Manufacturing was mostly ordering 8020, cutting it to size on the cold saw, then drilling and tapping holes in the proper places — every part personally made by the team, with me leading the manufacturing. Each hole position was measured to within 1/16 in, with a tolerance stack-up at the places where components interface — every drilled feature had to land within tolerance for the all-bolted frame to go together easily and properly. Tapping was the workflow that broke first: the initial design required tapping the ends of many 8020 bars, which took too much time and effort, so I moved joints to the side of the extrusion instead of the end and used slide-in T-slot hardware instead of tapped holes. 3D printing shipped as real hardware here too — the pin-actuation connectors joining both bench pins to one beam are printed parts.

Vendors were almost entirely McMaster-Carr, plus the gym benches from Amazon — both with very low lead times — and the acrylic quoted over email from Port Plastics and picked up. The device was machined and assembled over the course of a month. There was only budget for one physical build, so iteration happened in design: versions with and without the late steel supports, and a camera system that went through a lot of iterations.

$2,500 target, $2,926.21 on the BOM

In September the budget was iffy to say the least: $500 from BU, no stated figure from Vibram, and their guidance was to pitch material purchases and design decisions with justifications whenever we asked for funding. So I assembled the budget in stages: $100 per person from the class — $500 for the five of us — an extra $700 I secured from the school that other projects weren’t using, and the remainder from Vibram after I presented the bill of materials and the design justifications.

The final numbers disagree. The 2025-03-10 BOM, with buffer, totals $1,767.62 on Vibram’s side and $1,158.59 on BU’s — $2,926.21 combined; I remembered it as $2,500 of material. The BOM is the document, so $2,926.21 is the number I stand behind, above the $2,500 goal. The transparent plate the whole machine exists around was $278.00 of that. The line items, split by who paid, are in the bill of materials.

Load testing, lightest walker first

The validation plan was quite simply using it. Once it was assembled and all the supports were properly placed, I walked on it — I was the lightest member of the team, so I went first, and it held just fine. Then our heaviest member, at about 300 pounds, walked on it and did great. Two people stood on it at about 450 pounds total, and on that basis I felt comfortable checking off the safety-factor box. What we observed compared well with the hand-calc and FEA predictions, though nothing was recorded as a number on test day.

What the walk-on never proved

The walkway was never physically loaded to the full 500 lb design load; I overdesigned the safety requirements, and the final result was not tested up to the point FEA said was possible. Force to lift had a 40 lbf House of Quality goal and mean time before failure a 1-year goal, weighted 13%; neither was ever measured, and the finished unit was never weighed against its 350 lbf goal.

Delivered to the Boston Connection Lab

What went out the door in May 2025: the 2 × 5.5 ft, 0.75-inch cast acrylic walkway in its 8020 fixture with parallelogram handrails, gym-bench lift, steel supports, gas struts, casters, and the camera rail. Pretty much every single part was off the shelf — this was designed on a budget that could not use custom parts. That was deliberate: repairability was a selection criterion, and we handed over all documents and information at year end so Vibram could fix anything on their own.

The ramp 100% did its job — it supported weight and let video be filmed underneath without anyone reaching under it — and it is still in use at the Boston Connection Lab. Vibram’s contacts were engineers, we ran design reviews with them and our faculty advisor throughout, and they never said the work wasn’t what they wanted.

What this project taught me

My biggest takeaway was bias to action: in meetings where people were uncertain, step up with a plan. The decision I’m proudest of is killing the beach chair: I’m not gonna go into the sunk cost fallacy here — even though we spent time designing this, objectively it will not work, we have to pivot. Technically, I learned a lot about assembly and tolerance, and about designing for manufacturability. Major decisions we made together; individuals were empowered to go off, make things happen, and come back prepared — infinitely better than sitting on your hands waiting for someone else to give you an answer. What I would do differently is be more decisive in design decisions — that was a big learning experience of this project. We could talk and talk about potential solutions, but in my experience the best way to get there is to try and fail, or try and succeed; failing fast is better than not trying at all. We were limited heavily by budget, so I do understand the hesitancy, but being more decisive would have saved a lot of time — time we could have used continuously improving the project.

Gallery

The documents

• Spring 2025 final presentation
• Fall final report (complete)
• Fall halfway report (ME 460)
• House of Quality report
• Bill of materials, 2025-03-10