Graduation project — 01

Shio

A modular public bench pressed from crushed oyster shell waste and recycled HDPE — asking what a discarded material is still capable of.

The finished Shio bench in three-quarter view, its speckled composite surface catching the light.

Where it started

The idea came from a walk through Amanohashidate, Japan, past mounds of discarded oyster shell — a by-product with almost nowhere to go.

Back in Sydney, Broken Bay Pearl Farm became the source of the same problem at home: crushed shell waste with real structural potential, if it could be bound into something load-bearing.

Discarded oyster shell banked along the waterline at Amanohashidate, Japan, with a timber house and pines behind.
Amanohashidate, Japan — where the idea came from
Stacked oyster racks and shell waste at Broken Bay Pearl Farm.
Broken Bay Pearl Farm, NSW — where the material came from

Testing the composite

Fourteen tiles before one of them held.

The whole project rested on a material that did not exist yet. Seven batches, each one pressed, labelled and broken under a three-point bend rig.

Twelve composite test tiles laid out on a concrete floor, each labelled in marker with its HDPE-to-shell ratio — 100, 85/15, 80/20, 75/25, 70/30, 65/35, 60/40, 55/45.
Every tile, labelled with its ratio
BatchMixPeak
01HDPE + shell powderPowder would not bond — HDPE is hydrophobic, shell powder hydrophilic
02100 / 0 controlPure recycled HDPE, the benchmark to beat2750
0285 / 15 crushedCrushed shell bonded where powder failed2250
0380 / 20 powderFused on the surface, voids underneath<750
0370 / 30 powderBent by hand — no strength test needed<750
0465 / 35 pearl shellAkoya shell is aragonite, not calcite — less stable<1550
0470 / 30 pearl shellBrittle, but heat turned the surface gold<1250
0555 / 45 rock oysterToo little plastic to bind — decorative only<1000
0560 / 40 rock oyster40% is the ceiling for structural use<1000
0670 / 30 hand-groundInconsistent pellet size left air pockets<750
0770 / 30 controlledSame ratio, graded pellets — three times the strength2300
0775 / 25 controlledStrongest result, and the marbled surface was an accident2500

Peak load in psi, three-point bend

A composite tile snapped into a shallow V under the head of a hydraulic press, crushed shell visible along the fracture.
Three-point bend — every tile taken to failure

Powder was the wrong instinct

Fine shell powder never bonded. HDPE is hydrophobic and shell powder is hydrophilic, so the powder sank or sat on the surface — tiles that looked fused blew apart under an air gun. Switching to crushed shell aggregate tripled the strength at the same ratio.

The shell species mattered

Akoya pearl shell is aragonite; rock oyster is calcite. The pearl tiles were the most beautiful — heat turned the surface a gold iridescence — and the most brittle. Aragonite is the less stable crystal, and the tiles snapped well under the benchmark.

Preparation beat ratio

Batch six and batch seven ran the same 70/30 mix. The first, ground by hand into uneven pellets, failed under 750 psi. The second, with graded pellet size and even distribution, held 2300. How the material was prepared mattered more than what was in it.

Form development

Finding a shape the press could actually make.

  • Concept sketches exploring bench forms, leg geometry and shell-composite panel joints.
  • Concept sketches exploring bench forms, leg geometry and shell-composite panel joints.
  • Concept sketches exploring bench forms, leg geometry and shell-composite panel joints.
  • Concept sketches exploring bench forms, leg geometry and shell-composite panel joints.

Developing the composite

The bench is sheet-pressed from a 75/25 blend of recycled HDPE and crushed shell, engineered to a safety factor of 3 and tested well beyond that — to a safety factor of 10, reaching a compressive strength of roughly 2500 psi.

Because HDPE is usually sheet-pressed and CNC-cut, the shell introduced a new manufacturing constraint. With limited time I pivoted to framed final forms and experimented with sheet-pressing curved arcs — something rarely attempted, because they warp. The embedded shell surprised me: it stabilised the material and prevented the distortion.

Modularity meant the bench could be assembled, repaired and reconfigured in public without specialist tools.

The four flat-pack components of the bench laid out, showing the terrazzo-like shell composite.
Flat-pack components — two legs, a beam, a seat

The part that might not have worked

Nobody presses HDPE into a curve, because it warps.

The form depended on curved panels, and sheet-pressed HDPE is known to shrink and pull out of shape as it cools. Rather than scale up a model that might not be manufacturable, the DFL staff and I built a single arc frame to test the principle first.

The frame had to be sandblasted before it went near the press. At 200 °C paint doesn't just burn and release fumes — it transfers onto the sheet.

It worked, and for a reason I hadn't predicted: the shell was doing the stabilising. The aggregate held the sheet against its own shrinkage and the arc came out of the frame true. The material I'd spent seven batches trying to make strong turned out to solve the forming problem too.

A welded steel pressing frame on a workbench — a rectangle with an arc bridging its base, the paint sandblasted back to bright metal at every weld.
The proof-of-concept frame, welded to test one question

Project specs

Composite ratio
75 / 25 HDPE–shell
Compressive strength
~2500 psi at 10 mm
Test method
Three-point bend
Engineered safety factor
3
Held to
10 before bending, 15 before failure
Panel thickness
20 mm
Length
1200 mm — seats two
Shell source
Broken Bay Pearl Farm, NSW
HDPE source
Broken Bay Pearl Farm — discarded baskets
Form
Modular public bench, 1:2 scale model

Side elevation of the finished bench showing the arc of the seat and the beam joint.
The finished bench in three-quarter view, brass fixings set into the speckled shell composite.

Exhibited at BEGradEX, White Bay Power Station

Where it goes next

The joinery is still the open problem.

The panels press cleanly. Joining them afterwards without mechanical fixings is the part I have not solved — the brass here is a working answer, not the intended one.

Three directions worth testing: bio-based adhesives, interlocking geometry cut into the panel edges, and connectors set into the sheet during the press.