Bulkheads and seawalls built for the pressure behind the wall.

An open bulkhead is a cantilever. Soil and water push on one face, and the wall resists by what is buried below the mudline plus tiebacks running to deadmen set back in stable ground. Close that same run into a rectangle and the mechanics invert: nothing pushes the box sideways, because the fill inside pushes outward on all four faces at once. That outward pressure is carried around the perimeter as tension through the interlocks. The mechanism is the same in kind as the one cofferdam cells use — though neither the label nor the product is. A cofferdam is temporary by definition, built to keep water out of a hole so something permanent can be poured dry; this box is permanent and it stays full. And true cellular structures are built from hot-rolled straight-web piling whose interlocks carry a published, guaranteed tensile strength, which cold-formed light-gauge sheet does not. That matters less here than it sounds, because hoop tension scales with the depth of fill, and a cell this shallow generates very little of it.

Which makes the fill the structure, and the steel the membrane that holds it in. How much the box weighs, how stiff it is, and what it can carry are all questions about what went inside. Loose sand is the cheap answer and the wrong one in a closed cell. You cannot get a compactor into the corners of a box, so it never reaches proper density; uncompacted granular fill keeps settling under load for years, which shows up later as a deck that slopes and moves; and fines can migrate out through an interlock gap or a leak at the toe until there is a void under the deck that nobody can see from above.

What went in instead was a stabilized mix — sand blended with cement and a plasticized mortar, proportioned in advance and mixed to flow rather than to be compacted, so it fills the corners and the space under the deck frame by running into them. It is honest to call that fill and dishonest to call it concrete: it is not a footing, not a slab, and not a structural pour. A self-compacting cemented fill in the 50 to 100 psi range gives roughly the allowable bearing pressure of a well-compacted fill — which is precisely the performance you cannot get with a plate compactor inside a closed cell. It also cannot be poured through standing water, because the cement washes straight out of the mix, so a cell like this gets filled dry. Worth noting too: a cemented fill is far less permeable than the free-draining granular fill a conventional cell assumes, so a relief path out of the box deserves the same thought this page gives drainage behind any other wall.

The reinforcement in that fill is composite bar, not steel. Concrete protects steel chemically rather than physically. Fresh concrete pore water is strongly alkaline, and at that alkalinity a passive oxide film only a few nanometres thick forms on the bar and holds corrosion off. Steel in concrete is not safe because it is covered up; it is safe because it is sitting in a chemical bath. Chlorides working in through the cover break that film in spots, and once rust starts it does the real damage: corrosion product occupies roughly two to six times the volume of the steel it consumed, and locked inside hard concrete it works like a slow jack — the cover cracks, delaminates, and sheets off, exposing more bar to finish the job. Glass-fibre composite bar has no iron in it. There is no corrosion cell to start and no expanding rust to split the fill apart from the inside.

That is a real advantage and it is not a free one. Composite bar is not a drop-in substitute for steel: its stiffness is about a quarter of steel's, so members get sized by deflection and crack width rather than by strength, and the usual result is more bar and more section, not less. It cannot be bent on site — the resin cures once, at the factory, so every hook and bend has to be ordered made, and a factory bend is meaningfully weaker than the straight bar. "Composite rebar" is also not a regulated term; the certification worth asking any contractor for is ASTM D7957. As for how long it lasts: bars cut out of 15- to 20-year-old bridge decks show very little measurable loss, which is genuinely good evidence and is not the same thing as a century of proof. Nobody has that yet.

The gate and the ramped-down side of the deck land on thick-wall drill-stem pipe piles welded to angle iron — surplus oilfield pipe, used here for the reason marine contractors use it everywhere on this coast. Thick wall means a large corrosion allowance, not steel that resists corrosion; high-grade pipe rusts at much the same rate as ordinary structural steel in the same water. It simply starts with far more metal to lose. Everything above the fill line — the framing, the stair, the railings, the upper level — is the same deck and pier work Shoretech builds on any shoreline.