Designing a hull to hit a density rather than a strength

Optimising for a target the textbook does not have a method for.

Designing a hull to hit a density means treating the whole finished assembly as though it were a single material with one property that has to land exactly on a chosen value, average density. Most mechanical design proceeds by picking materials and thicknesses that clear a strength requirement with margin to spare, and a structure that is merely strong enough says nothing about whether it will float where it is meant to float.

A target with no safe side to miss on

Almost every other part of engineering rewards a comfortable margin. A beam that carries twice its expected load is safer than one carrying exactly its expected load, and the instinct to build in headroom is trained into an engineer early. Density offers no safe direction to overshoot in. A hull that comes out ten percent too dense sinks, and one that comes out ten percent too light bobs uselessly high, so the usual one-sided requirement that just has to be cleared becomes a two-sided one that has to be hit, across a whole assembly of parts each contributing weight and displaced volume.

An old-fashioned two-pan scale, with flour spooned onto one side while known weights sit on the other, is the same kind of target. The beam needs the two sides to land level, and a spoonful past the balance point tips it the other way just as surely as too little left it tipped the first way. Nobody using such a scale reaches for extra flour the way an engineer reaches for extra material on a bracket, and the same caution applies to every gram added to a hull once its running weight and running displacement are close to matching. It also explains why the final correction goes on in small amounts near the end.

Two running totals for every part

Every part in a hull, a motor, a battery, a housing wall, a cable, adds two numbers to the running totals: its weight, and the volume of water it displaces directly or lets the enclosing shell displace on its behalf. A lighter but smaller component can move the ratio in either direction, depending on how much air-filled volume it gives up along with the weight it saves. A thicker wall, the ordinary reflex for extra strength, adds weight with no extra displaced volume if the wall already sat inside the hull's outer envelope, so a choice that reads as conservative structurally is a direct threat to the target density. A spreadsheet tracking both totals, updated every time a bracket changes, sounds like overkill until a finished hull is floated and a dozen individually sensible decisions turn out to have summed to a vehicle sitting visibly too high or too low.

Because density belongs to the assembly, no single component can be optimised in isolation. A lighter motor or a thinner bulkhead has to be checked against the whole vehicle's totals before it counts as an improvement, since a part that gives up more displaced volume than its weight saving is worth moves the assembly further from its target.

Aiming on the light side

The target loosens once ballast is planned in from the start. A hull that lands a little light is corrected by adding a small weight low in the structure, a cheap and well-understood fix. A hull that lands heavy has no equally cheap fix, since structure required for strength can rarely come out, and adding buoyant foam to claw back displaced volume costs space and money. An experienced designer therefore aims deliberately on the light side of the line, a bias that looks odd to anyone trained only on strength margins.

My key error with this

My instinct was to let the model tell me the mass, on the reasoning that the CAD knows the volume and the materials and can therefore do the arithmetic more reliably than I can. Working that way meant I only found out where I stood after a shape was finished, so each discovery that I was above or below the density target arrived attached to a design I had already spent effort on, and the fix was another full revision. Doing the arithmetic by hand first, roughly, before the geometry existed, and then treating the completed model as a check against that hand calculation rather than as the source of it, changed the shape of the whole exercise. The manual numbers set the budget, the CAD confirmed whether I had spent it correctly, and disagreements between the two surfaced early enough to be cheap. What replaced the belief is that the model is a good auditor and a poor planner, and that letting it do the planning is what turns a target into a sequence of expensive iterations.