ParcelStore

Everything that walks a parcel's depth reads this rather than the substance's declared count, so a Distant log is one shell everywhere at once and nothing has to be told which tier it is at.

§14: "a tier change does not lose or gain mass or energy... which is the property that makes the LOD safe to apply to a burning object mid-burn." Shells are equal-mass by construction, so a fold is a proportional redistribution — target j takes source shells [j·n/m, (j+1)·n/m), splitting one where the ratio is not an integer — and the targets come out equal-mass too.

What it loses is the profile inside a fold, which is the point of a tier: two shells at 400 K and 300 K become one at their mass-weighted mean, and going back up hands each of them the same mean. Resolution in depth is what a distant object gives away.

surfaceTemperatureOf is what decides ignition and coreTemperatureOf is what decides doneness, and the gap between them is the whole reason shells exist.

A read rather than a stored field, and that is the design. Nothing writes a temperature except setTemperature, which writes the enthalpy that produces one — so a parcel on a plateau simply does not move, and no code anywhere had to know a plateau exists.

The temperature of the total enthalpy, not the mean of the shell temperatures. The two differ wherever the curve bends, and only this one has the property that matters: heat in, temperature out, with nothing lost between them.

Every radiative, convective and contact term in CH-5 lands here. The difference between this and addHeat is the difference between a log in front of a fire and a log in an oven.

Conserves this parcel's total enthalpy exactly — the arithmetic is in conduction.ts and so is the reason.

For authoring and for tests. Not for a tick: it discards whatever enthalpy the parcel held, so a boiling parcel set to its own boiling point loses however much of the plateau it had crossed, and the energy budget does not close across the call.

Neither this nor the thermal enthalpy is conserved on its own — a reaction moves energy between them. Their sum is, exactly, because the same number is added to one and taken from the other in reactShell.

§14's thermal quiescence, and the vocabulary is deliberately Track B's. A cold stone floor is a thousand of these and costs a bounded scan.

Called by everything that could change what a parcel is: heat arriving, water poured on, mass added, a contact, or a source coming within range. Cheap enough — a flag and an integer — that a caller never has to decide whether it is worth it.

A burning or smouldering parcel never sleeps, whatever its numbers say. A steady flame is a steady parcel by every measure here, and sleeping one would put out a fire by optimising it.

A Far parcel is stepped every fourth tick with four ticks of dt, so the same seconds are integrated either way — which is what makes a cadence conserving for free rather than by care.

Ignition is five conditions, and a pilot is one of them. Holding a match to wet wood does exactly nothing here, which is correct and is the API telling the truth about what a match is.

The two differ by exactly what a wet log gives off. Steam leaving a surface counts toward diluting the mixture above it and not toward feeding it, which is why 60% moisture stops a log catching in front of a fire that lights a seasoned one in eighty seconds.

A fire's size is its heat release rate, which is what every figure in the design is quoted against and is not the temperature. A pyrolysing surface reads negative here while it is gasifying, which is right and is why a flame has to keep feeding heat back to sustain one.

"12% moisture content" means twelve kilograms of water per hundred of dry wood, and wetComposition folded that into a wet-basis 0.107 to store it. Undoing the same conversion here is what makes a script comparing against the 0.25 that decides whether a log is worth burning read the quantity the author wrote.

Ice counts as water. A frozen log is not a dry one.

Mixed is a real answer rather than a failure to decide: wet wood is wood and water at once, and calling it either would be a lie a consumer would then draw.

Free, meaning beyond what the material holds as its own moisture, and that distinction is the whole reading. A seasoned oak log carries 12% water on a dry basis and is not glossy; a log somebody threw a bucket over is. Reporting the shell's total water would make the two identical and every piece of wood in a scene look rained on.

Derived rather than a column, because wet puts real water in the surface shell rather than setting a flag — so §17's wet-film roughness reads a quantity the boiling reaction is simultaneously consuming, which is what makes drying visible.

A caller pouring rain over everything asks first: wet refuses a material with no water in its model, and refusing is right, but a loop over every parcel in the rain had no way to tell an iron nail from a log before it was told.

Not a flag, and that is the whole point. The heat sink §12 describes is the same 2.44 MJ/kg every other drop of water in this model carries, so a doused fire goes out for the reason a real one does and a light sprinkling on a hot ember does not.

Refused, naming the substance, where that substance cannot hold water at all — which means it declares no reaction that touches it. Silently doing nothing would make "I poured a bucket on it" and "this material has no water in its model" indistinguishable.

The shells that are mostly carbon, summed by their own thickness. A shell counts once it is past half char, which is a threshold on a continuum and is said out loud: a finer answer needs a char front within a shell, which is a sub-shell quantity this model does not carry.

§18 refuses structural failure in writing and this is the refusal reporting. A beam whose char depth has eaten a fraction of its thickness has lost that fraction of its strength, and what happens next — a joint releasing, a body splitting, a building coming down — belongs to whoever owns the solver. This package writes the scalar, emits an event at a threshold, and breaks nothing.

Both must be the same substance, and two that are not are refused naming both. Two substances are two species sets and two shell counts, so mixing them would silently discard whichever set was narrower — and a stew is not a substance change. What a consumer combining two genuinely different materials wants is a third substance that is the mixture, authored as one.

The difference between this and addSurfaceHeat is a microwave and an oven: one deposits where the field penetrates to, the other at the surface. Nothing here decides which is physical.

The number a shrinking log is drawn from. A parcel's volume is its mass over its substance's density, and the density does not change — so this is simply what share of the mass is still here, which falls as volatiles leave and rises for nothing.

What every explicit surface term has to be bounded against. A flux times an area times a step is joules, and joules into a small enough shell is an arbitrarily large temperature change — so a term that does not know this number cannot tell the difference between heating something and destroying it. ChemistryWorld uses it to stop a radiative or convective step carrying a surface past the thing driving it.

Its density, porosity, emissivity, shell count and the species it can hold — enough that two libraries differing anywhere disagree here, and cheap enough to compute once a tick. §15 asks for the registry to be covered because two worlds with different substances are different worlds, and a replay crossing them should say so rather than diverge later.