Validation.

How close Full solve’s radiation comes to cases with an exact answer. Hot surfaces lose heat by radiation as well as to the air, and in still air or a sealed box it can be most of it. These cases are small enough to work out by hand, and they run through the same radiation code, at the same settings, as every Full solve with radiation on.

8 of 8 values within tolerance. Run on v3.928, 2026-11-13. The tolerances were set before the runs and not moved after them.

Two grey platesNet exchange between parallel surfaces

Two large flat plates facing each other across clear air, one hot and one cool. The heat that passes between them by radiation alone has a closed form, so it is the cleanest test of the radiation itself. Two pairs of finishes: ordinary grey surfaces, and bare metal (ε 0.1) facing paint (ε 0.92), where a shiny surface cuts the exchange to a fifth. The third row puts the plate in front of an opening whose air is warmer than the room it opens to.

casevaluereferenceMeshWeaveerrortoleranceverdict
ε 0.8 at 400 K facing ε 0.6 at 300 K, 20 mm apartnet radiative flux, hot plate517.73 W/m² σ(T₁⁴ − T₂⁴) / (1/ε₁ + 1/ε₂ − 1)517.73 W/m²+0.0006 %±5 %within
ε 0.1 at 400 K facing ε 0.92 at 300 K, 20 mm apartnet radiative flux, hot plate98.38 W/m² σ(T₁⁴ − T₂⁴) / (1/ε₁ + 1/ε₂ − 1)98.38 W/m²+0.0006 %±5 %within
ε 0.8 plate at 400 K facing an opening; the air there at 350 K, the surroundings at 300 Knet radiative flux, hot plate793.85 W/m² εσ(T⁴ − T_surr⁴)793.86 W/m²+0.0006 %±5 %within

What it says: the exchange lands on the exact value for both pairs, and the two plates agree with each other. In front of an opening the plate radiates to the surroundings you set (the Surroundings field on the heat card), not to the air in the opening: 793.86 W/m², where reading the air’s temperature would have given 480.56 W/m².

A sealed box with a hot blockEnergy balance and radiative share

A 20 mm block with a fixed power inside a closed 100 mm box whose walls are held at 300 K, no air movement. All the heat must leave through the box walls, by conduction through the air and by radiation, so the first check is that nothing is lost or made. The second is the block’s radiated heat against the two-surface enclosure formula, using the block’s own surface temperatures from the run.

casevaluereferenceMeshWeaveerrortoleranceverdict
20 mm block at 1 W in a 100 mm box at 300 K; ε block 0.85, walls 0.85heat out through the box walls1.000 W the heat put in1.000 W+0.0017 %±2 %within
20 mm block at 1 W in a 100 mm box at 300 K; ε block 0.85, walls 0.85heat the block radiates0.7554 W σA_b(⟨T_b⁴⟩ − T_w⁴) / (1/ε_b + (A_b/A_w)(1/ε_w − 1))0.7463 W−1.20 %±5 %within
20 mm block at 0.5 W in a 100 mm box at 300 K; ε block 0.1, walls 0.92heat out through the box walls0.5000 W the heat put in0.5000 W−0.0013 %±2 %within
20 mm block at 0.5 W in a 100 mm box at 300 K; ε block 0.1, walls 0.92heat the block radiates0.1404 W σA_b(⟨T_b⁴⟩ − T_w⁴) / (1/ε_b + (A_b/A_w)(1/ε_w − 1))0.1403 W−0.08 %±5 %within
20 mm block at 1 W in a 100 mm box at 300 K; ε block 0.85, walls 0.85; radiation offheat out through the box walls1.000 W the heat put in1.000 W+0.0000 %±2 %within

What it says: the heat balances to a few thousandths of a percent, and the radiated share matches the formula within 1.2 % with painted-like surfaces and within 0.1 % with a bare metal block. With radiation on, the ε 0.85 block radiates 75 % of its heat and runs at 348 K; with radiation off the same block reaches 490 K, 142 K hotter. Leaving radiation out of a still-air case is not a small error.

How to read itSaid plainly

The error is MeshWeave’s value against the reference beside it. Radiative exchange is held to ±5 % and the energy balance to ±2 %. These are steady cases with the walls given their power or temperature; radiation in a time-varying run, through conducting walls, or with a liquid coolant is not covered here. A finished Full solve with radiation shows, on the part, each heated region’s radiated share and the run’s energy balance.