Blender Langstroth hive with instanced bees and a nest heat balance: shivering heat, wall and vent losses, fanning, evaporation and honey use.
GraphsNest and outside temperature over the last 48 simulated hours
Metabolic heat made by the bees vs heat lost to the outside
Steady nest temperature vs outside temperature for these settings (white dot: now)
The science
Lumped heat balance. The occupied part of the nest is one thermal mass with temperature T: C dT/dt = Pbase + Pheat + Psun - (UA + Gfan)(T - Tout) - Pevap. The heat capacity is C = 22 kJ/K + 0.35 J/K per bee (comb, honey, wax, wood and bodies). The 1 s RK2 time step is far below the time constant C/UA (several hours).
Bee heat and control. Each bee releases about 1.5 mW at rest (Pbase = N p0). A proportional controller adds shivering heat Pheat = Kh (Tset - T), limited to 30 mW per bee. With brood the set point is 34.5 C; without brood the winter cluster core is held near 20 C. These control constants are assumptions chosen to give realistic magnitudes, not measurements.
Walls. U = 1 / (Rsi + Σ di/ki + Rso) with pine k = 0.12 W/m K and foam k = 0.035. The outside film resistance falls with wind, Rso = 0.04 / (1 + 0.3 v). The wall area that really loses heat is the occupied nest envelope (0.3 m²); a fixed 0.15 W/K stands for the entrance and gaps between boxes.
Cooling. Above the set point, bees fan at the entrance. Each fanning bee moves about 0.15 L/s of air, adding gfan = ρ cp · 0.15×10-3 = 0.18 W/K, and water evaporation removes up to 25 W (latent heat 2.26 MJ/kg). Fanning bees also burn honey. Above about 40 C the model overheats, as real colonies can in extreme heat.
Honey ledger. Metabolic power is Pmet = Pbase + Pheat + 0.02 W per fanner; honey burned is ∫ Pmet dt / 13.4 MJ per kg.
Independent checks. (1) The time-stepped temperature after twelve time constants equals the root of the net-power balance found by bisection. (2) A dead colony (no bees, no sun) cools exponentially; a log-linear fit gives τ = C/UA. (3) The honey burned over a simulated day at steady state equals Pmet × 86400 s / 13.4 MJ/kg. (4) A finite-difference solution of 1D conduction through the layered wall and both films reproduces the series-resistance U-value.
Honest simplifications. One temperature for the whole nest (the real cluster has a 35 C core and a colder mantle, drawn only as a Gaussian map), a proportional controller instead of real bee behaviour, no humidity or CO2 model, no brood development, no comb heat storage lag, no radiation to the sky. Use the numbers for orientation: they show which factors matter (wall insulation, wind, colony size), not an exact honey budget.
What to try: compare Winter cluster with Insulated winter and read the honey use per day; raise the outside temperature past 30 C and watch the fanning bees at the entrance; remove all bees (Dead colony) and see the nest cool with its time constant; shrink the colony in a cold snap until the shivering heat saturates and the nest temperature falls.