Floating paper lanterns

Hot air in a paper envelope: heat balance, Archimedes lift, a float ceiling and a wind that carries them away.

Switch to 3D

Presets

Playback

Lantern

Volume, mass, burner power and heat loss decide how fast it lifts and how high it floats.

Air and wind

Ground temperature, a power-law wind profile with veer and gusts.

Fleet

Lanterns are released one after another with a little spread.

Actions

Time speed goes up to 60x so a kilometre of flight takes about a minute.

Layers

Model checks

The science

Heat balance of the air inside

The burner puts power Q into the air in the envelope, which loses heat to the outside through the paper and the open mouth with an effective conductance U A (A is the envelope surface, U an effective coefficient that lumps convection, radiation and spill of warm air). With a lumped heat capacity C = rho V c_p + C_paper the temperature T obeys C dT/dt = Q - U A (T - T_a). It approaches T_a + Q/(U A) exponentially with the time constant tau = C/(U A), which is the 30 to 90 seconds you wait before a real lantern lifts off.

Lift and weight

The envelope is open at the bottom, so the hot air is at the ambient pressure p and has the density rho_in = p/(R T) against rho_a = p/(R T_a) outside. Archimedes gives a net upward force F = (rho_a - rho_in) V g - m g, where m is the paper, frame and fuel; the weight of the hot air itself is already inside the term rho_in V g. The fuel cell burns away, so m falls by about 12 g over the burn time.

Motion

Vertically (m + rho_in V + C_a rho_a V) dv/dt = F - (1/2) rho_a C_D A_v |v| v, with an added-mass coefficient C_a = 0.5. Where the drag balances F the climb speed is the terminal value v = sqrt(2F/(rho_a C_D A_v)), about 1 to 2 m/s. In the horizontal plane the lantern relaxes to the local wind with the same quadratic law on the side area A_s. The wind follows a power law u(z) = u_10 (z/10 m)^alpha, turns with height by a set number of degrees per kilometre (the Ekman spiral), and carries gusts modelled as an Ornstein-Uhlenbeck process with a 15 s correlation time.

Atmosphere and the float ceiling

The standard atmosphere has T_a(z) = T_0 - 0.0065 z and p(z) = p_0 (T_a/T_0)^(g/(R L)), so the density falls with height while the temperature excess Q/(U A) stays the same. Lift therefore shrinks with altitude and vanishes at the ceiling where (rho_a - rho_in) V = m; the simulation finds it by bisection and the lantern settles there if the fuel lasts. When the fuel is gone the air cools with time constant tau, lift turns negative and the lantern sinks.

Numerical checks on this page

(1) The simulated climb speed with fixed hot air against the closed form above; (2) the stored heat against the time integral of burner power minus losses, and the steady and decaying temperatures against Q/(U A) and exp(-t/tau); (3) the settling altitude in calm air against the root of lift = weight, and the model density at 11 km against the tabulated standard-atmosphere value 0.36391 kg/m3.

Honest simplifications: a lumped (single-temperature) model of the air; U is an effective, tunable coefficient, not a measured property of paper; no rotation, swinging or flame flicker dynamics; no paper stretch, wetting or tipping over; the wind is a smooth statistical profile, not a weather forecast; no rain, clouds or terrain. Real lanterns also vary in shape and fuel. Trends and orders of magnitude are right, individual trajectories are not forecasts. Lanterns are a fire and litter hazard in reality and are banned in many places.

What to try
  1. Press the first preset and watch the first lantern wait on the pad while the air heats; graph 2 shows the temperature rising toward the dashed equilibrium and the lantern lifting off when the net lift turns positive.
  2. Choose the weak burner preset: the float ceiling reads "none: too heavy" and the lanterns never leave the pad. Raise the burner power or the volume until lift appears.
  3. Choose the calm night preset: with no wind the lanterns climb straight up and settle at the green ceiling line of graph 1 (compare it with graph 3 where lift crosses zero).
  4. Compare the winter and summer presets: the colder ground gives a larger lift for the same temperature excess, so a cold night lifts higher.
  5. Set the wind turning to 70 degrees per km in the strong wind preset and watch the plan view: lanterns at different heights are carried in different directions.
  6. Use the short burn preset or "Cut the burner for 60 s" and see the lantern descend as the air cools.
Altitude of every lantern
Air inside the lead lantern: excess temperature
Lift at the equilibrium temperature against altitude (the ceiling is where it crosses zero)
Plan view of the drift (north up)