Why a chimney draws
Hot flue gas is lighter than the cold air outside. The two columns of different density weigh differently and the difference pushes gas up the pipe: Δp = g H (ρout - ρflue) with the ideal-gas density ρ = p M / (R T). The model walks along the pipe in 25 cm steps, cools the gas with T(s) = Tamb + (T - Tamb) exp(-U π D Δs / (ṁ cp)) and adds up the weight difference for every vertical piece (against room air inside, against outside air above the wall). A 4 kW fire and a 3.5 m insulated riser give about 13 Pa, the value stove makers ask for; a horizontal run gives no draft. The draft is used up by the flow resistance Δp = ṁ² Σ K / (2 ρ A²) of the air inlet, fuel bed, damper, elbows, wall friction, soot blockage and the exit. Closing the damper makes K = (1/(0.7 φ) - 1)² huge, so the flow falls.
Combustion and carbon monoxide
The fuel is described by its ultimate analysis (C, H, O, ash, moisture). Per kg it needs nO2 = nC + nH/4 - nO/2 moles of oxygen, 6.0 kg of air for dry wood and 17.2 kg for methane. The air ratio is λ = O2 supplied / O2 needed. With enough air all carbon becomes CO2. When oxygen is short, carbon can only go to CO (needs half the oxygen) or stay as soot, tar and unburnt hydrocarbons; the model gives hydrogen the first oxygen, then splits carbon between CO2 and CO, and a third of the starved carbon leaves as CO, capped at 30 % of the fuel carbon as measured for smouldering stoves. Even with excess air, poor mixing leaves a little CO that rises steeply below λ ≈ 1.3: ε(λ) = ε0 (1 + 6 exp(-(λ - 1)/0.35)) with ε0 = 1.5 % for a simple stove, 0.4 % for a modern one and 0.04 % for a gas burner. This is a heuristic fitted to the size of measured emissions, not a flame-kinetics code. Wet fuel burns cooler and dirtier, heating value LHV = LHVdry(1 - w) - 2.44 w (18.4 MJ/kg dry wood, about 14 MJ/kg at 20 % moisture). Unburnt CO (283 kJ/mol) and soot (394 kJ/mol) are subtracted from the heat.
What the stove really does with the heat
The hot gas gives heat to the steel body by the effectiveness-NTU law (NTU = UAg / (ṁ cp)); the body loses it to the room by natural convection plus radiation, and its temperature follows Cs dTw/dt = Qgas - Qloss. The bare pipe in the room is a radiator too. The room warms by C dT/dt = Q - (UA + ρ cp Qair)(T - Tout), so every cubic metre of ventilation air costs heat: this is the trade-off that makes people close vents.
Why a closed damper, a sealed room or a fan are dangerous
Fire in a stove is not switched off by closing the air: a hot bed of embers keeps releasing flammable gases whatever the air supply (the model keeps 25 to 30 % of the set rate for as long as the bed is hot). If the chimney cannot carry the gas away, the surplus leaves through the stove door and air inlet into the room. The same happens when the room is so tight that a stove (and a kitchen extractor fan, which loses flow as the underpressure rises) lowers the room pressure by as much as the draft: Δpnet = Δpdraft + proom falls to zero and the chimney runs backwards. Even a weak positive draft is not safe: wind gusts of a few pascals push the flow backwards part of the time. The model takes that as a normal distribution with spread 0.3 + 0.25 Δpwind and lets 60 % of the gas roll out for the share of time the net draft is negative. Room pressure comes from a mass balance of leakage in, leakage out, fan and stove, solved by bisection.
Carbon monoxide in the room
CO is colourless and odourless and has almost the density of air. In a well-mixed volume dC/dt = G/V - ACH·C, so it grows as C(t) = (G/(V ACH))(1 - e-ACH t) toward Css = G/(V ACH); the ventilation needed to hold a limit is ACH = G/(V Climit). The page uses three connected zones (near the stove, the rest of the room, the bedroom behind the door) with an exchange 0.5 Cd A √(g' H) through the door gap or opening. Guidelines: WHO indoor air quality guidelines 2010 (100 mg/m³ 15 min, 35 mg/m³ 1 h, 10 mg/m³ 8 h, 7 mg/m³ 24 h; at 20 °C 1 ppm = 1.165 mg/m³) and the NIOSH limits (35 ppm 8-h average, 200 ppm ceiling, 1200 ppm immediately dangerous). Chronic and acute effects depend on the person; the model does not predict health outcomes.
Limits of this model
One cabin, one steel stove and a round pipe, calibrated by physical reasoning rather than by measurements of a particular appliance. Heat transfer coefficients, the CO partition and the spill rules are simplified; real chimneys have unsteady puffs, thermal stratification, temperature inversions, wind pressure on the roof, leaks at joints and the human factor. Products of incomplete combustion other than CO (soot, tar, nitrogen oxides, fine particles) are not modelled. No appliance, rule or setting in this page is a safety recommendation: follow the maker's instructions, a qualified installer and local regulations. The cabin is an original model; names of appliance types are generic.