The core idea
A heat load is the rate at which heat enters a room — through walls, windows, people and equipment — measured in watts. A cooling unit must remove heat at that same rate to hold a steady indoor temperature. The envelope part is simple conduction:
Q = U · A · ΔT
where U (W/m²K) is how leaky the construction is, A the surface area (m²), and ΔT the indoor–outdoor temperature difference (K). A wall with U = 0.4, 12 m² of area and 8 K across it passes 0.4 × 12 × 8 ≈ 38 W — a single person often beats an entire wall.
The heavy hitter is solar gain through glass: roughly 300–400 W per m² of sunlit window at midday. Add people (≈90 W sensible each when seated), equipment (nameplate or metered watts), and ventilation air. For the air side: Q ≈ 0.33 × V̇(L/s) × ΔT — every litre per second of outdoor air at 10 K warmer than the room sneaks in about 3 W. Size the unit to the total, with a modest margin for latent (moisture) load.
Real-world example
A 30 m² top-floor living room on a 32 °C afternoon, held at 24 °C: 3 m² of west glazing in full sun contributes 350 × 3 = 1 050 W; the wall (U 0.4, 12 m² net) just 38 W and the roof (U 0.25, 30 m²) 60 W; two people 180 W; a TV, router and lights 150 W; ventilation of 60 m³/h ≈ 17 L/s about 44 W. Total sensible ≈ 1 520 W; with a ~15 % latent allowance, a 1.75 kW duty is the right size. The window alone is over half the load — which is why external shading beats a bigger compressor every time.
Common pitfall
Sizing by floor area rules of thumb ("100 W per m² → 3 kW for the room") instead of doing the load. The shortcut ignores the one thing that dominates: sunlit glazing. It oversizes shaded rooms — so the unit short-cycles, cools the air but never dehumidifies, and leaves the room cold and clammy — while undersizing the very room the rule was meant for. Count the watts, not the square metres.