When we size a cooling system, the first number that matters is conduction through the mold steel. Fourier's law gives us Q = k·A·ΔT/L, where k for P20 mold steel sits around 29 W/m·K, and for H13 it drops to roughly 24 W/m·K. That difference is not academic — it changes cycle time. On a recent 2-cavity ABS part with a 3 mm wall, we calculated the heat load at about 1.8 kW per cavity. Plugging that into the conduction formula told us the steel thickness between the cavity surface and the cooling line had to stay under 45 mm, otherwise the temperature gradient across the tool would exceed 12°C and we'd start seeing warpage on the long axis. In the shop, that meant moving baffles closer to the core insert and accepting a slightly tighter waterline layout.

Convection is where most of the actual heat leaves the mold. Newton's law of cooling, Q = h·A·(T_wall − T_water), drives the design of the cooling circuit. For turbulent water flow at 2.5 m/s in a 10 mm diameter channel, h typically lands between 8,000 and 12,000 W/m²·K. Drop the flow to 1.2 m/s and you're in the transition zone — h falls to maybe 4,500 W/m²·K, and cycle time stretches by 15–20%. We've seen this on large automotive bumper molds where manifold pressure was undersized. The fix was not more cooling lines but higher flow: a 25 mm main manifold feeding 12 mm branches, with a 3.5 bar pressure drop budget. Radiation, by contrast, contributes less than 5% in a closed mold, so we treat it as a minor correction — except on hot runner manifolds, where surface emissivity and shielding actually matter for thermal balance.

The real skill is not solving the equations but knowing which one dominates at each stage of the cycle. During injection, conduction through the polymer melt controls the skin freeze. During packing, convection in the cooling channels takes over. During ejection, radiation from the exposed core can add a few degrees of uneven shrinkage. We keep a simple spreadsheet on the floor: input wall thickness, material k-value, coolant temperature, and flow rate, and it spits out an estimated cycle time within 8% of what the press actually does. That's the gap between theory and practice — not closing it perfectly, but closing it enough to quote a job without gambling. For more mold sourcing and technical resources, visit MoldWorld at www.moldw.com.