A part isn’t one size — it’s one size at one temperature. Move it 10°C and it’s physically a different length, not a measurement error, and the amount it moves depends entirely on what it’s made of.
Every material has a coefficient of linear thermal expansion (CTE) — how much a given length grows per degree of temperature rise. The formula is simple, ΔL = L₀ × α × ΔT, but the consequence isn’t small: a 2000 mm steel weldment measured 10°C above a 20°C reference is genuinely about 0.24 mm longer than its drawing dimension, before anyone made a measurement mistake. That’s why every dimensional measurement is implicitly referenced to 20°C under ISO 1 — it’s not an arbitrary number, it was adopted internationally in 1931 specifically because it’s close enough to normal room temperature to hold in a lab without extreme heating or cooling.
| Material | CTE (µm/m·°C) | 1000 mm part, +50°C |
|---|---|---|
| Structural steel | ~12 | 0.60 mm |
| Stainless 316 | ~16 | 0.80 mm |
| Stainless 304 | ~17.3 | 0.86 mm |
| Aluminium (6061 / 5052) | 23.4 / 23.8 | 1.17 / 1.19 mm |
| Common 3D-print plastics (ABS, nylon) | 70–100+ | 3.5–5.0+ mm |
Aluminium expands roughly twice as fast as steel for the same temperature change. Bolt a steel fastener through an aluminium housing and heat the assembly, and the aluminium grows away from the bolt faster than the bolt does — which is a well-documented cause of gradual clamp-load loss in thermally cycled bolted joints, not a manufacturing defect. The same mismatch is what makes a bimetallic strip bend: two different metals bonded together, one expanding faster than the other, with nowhere to go but to curve.
Nobody’s checking gauge accuracy is usually fine — the part actually is a different size than it was that morning, and it’s genuinely a different size than a part measured in a 20°C metrology lab. I quote dimensions the way every drawing implicitly does — at 20°C — and when a part is going somewhere with real temperature swings (outdoor equipment, anything near heat sources, a shop with no climate control), the material’s CTE is part of the design decision, not an afterthought.