Failure mode
Anodising Colour Variation
Two extrusions from the same order look different once assembled, or the second shipment does not match the first. It is visible only after installation, which is the worst time to discover it.
What it actually is
Two extrusions from the same order look different once assembled, or the second shipment does not match the first. It is visible only after installation, which is the worst time to discover it.
Mechanism
Why it happens
Four causes account for most of it. They are not equally common, and they are not equally within the mill's control.
Alloy composition drift
The silicon-to-magnesium ratio in 6063 decides how the oxide grows. Small composition differences between heats show up as visible colour differences.
Extrusion parameters
Billet temperature and exit speed set the grain structure. Coarse grain scatters light differently and reads as a different shade.
Bath control
Temperature and acid concentration in the anodising bath determine film growth rate. A drifting bath gives a drifting film thickness.
Mixed heats in one order
Anodising one load made up of two heats is the most common cause of within-order colour spread.
Verification
How it is measured
Film thickness is measured by eddy current, ISO 2360. Sealing quality is checked by the phosphoric-chromic acid mass loss method, ISO 3210. Colour difference is quantified as delta E with a colorimeter under a defined light source and observer angle — the number matters because "looks the same" is not enforceable.
Specification
How to specify around it
Most of these cost nothing to write into an order and are difficult to enforce afterwards. This is the part worth copying into your next RFQ.
1
State the film thickness class
AA10, AA15, AA20 — name the class in microns. It is the single most effective control on appearance and durability.
2
Require one heat per order
Ask that an order be anodised from a single heat, or that heats be kept separate and labelled.
3
Sign a master sample
Both parties retain a signed reference sample. Acceptance is measured against it, not against a description.
4
Set a delta E limit
Agree a numerical tolerance, commonly in the range of 1.0 to 1.5 delta E, and the light source it is judged under.
For anodised orders we ask whether the quantity can be run from a single heat, and we raise the master sample question before production rather than after delivery. Where the finish is architectural and visible, we say that the sample step is not optional.
We do not claim this never happens. Metal is rolled, shipped across an ocean and unloaded by someone else. What we do is ask the question before the order runs, put the answer in writing, and be the single party that answers for it afterwards.