Thickness tolerance and its effect on 5052 aluminium parts

A folded 5052 box comes back 0.8 mm long on the outside dimensions, and nothing on the route changed: same laser program, same tooling, same operator. The coil did. A 1.5 mm sheet that sits 0.09 mm heavy at one end of the mill band bends to a different developed length than one at nominal, and on four bends that difference is 0.94 mm.

5052 aluminium thickness tolerance is therefore a forming variable, not a purchasing detail. It moves bend deduction, blank length, formed dimensions, mass, press tonnage and the stack-up of a welded frame. The band is legal, repeatable and invisible on the print, which is why it gets blamed on the press brake.

This page gives the band, the arithmetic that turns it into a dimension, the measurement method that finds it, and the drawing notes that stop it becoming a non-conformance. The alloy and forming basics are in 5052 aluminium for sheet metal parts.

What a mill can actually guarantee on 5052 thickness

A rolled sheet is not a ground plate. The mill holds thickness within a band that widens with nominal thickness and with width, and every coil sits somewhere inside that band. For 5052 sheet between 1.0 mm and 3.0 mm at common widths, the permitted variation is of the order of ±0.07 to ±0.10 mm, about 5–7% of nominal. ASTM B209 defines the permitted variation by thickness and width class, and the mill certificate states the coil each certificate covers.

Two features of the band matter more than its size. The first is coil-to-coil variation: consecutive coils from the same mill can sit at opposite ends of the band, so a part made last month and a part made this month can differ by 0.18 mm in thickness with both inside specification. The second is crown: a rolled coil is slightly thicker at mid-width than at the edges, typically by 0.02–0.03 mm, so the same sheet yields a different bend near the centreline than at the edge.

A third point is often confused in review. ISO 2768-1 covers general tolerances for linear and angular dimensions where the drawing does not call out an individual tolerance. It says nothing about raw material thickness. A drawing that invokes ISO 2768-1 and specifies "1.5 mm" without a material tolerance has said nothing enforceable about the sheet, and the mill band applies by default.

The band, turned into numbers

Bend deduction in the table is for a 90° bend at a 1.5 × t inside radius with K = 0.44, which reduces to 1.953 × t. The "band" column is the half-width of the change across the permitted range, not the mill tolerance doubled.

How 5052 aluminium thickness tolerance moves the bend

The bend allowance and the setback both move

Bend allowance is (π/180) × angle × (R + Kt). With a 90° bend at a 1.5 × t inside radius — 2.25 mm on 1.5 mm sheet — and K = 0.44, the thickness term contributes 0.44 × t inside the bracket, so bend allowance changes by 1.5708 × 0.44 = 0.69 mm for every 1 mm of thickness. Over a ±0.09 mm band that is ±0.062 mm.

The outside setback is (R + t) tan(angle/2), which for a 90° bend is simply R + t. It changes by the full thickness change, ±0.09 mm, with nothing to damp it. Bend deduction is twice the setback minus the allowance, so it moves 2 × 0.09 − 0.062 = ±0.118 mm, or 0.236 mm across the whole band.

The K-factor partly compensates, and not enough

K is not a constant. As the sheet gets thicker at a fixed punch radius, the radius-to-thickness ratio falls and the neutral axis migrates towards the inside of the bend, so K drifts down from about 0.44 towards 0.40. That reduces the allowance slightly and takes a little back from the setback change.

It is a partial correction. In practice the net movement across a ±0.09 mm band on a 90° bend at R = 1.5t is about 0.09–0.12 mm of bend deduction either side of nominal, which matches the table rather than cancelling it.

A worked four-bend box

Take a 1.5 mm 5052 box with four 90° bends, flanges 60, 40, 60 and 40 mm, a 2.25 mm inside radius and K 0.44. Nominal bend deduction is 2.93 mm, so the flat length is 200 − 4 × 2.93 = 188.3 mm.

Now develop the same part for a coil at 1.59 mm and one at 1.41 mm:

• At 1.59 mm: allowance 1.5708 × (2.25 + 0.44 × 1.59) = 4.63 mm; setback 2.25 + 1.59 = 3.84 mm; deduction 2 × 3.84 − 4.63 = 3.05 mm. Four bends need 200 − 12.19 = 187.8 mm of flat.
• At 1.41 mm: allowance 4.51 mm; setback 3.66 mm; deduction 2 × 3.66 − 4.51 = 2.81 mm, so four bends need 188.8 mm.

The 188.3 mm blank developed for nominal, formed on the thick coil, produces a box 0.47 mm oversize across the four flanges — about 0.12 mm per flange. On the thin coil the same blank produces a box 0.48 mm undersize. The spread between the two coils is 0.94 mm, or ±0.47 mm on an overall dimension that spans all four flanges, and none of it is a machine error.

Mass, tonnage and stack-up follow the same number

Mass scales linearly. A 2 m² panel in 1.5 mm 5052 weighs 8.04 kg at nominal, 8.52 kg at the top of the band and 7.56 kg at the bottom — a 0.96 kg spread on one panel. On a lift car door with a specified mass for the gearless machine, that spread has to be inside the allowance.

Tonnage scales with thickness squared. Air-bending force per metre is 1.42 σu t² / V, so for 2.0 mm 5052-H32 at 215 MPa with a 16 mm V die the nominal force is 76 kN/m. At the top of a ±0.10 mm band it is 84 kN/m and at the bottom 69 kN/m — 21% either side of nominal. On a 4,450 mm bed that is still far inside a 200 T brake, but on a smaller machine the top of the band is the case that has to be checked.

Stack-up is where a welded frame pays. Six 1.5 mm 5052 panels clamped into a rail give a nominal 9.00 mm; at the top of the band the stack is 9.54 mm and at the bottom 8.46 mm. A ±0.54 mm stack against a ±0.5 mm rail gap is a fit that works on average and fails at the extremes. Gauge the stack, or specify a shim, or set the gap from the measured stack rather than from nominal.

Measuring it so the number is usable

A micrometer with flat anvils is the instrument for sheet. Take five readings: one at the centre and one 25 mm in from each corner. Do not measure at the sheared edge, where the burr reads high and the shear zone reads low.

The instrument has to be fit for the tolerance it is judging. A 0–25 mm micrometer reads to 0.01 mm, which is adequate against a ±0.09 mm band. A vernier caliper resolves to 0.02 mm at best and its jaws rock on a sheet, so its repeatability is a large part of the band being measured. Check the micrometer against a gauge block at the start of each shift and record the zero error, because a 0.02 mm zero error is a fifth of the thickness tolerance in question.

For a coated or a very large panel, use an ultrasonic thickness gauge; eddy-current instruments measure coating over an aluminium substrate, not the substrate itself, so they answer a different question. Record the coil number, the mill certificate value and the five readings on the inspection record, and compare them — a coil that reads 0.06 mm above its certificate is worth a conversation with the mill before it becomes a batch problem.

Frequency follows risk. Every new coil at goods-inward, then a per-batch spot check on the formed part. If the part is developed to ±0.5 mm and the coil band alone uses ±0.47 mm of it on a four-bend box, the sizing check at goods-inward is doing real work.

On the Ganesh route

Thickness is recorded at five points on the first article inspection, against the coil certificate and the drawing note. Parts are cut on the 12 kW fibre laser with its 40 ft × 16 ft bed, formed on the 200 T 5-axis CNC press brake to 10 mm over a 4,450 mm bed, and where the part is coated, the DFT gauge and cross-hatch test are recorded per batch, with PPAP and first article inspection available on request.

Typical sheet metal tolerance is ±0.5 mm, tighter on critical features by agreement, and both plants hold ISO 9001:2015 and ISO 14001:2015. Where a 5052 part is dimensionally sensitive, the flat pattern is issued with the K-factor and the thickness basis written on it, so the next coil change is a recalculated blank rather than a press-brake adjustment found by trial. The materials and grades hub holds the other alloy pages, and the custom parts range covers 0.8–40 mm across CRCA, stainless, aluminium, brass and copper.

What to put on the drawing

1. Nominal thickness and the permitted tolerance, or a reference to the mill standard and thickness class.
2. The measurement basis — five readings, 25 mm in from the edges — if the tolerance is measured rather than assumed.
3. Which dimensions are formed and which are cut, so the tolerance block does not get applied to a developed length.
4. Bend radius, bend angle and the permitted angular tolerance for every bend.
5. Flat pattern ownership: whether the supplier develops it and on which K-factor.
6. Flatness and twist limits on the formed part, in millimetres over the part length.
7. Fastener grip range, where a rivet or bolt passes through the sheet, covering the whole thickness band.

The width of the tolerance block on a formed part is a separate subject, and the inspection method for it is set out in bending tolerances and inspection. The K-factor that turns the flat into the bend is derived in how to calculate bend allowance.

When the thickness band cannot be the fix

If a part genuinely needs ±0.1 mm on a formed dimension in 1.5 mm sheet, a tighter material call-out is not the route. The workable options are a checking fixture that holds the critical dimension through welding, a machined feature added after forming, or a process change to press forming where the tool sets the geometry. Grinding sheet to thickness is a different industry and we do not do it.

Parts that are flat-critical rather than formed-critical are a different conversation, because the flat pattern is then cut directly and no bend deduction intervenes.

Send the drawing, the dimension that drifts and the coil details, through request a quote, and the review will state the thickness basis the flat pattern assumes before the parts are made.