How to Reduce Aluminum Die Casting Cost

Most requests to reduce die casting cost arrive as a request to reduce the quote. That conversation has a floor, and it is a low one: metal price, machine time and labour are what they are, and a supplier who cuts below them is cutting something you will meet later as a reject or a tooling correction.

The savings sit upstream of the quote, in decisions about geometry, volume and specification. On the housing worked through at the end of this article, none of the 39% reduction came from the price list.

Where die casting cost actually comes from

This is a build-up for a 480 g (1.06 lb) A380 housing at 20,000 units per year: single-cavity die with two slides, six machined faces, clear anodized finish. Cell rates are illustrative but internally consistent at $54 per hour for a loaded casting cell and $38 per hour for a machining centre including fixture and perishable tooling.

Two things in that table are worth sitting with. The casting itself, the operation the supplier is nominally selling, is $0.93 of $11.20, while machining and finishing together are $4.50. Almost half the price is spent removing and covering material the die already formed, which is why cost reduction on a casting is mostly a conversation about the drawing.

And metal exceeds casting conversion. It usually does, because runners and overflows are trimmed and returned to the melt, so only net part weight plus melt loss is consumed. Weight reduction therefore attacks the second-largest controllable line directly.

The cost levers, ranked by impact

The ordering matters more than the individual percentages. The four largest levers are design decisions, and three of them close permanently once the die is cut.

Volume and cavity count

The cavity arithmetic

Adding a cavity does not halve cycle time; it lengthens it slightly, because the shot is bigger and the die carries more heat. But it yields two parts per cycle instead of one, so conversion cost per part still falls sharply.

Read the last two columns together. One cavity to two saves $0.36 per part for $6,500 more tooling, paying back at about 18,000 parts: a nine-month payback at 25,000 units a year, and a four-and-a-half-year mistake at 4,000. Two cavities to four saves a further $0.19 against $11,500, needs roughly 60,500 parts, and usually needs a larger machine because clamping force scales with projected area.

Those break-even figures are lifetime, not annual. A die survives 80,000 to 150,000 shots, so a four-cavity tool on a program with 60,000 lifetime units is over-tooled however healthy the annual number looks. Cost drivers by cavity and slide count are broken down on the die casting tooling page.

Committing to an annual volume

The volume figure on your RFQ does more work than any other input. It sets cavity count, machine selection, whether extraction is manual or automated, and whether the machining fixture is manual-clamp or hydraulic.

• State the real annual volume, not a cautious one. A quote built on 3,000 units per year is a single-cavity quote. If demand turns out to be 20,000, you are locked into that tool and paying $0.36 per part more for its life, which is $7,200 a year against $6,500 of one-time tooling.
• State the three-year outlook separately. A tool can be designed as a two-cavity unit die with provision for a second insert set later, which is a different decision from a plain single-cavity die.
• Consolidate releases. Twelve monthly releases of 1,700 parts and four quarterly releases of 5,000 have the same annual volume and different costs, because each run carries two to four hours of cell setup and thermal stabilisation.

Design levers

Part consolidation

The largest lever available, and the one most often left on the table because it requires looking at an assembly rather than a part. Geometry lives in the steel, so brackets, bosses, ribs, cable routes, gasket lands and standoffs that currently exist as separate parts can become features of one casting at no incremental piece cost.

Here is what the accounting looks like when a five-piece welded bracket becomes one casting at 10,000 units per year.

Most of that is not material. It is operations that stopped existing. The same pattern appears when a casting replaces a bolted sub-assembly or an extrusion bonded to a housing. Our brackets and mounts page covers the weldment case in more detail.

Consolidation has a limit worth naming: every part you absorb makes the casting bigger, and it stops paying the moment it pushes past a machine tonnage step or forces a wall thickness increase to keep the flow path filling.

Wall thickness reduction

Wall thickness pays twice. Material weight scales linearly, so a wall-dominated part taken from 4.0 mm to 2.5 mm loses 30 to 35% of its weight. Solidification time scales with roughly the square of section thickness, so a 2.5 mm wall freezes in about 39% of the time a 4.0 mm wall needs and the dwell portion of the cycle collapses.

The rule that makes this safe is to recover stiffness with ribs rather than thickness. Ribs at 60 to 80% of the adjoining wall with 1 mm minimum internal fillets add second moment of area without creating the isolated thick sections that become shrinkage porosity. A 1.4 mm rib grid on a 2.0 mm wall out-stiffens a plain 3.2 mm wall at lower weight and shorter cycle. The floor is 1.5 mm typical, 1.0 mm on small parts with short flow paths, and the target values are in the aluminum die casting design guide.

Eliminating slides and undercuts

Every slide is $2,000 to $5,000 of tooling, seconds of slide-pull time on every cycle, an extra flash line to deburr, and a wear item to maintain across the tool’s life. Three questions clear most of them out of a design.

1. Can the feature rotate into the draw direction? A side-entry cable boss often works equally well angled 30° toward the parting line, where it pulls straight.
2. Can the undercut become a machined feature? If the face is already being machined in that setup, a milled slot can beat a slide outright below about 15,000 units per year.
3. Can the parting line move? A stepped parting line costs more to cut than a flat one and far less than two slides.

Specification levers

Machining only what function requires

Machining is the largest single line in most cast piece prices, and the drawing sets it, not the process.

The expensive pattern is not a tight tolerance. It is tight tolerances scattered across faces that cannot be reached in one fixture, because setup count drives cost harder than cycle time. One bearing bore and one sealing face is routine; the same two plus four holes on a third face turns one setup into three.

Delete every tolerance that exists because a template had it, group the features that genuinely need precision onto one or two faces at design time, and use small machined datum pads instead of machining a whole face where all you need is a repeatable three-point reference. The mechanics are covered in CNC machining after die casting.

Choosing the cheaper alloy where the application allows

The guidance runs one way. If the part is an internal bracket or an unsealed housing, A380 is correct and anything else is money spent on properties nobody will measure. If it must hold pressure, live outdoors or move heat, the premium grade earns its cost. The full comparison sits on the materials hub.

Finish selection

Every casting ships trimmed, deburred and shot blasted to a uniform matte at Ra 1.6-3.2 µm, so a part needing neither protection nor colour needs no finish line in the quote. The two most common overspecifications are a cosmetic finish on a component sealed inside another enclosure, and anodizing on a die casting, which costs more than powder coat and produces a darker, patchier result because 7.5 to 12% silicon does not convert to oxide. The surface finishing page carries the full comparison.

Commercial levers

Tooling amortisation versus upfront payment

Upfront is cheaper in total, typically by 3 to 6% of piece price, because amortisation transfers financing cost and volume risk to the supplier and both are priced in. It is still the right answer when volume is genuinely uncertain, or when a capital request is harder to approve than a unit cost increase.

Two related points. State ownership explicitly, including that the die is your property once paid for and can be transferred. And scope tooling against program life rather than the first order, because a die specified for 40,000 shots on a program that will run 200,000 needs replacing at a worse moment than the original quote implied.

Packaging and freight consolidation

The smallest lever, and the one most often ignored because it sits with a different department.

• Design the carton around the part. A 15% gain in cube fill is a 15% reduction in freight per part on a volumetric shipment.
• Ship full pallets and, where possible, full containers, since part-container freight carries a disproportionate handling cost.
• Consolidate multiple part numbers into one shipment on a common release schedule.
• Where parts nest, design them to nest, which is a second quiet dividend from thinner walls.

A worked example: $11.20 to $6.80

The housing from the first table, taken through three changes. Commercial terms did not change and volume stayed at 20,000 units per year.

Change 1: wall thickness. Nominal wall from 4.0 mm to 2.5 mm, stiffness recovered with a 1.7 mm rib grid at 14 mm pitch. Part weight fell from 480 g to 330 g, a 31% reduction, and solidification dwell fell with it.

Change 2: slides eliminated, cavity count doubled. The side-entry cable boss rotated 34° into the draw direction and the side undercut became a milled slot on a face already being machined. Removing both slides made a two-cavity straight-pull die affordable at $19,600 against $14,800 for the single-cavity die with slides. Cycle time went from 62 s to 52 s, so per-part machine time fell from 62 s to 26 s.

Change 3: tolerances and finish rationalised. A blanket ±0.05 mm on 34 dimensions became ±0.05 mm on the six that carry function, with everything else at ±0.25 mm as-cast. Machining went from two setups and 4.8 min to one setup and 2.4 min. The clear anodize became shot blast plus trivalent chromate, because the part sits inside a sealed enclosure and is never seen.

At 20,000 units per year that is $88,000 annually, and the extra $4,800 of tooling repaid itself in about 1,100 parts, roughly three weeks of production.

Notice which lines moved most. Machining and finish account for $2.53 of the $4.40 and neither required touching the die, so a frozen design would still have delivered 23%.

False economies to avoid

Chasing the lowest quote. The cheapest quote is usually the one that assumed the least: a lighter die base, no slide allowance, no DFM review, sampled rather than full inspection, coating to a generic specification, tool life scoped for a fraction of the program. If one quote sits 30% below the others on the same drawing, the gap is a scope difference and it is worth finding before the purchase order rather than after.

Over-tolerancing as insurance. Applying ±0.05 mm everywhere because it feels safer is the most expensive habit in the business. It converts free as-cast features into machined ones, multiplies setups, adds gauges, and creates dimensional rejects on features nobody would have measured.

Cosmetic finishes on hidden parts. A coating callout inherited from a previous revision, on a component that lives inside a sealed housing, is $0.60 to $2.00 per part for nothing, plus masking, a batch record and a cosmetic reject category that would not otherwise exist.

Downgrading alloy where the property is load-bearing. The mirror image. Substituting A380 for A360 on a housing that must hold 6 bar saves $0.20 per part and risks leak rejects at 100% inspection, field returns and a requalification. Downgrade alloy where the property is unused, never where it is the requirement.

One more that behaves like a false economy without quite being one: cutting volume commitments to preserve flexibility. It is a legitimate choice made knowing the price. Below roughly 1,000 to 2,000 parts per year tooling amortisation dominates everything else and machining from billet is often cheaper outright, which is the calculation in die casting versus CNC machining.

How to run this on your own part

1. Get the piece price broken into the ten lines in the first table. If it will not break out, you cannot target anything.
2. Count the dimensions tighter than ±0.1 mm, then count the ones a functional requirement actually depends on. The ratio is usually revealing.
3. Ask whether the nominal wall exists for stiffness or for habit, and whether ribs would do it lighter.
4. Look at the assembly rather than the part, and count the adjacent pieces that could become features.
5. Count the slides, and ask the three questions above about each one.
6. Put the real annual volume and three-year outlook on the RFQ, and ask for cavity count options with the break-even volume against each.

Send the model and drawing through contact and the written DFM report comes back with the levers that apply to your geometry, the piece price at each cavity count and the tooling cost against each. Where the honest answer is that your volume does not justify a die at all, that is what the report will say.