High pressure die casting injects molten aluminum into a steel die in three phases: a slow shot at 0.1-0.5 m/s to clear the sleeve without trapping air, a fast shot at 2-6 m/s that fills the cavity in 20-100 milliseconds, and intensification at 40-100 MPa held through solidification. Aluminum must run on cold chamber machines because it dissolves the submerged steel of a hot chamber injection system. Machine tonnage is calculated as total projected area multiplied by cavity pressure plus a 15-25% safety factor, which puts a 459 cm² shot at 55 MPa on a 400 ton machine. Cycle times run 30-90 seconds and production dies last 80,000-150,000 shots.
High pressure die casting is often described as injection moulding with metal, which gets the shape of the idea right and the physics wrong. Plastic fills a mould in a second or two and cools over half a minute. Aluminum fills a die in 20 to 100 milliseconds and starts freezing on contact, which means the whole process is a race between a metal front and a solidification front. Almost every decision in HPDC, from plunger velocity to where the water lines are drilled, exists to win that race.
This article goes past the process overview in how aluminum die casting works and into the engineering decisions: shot profile, tonnage selection with the arithmetic shown, cavity count economics, vacuum assist, and where the process genuinely runs out.
Cold chamber and hot chamber
Two machine architectures exist, and for aluminum the choice is made for you.
| Cold chamber | Hot chamber | |
|---|---|---|
| Injection system location | Outside the melt | Submerged in the melt |
| Metal delivery | Ladled dose per shot | Drawn through a gooseneck |
| Alloys | Aluminum, magnesium, brass | Zinc, magnesium, lead |
| Typical cycle | 30-90 s | 5-30 s |
| Injection pressure | 40-100 MPa | 15-40 MPa |
| Air entrainment risk | Higher, open sleeve | Lower, closed system |
| Machine clamping range | 160-4,000 tons | 20-800 tons |
Aluminum must use cold chamber, and the reason is chemistry rather than preference. Molten aluminum dissolves iron readily, so a steel gooseneck and plunger sitting permanently in an aluminum bath is consumed rather than used. Zinc does not attack steel the same way, which is why zinc gets the faster architecture.
The penalty is measurable. Every cold chamber shot needs a ladle transfer into a horizontal sleeve that is open to atmosphere, so seconds are added and the metal has an opportunity to pick up oxide and entrain air before the plunger even moves. That single architectural constraint is responsible for the porosity characteristic of aluminum die castings and for cycle times two to three times longer than zinc.
We run 160 to 1,250 ton cold chamber machines, which covers castings from 20 g to 12 kg.
The three injection phases
Injection is not one movement. It is three, and where each transitions into the next determines quality more than any other machine setting.
| Phase | Plunger velocity | Duration | Purpose |
|---|---|---|---|
| Slow shot | 0.1-0.5 m/s | 1-3 s | Move metal to the gate without breaking a wave over the plunger |
| Fast shot | 2-6 m/s | 20-100 ms | Fill the cavity before the front freezes |
| Intensification | Static | 8-25 s | Hold 40-100 MPa while the part solidifies |
Phase 1: slow shot
The shot sleeve is a horizontal cylinder only partly filled with metal, which means there is a free surface, and a free surface can break. Push the plunger too quickly and the metal rolls forward as a wave, curls over on itself and traps a slug of air that then gets injected into the cavity ahead of the metal.
Correct slow-shot velocity keeps the wave rising smoothly to the sleeve roof just as the metal reaches the gate. The optimum is a function of sleeve diameter and fill fraction, and it typically lands between 0.1 and 0.5 m/s. Getting this phase wrong produces gas porosity that no amount of intensification pressure will remove, because the air is already inside the metal.
Phase 2: fast shot
Once metal reaches the gate, the plunger accelerates to 2 to 6 m/s. Because the gate cross-section is small compared with the plunger face, this multiplies into a gate velocity of 30 to 50 m/s, and the cavity fills in 20 to 100 milliseconds.
Gate velocity is a genuine optimum rather than a value to maximise:
- Too slow and the flow front loses heat, reaches 20 to 30% solid and stalls, giving cold shuts and misruns.
- Too fast and the stream atomises into a spray on entry, mixing air into the metal and eroding the gate area of the die.
Fill time can be estimated as cavity volume divided by gate area times gate velocity, but the useful discipline is the reverse calculation: decide the fill time the thinnest wall requires, and size the gate to achieve it.
Phase 3: intensification
Aluminum contracts about 6% by volume as it solidifies. Intensification is the answer to that. Once the cavity is full, hydraulic pressure is stepped up to 40 to 100 MPa (6,000 to 15,000 psi) and held while the casting freezes, feeding liquid metal into the shrinking interior and pressing the skin against the die wall.
Two consequences follow. Castings reproduce die surface texture faithfully and reach Ra 1.6 to 3.2 µm as-cast. And density in thin sections is high enough that the as-cast skin is the soundest material on the part, which is why machining allowance should be the minimum that clears draft and distortion rather than a generous margin.
Higher intensification is not free: it increases flash, loads the die, and accelerates heat checking at gates and corners.
Selecting machine tonnage
Machine clamping force has to exceed the force trying to blow the die open during injection. Undersize it and the die opens a fraction of a millimetre, which produces heavy flash, dimensional drift and eventually a damaged parting face.
The calculation
Separating force = total projected area × cavity pressure
Required clamp = separating force × safety factor (1.15 to 1.25)Total projected area is the shadow the whole shot casts on the parting plane: part, runner, biscuit and overflows. Cavity pressure is chosen from the part class:
| Part class | Cavity pressure |
|---|---|
| Simple, thick-walled, non-critical | 30-40 MPa |
| General structural housings and brackets | 45-60 MPa |
| Thin wall, leak-tight, high integrity | 70-90 MPa |
A worked example
Take the 400 g housing, 240 × 160 mm in footprint, 2.5 mm nominal wall, one cavity, specified as a general structural housing.
| Step | Calculation | Result |
|---|---|---|
| Part projected area | 24.0 cm × 16.0 cm | 384 cm² |
| Runner, biscuit and overflows | Estimated from layout | 75 cm² |
| Total projected area | 384 + 75 | 459 cm² |
| Convert to mm² | 459 × 100 | 45,900 mm² |
| Cavity pressure | Structural housing | 55 MPa = 55 N/mm² |
| Separating force | 45,900 × 55 | 2,524,500 N = 2,524 kN |
| In metric tonnes | 2,524 ÷ 9.807 | 257 tonnes |
| Safety factor | 257 × 1.20 | 309 tonnes |
| Machine selected | Next size above 309 | 400 ton |
The 280 ton machine in our range is 29 tonnes short of the calculated requirement and would flash the parting line, so the part goes on a 400 ton machine with 91 tonnes of margin. That margin is useful, not waste: it absorbs a cavity pressure increase if porosity turns out to need one.
One quotation trap worth knowing. Machine sizes are quoted in metric tonnes-force in Asia and Europe and often in US short tons in North America, a 10% difference. A “400 ton” machine is either 400 tonnes (3,923 kN) or 363 tonnes (3,559 kN) depending on who wrote the specification.
Rules of thumb and where they fail
The shorthand is 0.4 to 0.8 tonnes of clamping force per cm² of projected area, spanning the 40 to 80 MPa cavity pressure band. The example above works out at 0.56 tonnes/cm² before the safety factor.
It fails in two situations. Runners and overflows add 15 to 40% to projected area, so a rule applied to part area alone undersizes the machine. And a thin-wall or leak-tight part needing 80 MPa lands near the top of the band regardless of size, so a small part can need a surprisingly large machine.
Cavity count
More cavities divide the conversion cost but multiply projected area, which forces a larger machine and a more expensive tool. The trade has a break-even.
Using the same housing:
| Single cavity | Two cavities | |
|---|---|---|
| Total projected area | 459 cm² | 898 cm² |
| Required clamp at 55 MPa, 1.2 factor | 309 tonnes | 604 tonnes |
| Machine | 400 ton | 630 ton |
| Machine rate, all-in | $58/hour | $78/hour |
| Cycle time | 55 s | 60 s |
| Parts per hour | 65 | 120 |
| Conversion cost per part | $0.89 | $0.65 |
| Tool cost | $9,000 | $13,000 |
The second cavity saves $0.24 per part and costs $4,000, so it pays back at roughly 16,700 pieces. At the 2,000 pieces per year this housing actually runs, a single cavity is clearly correct. Above about 20,000 pieces per year, the second cavity returns its cost inside the first year and also doubles the parts obtained from an 80,000 to 150,000 shot tool life.
Two further considerations argue for restraint on cavity count. Balancing fill between four or more cavities is genuinely difficult, and unbalanced cavities produce part-to-part property and dimensional variation that is hard to diagnose. And a single-cavity tool holds tighter dimensional control, which matters when the part has a critical as-cast feature.
Vacuum assisted die casting
A vacuum valve at the far end of the fill path, connected to an evacuated tank, pulls the cavity down to roughly 50 to 150 mbar before and during injection. The metal front then advances into near-vacuum instead of compressing a cavity full of air.
| Conventional HPDC | Vacuum assisted | |
|---|---|---|
| Gas content, typical | 10-30 cm³ per 100 g | Under 5 cm³ per 100 g |
| Added tool cost | — | $3,000-8,000 |
| Added maintenance | — | Valve is a consumable, die must seal |
| Thin wall capability | Good | Better, helps below 1.2 mm |
| Blister risk on heating | Present | Substantially reduced |
| Weldability | Poor | Fair to good in low-copper alloys |
It is worth the cost when the part must hold pressure without impregnation, when it must be welded or brazed, when it is a structural safety component with fatigue requirements, when walls are below 1.2 mm and back-pressure is limiting fill, or when a machined face repeatedly exposes porosity. This is why vacuum shows up most often on automotive structural and fluid-handling parts.
It is not worth the cost on a cosmetic housing, a bracket that carries no pressure, or a part where conventional castings already pass. Vacuum reduces porosity; it does not eliminate it, and it does not turn a die casting into a forging.
Die thermal management
The die is a heat exchanger that happens to have a part-shaped hole in it.
The arithmetic is worth doing once. Cooling aluminum from 690 °C to a 200 °C ejection temperature releases roughly 875 kJ/kg: superheat above the liquidus, then about 390 kJ/kg of latent heat of fusion, then solid-state cooling. A 640 g shot therefore dumps about 560 kJ into the die, and at 65 shots per hour that is 36 MJ/hour, or a continuous 10 kW. A die casting cell is a 10 kW heater that has to hold the die surface between 180 and 280 °C.
How that heat is managed:
- Cooling channels. Typically 8 to 12 mm diameter, drilled with the centreline 1.5 to 2.5 line diameters from the cavity surface and spaced 3 to 5 diameters apart. Closer gives faster cooling and a shorter cycle, but too close weakens the steel and invites cracking.
- Jet cooling and bubblers. Cores, pins and boss formers cannot take a through-drilled channel, so they get an internal tube that delivers coolant to the tip. Without it, core pins run hot, solder and break.
- High-conductivity inserts. Beryllium-copper or equivalent inserts at local hot spots pull heat out of geometry that water cannot reach.
- Die spray. The water-based release agent applied between shots both lubricates ejection and removes 10 to 25% of the total heat load. It is also a process variable: spray volume and dwell drift over a shift and take dimensions with them.
- Thermal equilibrium. A cold die produces different parts from a hot one, so 20 to 50 shots are run to stabilise the die before first-off inspection. This is why short runs carry disproportionate setup cost.
Getting this wrong is directly visible on parts. Too cold and you get cold shut, misrun and poor surface reproduction. Too hot and you get soldering, longer solidification dwell, more flash, and dimensional drift across the run. Both sets of symptoms are catalogued in common aluminum die casting defects.
Cycle time and how it reaches piece price
A representative 55 second cycle for a mid-size housing:
| Element | Time |
|---|---|
| Ladle and pour into shot sleeve | 5 s |
| Slow shot | 2 s |
| Fast shot, cavity fill | 0.05 s |
| Intensification and solidification dwell | 23 s |
| Die unlock and open | 3 s |
| Ejection and part extraction | 8 s |
| Die spray and blow-off | 9 s |
| Die close and lock | 5 s |
| Total | 55 s |
Cavity filling, the part everyone focuses on, is 0.1% of the cycle. Solidification dwell is 42%, and it is the only element that part design controls, because solidification time scales with the square of section thickness.
Conversion cost follows directly:
parts per hour = 3600 ÷ cycle seconds × cavities
conversion per part = machine hourly rate ÷ parts per hourFor the single-cavity housing: 3600 ÷ 55 = 65 parts per hour, and $58 ÷ 65 = $0.89 per part. Take the nominal wall from 3.0 mm to 2.2 mm and dwell falls from 23 to about 12 seconds, giving a 44 second cycle, 82 parts per hour and $0.71 per part. That is $0.18 saved, worth $18,000 a year at 100,000 pieces.
Keep it in proportion. On the same part, deleting one machined face saves more than $1.00. Cycle time is worth engineering at high volume and is rarely the largest lever at moderate volume.
HPDC compared with other casting processes
| Attribute | High pressure die casting | Permanent mould (gravity) | Sand casting |
|---|---|---|---|
| Tooling cost | $8,000-45,000 | $4,000-15,000 | $500-4,000 |
| Tooling lead time | 25-40 days | 15-25 days | 5-12 days |
| Minimum wall | 1.0-1.5 mm | 3-4 mm | 4-6 mm |
| As-cast linear tolerance | ±0.1 mm | ±0.3-0.5 mm | ±0.8-1.5 mm |
| Surface roughness | Ra 1.6-3.2 µm | Ra 6-12 µm | Ra 12-25 µm |
| Cycle time | 30-90 s | 3-10 min | 10-30 min |
| Economic annual volume | 1,000-500,000+ | 300-10,000 | 1-1,000 |
| Porosity character | Fine gas porosity typical | Less gas, more shrinkage | Coarser, more of both |
| T6 heat treatable | No | Yes | Yes |
| Typical alloys | A380, ADC12, A360, A413 | A356, 355 | Widest range |
The pattern is consistent: HPDC buys thin walls, tight as-cast tolerance, fine surface and short cycles, and pays with tooling cost, lead time and entrapped gas. Sand casting inverts every one.
What HPDC cannot do
Being direct about the limits saves programs from failing late.
- Porosity is always present. It can be controlled to a specified X-ray level and it is often irrelevant, but a machining cut into a thick section can expose it and a pressure boundary through it can leak.
- No T6 heat treatment. Solution treating at 500 to 540 °C blisters the surface as trapped gas expands. T5 stress relief at 200 to 250 °C is the practical ceiling.
- Poor structural weldability. Trapped gas boils into the weld pool. Design for fasteners, staking or adhesive instead.
- No wrought alloys. A380, ADC12, A360 and A413 are the practical list. 6061 and 7075 cannot be die cast.
- Sections above about 6 mm. Thick regions develop shrinkage porosity at their thermal centre and need to be redesigned as ribbed thin walls.
- Tooling commitment. Tooling at 25 to 40 days and production at 15 to 25 days means the geometry has to be frozen well before the first production part exists.
None of these are tuning problems. They are properties of injecting metal fast into steel, and the correct response is to design around them or choose a different process.
Our high pressure die casting page details the machine range, shot control and process monitoring we run, and the tooling page covers cavity count and tool construction. Send a 3D model through contact and the quotation will state the projected area, cavity pressure, machine tonnage, cavity count and cycle time behind the price, so the number can be checked rather than trusted.
