Billet Temperature Variance Is Upstream of More Defects Than You Think

Why Billet Temperature Rarely Gets Measured - and What That Costs

In aluminum die casting operations, the shot sleeve receives molten metal poured from a ladle or automated dosing system. The billet (the slug of metal in the sleeve before the shot) has a temperature that varies with pour temperature, ladle-to-sleeve transfer time, sleeve temperature, and dwell time between pour and shot. In theory, pour temperature is controlled and constant. In practice, it varies by 15-40 degrees Celsius shot to shot in many operations, and nobody is measuring the billet directly.

Temperature control attention goes to the furnace setpoint, which is controlled. The furnace is at 680C, the alarm trips if it goes below 675C, and the process log shows furnace temperature within spec for the entire shift. What is not logged is the actual metal temperature at the moment of injection, which is a function of furnace temperature minus all the heat lost between furnace and die cavity.

Billet temperature at injection varies with ladle preheat temperature, ladle fill time, transfer distance, pour height (which affects air cooling of the stream), sleeve temperature, and delay between ladle completion and shot activation. Each of these introduces variance. A 20-degree drop in injection temperature for an aluminum alloy like A380 reduces fluidity measurably - the metal's ability to fill thin sections and long flow paths before the solidification front advances.

The Cold Shut and Misrun Connection

Cold shuts occur when two metal flow fronts meet in the die cavity after one or both have partially solidified. The solidification skin on each front prevents proper fusion, leaving a seam that looks like a crack but is actually an unfused interface. Misruns occur when metal solidifies before the cavity is completely filled, leaving a section of the part missing or incompletely formed.

The standard diagnostic pathway for cold shut and misrun is to check fill time, gate velocity, venting adequacy, and die temperature. These are the right variables to examine. But they are not the complete variable set, and billet temperature is often the one that explains the variance the standard variables leave unexplained.

Consider a production run where shot parameters are held constant - gate velocity at spec, fill time at spec, die temperature at spec. If billet temperature variance is 25 degrees shot to shot, you will see random cold shut events that correlate poorly with any logged process variable. The engineering team adjusts gate velocity up by 3% based on a best guess, the cold shut rate drops for two hours, then resumes. The real variable - billet temperature - was never measured and therefore never appears in the analysis.

How the Pattern Shows Up in Process Correlation Data

When ForgePuls deploys on an HPDC line, the process correlation engine ingests shot parameters from the die casting machine OPC-UA server and links them to inspection results. In lines where billet temperature is not directly measured, we see a characteristic pattern: cold shut defect clusters that correlate with pour timing intervals longer than nominal but that show no corresponding anomaly in gate velocity or fill time.

The signature is a delayed shot - a pour was completed, the shot activation was delayed by 8-12 seconds beyond normal dwell time, possibly because of a robot interrupt, a part handling pause, or a manual intervention on the die. The machine records normal shot parameters because the shot itself was executed correctly. The defect occurs because the metal was cooler at injection. The process log shows nothing unusual. The machine OEE calculation is not affected. The defect looks random.

Identifying this pattern requires correlating inspection outcomes against cycle time - specifically, the interval between pour completion and shot activation. This timing signal is available in most die casting machine OPC-UA servers as part of the cycle data, but it is not a parameter most foundries have included in their SPC monitoring. When we add it, the correlation with cold shut clustering becomes evident within a few hundred shots.

The Practical Fix Hierarchy

Once billet temperature variance is identified as a contributing factor, the response options range from measurement-only to process change:

The lowest-intervention approach is to add dwell time monitoring to the SPC system and set an alert when shot delay exceeds a threshold - say, 10 seconds beyond nominal. The alert triggers a review of that shot's inspection result. If the delay correlated with a defect, the part gets flagged for additional inspection before shipment. This costs nothing in process change and provides immediate data integrity improvement.

The next level is billet temperature measurement. A non-contact IR pyrometer aimed at the shot sleeve after pour can measure the metal surface temperature before shot activation. This adds a direct variable rather than a proxy (cycle timing). Several die casting machine vendors offer this as an option on newer machines; on legacy equipment, a standalone pyrometer with OPC-UA or Modbus output can be integrated into the process data stream.

The highest-intervention option is direct temperature control: heated shot sleeves that maintain metal temperature within a tighter band during dwell, or automated pour systems that control ladle temperature and transfer time more precisely. This is the right answer for high-volume, safety-critical applications where cold shut has strict zero-tolerance acceptance criteria, but it requires capital investment and process qualification time.

The Cpk Implications

From a quality system perspective, billet temperature variance that is unmeasured and uncontrolled is a source of process capability degradation that does not appear in your Cpk calculations. If your Cpk is calculated on gate velocity and fill time, and those parameters are controlled, your quality statistics look better than they are. The cold shut events appear as unexplained noise in your defect rate, reducing first-pass yield below what the controlled parameters alone would predict.

Under IATF 16949 requirements for production part approval and ongoing process monitoring, automotive customers increasingly ask about the complete set of controlled process parameters and their Cpk values. "We control shot parameters" is true and insufficient if the thermal state of the metal at injection is uncontrolled and contributing to rejects. Regulators and customers under customer-specific requirements (CSRs) from OEMs are becoming more specific about what constitutes adequate process characterization.

Adding billet thermal state to your process control plan - even as a monitored rather than controlled parameter - demonstrates process understanding and provides data for root cause analysis when cold shut events occur.

What to Measure and What to Ignore

Not every thermal variable is worth adding to process monitoring. The goal is to capture variance that actually correlates with defect outcomes - not to build an ever-larger process database that no one analyzes. From our process correlation data, the variables with the strongest evidence for cold shut and misrun correlation are: shot dwell time (interval from pour completion to shot activation), first shot after die open more than 3 minutes (cold die cycle), and night shift pour sequence intervals (typically longer due to staffing pace).

Variables that are worth measuring but show weaker direct correlation: furnace temperature (too buffered from injection to show strong shot-to-shot correlation), ladle preheat temperature (variable but less impactful than dwell time).

The highest-leverage addition for most HPDC operations that are not currently monitoring thermal state is shot dwell time. It is available in machine data, requires no new sensors, and correlates with cold shut in the pattern we described above. Start there before adding sensors.

For further reading on how process variables are correlated in the ForgePuls inspection data model, see our article on OPC-UA integration in foundry environments.