Preform Weight Tolerance: Control Limits and SPC

Preform weight is the cheapest measurement in a PET plant and the fastest indicator of process drift. It responds to screw recovery, cushion stability, non-return valve condition, hot runner temperature, mould cooling and back pressure all at once, and it can be measured in seconds by anyone on the shift. Yet in most plants the weight is checked when a customer complains and not before — which turns the single most useful early-warning signal into a postmortem.

The cost of that habit is measurable in three places. First, giveaway: a process running at the top of its band because nobody knows where the centre is spends resin on every preform, all shift, all year. Second, escapes: a gradual drift that would have been caught as a trend on a control chart instead becomes a batch of lightweight preforms that blow into thin-walled bottles and fail a top-load or drop test further downstream. Third, arguments: without a control chart, every quality discussion starts from a single sample and an opinion.

Sailwin has built PET injection moulding machines for 15+ years, with 500+ machines delivered into 60+ countries, CE marking, ISO 9001:2015 manufacturing and a 2-year whole-machine warranty. The SW-P series covers 14 models from 170 kN to 5500 kN and supports valve-gate hot runners up to 64 cavities, with dedicated PET screws, far-infrared nano heating coils, PID temperature control to about ±1 °C and continuous logging of 40+ process parameters. That logging only pays off if somebody is plotting the weight. This article sets out how to derive the tolerance, how to run SPC on it without overreacting, and what each drift pattern tells you about the machine.

1. Why Preform Weight Is the Fastest Signal You Have

Weight is attractive as a control variable for an unglamorous reason: it is a total mass measurement, so it integrates everything that happened to the shot. It does not tell you which variable moved, but it tells you that something moved, immediately and cheaply. Other measurements — wall thickness distribution, crystallinity, acetaldehyde, dimensional checks on the neck — are slower, need more equipment, and in the case of AA testing need a laboratory.

In PET injection moulding the weight signal is unusually clean because each cavity produces one discrete part with a short cycle. The variables that most commonly move it are well known: shot size and cushion stability, the condition of the non-return valve on the screw tip, hot runner tip temperature and its stability, mould and core cooling temperature, back pressure and screw speed, and the proportion and consistency of regrind. A stable process holds weight inside a narrow band for a whole shift; anything that changes one of those inputs shows up as a shift in the mean or a change in the spread.

There is a second, less obvious benefit. Weight is measurable on the finished article without destroying it, which means the same sample can be weighed, dimensionally checked and then blown. A plant that weighs every cavity once per hour, records it, and plots it is generating a quality record that costs almost nothing and covers the entire production of the tool. At 64 cavities, that record is the only practical way to notice that one cavity is drifting while the other sixty-three are perfect.

2. Setting the Preform Weight Tolerance: Working Backwards From the Bottle

A preform weight tolerance quoted on its own has no meaning. It has to be derived from what the customer was promised. The chain below shows how that derivation normally runs; the percentage figures are industry typical values for a well-controlled PET process rather than fixed rules, and they tighten as bottles are lightweighted.

One practical warning about the last row. If the sampling plan weighs five preforms from the same cavity, the chart will be beautifully flat and completely blind to the problem that actually matters at high cavitation. At 64 cavities a complete round takes longer, but it is the only sample that can expose a single drifting tip. Methods that weigh the whole shot in one go give the mean and nothing else, which is why a robot take-out with an inline weighing station is worth considering on high-cavity tools — the EUROMAP 67 interface on Sailwin SW-P machines is designed to make that kind of peripheral integration straightforward.

3. Running SPC on Preform Weight Without Overreacting

A preform weight control chart has two jobs: to separate normal variation from a real change, and to do it fast enough that the change can still be corrected cheaply. The mechanics are standard. Collect a subgroup of consecutive preforms at a fixed interval, record the subgroup mean and the subgroup range on an X-bar and R chart, and calculate the control limits from the process itself once it is running steadily. Recalculate them when a genuine change is made — a new tool, a new screw, a different resin batch — and not when a point looks inconvenient.

The most common mistake is charting against the specification band instead of against the process. The specification band is what the customer will accept; the control limits are what the machine can do today. Overlaying them on the same chart encourages operators to adjust the machine whenever a point approaches the specification limit, and each adjustment adds variation instead of removing it. A process that is comfortably capable but running slightly off centre is not a problem to be corrected every hour; it is a fact to be recorded and reconsidered if it becomes a trend.

The second mistake is ignoring the range chart. A process can hold its mean perfectly while its spread widens, and a widening spread is an earlier and more serious warning than a drifting centre, because it usually means a mechanical change: a non-return valve that is starting to leak, a hot runner tip with a developing blockage, or inconsistent regrind. Reading only the mean chart hides the signal until the spread is wide enough to breach a limit on the mean as well.

A weight chart is not a report to be filed. It is a work instruction. If the person who can change the hot runner temperature never sees the trend until the end of the shift, the chart has told nobody anything in time to act.

4. Reading the Chart: What Each Preform Weight Drift Pattern Tells You

The value of plotting weight continuously is that drift has a shape, and the shape narrows the search. The table below maps the patterns most commonly seen on PET preform tools to the subsystems that produce them. It is a diagnostic starting point, not a substitute for measuring the machine parameters — Sailwin machines log 40+ of them in real time, so the log and the chart should always be read together.

Whichever pattern appears, the second question is always capability. A process whose normal variation nearly fills the specification band will keep generating alarms no matter how well it is charted, because it has no room to absorb normal variation. That is a machine and tool setup problem — hot runner balance, cooling design, screw and non-return valve condition — and it is the point at which a conversation with the machine supplier becomes more productive than another round of operator adjustments.

5. Case Study: A Single Cavity Drifting Inside a Compliant Tool

A producer running a 48-cavity preform tool passed every weight check for months, because the procedure weighed five preforms taken from the same position on the take-out plate. Field complaints about neck distortion started arriving from a single filling site, and the tool still tested inside specification on the bench.

Scenario based on a Sailwin customer project; site-specific figures available on request during engineering review.

Frequently Asked Questions