Walk into most moulding shops that have been running for more than a decade and you will find the same quiet arrangement: a handful of veteran setters who know which machine likes which numbers, which tool needs a nudge on hold pressure first thing in the morning, and which cavity always runs a touch short. That knowledge is real, hard-won and usually correct. It is also invisible to anyone auditing the process, and it walks out of the building every evening.
Across two decades in automotive and aerospace quality, I have sat through customer audits where the process documentation said one thing, the machine screen said another, and the setter's head contained the actual truth of how parts were made. The gap between those three sources is where scrap lives. When the good setter takes holiday, the night shift inherits a process that only works if you know the folklore.
Scientific moulding exists, in large part, to convert that tribal memory into measured variables anyone can reproduce. The point is not to disparage experienced setters. Their intuition is grounded in physics they were never taught formally — they have observed the consequences of shear thinning, gate freeze-off and thermal drift thousands of times. What the methodology offers is the language and instrumentation to explain what they already sense, and to test whether their beliefs still hold after a tool repair, a resin lot change or a move to a different press.
What Cavity Pressure Actually Tells You
Cavity pressure is the most information-rich variable in injection moulding, and it is the one most shops never measure directly. Melt pressure at the nozzle tells you what the machine is doing; cavity pressure tells you what the plastic is doing inside the tool, which is the only thing the part cares about. A transducer behind a pin, or a flush-mounted strain gauge at the end of fill, gives you the actual story.
The typical cavity pressure trace rises steeply during volumetric filling, peaks around switchover to hold, then decays as the gate freezes and the melt contracts. Part weight, dimensions, sink, voids and warp all correlate with the area under that curve far better than with any machine setting. Two identical machines running identical set-points can produce different parts because screw wear, check-ring leakage and heater band behaviour differ. Cavity pressure erases that ambiguity.
Where this pays for itself is fault detection. A stuck check ring shows up as a shifted fill peak within a shot or two. A blocked gate on a multi-cavity tool appears as a collapsed trace on one cavity. Material lot variation shows as a change in the pressure rise slope during filling — effectively a viscosity shift in real time. Setting alarm limits on peak pressure and the integral of the curve turns the mould into its own inspection device, shot after shot.
The Viscosity Curve: Selecting an Inherently Robust Set-Point
The viscosity curve, sometimes called the rheology curve, is the first study to run when a tool comes to a new press. The method: choose a melt temperature in the middle of the resin's range, set switchover by position at around ninety-five percent of fill, then run a series of shots while increasing injection velocity step by step and recording peak injection pressure at each step.
Plot relative viscosity — peak pressure multiplied by velocity, a proxy for shear stress over shear rate — against velocity on a log-log scale, and you get a curve that falls steeply and then flattens. The flat region is where you want to run. There, small variations in injection speed from machine drift or hydraulic noise produce almost no change in melt viscosity, so the process is inherently robust. On the steep part of the curve, the same drift changes viscosity dramatically and dimensions wander with it.
Most setters pick injection speed by feel, on whatever machine they learned on, and nobody questions it. I have seen tools moved between presses where the original speed sat on the steep part of the curve, and every lot change produced a nervous week of tweaking. Repeating the study on the new machine and selecting a velocity in the flat region usually settles the process without touching anything else.
Two caveats. The curve is valid only for that melt temperature, that resin and that tool geometry — redo it after any significant change. And velocity alone is not the whole answer: a cavity pressure balance study on multi-cavity tools follows, because even a well-chosen ram speed cannot rescue a runner system that fills cavities unevenly.
Viscosity curve study, step by step
- 01Fix melt temperatureMid-range of the resin specification; everything downstream is conditional on this.
- 02Set position switchoverAround 95% of fill, so packing does not contaminate the fill measurement.
- 03Step injection velocityIncrease speed in increments, recording peak injection pressure at each step.
- 04Plot log-log curvePeak pressure × velocity against velocity; the curve falls, then flattens.
- 05Select flat-region velocityMachine drift there produces almost no viscosity change — inherent robustness.
Mould Qualification That Means Something
Qualifying a mould by running a few hundred shots, measuring them and declaring victory tells you almost nothing about robustness. A proper qualification deliberately challenges the process. After the viscosity curve, the standard approach is a pressure-temperature study: run a matrix of hold pressures and melt temperatures, mould a sample at each corner and centre point, measure the parts, and find the set-point that sits in the middle of the process window, not merely inside it.
The distinction matters. A set-point at the edge of the window survives only until the first heater band ages or the first resin lot shifts. A set-point at the centre absorbs normal variation without producing out-of-tolerance parts. Parts from the window corners should be measured and documented so you know the direction each dimension moves with each variable — that knowledge lets you diagnose a drift in seconds instead of evenings.
Gate seal studies belong here too. Increase hold time shot by shot, weigh the parts, and find the point where weight stops increasing — the gate is frozen. Set hold time above that point with margin. Parts weighed just below and just above seal differ visibly in sink and dimensional stability, and running without gate seal means hold time variation translates directly into part variation.
Documentation from these studies is the antidote to setter folklore. When the process engineer leaves, when the tool returns from refurbishment, when a customer asks for evidence the process is capable — the viscosity curve, the window study and the gate seal data answer questions that memory cannot. Not replacing judgement, but recording the basis for it.
A set-point at the edge of the window survives only until the first heater band ages or the first resin lot shifts.
Cavity-to-Cavity Variation: Where Folklore Fails Hardest
Multi-cavity tools are where folklore fails hardest. Setters talk about "the number three cavity" running heavy, and everyone adjusts the process around it — usually by adding hold pressure until the worst cavity is acceptable, which overpacks the best one and creates flash or dimensional trouble at the other end. The tool is unbalanced, and the process is being used as a crutch.
The right method is a cavity balance study: short-shot each cavity progressively, filling perhaps five percent at a time, and weigh or measure each cavity's contribution at every stage. This reveals whether the imbalance exists from the start of fill — a runner or gating problem — or develops during packing, which points to gate size, gate land length or thermal differences between cavity halves. Flow group analysis, filling cavities in groups by runner hierarchy, separates runner imbalance from cavity-specific issues.
Once you know the cause, the fixes are physical: runner dimension adjustments, gate resizing, or deliberately restrictive flow elements to balance naturally unbalanced layouts. What you must not do is paper over it with process settings, because the balance you impose is valid for one viscosity only — one melt temperature, one resin lot, one shear history. Change any of those, which you will, and the imbalance returns in a pattern the setter does not recognise.
Cavity pressure sensors make this continuous. A transducer in each cavity, or at least in the worst and best cavity, lets you watch the balance drift in real time and alarm on it. Without sensors, your only cavity-level signal is part measurement downstream, which arrives late and only if you measure cavities individually — which many shops, measuring a mixed sample, quietly do not.
Process-setting compensation versus physical correction
What folklore does
- Raise hold pressure until the worst cavity passes
- Overpack the best cavity into flash and dimensional trouble
- Re-tune after every resin lot or temperature change
- Balance is invisible until downstream measurement complains
What the studies do
- Short-shot each cavity in 5% increments and weigh contributions
- Separate runner imbalance from gate and thermal effects
- Fix the tool: runner sizing, gate resizing, flow restrictors
- Monitor worst and best cavity with pressure sensors and alarms
Making the Transition Without Losing the Knowledge
Resentment kills these projects. A setter who has spent twenty years being the person everyone calls when the tool misbehaves will not hand that status to a spreadsheet, especially if it is presented as a correction of his ignorance. Frame it the opposite way: the studies formalise what he already knows, and the sensors give him better tools than his ears and his fingertips ever did.
Start with the problem tools — the ones that require the most folklore to keep running. They give the fastest, most visible payback, and the setters involved become advocates once they see cavity pressure traces explaining a fault they had been chasing by instinct. Bring the best setters into the studies explicitly; their observations about which parameters matter, gathered over years, often direct the instrumentation to the right place on the first attempt.
Then write it down. Process documentation that captures the viscosity curve, the window study, the cavity balance data and the alarm limits is the control plan the setter's memory was silently standing in for — better than nothing, but undocumented, unversioned and unrecoverable once its holder retires. In plants where I have watched that person leave without a replacement system, the shop paid for it in scrap, in customer complaints and in months of rediscovery. The instruments and studies of scientific moulding are not expensive compared with that.