By PolyTakt Team · 29/7/2026
Decoupled moulding is often reduced to “fill to 95% and transfer.” The percentage is real training guidance — but schools interpret fill-only differently, and first stage is often a profile that must preserve the fill time from the speed study. This article explains why the methodology is misunderstood when it is treated as a recipe instead of a way of separating and understanding process variables.
Few concepts in scientific injection moulding are discussed more frequently than decoupled moulding.
And few are misunderstood more often.
For many processors, decoupled moulding has become reduced to:
But decoupled moulding is not a recipe.
It is a methodology intended to separate process variables so they can be understood and controlled independently.
When applied correctly, it creates stability.
When misunderstood, it simply creates another collection of machine settings copied from one process to another without understanding.
The fundamental principle is simple:
Different stages of the moulding process should control different outcomes.
For example:
By separating these stages, the process becomes more predictable and easier to optimise scientifically.
The methodology itself is sound.
The problem is how often it becomes oversimplified.
Most scientific moulding training courses teach a first-stage fill of approximately 95% to 98% before transfer.
That guidance is real. RJG and related Decoupled Moulding teaching have used it for decades. It is not superstition.
The misunderstanding is treating the percentage as a finished recipe — and assuming everyone means the same thing by it.
The purpose of a fill-only first stage is to keep filling under velocity control, then hand the process to pack and hold before the cavity is packed out under first-stage force.
Transfer too late, and fill begins to pack the part. Flash risk rises. The stages are coupled again.
Transfer too early, and pack must finish the fill as well as densify the part. Weight and dimensions become hold-sensitive in unpredictable ways.
The percentage is a taught starting convention for finding that handoff. It is not a universal law.
Even when courses quote the same numbers, interpretation differs.
Some approaches — including the RJG-style volumetric fill-only method many processors use in practice — aim for a part that is volumetrically full at transfer but not packed: visually complete, with sinks and voids still present, then finished under pack and hold.
Others treat first stage as a visual fill that remains short, with pack completing the last of the cavity volume as well as densifying the part.
Both camps are trying to protect the same principle: do not pack under first-stage velocity and available pressure. They disagree on what the fill-only part should look like when you get there.
Geometry, wall thickness, venting, and multi-cavity balance all influence which interpretation works. Thin-wall parts may need a fuller first stage because the flow front will not restart once it freezes. Thick sections can mislead if judged only by a visual short. There is no single appearance rule that fits every tool.
Initial development is usually done at a single injection speed.
That is deliberate. A constant speed lets you establish:
Once those are known, a velocity profile is often introduced — typically to decelerate into end of fill and avoid slamming the cavity, and sometimes to slow through the gate to reduce blush.
The critical constraint is that the resulting fill time matches the fill time established in the single-speed study.
If the end of fill is slowed, earlier segments must be increased to compensate. What you are protecting is the average volumetric fill rate — and therefore the rheology characterised during development — not the shape of the speed graph.
Alter the profile and change the fill time, and you are no longer running the fill you studied. You are running a different process dressed up as the same setup.
Velocity control requires available injection pressure above what the fill actually uses. Without that headroom, the machine becomes pressure-limited and fill rate drifts with viscosity.
In practice, many materials and geometries run first-stage peak pressure somewhere around half to two-thirds of available capacity. Others need a much harder fill — sometimes leaving little reserve, and occasionally demanding far more pressure than a generic “rule of thumb” assumes — simply to get enough plastic into the cavity before pack.
Material, wall thickness, flow length, gate restriction, and melt temperature all decide what is required.
There is no one percentage that fits all.
Blindly applying fixed fill percentages, single speeds, or pressure rules without understanding the process objective defeats the purpose of scientific moulding entirely.
The number itself is not the science.
Understanding why the number matters is.
Decoupled moulding assumes the machine can:
But not all machines behave equally.
A process developed scientifically on one machine may become unstable on another if:
The methodology has not failed.
The assumptions behind the process have changed.
Even the best decoupled process cannot overcome:
Scientific moulding is often treated as though processing alone can solve instability.
It cannot.
The process remains constrained by the physical realities of the tool and machine.
One of the most damaging habits in moulding is the belief that a process can simply be copied:
Without understanding the interaction between:
This transforms scientific moulding into little more than technical superstition.
The process may run.
But the understanding is missing.
The true purpose of decoupled moulding is not to create standardised settings.
It is to:
This requires engineering discipline, not just machine knowledge.
Decoupled moulding is not misunderstood because the methodology is complicated.
It is misunderstood because many organisations seek shortcuts instead of understanding.
Scientific moulding cannot be reduced to a handful of percentages and machine settings.
95% to 98% is useful teaching. Single-speed development is useful discipline. Velocity profiles are useful tools.
None of them replace understanding fill state, fill time, machine response, and why the process behaves the way it does.
Without that understanding, even the most advanced methodology eventually becomes reactive processing with more technical language attached to it.
