Scientific Molding: Build a Robust Process Window

Most injection molding processes are set by a molder who adjusts until the parts look acceptable. That process works until the machine changes, the resin lot changes, or the tool moves to another plant — then it has to be rebuilt from scratch, and nobody knows which of the forty settings actually mattered.

Figure 1. Scientific Molding: Build a Robust Process Window.

Scientific molding replaces that with a small set of measurements that define a process in terms of physics rather than dial positions. The output is a process window: a documented range of conditions inside which the part is good, wide enough to absorb normal variation.

The four variables

Almost every press setting maps back to four physical variables. Control these four and the rest follows:

Process technicians deal with thirty settings; a scientific molder deals with four numbers and knows the acceptable range for each.

Step 1 — The viscosity curve (melt temperature and flow rate)

The viscosity curve answers: how does the pressure needed to fill change with injection speed?

Procedure:

1. Set the machine to fill 95–98% of the part by weight, on velocity control, with no hold pressure.
2. Run a series of shots at increasing injection speeds, from very slow to the machine's maximum or until the pressure limit is reached.
3. Record the peak injection pressure required for each speed (corrected to a common melt temperature).
4. Plot pressure against fill time.

Two features appear on the curve:

• A steep region at high speed — shear thinning. Above a certain speed the required pressure rises sharply; the machine is at its limit and the process is unstable.
• A steep region at low speed — the melt is cooling as it fills, so viscosity rises with fill time.

The usable window is the flat region between them. Operating in the middle of the flat region makes the process insensitive to small speed variations, which is exactly what you want in production.

Step 2 — Pressure drop study

The pressure drop study separates the contribution of the machine, the nozzle, the runner and the cavity:

1. Inject into the air (or into a short-shot) to establish the peak pressure at the nozzle.
2. Transfer to hold and read the pressure at the nozzle, then at the sprue, then at the cavity.
3. Repeat at several flow rates.

This produces the data that tells you whether a pressure problem is the machine, the hot runner, the gate, or the part. It is also how you establish the pressure limit of the tool honestly: if 90% of the available pressure is used just to reach the gate, the process has no margin.

Step 3 — Gate seal study

The gate seal study answers: how long does hold pressure need to be applied for the gate to freeze off and the part to stop gaining weight?

Procedure:

1. Establish a stable fill and a reasonable hold pressure.
2. Run a series of shots with hold times increasing in 0.5 s steps.
3. Weigh the parts, including the runner (or the part with the gate attached).
4. Plot part weight against hold time.

Weight rises with hold time and then plateaus. The knee of that curve is the gate seal time. Production hold time should be set at or slightly above the seal time — never below, because a part that has not sealed continues to shrink unpredictably, and never far above, because the extra hold time adds cycle time for no benefit.

Step 4 — Cooling time and mold temperature

Cooling is typically 50–70% of the cycle, so it deserves the same rigor:

• Mold temperature should be set from the resin supplier's range and controlled to ±2 °C if the part has cosmetic or dimensional requirements. Both halves matter: a 15 °C difference between cavity and core sides bends the part.
• Cooling time should be determined by part ejection temperature, not by habit. Measure the part temperature at ejection and reduce cooling time until the part is just stable enough to eject without distortion, then add a safety margin.
• Wall thickness drives cooling time with a squared relationship: a 2.0 mm wall needs roughly 1.6 times the cooling time of a 1.6 mm wall.

Step 5 — Build the window, not a point

A single "good" set of numbers is not a process; it is a point. Production needs a box:

1. Pick a quality criterion that is measurable — part weight, a critical dimension, flatness, weld line visibility.
2. Run a designed experiment (DOE) varying the four variables across a range, typically a fractional factorial or a response surface design.
3. Identify which variables actually affect the criterion, and the range in which the criterion stays inside spec.
4. Define the production window as the region where the part passes, with margin on all sides.

A common and useful result: part weight and critical dimensions depend strongly on hold pressure, melt temperature and fill speed, and only weakly on hold time — provided hold time is above the gate seal time. That single finding typically removes half the variability the molder was fighting.

Decoupled molding: control what actually matters

Standard machine control is a cascade: velocity control until transfer, then pressure control. The weakness is that the switchover point moves, so the amount of material in the cavity at transfer varies shot to shot.

Decoupled molding separates the process into three independent phases:

1. Fill — velocity controlled, until the cavity is roughly 95–98% full.
2. Pack — a small, controlled amount of material pushed in at high pressure for a short time.
3. Hold — lower pressure to compensate shrinkage until the gate seals.

Because the fill phase is fixed, the cavity pressure at transfer becomes repeatable, and the part becomes repeatable. The practical way to implement decoupled molding is with a cavity pressure sensor and a short-shot study to find the transfer position; without a sensor, the position transfer can be set from the short-shot weight curve, which is less precise but still better than a hydraulic timer.

Machine independence

The real test of a scientific process is portability. A process defined as "injection speed 45%, hold pressure 380 bar, screen 3" is not portable — the same numbers on a different machine, with a different screw, will not give the same melt. A process defined as:

• Melt temperature 235 °C
• Fill time 0.85 s to a 96% short shot
• Pack at 55 MPa for 0.4 s
• Hold at 30 MPa until gate seal at 4.5 s
• Mold temperature 60 °C / 60 °C, cooling 12 s

...is portable, because each number describes the plastic rather than the machine. Documenting the process this way, with the viscosity curve, the gate seal study and the pressure drop study attached, is what lets a tool move between machines or plants without a week of re-qualification.

Minimum documentation for a scientific process

• Viscosity curve with the selected operating point marked
• Pressure drop study at the production flow rate
• Gate seal curve with the production hold time marked
• Cooling time basis and measured ejection temperature
• Mold temperature on both halves, with tolerance
• Short-shot series documenting the transfer position
• Process window chart: the four variables, nominal values and limits
• Part weight and critical dimension at the nominal point, with capability figures
• Reaction plan: what to check first when a part goes out of spec

Common pitfalls

Scientific molding is not a training course or a software package. It is the discipline of measuring four variables, writing down the range in which the part is good, and then running the press inside that range. The measurement effort is a few shifts per tool; the payoff is a process that survives resin lots, machine changes and shift handovers.

Figure 2. Quick reference: the key checks and reference values from this guide.