Last updated: 2026-09-02 · 11 min read

How Do You Build a Process Window That Survives Resin Lot Changes?

A process window that survives resin lot changes is built by deliberately pushing every process axis — melt temperature, mold temperature, hold pressure, hold time, injection speed and cooling time — to its failure boundary during development, measuring where defects start, and then setting production at the center of that measured window with part weight as the control variable. A process that merely “runs well on the trial day” has a window of unknown size; the first resin lot with a different melt flow index finds its edge, and the scrap rate finds it too.

Key Facts

  • A process window is the range of each process setting inside which the part passes every specification. It is measured by boundary-hunting during development — deliberately driving each axis toward flash on one side and short shots or sinks on the other — never by assuming that the values printed on a data sheet define it.
  • Resin lots vary by design, not by accident: melt flow index within a grade’s specification can swing enough to move the viscosity curve by a meaningful percentage. A window with thin margins on holding pressure or injection speed absorbs that swing silently; a marginal window turns the same swing into flash, short shots, or drifting dimensions.
  • The center-point discipline is the whole game: production setpoints should sit at the measured center of every axis, so that any lot-to-lot drift has equal distance to travel before it becomes a defect in either direction. Setpoints parked near one edge give away half the window before the first shot of a new lot.
  • Part weight is the most honest control metric on the floor: it responds to viscosity, hold pressure and check-ring behavior faster than visual inspection, and it can be trended shot to shot. A validated weight range, tied to the window study, gives the operator a number to watch when a new lot enters the hopper.
  • The window study is documentation, not folklore: edges, setpoints and the defect that marks each boundary go on the process sheet, so that a process engineer years later can see what was tested and what was never approached.

    • Gate vestige — The visible mark left on the part surface where the gate was trimmed. Edge gate: 0.5–2.0mm vestige. Pin-point gate: 0.2–0.5mm. Valve gate: 0.05–0.1mm (flush). Vestige size affects cosmetic appearance and may require secondary trimming. The related guide covers adjacent topics.

    What a Process Window Actually Is

    Every molding process lives in a space defined by its adjustable settings. Inside a certain region of that space, parts pass — dimensions in tolerance, cosmetic surfaces approved, mechanical properties met. Outside it, in every direction, a specific defect appears: raise melt temperature far enough and you meet splay or degradation; lower it far enough and you meet short shots or weld-line weakness. That passing region is the process window, and the distance from your setpoint to each boundary is your margin.

    The definition sounds trivial, but its consequence is not: margins only exist if someone measured where the boundaries are. Many programs never do. The mold trials, runs, produces parts that pass, and the process sheet freezes a set of numbers that looked good that week. Nobody knows how far the process can move before it fails — until a resin lot change, a new machine, or a seasonal shop-temperature shift moves it, and the answer arrives as scrap.

    Why Some Windows Collapse on the Next Lot

    Resin is a natural-product-adjacent material: polymerization batches differ, additive dispersions differ, and regrind ratios on the customer side differ. Grade specifications anticipate this — melt flow index ranges, moisture ceilings and lot certificates exist precisely because lot-to-lot variation is expected and accepted within limits.
    A resin that sits at the high end of its MFI range flows differently at the same barrel settings than a lot at the low end: lower viscosity means less shear heating, different gate freeze behavior, and a holding phase that packs differently through the same runners.

    A robust process shrugs this off. A fragile one does not, and the failure modes are predictable: windows set near the flash edge flash on the low-viscosity lot; windows set near the short-shot edge short on the high-viscosity one; windows balanced with barely enough holding time lose gate seal first on the hotter-running lot. In each case the process did not “go bad” — it was always sitting near the cliff, and the lot change pushed it one step closer.

    The Six Axes and Their Failure Boundaries

    Axis Boundary One (High) Boundary Two (Low)
    Melt temperature Splay, degradation, long cycle from extended cooling Short shots, weak weld lines, high injection pressure
    Mold temperature Sticking, cycle stretch, warp direction shifts Flow marks, high residual stress, dimensional drift
    Hold pressure Flash, over-packed brittle zones, ejection damage Sinks, voids, undersized dimensions
    Hold (screw) time Wasted cycle; weight plateau hides process drift Weight and dimension variation before gate seal
    Injection speed Burn marks, gas traps, shear splay Flow marks, hesitation, weld-line weakness
    Cooling time Pure cycle waste — a cost boundary, not a defect one Ejection distortion, post-mold warp, dimensional creep

    Two practical notes belong next to this table. First, some boundaries are cosmetic and some are dimensional — the study should record which specification closes each edge, because a window bounded only by appearance defects may still be wide enough for function but fail audits, and vice versa. Second, boundaries interact: hotter melt widens the hold-pressure window on one edge and narrows the cooling-time window on the other.
    A six-axis study run one axis at a time, corners untouched, understates the true shape of the space. The standard compromise is a one-factor boundary hunt on all six axes, then a deliberate corner check on the two or three axes the mold is most sensitive to.

    Building the Window Step by Step

    1

    Baseline and instrument first

    Run the current candidate process until stable, then record everything the machine can offer: actual melt temperature by probe, not barrel setting; cavity or nozzle pressure traces; screw position at hold switch-over; and shot weight over at least ten consecutive shots. The study’s conclusions will only be as good as this baseline, because every boundary you later find is expressed as a change from these signatures.

    2

    Hunt one boundary at a time

    Pick an axis, step it in fixed increments toward its failure mode, and record the step where the defect first appears — then step back inside and confirm recovery. Six axes, two directions, documented. The increments should be fine enough that the boundary is bracketed within one step: a window edge known only as “somewhere between 40 and 60 percent speed” is not yet a measurement.

    3

    Check the critical corners

    For the two or three axes with the thinnest margins, run the worst-case combinations: highest melt with lowest hold, coldest mold with fastest fill. Corners are where interacting limits hide, and a window that passes its corners is far more robust than one validated axis by axis.

    4

    Set the center and lock the weight band

    Compute the midpoint of every axis between its measured boundaries and move the production setpoints there. Then define the validated shot-weight band — the weights observed across the window study at passing conditions — and write it on the process sheet as the floor’s daily control number with an action rule: outside the band, hold production and investigate before adjusting.

    5

    Prove it with a stress lot, not just the study

    When a lot arrives with an MFI at the opposite end of the grade’s certificate from the development lot, run it against the window: the setpoints should not need to move. That one verification run is the most convincing evidence the window will survive production, and the comparison belongs in the qualification record.

    Gate vestige — The visible mark left on the part surface where the gate was trimmed. Edge gate: 0.5–2.0mm vestige. Pin-point gate: 0.2–0.5mm. Valve gate: 0.05–0.1mm (flush). Vestige size affects cosmetic appearance and may require secondary trimming. See also related guide.

    The Sensitivities That Catch Programs Out

  • Hold-pressure margin on thin-wall parts: gate freeze amplifies viscosity changes, and windows measured on one lot can halve on the next. The boundary hunt on this axis deserves the finest increments.
  • Check-ring health during the study: a worn check ring shifts hold efficiency silently, and every boundary measured with it is contaminated. Verify repeatability of screw recovery before trusting any numbers.
  • Machine-to-machine translation: a window validated on one press is a hypothesis on another of different tie-bar stiffness, shot capacity or screw geometry. The center-point setpoints travel; the boundaries must be re-measured or formally confirmed.
  • Moisture as a hidden axis: dried resin at the moisture ceiling versus at the floor behaves like two different lots. The study should specify the drying state it was run at, and production must hold it.
  • Regrind ratio on the customer side: a window validated at virgin resin narrows when 25 percent regrind enters the stream. State the ratio the window is valid for, or re-validate at the ratio that will actually run.
  • Colorant and additive lots: masterbatch carriers shift viscosity too. A natural-resin window is a starting point, not a certificate, for a colored program.

    • Documenting the Window So It Survives Handover

    A window that lives in one engineer’s head leaves with the engineer. The durable form is a one-page window record attached to the process sheet: each axis, its measured low and high boundaries, the defect that marks each boundary, the chosen center setpoint, and the validated weight band.
    A short section lists what was not tested — corners not reached, machines not tried, lots not stressed — because the honest record of unknowns is what lets the next engineer re-run exactly the right experiment when something drifts.

    The process sheet itself then carries two layers: the setpoints the operator runs, and the alarm limits the weight trend is judged against. Keeping them visibly distinct prevents the most common maintenance error — operators nudging setpoints back toward an edge to chase a weight number, quietly spending the margin the study paid for.

    Frequently Asked Questions

    How long does a proper window study take?

    For a B-level tool with six axes and corner checks, plan one to two press-days including documentation, more if the part has tight cosmetic or dimensional specifications that force fine steps. Spread against the life of a production mold and the scrap it prevents, it is among the least expensive qualification work on the program.

    Do we need a full DOE instead of a boundary hunt?

    A designed experiment earns its keep when interactions are expected to dominate — multi-cavity imbalance, reactive materials, tight flatness. For most parts, the boundary hunt plus targeted corner checks finds the same edges in a fraction of the shots. The discipline that matters is fixed increments, documented boundaries and recovery confirmation — not the statistical pedigree of the design.

    Which axis should be centered first if time is short?

    Hold pressure and hold time, in that order: they control part weight and dimensional stability and are the axes most sensitive to viscosity swings between lots. Melt and mold temperature margins usually come broader from the resin supplier’s recommended range; the holding phase is where thin windows hide.

    How does the window interact with scientific molding data?

    It is built on it. Cavity pressure curves, viscosity studies and gate-seal analyses are the instruments that make boundaries visible early and precisely; the window study then expresses those findings as machine setpoints with measured margins. A window without the underlying signatures is harder to defend when something drifts.

    What changes when the mold moves to another molder?

    The documented window is the transfer asset: the new shop starts from the same center points, verifies the weight band reproduces, and re-brackets only the axes that fail to reproduce. With the record, a transfer is a verification exercise; without it, every boundary is rediscovered in production, one scrap incident at a time.

    Should the window be re-validated periodically?

    After any meaningful change — press, screw, runner modification, resin grade change, or a persistent weight-trend shift — re-bracket the affected axes. Otherwise the record stands until the process proves it wrong. Annual re-testing of a stable process adds cost without information; trending the weight band does the monitoring for free.

    Key Takeaways

    1. A process window is measured, not assumed: six axes, two directions, boundaries recorded with the defect that marks each one.
    2. Resin lots vary inside their own specifications; center-point setpoints are what let the process absorb that variation without touching the edges.
    3. Part weight, validated across the window study, is the floor’s early-warning instrument when a new lot or a new day begins.
    4. Corners matter: interacting limits hide in combinations, and the thinnest axis deserves corner checks, not just single-axis brackets.
    5. The window record — boundaries, defects, setpoints, weight band, untested areas — is what makes the process transferable and auditable years later.

    aCatMould runs the six-axis window study as a standard part of trial qualification on production tools, and ships the window record with the process sheet — boundaries, center points, validated weight band and the untested areas listed honestly. When your first replacement lot arrives, the question it answers is not “will this run” but “watch the weight band, and expect nothing else to move.”

    Sources
    RJG — scientific molding process documentation practice
    Plastics Technology — process window development
    BASF — resin lot variability and processing guidance

    Sources: Plastics Technology · Plastics Technology Online · Society of Plastics Engineers (SPE) · ThomasNet