What Is Mold Flow Analysis and When Is It Worth It?

Last updated: 2026-09-01 · Reading time: about 10 min

Mold flow analysis is simulation: the melt’s journey through your part, computed before any steel is cut.
It predicts how the fill, the packing and the cooling will behave — and therefore where weld lines, sinks and warpage will appear — which moves those discoveries from the trial bench to the design desk, where they cost edits instead of rework. It is worth paying for when the part’s complexity, tolerances or material cost make the surprises expensive; it is overhead on a simple, forgiving geometry. This article tells the two cases apart.

Key facts

<>

  • A flow study simulates filling, packing, cooling and the resulting warp — a chain of physics, not a guarantee of parts; its outputs are predictions with confidence, ranked by which one matters to your drawing.
  • Its strongest predictions are comparative: where weld lines land, where gas traps form, where sinks will show — relative statements that DFM acts on immediately.
  • Its weakest predictions are absolute numbers: exact cycle time and guaranteed dimensions belong to T1 measurement, not to software. The trial stages remain the verdict.
  • The study pays when its findings change the design — gate locations, wall transitions, rib layout — which is why it belongs inside the DFM review, not after it.
  • Complexity, tolerance and material cost drive the break-even; a study that ends in “everything looks fine” was worth what you learned in confidence, and no more.

    • Cavity — The hollow space inside the mold that forms the part shape. Multi-cavity molds produce multiple parts per shot. Cavity count determines production output but also mold cost and complexity. Each cavity must be balanced for uniform fill. See also related guide.

    What does a flow study actually simulate?

    Mold flow analysis
    Computer simulation of the injection molding process — melt front advancement, pressure and packing, cooling, and the shrinkage that produces warp — run on the part geometry, resin data and gate layout before tooling is built.

    The simulation runs in stages, and each stage answers a different question. Fill: where does the melt go, where do flow fronts meet — the weld lines — and where does air get trapped? Packing: how does the applied pressure compensate shrinkage, and where will the sinks form on thick or ribbed sections? Cooling: how do temperatures distribute, how long until ejection, and where does uneven cooling set up stress?
    Warp: the sum of the shrinkage history, predicted as a shape change with direction and magnitude. The physics is real, the inputs are resin data and geometry, and the outputs are maps a tooling engineer can act on — gate moved, wall thickened, rib tied to a wall, vent added. What the study cannot do is hold tolerances for you: it ranks risks, it does not certify dimensions. That verdict stays with the measured evidence at trial.
    Understanding this division — predictions guide the design, measurements prove the tool — is the difference between using simulation and believing it.

    What does it predict well, poorly, and who verifies?

    Outcome Prediction quality Who verifies it
    Fill pattern and weld line locations Strong — the single most actionable output. Seen live at T0; confirmed in the T1 parts.
    Air traps and venting needs Strong — locations map directly to vent planning. Burn marks at trial confirm or refute.
    Sink risk on thick sections Good — comparative between design variants. Surface inspection on T1 parts.
    Warpage direction and shape Moderate — trends reliable, magnitudes approximate. Measured against the drawing at T1; the warp record closes it.
    Cycle time Approximate — a planning number, not a promise. The real cycle, timed in production.
    Clamp force estimate Good — projection area and pressure give a sound estimate. The press itself, at first shot.
    Dimensional guarantees Not a prediction at all. CMM measurement against the drawing — software never certifies dimensions.

    The last row exists because it is the most misused: buyers occasionally meet a proposal where simulation is presented as assurance.
    It is not assurance; it is a design instrument. Its honest role is to make the first steel cut smarter — to move the arguments about gates, walls and ribs into a format everyone can see, before they are expensive to change. The rows where it excels are exactly the rows where DFM argues best: comparative, visual, actionable.
    A study whose findings do not change the design was still worth running — sometimes the finding is that the design was already sound — but a study that changes nothing and explains nothing should make you ask how hard anyone looked.

    How do you turn a study into better steel?

    1Send real inputs. The 3D model, the candidate resin with its data, the gate candidates and the drawing’s tight features; a study on assumed inputs produces confident nonsense.
    The datasheet work from the material conversation feeds directly here.
    2Ask for the findings, not the pictures. Every colored fill plot needs a sentence: what it means for this drawing, which callout it threatens, what change it suggests. Reports that narrate are reports you can act on.
    3Route findings into DFM. Weld line across a cosmetic face becomes a gate discussion; sink behind a rib becomes a wall discussion.
    The study earns its cost when its findings change geometry — on purpose, before steel.
    4Re-run what changed. A revised design deserves a confirming pass on the changed regions; a study of the old design annotating the new one is a document, not a tool.
    5Keep the report beside the T1 record. Prediction against measurement is how both sides calibrate; the pairs of documents make the next program’s simulation sharper and the trial conversation shorter.

    When is the study clearly worth it?

  • Cosmetic surfaces with named classes — where weld lines and gate blush are appearance decisions, not tolerances to be sorted later.
  • Multi-cavity balance — each added cavity multiplies the fill questions; the study prices balance before the tool does.
    The cavitation decision and the study travel together.
  • Tight tolerances on warp-sensitive geometry — long spans, asymmetric sections; warp direction known in advance is a drawing saved.
  • Expensive or difficult resins — where fill uncertainty is paid for in every rejected shot; the material conversation wants simulation as its witness.
  • Thin-wall or long-flow parts — the regimes where fill behavior is least intuitive and the software earns its keep fastest.
  • A new supplier relationship — the study doubles as a communication instrument: it shows how the shop reasons about your part before the steel proves it.

    • When is it genuinely not worth it?

    The honest counter-case exists, and pretending otherwise sells studies instead of engineering.
    Simple, forgiving geometry — a thick, short-flow part in a known resin, non-cosmetic, generous tolerances — usually survives its first trial without needing software’s opinion first, and the trial itself is the more direct oracle. Mature designs being re-cut with known behavior, prototype tools with deliberately short missions, parts whose first article costs less than the study — all legitimate skips. The judgment is the same one that runs the whole program: what would the surprise cost?
    A study is worth its price when the answer is “more than the study.” Programs get into trouble in two symmetrical ways: buying simulation as insurance on trivial parts, and skipping it on parts whose failure modes were expensive and foreseeable. Neither error is about software; both are about matching instruments to stakes.
    Ask for the recommendation either way, in writing, with the reasoning — a shop that says “this one doesn’t need it, and here is why” is demonstrating the same judgment you are actually buying. The itemized quote that lists the study as a line is the one whose engineering you can audit.

    “Flow analysis doesn’t tell you the part will be good. It tells you where the part is planning to hurt you — while the fix is still a drawing edit instead of a re-cut.”
    — Arthur, Tooling Engineer at aCatMould

    Frequently asked questions

    Does everyone run flow analysis?

    No — mature shops run it selectively, where stakes justify instruments. The disciplined question is never “do you always?” but “when do you, and what changes because of it?” Shops that answer with criteria and examples are shops whose studies change designs; shops that answer with slogans run software as ceremony. Judge the answer the way you judge any scope line — by whether it names its mechanism.

    What inputs make or break the study?

    Resin data and gate candidates. A study on generic material values produces generic conclusions; the grade-specific data — rheology, thermal, shrinkage — is what makes the predictions belong to your part. Gate location candidates matter because the study’s most useful outputs are comparative: this gate versus that one. Provide both, and the study answers questions you actually asked; provide neither, and it answers questions nobody asked.

    Can simulation replace the trial?

    No — and any proposal that slides that way should end the conversation. Simulation predicts; the trial measures. The two are sequential instruments: the study makes the first steel smarter, the trial proves the steel. The stage system survives every advance in software for the same reason inspection survives every advance in machining — reality is the only auditor that signs.

    How long does a study take?

    Days for a single-gate part with clean inputs; longer for multi-gate, multi-cavity or iterative studies where designs change and re-runs follow. The schedule conversation matters: a study that arrives after the design freeze is a document, while one that arrives inside DFM is a decision instrument. Time it like any scope line — with its purpose named, ahead of the freeze it is meant to inform.

    What should a good study report contain?

    Inputs listed (model, resin, gate candidates), the four stages’ findings, each finding tied to a drawing feature, and recommendations with their reasoning. Pictures alone are not findings. The best reports read like engineering correspondence: here is what the melt will do, here is what it threatens, here is what we would change. The report you can argue with is the report worth paying for.

    How does the cost of a study compare with the rework it prevents?

    The comparison is structural, not numeric: a study is priced once, while the rework it prevents is priced in steel hours, trial days and program delay — categories that grow with the part’s complexity and schedule pressure. The break-even sits where those categories begin to outweigh the study’s line, which is why high-cavitation, tight-tolerance and cosmetic work clears it so easily and simple parts rarely do.
    Run the comparison with the quotes you actually hold; the answer is specific to the program, and the shop that helps you run it honestly is demonstrating the judgment you are buying either way.

    Who owns the study — buyer or supplier?

    Either can commission it; what matters is that both see it. A buyer-side study is an independent second opinion; a supplier-side study shows how your molder reasons. The failure mode is ownership as secrecy — findings that never the DFM table inform nobody. Wherever it runs, the report belongs in the program record, next to the tool’s life documents, where the next program can learn from it.

    Key takeaways

    1. Flow analysis simulates fill, pack, cooling and warp — predictions that move discoveries from the trial bench to the design desk.
    2. Its strongest outputs are comparative — weld lines, gas traps, sink risk — exactly the findings DFM acts on.
    3. It never certifies dimensions; T1 measurement remains the verdict, simulation the guide.
    4. Worth it where complexity, tolerance or material cost make surprises expensive; overhead on forgiving geometry.
    5. Real inputs, narrated findings, routed into DFM, kept beside the T1 record — that is how a study pays.

    If your part carries cosmetic surfaces, tight callouts or an expensive resin, send us the 3D and the drawing. You will get the study recommendation with its reasoning — or the honest case for skipping it — from a named tooling engineer, before any steel is ordered.

    Sources
    Autodesk Moldflow — reference for mold flow simulation capabilities and methodology
    ISO 20457:2018 — Plastics moulded parts: Tolerances and acceptance conditions, International Organization for Standardization
    Plastics Industry Association — industry resources on molding simulation practice
    MatWeb — material property database, reference for resin input data in simulation

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