What Happens at T0, T1 and T2 in Mold Trials?

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

Trial stages exist so that a tool proves itself in steps, not in a single hopeful shot.
T0 is the first look — the tool on the machine, filling checked, defects seen while they are still geometry, fixable with edits rather than programs; T1 is the formal trial the schedule promised, producing parts measured against the drawing; T2 is the confirming trial after corrections, the one that converts findings into an approved tool. Knowing what each stage owes you turns the trial calendar from jargon into an acceptance plan.

Key facts

<>

  • T0, T1 and T2 answer different questions: does it fill, does it meet the drawing, and does it repeat — in that order.
  • T1 is the contractual center of gravity: its full-dimensional report and evidence package are the deliverables the acceptance criteria point at.
  • Trial count is a scope item like steel or cooling — quotes that name how many trials are included and who pays for corrections are quotes that finish.
  • Defects found at T0 are geometry edits; the same defects found after delivery are change orders — the stage system exists to keep them on the geometry end of that curve.
  • The trial stages slot directly into the lead-time threads; programs that schedule the loop explicitly ship on calmer dates.

    • Process parameter — A controllable variable in the injection molding process: melt temperature, injection speed, holding pressure, cooling time, mold temperature. Each parameter affects part quality — deviations cause defects. Parameters are set during T1 and locked for production. See also related guide.

    What actually happens at each stage?

    T1 trial
    The first formal molding trial of a completed tool, run to a written setup, producing parts that are measured against the drawing — the trial whose dimensional report and evidence package form the basis of tool acceptance.

    T0 is an engineering look: the tool mounts, the process runs short, and everyone watches the melt fill, pack and eject for the first time. Its output is a list — short shots, flash, warp, gate quality — and its value is timing: findings here are still steel-and-hours, not program crises.
    T1 is the formal event: a documented setup, a defined sample count, parts measured against the drawing on agreed features, and the evidence package — report, videos, measurements — that the T1 acceptance process consumes. T2 exists because T1 is honest: real tools find real corrections, and the confirming trial proves the fixes held and the process repeats.
    Programs that treat T2 as failure have misread the system — a tool that needed one correction round and proved it at T2 is a tool that worked exactly as stages intend. The delay sources begin when the loop is unbudgeted, not when the loop happens.

    How do the three stages differ in the ledger?

    Dimension T0 — first look T1 — formal trial T2 — confirmation
    Purpose See the fill; catch gross errors. Prove the tool against the drawing. Prove the corrections and the repeatability.
    Setup Short runs, exploratory settings. Documented process window, agreed parameters. Same window as T1; changes only with reasons.
    Samples A few, often imperfect. Agreed count, full documentation. Confirming count from stable conditions.
    Measurement Visual and functional checks. Full-dimensional report on named features. Key features re-measured; drift checked.
    Evidence Findings list, photos. Report, video, raw data — the acceptance package. Confirmation report against the T1 findings.
    Defect value Highest — geometry-stage fixes. Formal findings with owners and dates. Verdict: held or re-opened.
    Typical follow-up Edits before formal acceptance. Correction list, T2 planning. Sign-off and release to production.

    The column to internalize is evidence.
    T0 can be run on memory and handshakes; T1 cannot — its report is the artifact your acceptance decision reads, which is why its setup deserves as much care as its execution. A T1 run without a documented process window produces parts nobody can reproduce, and a report without named features measures enthusiasm, not geometry.
    When you review a trial plan, the question that sorts professionals from improvisers is simple: “what will the T1 report contain, feature by feature?” The criteria list and the answer should match line for line. A trial plan that reads like this table — each stage with a purpose, an evidence type and a verdict — is the difference between watching a tool develop and discovering one.

    How do you prepare for a T1 that decides things?

    1Freeze the measurement plan. Which features, on which instrument, against which drawing revision — decided before the trial, not negotiated over samples afterward.
    The plan is half the trial’s value.
    2Confirm the sample count and disposition. How many shots, who keeps what, and which samples ship where; ambiguity here costs weeks of shipping and re-shooting later.
    3Attend or delegate with authority. Someone present can see defects live and trade decisions in real time; decisions relayed across time zones convert hours of trial into weeks of email.
    The DFM discipline of named owners applies to the bench, too.
    4Record the process, not just the parts. Settings, cycle, melt and mold temperatures — the numbers that let corrections be reasoned instead of guessed; the warp conversations especially need them.
    5Decide the finding workflow in advance. Every finding gets an owner, a fix description and a re-verification method; the correction list that leaves the trial is a plan, and the change-order terms for out-of-scope work are already agreed.

    What must the T1 evidence package contain?

  • The full-dimensional report — every named feature, measured value and verdict, in a format your quality team can audit; the core of acceptance.
  • Raw measurement data — the machine’s own output, not only the summary; disputes are settled by raw numbers, not by excerpts.
  • Dated trial video — the tool running, labeled with tool number and date; it proves the parts came from your tool on your setup.
  • The process record — settings and conditions behind the samples, so the parts can be reproduced and the corrections reasoned.
  • A findings list with owners — each defect, its proposed fix, who executes it and when; the document T2 will be graded against.
  • Material traceability — resin grade and lot records for the trial shots; a report on unknown material measures nothing.
    The datasheet conversation closes the loop.

    • What does the number of trials really tell you?

    Trial count is information about the program, not a scoreboard. A tool that runs clean at T1 usually met a DFM process that took the geometry seriously — fill analysis, honest tolerances, gate locations argued before steel.
    A tool that needs two correction rounds is not cursed; it is a tool whose complexity surfaced honestly, and its T2 verdict means more than a rubber-stamped first pass would have.
    What deserves scrutiny is not the count but the accounting: quotes that include one trial and price the rest as surprises are transferring their planning gap to your budget, while quotes that state the loop — trials included, corrections priced, re-verification method named — are quoting a process they have run before. The same logic separates the tools themselves: a shop that arrives at T2 with documented findings and held fixes is demonstrating exactly the engineering you are buying.
    The stage system, read correctly, is not bureaucracy; it is the only mechanism that lets a buyer watch thinking happen before the tool ships. The quotes that differ most on trial scope are usually the ones worth interrogating first. Count the loop in the schedule the way you count the steel in the quote: named, priced and owned.

    “T0 finds what the design hid. T1 finds what the steel hid. T2 finds whether anyone was paying attention. A program that skips stages is not saving trials; it is relocating them into production.”
    — Arthur, Tooling Engineer at aCatMould

    Frequently asked questions

    Can we skip T0 and go straight to T1?

    On a simple, well-analyzed tool, sometimes yes — and the savings are real. The risk is arithmetic: findings that would have been geometry edits at T0 arrive as formal T1 findings, with re-trial, re-measurement and re-shipping attached. The decision belongs to the complexity profile; ask the shop where they would place your tool and why. The answer, in writing, tells you how they think about your program.

    Who pays for a second trial?

    Whatever the quotation says — which is exactly why it must say something. The professional split: corrections for work inside the quoted scope are the shop’s; changes the buyer introduces afterward are change orders; re-verification after either is priced as the trial loop the quote should have named. The payment schedule usually leans on T1 acceptance — one more reason the report must be real.

    How many T1 defects are normal?

    There is no universal count — complexity sets the number. What is normal is structure: findings that carry owners, fixes and re-verification methods. A short findings list with no evidence behind it is worse than a long one with a plan; the first hides work, the second schedules it. Grade the trial by its documents, not by its defect count.

    Should the molder or the buyer run the measurement?

    The shop measures and supplies raw data; the buyer audits against the drawing and the named features. Independent third-party measurement exists for disputed or high-stakes tools, at a price that buys both sides a common number. What does not work is measurement without raw data — a summary nobody can re-derive is a claim, not a record.

    What if T2 shows a regression?

    Then the loop continues, with one addition: the regression gets diagnosed against the T1 record — process drift, correction side-effects, or a fix that traded one defect for another. This is why the process record matters; a T2 run without it cannot even say what changed. Regressions are ordinary engineering events; undocumented ones are schedule events.

    Do stage names differ between shops?

    They do — some shops call the exploratory run T0, others call it a “setup trial”; some number the confirming shot T2, others T3 and beyond. The letters are less important than the sequence of questions behind them: fill, meet-the-drawing, repeat. When you compare quotes, map each shop’s stage names to those three questions and the plans become comparable — a quote that cannot say which of its trials proves the drawing is quoting theater, not trials.

    How do trials interact with mold acceptance?

    Directly: the acceptance criteria name what T1 (and T2, where applicable) must demonstrate, and the evidence package is the proof. A tool is accepted when the documents agree with the criteria — not when everyone is tired. Programs that tie acceptance to evidence rather than to arrival dates end their projects with a tool they can defend, and the life of that tool starts audited. Criteria, stages and evidence are one document under three names; programs that write it once never argue it twice.

    Key takeaways

    1. T0 asks does it fill, T1 asks does it meet the drawing, T2 asks does it repeat — three questions, three stages.
    2. T1 is the contractual center: its documented setup and full-dimensional evidence package are what acceptance reads.
    3. Prepare the measurement plan, sample disposition and finding workflow before the trial, not over its parts.
    4. Trial count is scope: quotes that name the loop and its pricing are quoting a process they have run.
    5. Stages exist to keep findings on the geometry end of the cost curve; skipping them relocates the work into production.

    If your trial plan has more dates than documents, send us the drawing and the tooling quote. You will get the stage plan your tool actually needs — measurement plan, evidence list and finding workflow included — from a named tooling engineer, before any steel is ordered. Bring the trial calendar with you; the answers belong on it as much as the steel does.

    Sources
    ISO 20457:2018 — Plastics moulded parts: Tolerances and acceptance conditions, International Organization for Standardization
    Plastics Industry Association — industry resources on molding and tooling practice
    LKM — mold base supplier, reference for tool architecture in trial planning
    DME — mold technologies and components, reference for trial setup practice

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