Last updated: 2026-09-04 · 6 min read
What Is a Run-at-Rate Test? — Proving a Mold Holds Production Speed Before You Accept It
A mold that makes one good part differs from one making thousands per hour, and the run-at-rate test bridges that gap. A functional trial asks whether the tool makes a good part; a run-at-rate asks whether tool, press, material and operator produce good parts at assumed cycle and yield over a sustained window. Most defects hide in short runs and surface in long ones — cooling that cannot keep up, slides sticking hot, drifting dimensions.
Key Facts
<>
Runner system — The channel network that delivers melt from the machine nozzle to each cavity. Cold runner: solidifies between shots, produces runner waste. Hot runner: stays molten, eliminates runner waste but adds $15k–$40k to mold cost. The related guide covers adjacent topics.
What Does Run-at-Rate Prove That a Trial Does Not?
Five questions, each answered differently by a short trial versus a sustained run:
| Question | Functional Trial | Run-at-Rate |
|---|---|---|
| Makes a good part? | Yes — its whole job | Assumed; goes further |
| Holds the cycle time? | Not measured long | Sustained at target rate |
| Stays in tolerance hot? | A few parts, often cool | First/middle/end, fully soaked |
| Yield acceptable? | Sample too small | Scrap and rework counted |
| Line runs unattended? | No — trials are attended | Feeding, drying, downtime tested |
The right column is one idea five ways: everything that only shows up under sustained, hot, representative production. A functional trial is a short attended burst — necessary but nowhere near sufficient — because it runs cool, is watched hand-to-mouth by a technician who fixes each hiccup, and makes too few parts to reveal a one-in-a-hundred defect. The run-at-rate removes those crutches.
Running for hours lets the mold thermal steady state, where cooling inadequacy surfaces as creeping cycle or shifting dimensions. Counting scrap across thousands of shots reveals intermittent defects invisible in twenty parts. Testing the whole cell — not just the mold — exposes throughput killers: a dryer that cannot keep the hopper filled, a robot slower than assumed, a cycle needing operator flash removal. Measuring cycle as it actually held turns the datasheet into real cost per part.
How Do You Structure a Defensible Run-at-Rate?
Four rules, in order:
- Agree the pass numbers before you start the machine. A run-at-rate without pre-set acceptance criteria is a demo, not a test. Fix in advance the target cycle, the duration or shot count, the required uptime, the maximum scrap and rework rate, the dimensional capability on key features, and the sampling points. Write them into the acceptance document so nobody negotiates what “good” means after seeing a mediocre result. This is also where the business case cost-per-part ties directly to the measured cycle and yield.
- Run the real production configuration, not a lab setup. Run the actual mold in its cavitation, on the target press, with the real robot, dryer, feeder, mold-temperature controllers, material lot and color, and the intended cycle and part-handling sequence. A single-cavity shortcut or a hand-fed trial that passes tells you nothing about the eight-cavity line you will depend on. Include the changeover and in-process checks you will genuinely run. The closer the test resembles the shipping line, the more its result transfers to daily output.
- Measure the drift the short trial hides. Collect parts at defined intervals — start, middle, end of soak, and after any interruption — and measure critical dimensions at each so you catch thermal creep rather than trusting the first good shot. Track every stoppage, its cause and duration. Count scrap and rework continuously so intermittent defects and real uptime surface in numbers, not impressions. Compute capability from the run’s own measurements so acceptance rests on demonstrated consistency.
- Document the result as the acceptance record. The run should end in a signed report stating each pre-agreed criterion, the measured value, and a clear pass or fail with supporting data: cycle-time log, uptime and downtime list, scrap tally, interval dimensional results. That document, not a verbal “it ran fine,” releases final payment, opens a production order and protects both sides if a launch later disputes whether the tool performed.
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. The related guide covers adjacent topics.
A run-at-rate is where “the mold works” becomes “the mold works as a production asset that hits the numbers the program was funded on.” Everything else is a demonstration.
Frequently Asked Questions
How long is a typical run-at-rate?
Long enough to thermal steady state and gather meaningful scrap and downtime data — commonly two to eight hours continuous, or a defined shot count in the low thousands. The goal is to exceed the transient where the mold is still soaking, because that is precisely the regime a short trial lives in and a production line does not.
Too short and a marginal cooling or feeding problem stays hidden; absurdly long adds cost without adding signal once metrics have stabilized. The right duration is the one that lets you measure cycle, yield and dimensions after the tool is hot and still running — set it up front and tie it to the sampling plan.
Who runs it — the mold maker or the customer?
Ideally on the production setup the parts will ship from, witnessed by whoever is accepting them. The mold maker should run the tool competently — that is part of what you validate — but acceptance is meaningful only to the party whose program depends on throughput. For a customer who cannot be present, the defensible path is a pre-agreed protocol, a documented run, retained samples shipped for independent measurement, and raw cycle and dimensional data.
Whoever decides pass or fail must see the measured numbers behind the claim, not just a good-looking first-off part. The discipline is the same either way.
Can a mold pass T1 and still fail run-at-rate?
Frequently — that gap is the entire reason the run-at-rate exists. A tool can produce an in-spec part on an attended, cool, short trial and then fail once it is hot and unattended: the cycle creeps past target as cooling saturates, a slide that ran cold begins to drag, dimensions drift once the steel soaks, or an intermittent defect appears only across thousands of shots.
A mold that passes T1 but misses run-at-rate is not cosmetically defective — it is not yet a production asset. The fix is usually a cooling, gating, ejection or ancillary change best caught here, before you release final payment and commit volume to a tool that makes good parts only while watched.
Takeaways
<>
Run-at-rate is the last gate before you trust a mold with a real program.
Run it with the discipline of an acceptance test, not the optimism of a demonstration, and you will know — with data — whether the tool is a production asset or a liability.
Sources
- Society of Plastics Engineers — Technical Papers
- ENGEL — Injection Molding Technology
- Arburg — Practical Injection Molding
- Plastics Technology — Processing Tips
- RJP Inc — Mold Validation Resources
Sources: Plastics Technology · Plastics Technology Online · Society of Plastics Engineers (SPE) · ThomasNet