This program exists so a number on an engineering page can trace back to a coupon and its raw data, not just to a search result. Each experiment writes down its method and its limits before it names any number, so a reader can check the work instead of taking our word for it.
Status of the program
No experiment in this program has run yet, and no result is published yet. Every engineering number on this site today comes from a named outside source, not from a coupon we printed and checked ourselves. This page will change on the day that is no longer true.
At what thickness does a thin wall print clean, and does the printed wall match the number set in the slicer?
Whether a freestanding wall completes without collapsing, and how the printed thickness compares with the slicer's setting.
A wall set too thin can warp, gap, or fail to print at all. A buyer designing a thin feature wants a real floor, not a number copied from a spec sheet for a different process.
A reviewed result may support
The thinnest wall setting that printed clean in this test, on this printer, in this material.
A reviewed result may not support
A single wall thickness number that works for every material or printer.
How much weaker is a part printed standing up than the same part printed flat?
Load at break, compared between a part printed flat and the same part printed standing.
A part loaded across its layers can break at a much lower load than the same part loaded along its layers. A buyer choosing an orientation wants a real number for that gap, not a rule of thumb.
A reviewed result may support
The size of the strength gap between flat and standing, for one material, one printer, and one bar shape.
A reviewed result may not support
A strength ratio that applies to every material, part shape, or load direction.
How much force pulls a heat-set insert out of a printed boss, and does hole size change that?
Axial pull-out force on a heat-set insert, compared across hole diameters.
An insert that pulls out under load fails the whole assembly. A buyer sizing a boss wants a real pull-out number, not one borrowed from a different material or process.
A reviewed result may support
Pull-out force at the hole sizes tested, for one insert brand and size, one material, one printer.
A reviewed result may not support
A pull-out figure for a different insert brand, size, or material.
How much does an unsupported bridge sag as the span gets longer?
Sag depth on the underside of a bridge, measured across a set of span lengths.
A bridge that sags too far can touch whatever sits below it, or throw off the top surface. A buyer spanning a gap wants a real sag number for the span they need.
A reviewed result may support
How sag changed with span length, for one material, one printer, one nozzle.
A reviewed result may not support
A bridge span figure that works for every material or printer, without a coupon of your own.
How many times can a cantilever snap flex before it cracks?
The number of insert-and-remove cycles a cantilever snap survives, compared across deflection amounts.
A snap-fit that a customer opens and closes many times needs to survive that use. A buyer designing a snap wants a cycle count for the deflection they plan to use, not a borrowed strain table.
A reviewed result may support
Cycle count to cracking at the deflection amounts tested, for one material, one snap shape, one printer.
A reviewed result may not support
A cycle count for a different material, snap shape, or deflection amount.
Every experiment moves through the same stages, in this order, before it can publish a result.
Planned
Ready to run
Running
Measured, not yet reviewed
Reviewed
Reviewed, awaiting publication
Published
Invalidated
Every experiment records
Material, filament brand, and spool
Machine and nozzle size
Layer height, print temperature, and bed temperature
Print orientation on the plate
Slicer name, slicer version, and print profile
Coupon version and its exact geometry
Sample count, planned and then actual
Measurement tool, its resolution, and its calibration status
What counts as internal evidence, and what does not
Counts
A reviewed dataset with its method written down first
The raw CSV file, not only a summary number
The real sample count for that run
The resolution of the tool used to take the reading
A stated list of what the result can and cannot support
Does not count
One print, by itself
A photo of a part
A slicer preview
A simulation
A number pulled from a search result
A plan for an experiment that has not run yet
How raw data will be published
A published result links to its raw data as a CSV file, one row for every reading, with the sample id, the condition, the nominal value, the measured value, and a reason next to any reading that was excluded.
The column format is written down in our raw data format document, so a reader can check the arithmetic behind a summary table instead of trusting it on sight.
How limitations are handled
Every result carries a list of what it may conclude and a list of what it may not conclude, even when the data looks clean.
One printer and one spool of filament is one setup, not a rule for every printer and every spool.
A reading that gets excluded stays in the CSV, with the reason written next to it. Nothing is deleted from the file.
Planned experiments
These are plans for future runs, not results. None of them has run yet.
PlannedHole Compensation
Hole Compensation Ladder, PLA, 0.4 mm Nozzle
On one P1S-class printer in PLA at 0.4 mm, how far under nominal does a vertical-axis hole print at 3, 4, 5, 6 and 8 mm, and which offset step from the hole compensation ladder (0, 0.1, 0.2, 0.3 mm) lands closest?
At which diametral interference (0.00 to 0.30 mm in 0.05 steps) does a printed pin hold in a printed bore by hand, and at which does it split the boss?
At which slicer wall thickness (0.4 to 2.0 mm in 0.2 steps) does a freestanding wall print complete, and how does its measured thickness compare with the slicer's number?