A real case
A 48 mm deep pocket with a 3 mm internal corner passes every CAD check and stops the job at the machine six weeks later. Upload the STEP before the design review: manufacturability findings come back before a process is picked, and a voxel removal simulation quantifies and locates the stock your tooling cannot reach.
The situation
A 6082 aluminium mounting bracket, one STEP file, design review on Thursday and release the Friday after. The part carries a 48 mm deep pocket with a 3 mm internal corner radius, drawn that way because the mating boss needed the clearance and nothing in CAD objected to it. Six weeks after release a machinist rings: nothing in the crib bottoms out in that corner at that depth, the job turns into two setups, and the fix is now a revision, a re-quote and a date that moves. At the model it was one number and an afternoon — open the corner to 5 mm and lose nothing that mattered. The distance between those two costs is entirely a matter of when somebody looked.
Step by step
An STL or STEP model, a DXF or PDF drawing, or both together on one job go through the same first look, with no process chosen and no template set up first. Detected holes come back named, each carrying its own confidence, with the overall envelope and the part's face counts alongside them. The file that goes in is the one already open on your screen at the review.
General manufacturability issues are flagged for any uploaded part before a process has been chosen — counted high and medium, with the detail behind each finding rather than a single verdict on the part. That is the moment the corner radius stops being a modelling habit and becomes a question with something specific attached to it, six weeks ahead of the phone call from the machine.
Every specialist panel mounts against the same job, so moving from the findings into the machining view is a panel switch rather than a second upload. Machinable features come back detected, each with its own confidence, and a real voxel removal simulation runs against the actual tooling to quantify and locate the stock no cutter can reach. When an undercut genuinely needs a second setup, a 3+2 setup plan is generated, gouge-checked, indexed and priced into the quote instead of stopping at "needs 5-axis" — which is what a second setup discovered on the floor costs you, priced before it happens.
Every feasible process carries a should-cost curve across quantities, and every cost line on it traces to a named rate, posture setting or engine rule rather than an opaque total. Every result carries a confidence value that states its basis — a published prior, shop-calibrated actuals, a deterministic rule, or a flag that it needs human review. Open the corner to 5 mm, upload the changed model and link it as a new revision of the one already there: what moved comes back directly — findings added and resolved, weight, volume and bounding-box deltas, and a per-method cost-curve delta — without re-analyzing anything.
What you get
What you take into Thursday's review is short and specific: the findings the upload returned with their severity and the detail behind each, the stock the removal simulation could not reach and where it sits on the part, and a per-piece number whose every line names the rate, posture setting or rule that produced it. The corner-radius conversation then takes four minutes at a review instead of a change order in week six. Change the callout afterwards and the second upload links as a new revision of the first, so what that one number was worth reads off a single comparison — findings added and resolved, weight, volume and bounding-box deltas, and a per-method cost-curve delta, without re-analyzing anything. One thing worth being straight about: the removal simulation models the cut, not the fixture or the machine — it will not tell you the head is about to meet a clamp, and no collision or fixture modelling is claimed. Partimation is decision support, not a guarantee — every finding names what produced it, and your supplier and your own engineering judgment still sign it off before tooling is committed.
Already shipped
Built into PartDFM.
Built into MFGDFM.
Built into MachDFM.
Built into MachDFM.
Built into MachDFM.
Built into CostDFM.
Built into CostDFM.
Built into CostDFM.
Built into CostDFM.
Built into PartDFM.
FAQ
A real voxel removal simulation runs against the actual tooling and quantifies and locates the stock that cannot be reached, which is the check a CAD screen never performs. Machinable features come back detected first, each carrying its own confidence, and when an undercut genuinely needs a second setup a 3+2 setup plan is generated, gouge-checked, indexed and priced into the quote rather than stopping at "needs 5-axis".
The first upload is early enough. General manufacturability issues are flagged for any uploaded part before a process has been chosen, counted high and medium with the detail behind each finding, so a model can go in at a review where the wall thickness and the corner radius are still opinions rather than released geometry.
A revision delta answers precisely that. Upload the part with the callout changed, link it as a new revision of the one already there, and the difference comes back without re-analyzing anything — findings added and resolved, weight, volume and bounding-box deltas, and a per-method cost-curve delta. Every cost line still traces to a named rate, posture setting or engine rule, so the difference is readable line by line rather than as two totals.
A supplier review is still what commits tooling. Partimation is decision support, not a guarantee: the removal simulation models the cut rather than the fixture or the machine, no collision or fixture modelling is claimed, and every finding names what produced it precisely so you can put it to the supplier. Confirm it with them and with your own engineering judgment before anything is cut.
Where this fits
You design the part. Somebody else has to make it, and they tell you weeks late. Upload the model before release: manufacturability findings come back before a process is picked, and six process families are screened with the constraint behind each verdict and a should-cost curve across quantities.
Manufacturability feedback that arrives after release is a change order, not a review. Partimation returns findings on every upload, before a process has been chosen, and runs a real removal simulation against the tooling you actually run to show exactly which stock a cutter cannot reach.
Get started
Send a model or a drawing and book a walkthrough — or start free and run one through yourself.
Book a demo
Tell us about your shop and we'll set up a walkthrough on a real part.