Same part, very different strength
A layer-built part is not equally strong in every direction. Within a layer the material is continuous; between layers it is a bond formed while the previous layer was cooling. Loaded along the layers, a printed part behaves roughly like the bulk polymer. Loaded across them, it can fail at a fraction of that — and it fails suddenly rather than gracefully.
This is why orientation is a manufacturing decision and not a packing convenience. The same geometry, rotated ninety degrees, is a different part mechanically.
What the platform does about it
Every quote runs an orientation search: a fixed list of candidate orientations, each scored against versioned weights for support volume, contact area, overhang area and build height. The best-scoring orientation is recorded on the quote alongside the ruleset version that scored it, so it can be inspected. It is a recommendation: the part is sliced, priced and printed the way you uploaded it, so if orientation matters, upload the model already turned the way you want it built.
Two limitations are worth knowing. The search optimises for those criteria only: it does not know which face of your part is cosmetic or which axis carries load. And the metrics it scores are convex-hull approximations, which fill in concave features — good enough for ranking candidates, not a measurement. If load direction matters, the design advice below is the most dependable way to make it work.
How to design so orientation matters less
- Put load into compression or into the build plane wherever the geometry allows. A bracket loaded in shear along its layers is a much safer part than one loaded in tension across them.
- Add fillets at the base of any feature that will be loaded. A sharp internal corner concentrates stress exactly where the interlayer bond is weakest.
- Avoid a thin, tall boss that must resist a side load. That is the single most common printed-part failure, and it is a geometry problem, not a material one.
If it has to be strong in every direction
Powder-bed processes are much less directional (we do not run them — see the powder-bed guide). With FDM, the reliable route is to change the design so the load path follows the layers. Switching to a stronger material on its own rarely fixes it: the weakness is the bond between layers, not the material.