My DoE model predicted a peak dimensional accuracy of 98.289% at LH 0.261 mm, RW 0.563 mm, MH 3.631 mm. In confirmation prints at that setting, I observed 97.52%, 97.17%, and 97.71% (mean 97.47%, s.d. ≈ 0.27%). Despite the high value, bridges collapsed at this “mathematical optimum,” so it was not practically usable.

I adopted a more robust set—LH 0.20 mm, RW 0.45 mm, MH 2.25 mm—which printed reliably with stable bridges and consistent accuracy (mean ≈ 97.14%). This window traded ~1% of theoretical accuracy for bridge integrity and repeatability, and became my recommended setting.

Layer Height dominated the response in the quadratic model; Raster Width and Mover Levitation Height had secondary/no significant effects over the tested ranges. Goodness-of-fit and residual checks supported using the model to guide settings, but final selection prioritized print stability over the absolute numerical peak.