FuruRok

everyone looks at topology optimized spacecraft hardware and assumes it has to be 3d printed Well turns out when nasa goddard tested both on the same exoplanet telescope bracket CNC machining beat metal 3d printing on cost, speed, and ultimate strength ryan mcclelland fed the solver the exact geometry of a 1/8 inch milling cutter alongside the bolt pattern, 10g shock loads, and red keep-out zones for bolt installation the algorithm evolved 31 candidates in an hour, removing every low-stress voxel only from directions a milling tool could actually reach the result vs the 3d printed version: -> automated CNC (6061-T651 billet): $1,000, in hand in 3 days 147 Hz stiffness, 14.8 MPa max stress -> laser powder bed 3d print (AlSi10Mg): $2,000, took 3 weeks, required support removal and heat treat, and broke earlier in destructive load testing than the billet CNC part the wild part isn't just that the organic bracket was 3x stiffer per kilogram and had 7x lower stress than what two senior engineers spent two days drawing in CAD it's that modern 5-axis mills can carve alien bone geometry out of solid aluminum faster and cheaper than a printer can sinter it you stop drawing the part and start engineering the constraints