AM Defect Origin Study

Overview

A contract manufacturer producing LPBF 316L stainless steel components was experiencing recurring porosity concentrated near scan-track boundaries in thin-wall features. Standard parameter sets from the machine OEM had been validated on bulk geometries, but defect rates climbed above 2% when applied to walls below 1.5 mm. Parts were failing CT inspection at incoming quality, and the root cause was unknown.

Problem

Metallographic cross-sections revealed two distinct porosity populations:

• Spherical gas pores (10–40 μm) distributed uniformly — consistent with feedstock moisture or dissolved gas.
• Irregular lack-of-fusion voids (50–150 μm) concentrated at scan-track overlap zones in thin-wall regions.

The gas pores were within spec. The lack-of-fusion voids were not. Their spatial correlation with thin-wall features pointed to a geometry-dependent thermal condition rather than a global parameter deficiency.

Method

The analysis proceeded in three stages:

01 — Thermal History Reconstruction

A transient thermal model was built for the thin-wall cross-section using the actual scan strategy (67° rotation, 10 mm stripe width). Melt pool dimensions — depth, width, and length — were computed as a function of position along each scan vector. The model was calibrated against single-track melt pool measurements from the same powder lot and machine.

02 — Process Window Mapping

Melt pool depth-to-width ratios were mapped across the part cross-section to identify regions falling outside the conduction-mode processing window. In bulk regions, the thermal boundary conditions maintained stable melt pool geometry. In thin-wall regions, reduced heat conduction into surrounding powder caused localized overheating on interior scans and insufficient remelting depth at overlap zones where scan vectors reversed direction near the wall boundary.

03 — Defect Regime Correlation

The predicted lack-of-fusion boundary (where remelting depth fell below the layer thickness) was overlaid on the metallographic defect map. Spatial agreement was within one scan-track width (∼100 μm). The root cause was confirmed: scan-track overlap zones in thin walls experienced insufficient remelting due to altered thermal boundary conditions — not inadequate energy input.

Result

The fix was not higher power or slower speed — it was a geometry-aware scan strategy adjustment:

• Modified overlap fraction in thin-wall regions (increased from 30% to 45% where wall thickness < 1.5 mm).
• Adjusted scan vector ordering to prevent heat accumulation at reversal points near free surfaces.

Post-modification builds showed porosity below 0.1% across all wall thicknesses. Parts passed CT inspection on the first build with the revised parameters. No change to laser power, speed, or hatch spacing was required.

Deliverables

• Root cause report — documented thermal analysis, defect correlation, and corrective action with full traceability to calibration data.
• Process window map — melt pool stability boundaries as a function of local wall thickness, usable for future part qualification.
• Revised scan strategy specification — geometry-dependent overlap and vector ordering rules, ready for integration into build preparation software.