additivetools
All articles
13 May 202612 min read
lpbfprocess-developmentporosityparameter-optimisationdensity

LPBF Parameter Development: From First Principles to a Stable Process Window

Qualifying a new alloy or machine for Laser Powder Bed Fusion is not a matter of borrowing parameters from a datasheet and hoping for the best. Datasheets represent one machine family, one powder lot, one build plate temperature, and one layer thickness. Transfer to a different machine — even the same model from the same manufacturer — can shift the optimal energy window by 10–20%. This article sets out a reproducible, physics-grounded methodology for developing a robust process window from scratch.


Volumetric Energy Density: a useful proxy with real limits

The standard first-order parameter is Volumetric Energy Density (VED), defined as:

VED = P / (v × h × t)   [J/mm³]

where P is laser power (W), v is scan speed (mm/s), h is hatch spacing (mm), and t is layer thickness (mm). VED collapses four independent variables into one number, which makes it convenient for mapping porosity against a single axis.

In practice, two parameter sets with identical VED but different individual parameters can produce completely different melt pools. A high-power, high-speed combination at a given VED produces a shallower, wider melt pool than a low-power, low-speed combination. The melt pool geometry — not VED alone — determines grain structure, residual stress, and whether keyholing occurs.

Use VED to:

  • Set the initial search space (typically 40–100 J/mm³ for steels and nickel superalloys; 30–70 J/mm³ for aluminium alloys)
  • Identify the rough boundary between lack-of-fusion and keyhole regimes
  • Compare parameter sets within a single machine configuration

Do not use VED to compare across machine platforms, beam profiles, or powder particle size distributions. For that you need normalised enthalpy.


Normalised Enthalpy and the Keyhole Threshold

King et al. (2015, Acta Materialia, doi:10.1016/j.actamat.2014.10.028) introduced the normalised enthalpy ΔH/hs as a dimensionless measure of melting depth:

ΔH/hs = A·P / (π·ρ·Cp·Tm·√(α·v·d³))

where A is absorptivity, ρ is density, Cp is specific heat, Tm is melting temperature, α is thermal diffusivity, v is scan speed, and d is beam diameter. The keyhole transition occurs at ΔH/hs ≈ 6–8 for most metals (the exact threshold is material- and atmosphere-dependent; see Cunningham et al., 2019, Science, doi:10.1126/science.aav2550).

The practical value of this framework: it explains why the same VED causes keyholing in Ti-6Al-4V (low absorptivity at 1070 nm, high Tm) but not in 316L stainless steel (higher absorptivity). It also predicts how parameter sets will scale when changing beam diameter or layer thickness.

For a new alloy, calculate ΔH/hs across your planned parameter space before running any samples. This immediately identifies the unsafe high-energy region and narrows the experimental search space.


Defect Regimes: LOF vs Keyhole Porosity

Two distinct porosity mechanisms govern LPBF, and they look different in cross-section:

Lack-of-Fusion (LOF) Porosity

  • Cause: insufficient VED to melt through to the previous layer; incomplete fusion between tracks
  • Morphology: irregular, elongated voids aligned with layer boundaries; often contain unmelted powder particles
  • Cross-section appearance: jagged pores, typically 50–500 µm, in a roughly planar distribution
  • Location: concentrated at track boundaries and in the inter-layer region
  • Threshold: below roughly 40 J/mm³ for most metallic alloys at standard layer thickness

Keyhole Porosity

  • Cause: excessive VED creates a deep vapour cavity (keyhole) that collapses and traps gas
  • Morphology: nearly spherical pores, typically 10–100 µm, distributed through the bulk
  • Cross-section appearance: round, clean-edged pores with no evidence of unmelted powder
  • Location: distributed throughout the bulk, concentrated along scan tracks
  • Threshold: above ΔH/hs ≈ 6–8 (material-dependent)

Distinguishing them: In a polished cross-section under an optical microscope, LOF pores are immediately distinguishable from keyhole pores by shape alone. Synchrotron CT (Cunningham 2019) confirmed this morphological distinction; high-resolution benchtop CT (voxel size < 5 µm) can achieve the same result non-destructively. The Porosity Predictor models the boundary between these regimes as a function of your parameter set.

Between these two extremes lies the stable conduction mode window — typically 50–80 J/mm³ for steels, 55–90 J/mm³ for nickel superalloys, and 30–60 J/mm³ for aluminium alloys — where relative density exceeds 99.5%.


The Parameter Development Sequence

Step 1 — VED Scan Array

Print a matrix of single-layer or small cube specimens varying VED in steps of ~5 J/mm³ across the full range from estimated LOF boundary to estimated keyhole boundary. Fix layer thickness and hatch spacing; vary power and speed proportionally to isolate VED effects before introducing individual parameter sensitivity.

A 5×5 array (25 cubes, ~10 mm side) on a single build plate is sufficient for the initial scan. Cross-section in the build direction (XZ plane) to examine both inter-layer fusion and track-to-track bonding.

Step 2 — Melt Pool Characterisation

At promising VED values from Step 1, run single-track experiments at several power/speed combinations with equal VED. Measure melt pool width and depth from polished cross-sections. Cross-reference with the Melt Pool Geometry calculator predictions. Look for:

  • Track width consistent with designed hatch spacing (overlap 10–30%)
  • Melt depth 120–150% of layer thickness (to ensure bonding to the previous layer)
  • No evidence of humping (a Rayleigh–Taylor instability at very high scan speeds, v > 1500 mm/s for most alloys)

Step 3 — Sensitivity Matrix

For each candidate parameter set, build a small Design of Experiments (DoE) varying power ±10% and speed ±10% around the nominal values. This 9-point matrix (3×3) quantifies the sensitivity of density to parameter perturbations — directly analogous to the process capability index used in conventional manufacturing. Use the Laser Parameter Optimizer to generate the DoE matrix and estimate the stable process window.

A robust process window produces >99.5% relative density across all 9 points. A fragile window may hit 99.9% at the nominal point but drop to 98% with a ±5% power variation — unacceptable for production parts.

Step 4 — Porosity Measurement

Measure density by at least two methods to cross-validate:

MethodAccuracyResolutionDestructive?Notes
Archimedes (ASTM B311)±0.1%Bulk averageNoFast, cheap; misses isolated internal pores if part is sealed
Image analysis (cross-section)±0.05–0.2%Planar sliceYesDistinguishes LOF vs keyhole; sampling-dependent
CT scanning±0.01–0.05%Full volumetricNoGold standard; cost ~£300–1000/part; 5–10 µm voxel needed
Helium pycnometry±0.05%Bulk averageNoUnaffected by connected porosity; ASTM B923

Archimedes is the workhorse for DoE screening. CT scanning is mandatory for aerospace or fatigue-critical qualification where the spatial distribution of pores matters, not just bulk density. Cross-section image analysis using ImageJ or equivalent software quantifies LOF vs keyhole fractions if pore morphology classification is required.


Density Targets by Application

Not all applications require the same density:

  • Structural aerospace (fracture-critical): >99.9% relative density, zero LOF pores > 100 µm, confirmed by CT per AMS 7010 / ASTM F3122
  • General structural (non-fracture-critical): >99.5% relative density, Archimedes sufficient
  • Thermal management (conformal cooling inserts): >99.0% acceptable if internal channels are the primary functional feature
  • Research/development coupons: >99.0% relative density for valid mechanical test results

When to Use the LPBF Parameter Development Wizard

The LPBF Parameter Development wizard steps through this methodology interactively: it takes your alloy's thermal properties, beam diameter, and target layer thickness, computes the ΔH/hs map, predicts the LOF and keyhole boundaries, and outputs a prioritised DoE matrix for your first build. It also links directly to the VED calculator and Melt Pool Geometry tool so you can refine predictions as you gather experimental data.

Use the wizard as the starting point. Use the experimental sequence above to validate and extend it for your specific machine-alloy-atmosphere combination.


Documentation and Traceability

ISO/ASTM 52904:2019 (Process Characteristics and Performance — Making Metal Parts by Laser-Based Powder Bed Fusion) requires that process parameters be documented with traceability to the qualified window. As a minimum, record:

  • Laser power ± calibration tolerance (typically ±2–5%)
  • Scan speed ± controller accuracy
  • Layer thickness ± recoater calibration
  • Hatch spacing and scan strategy (e.g., 67° rotation between layers)
  • Powder batch number, PSD (D10/D50/D90), and Hall flow rate
  • Atmosphere oxygen level during build (typically < 500 ppm for Ti; < 1000 ppm for steel)
  • Build plate preheat temperature

This documentation constitutes the Qualified Process Parameter Set (QPPS) that all subsequent production builds must conform to.


Further reading

Continue reading — sign up free

Free account · no credit card · no marketing. Sign up to unlock the full library: 26 articles, 45 materials, 10 processes, 43 papers and more.