Metal Additive Manufacturing: The Complete Engineering Guide to LPBF, DMLS, SLM, EBM, and DED — Physics, Materials, Process Control, and Production Economics

A dense equation-driven engineering deep-dive into metal powder bed fusion and directed energy deposition — covering laser-powder interaction physics from…

·

Metal additive manufacturing has crossed the chasm from prototyping curiosity to production-qualified manufacturing process. GE's LEAP engine fuel nozzle — consolidated from 20 brazed parts into a single CoCr LPBF component — now flies on every Airbus A320neo and Boeing 737 MAX. Stryker's Tritanium spinal cages, built via SLM in Ti-6Al-4V with engineered lattice porosity, have been implanted in over 500,000 patients. SpaceX's SuperDraco thrust chambers are DED-deposited Inconel, printed in days instead of forged over months. This is not the future — this is today's production floor.

But metal AM is not "press print and walk away." The physics of melting micron-scale metal powder with a 70–100 μm focused laser spot at scan speeds of 500–2,500 mm/s produces thermal gradients exceeding 10^6 K/m, cooling rates of 10^5–10^7 K/s, and residual stresses that can tear a 5 mm Ti-6Al-4V build plate off its substrate. Getting it right requires understanding everything from Fresnel absorptivity to Kurdjumov-Sachs orientation relationships in the resulting microstructure.

This guide covers the full engineering stack: the physics at the melt pool, the metallurgy of the solidified bead, the mechanics of part-scale distortion, and the economics that determine whether a $700,000 DMLS machine pays for itself.

1. The Metal AM Technology Landscape

Metal additive manufacturing encompasses seven ASTM F42 categories, but four dominate commercially:

Process · Energy Source · Feedstock · Build Volume (typical) · Layer Thickness

LPBF (Laser Powder Bed Fusion) · Fiber laser, 200–1,000 W · Gas-atomized powder, 15–53 μm · 250 × 250 × 300 mm · 20–60 μm

EBM (Electron Beam Melting) · Electron beam, 3–6 kW · Gas-atomized powder, 45–106 μm · 200 × 200 × 380 mm (Arcam) · 50–100 μm

DED (Directed Energy Deposition) · Laser/IPA/EB, 1–4 kW · Powder or wire, 50–150 μm · 900 × 1,500 × 900 mm (DMG MORI) · 0.5–2.0 mm

Binder Jetting · IR lamps (curing) + furnace (sintering) · Powder, 5–35 μm · 800 × 400 × 500 mm (ExOne) · 30–100 μm

Terminology note: DMLS (Direct Metal Laser Sintering) is an EOS trademark. SLM (Selective Laser Melting) is an SLM Solutions / Nikon trademark. Both are LPBF — the laser fully melts the powder, not sinters it. I'll use LPBF throughout for scientific accuracy.

2. Laser-Powder Interaction Physics

The fundamental event in LPBF is the absorption of laser energy by a powder bed and the subsequent formation, convection, and solidification of a melt pool. At t = 0, a focused Gaussian beam strikes a bed of spherical metal particles. The physics unfolds in microseconds.

2.1 Laser Absorption in Metal Powders

For a flat metal surface, laser absorptivity \alpha at normal incidence follows the Hagen-Rubens relation for the infrared (fiber lasers operate at \lambda = 1,064 nm):

where \sigma is electrical conductivity and \rho_e is electrical resistivity. For pure copper at 1,064 nm, \alpha \approx 0.03 — reflective nightmare. For 316L stainless steel, \alpha \approx 0.35. For Ti-6Al-4V, \alpha \approx 0.55.

However, a powder bed behaves very differently from a flat surface. Multiple scattering between particles creates a "light trap" effect that can increase effective absorptivity to \alpha_{\text{eff}} = 0.7–0.85, depending on particle size distribution and packing density. The penetration depth of the laser into the powder bed can be modeled via:

where \beta_{\text{ext}} is the extinction coefficient, on the order of 10^5–10^6 m^{-1} for typical metal powder distributions. This means the laser energy is absorbed within the first 1–3 particle diameters.

2.2 Volumetric Energy Density: The Master Parameter

The most useful aggregate parameter for LPBF process control is the volumetric energy density E_v:

where P is laser power [W], v is scan speed [mm/s], h is hatch spacing [mm], and t is layer thickness [mm].

This single number — the energy delivered per unit volume of powder — correlates with melt pool morphology, density, and microstructure across most alloys. Empirical process windows cluster around:

Alloy · Optimal E_v [J/mm³] · P [W] · v [mm/s] · h [μm] · t [μm]

Ti-6Al-4V · 45–75 · 200–370 · 800–1,400 · 100–120 · 30–60

Inconel 718 · 50–90 · 250–370 · 700–1,200 · 90–110 · 40

AlSi10Mg · 50–70 · 200–400 · 800–1,600 · 100–150 · 30–50

316L SS · 55–100 · 150–380 · 600–1,200 · 80–120 · 30–50

Maraging Steel (MS1) · 60–80 · 200–370 · 800–1,100 · 90–110 · 40

CoCr (MP1) · 70–120 · 180–370 · 600–1,000 · 80–100 · 30–40

CuCr1Zr · 200–350 · 350–500 · 400–800 · 80–120 · 30–50

Below E_{v,\text{min}}: lack-of-fusion porosity (irregular, elongated voids between tracks). Above E_{v,\text{max}}: keyhole porosity (spherical gas-entrapped voids from unstable vapor depression). The process window is narrow — typically ±15–25% of optimal E_v.

2.3 Melt Pool Thermal Field: The Rosenthal Solution

For a point heat source moving at constant velocity over a semi-infinite medium, the Rosenthal equation gives the steady-state temperature field:

where:

The melt pool boundary occurs where T = T_m (melting temperature). The melt pool dimensions can be estimated from this isotherm. For typical LPBF conditions (200 W, 1,000 mm/s on 316L), the melt pool is approximately 80–140 μm wide, 50–80 μm deep, and 150–300 μm long — an elongated teardrop shape.

The cooling rate at the solidification front, critical for microstructure, is:

For Ti-6Al-4V LPBF: \dot{T} \approx 10^6–10^7 K/s — faster than any casting or forging process by three orders of magnitude. This is why LPBF microstructures are so fine: the dendrite arm spacing \lambda_2 scales with cooling rate as:

where n \approx 0.33–0.42 and A is alloy-dependent. For 316L at 10^6 K/s, \lambda_2 \approx 0.5–2 μm — producing yield strengths 30–50% higher than wrought material.

2.4 Marangoni Convection and Melt Pool Dynamics

Surface tension gradients drive fluid flow within the melt pool. The Marangoni number quantifies this:

where d\gamma/dT is the temperature coefficient of surface tension [N/m·K], \Delta T is the temperature difference across the pool, L is pool length, \mu is dynamic viscosity, and \alpha is thermal diffusivity.

For most liquid metals, d\gamma/dT < 0 (surface tension decreases with temperature). This drives outward flow from the hot center to the cooler edges — beneficial for wetting and spreading. However, oxygen and sulfur contamination can flip the sign: d\gamma/dT > 0 drives inward flow, deepening the melt pool and creating a narrower, less stable track. This is why atomization gas purity matters — oxygen must be < 500 ppm for reactive alloys.

2.5 Keyholing and Vapor Depression

At sufficiently high energy density, the recoil pressure from metal vaporization exceeds surface tension pressure, and the melt pool surface depresses into a vapor cavity (keyhole). The threshold laser intensity follows from balancing recoil pressure p_{\text{recoil}} \approx 0.54 p_{\text{sat}}(T) against Laplace pressure p_{\text{Laplace}} = \gamma / r:

For 316L at typical LPBF parameters, keyholing onsets around I \approx 2–4 MW/cm², corresponding to E_v \approx 100–150 J/mm³ for a 70 μm spot. Keyhole porosity — spherical pores 20–100 μm from vapor trapped during cavity collapse — is a primary defect mode at high energy density. X-ray imaging at synchrotrons (e.g., the Zhao et al. 2020 experiments at APS) shows that keyhole pores form when the keyhole rear wall collapses faster than the melt can backfill, trapping argon shielding gas.

3. Metallurgy of Laser-Powder-Bed-Fused Alloys

The extreme thermal conditions of LPBF produce microstructures fundamentally different from cast or wrought equivalents. Understanding these is essential for predicting mechanical properties.

3.1 Ti-6Al-4V (Grade 5 / Grade 23)

As-built microstructure: The 10^6 K/s cooling rate from above the \beta-transus (T_{\beta} = 995°C) produces a fully martensitic \alpha' phase — fine acicular (needle-like) laths within prior-\beta columnar grains that grow epitaxially across layers along the build direction (⟨001⟩ for bcc \beta). The prior-\beta grains are typically 50–200 μm wide and extend hundreds of microns in Z. The \alpha' lath width is 0.2–1.0 μm.

Mechanical properties (as-built):

HIP + heat treatment (920°C / 100 MPa / 2 hr + 730°C / 2 hr): HIP collapses internal porosity. The thermal cycle decomposes \alpha' → \alpha + \beta (lamellar), dropping strength to 930–980 MPa UTS but recovering elongation to 12–16%. This is the standard aerospace post-processing for fracture-critical parts.

Fatigue performance: As-built Ti-6Al-4V shows high-cycle fatigue (HCF) strength at 10^7 cycles of 200–350 MPa (R = 0.1), significantly lower than wrought (500–600 MPa). The drivers are surface roughness (R_a = 8–20 μm as-built, acting as crack initiation sites) and internal lack-of-fusion defects. Machining or shot peening can recover 400–550 MPa HCF strength.

3.2 Inconel 718

As-built LPBF Inconel 718 has a cellular-dendritic microstructure with strong ⟨001⟩ texture along the build direction. The cooling rate suppresses \gamma'' and \gamma' precipitation — the as-built matrix is a supersaturated \gamma (fcc) solid solution with Nb-rich Laves phase at interdendritic boundaries.

Direct-aged (no solution treatment, just 720°C / 8 hr + 620°C / 8 hr):

Full heat treatment (AMS 5664: 980°C/1 hr + aging): Recrystallizes the columnar grain structure and dissolves Laves phase, producing isotropic equiaxed grains. This is preferred for rotating aerospace components.

3.3 AlSi10Mg

The near-eutectic composition (Al–10Si–0.35Mg) is designed for casting/AM weldability. As-built microstructure: fine cellular α-Al cells (~500 nm) surrounded by a continuous Si network. This produces:

T6 heat treatment (540°C solution + 160°C age): The Si network spheroidizes and coarsens, and Mg₂Si precipitates form. This drops strength to 250–300 MPa UTS but dramatically improves ductility to 8–15% — essential for automotive crash structures.

The high thermal conductivity of aluminum (k \approx 150 W/m·K for AlSi10Mg) means the Rosenthal melt pool is wider and shallower than for steel — easier to achieve full density but harder to maintain fine feature resolution. Preheating the build plate to 150–200°C reduces thermal gradients and is standard practice.

3.4 316L Stainless Steel

As-built 316L exhibits a hierarchical microstructure: melt pools (50–100 μm) containing elongated columnar grains with cellular sub-structure (0.5–2 μm cell size). The cellular boundaries are enriched in Cr and Mo due to microsegregation. This cellular structure provides an unusual combination of strength and ductility:

316L also exhibits transformation-induced plasticity (TRIP) under deformation: metastable austenite → \alpha' martensite, providing work hardening. LPBF 316L shows enhanced TRIP effect due to the fine cellular structure providing more nucleation sites.

3.5 Maraging Steel (18Ni-300 / MS1)

The as-built microstructure is a soft martensitic matrix (Fe–Ni lath martensite) with high dislocation density. The magic happens during aging at 490°C / 6 hr: coherent Ni₃(Ti,Mo) and Fe₂Mo intermetallic precipitates form at the nanometer scale, producing:

Maraging steel is the go-to LPBF material for injection mold tooling — it machines easily in the soft as-built state (30–35 HRC), then hardens uniformly during aging with minimal distortion (0.05–0.10% dimensional change, isotropic). The combination of LPBF's geometric freedom (for conformal cooling) with maraging steel's predictable heat treatment makes it ideal for mold inserts.

4. Residual Stress and Distortion Mechanics

The thermal gradient mechanism (TGM) is the dominant source of residual stress in LPBF. As the laser scans a track, the heated region expands but is constrained by the surrounding cold material, producing compressive plastic strain. On cooling, the region contracts but is now too short, producing tensile stress. Each layer adds a new increment of tensile stress at the top surface, bending the part downward (tensile side up).

4.1 Inherent Strain Method

The inherent strain \epsilon^* is the plastic strain that remains after a thermal cycle. For a single laser track:

where \Delta T_p = T_{\text{max}} - T_0 is the peak temperature rise above ambient and f depends on the stiffness of the surrounding material (0 for free expansion, 1 for full constraint). In a powder bed, f \approx 0.7–0.9.

The part-scale residual stress field can be computed by applying \epsilon^* as an initial strain in a linear elastic FEA model. This is the basis of commercial build simulation software (ANSYS Additive, Simufact Additive, Amphyon).

4.2 Cantilever Deflection: A Design Rule Heuristic

For a simple cantilever of length L, thickness t, and width w built horizontally unsupported, the tip deflection \delta from residual stress can be estimated as:

where \sigma_r is the effective residual stress (typically 200–600 MPa for as-built Ti-6Al-4V) and E is Young's modulus. For Ti-6Al-4V (E = 110 GPa, \sigma_r \approx 400 MPa), a 10 mm long, 2 mm thick unsupported overhang would deflect:

This is why support structures are needed for overhangs below 45° (relative to the build plate) — the deflection exceeds typical dimensional tolerances.

4.3 Support Structure Design Rules

Supports serve three functions: thermal anchoring (conducting heat away from the melt pool), mechanical anchoring (resisting residual stress warpage), and powder removal pathways.

5. Electron Beam Melting (EBM)

EBM operates on similar powder bed principles but with fundamental differences in energy source, atmosphere, and thermal management.

5.1 Key Differences from LPBF

Parameter · LPBF · EBM

Energy source · Fiber laser, 1,064 nm, 70–100 μm spot · Electron beam, 100–200 μm spot (defocused for preheat)

Atmosphere · Argon/N₂, < 1,000 ppm O₂ · Vacuum, 10⁻³–10⁻⁵ mbar

Preheat temperature · 40–200°C (build plate) · 650–1,100°C (entire powder bed, maintained between layers)

Scan speed · 500–2,500 mm/s · 1,000–8,000 mm/s (electromagnetic deflection, no inertia)

Layer thickness · 20–60 μm · 50–100 μm

Powder size · 15–53 μm · 45–106 μm

Surface finish (Ra) · 5–15 μm · 15–35 μm (coarser powder, larger melt pool)

The elevated preheat temperature in EBM is the key advantage: it dramatically reduces thermal gradients and residual stresses. Ti-6Al-4V parts built by EBM at 700°C preheat have near-zero residual stress and can be cut off the build plate without stress relief. The preheat also decomposes \alpha' martensite in-situ, producing an \alpha + \beta lamellar microstructure directly — no post-build HIP required for many applications.

5.2 Powder Smoking: The Electrostatic Instability

A unique EBM failure mode: if the electron beam impinges on loose powder with insufficient pre-sintering, powder particles acquire negative charge, repel each other electrostatically, and "smoke" — a cloud of charged powder expands into the vacuum chamber, scattering the beam and potentially shorting the high-voltage supply (60 kV). The fix: a low-power, defocused beam pre-sinters each layer before the melt scan, creating a weakly bonded conductive network that prevents charge accumulation.

5.3 EBM Applications

EBM's niche is high-temperature alloys that benefit from the vacuum environment and elevated preheat:

6. Directed Energy Deposition (DED)

DED blasts metal powder (or wire) through a deposition nozzle into a laser or electron beam melt pool. Unlike LPBF, DED adds material to an existing substrate — making it ideal for repair, feature addition, and large-scale near-net-shape builds.

6.1 Clad Geometry Model

For a single DED track, the clad cross-sectional area A_c depends on the powder catchment efficiency \eta_p and the mass flow rate \dot{m}:

where \rho is material density and v is traverse speed. Typical \eta_p = 0.4–0.7 (40–70% of powder actually melts into the clad; the rest bounces off or is carried away by shielding gas). Width-to-height ratio of 3:1 to 5:1 is typical for a well-formed clad bead.

The dilution ratio D — the fraction of the clad that is remelted substrate — is controlled by laser power:

Optimal D \approx 0.10–0.30 for repair applications: enough for metallurgical bonding but not so much that substrate chemistry dilutes the clad alloy properties. For multi-layer builds, D \approx 0.30–0.50 ensures interlayer fusion.

6.2 DED vs LPBF: When to Use Which

Criterion · LPBF · DED

Resolution · 0.1–0.3 mm · 1–3 mm

Build rate · 5–20 cm³/hr · 50–500 cm³/hr

Max part size · 250–500 mm · 1–5 meters

Surface finish (Ra) · 5–15 μm · 20–50 μm

Multi-material · No · Yes (multiple hoppers, graded transitions)

Repair capability · No · Yes (deposit onto existing part)

Equipment cost · 400K–1.5M · 500K–2.5M

Cost per kg (Ti-6Al-4V) · 500–1,500 · 200–800

DED's killer applications:

7. Binder Jetting for Metals

Binder jetting sidesteps the thermal physics of LPBF entirely. An inkjet printhead selectively deposits a polymer binder onto a powder bed, layer by layer. The "green" part is then cured, depowdered, and sintered in a furnace — where it shrinks 15–20% isotropically to full density.

7.1 Shrinkage Compensation

The sintering shrinkage S (linear) is:

Typical green density is 50–60% (from powder packing + binder), and sintered density is 97–99%. For 316L: \rho_{\text{green}} \approx 4.0 g/cm³, \rho_{\text{sintered}} \approx 7.85 g/cm³, giving S \approx 0.18 (18% linear shrinkage). The part must be scaled up by 1/(1-S) \approx 1.22\times in all dimensions.

The challenge is anisotropic shrinkage: gravity, friction with the setter plate, and non-uniform green density cause different shrinkage in X/Y vs. Z. Parts over 100 mm in any dimension typically need trial builds and shrinkage factor iteration to hold ±0.5% dimensional tolerance.

Binder jetting's advantages: no supports needed (the powder bed supports itself), high throughput (800 × 400 mm bed produces hundreds of small parts simultaneously), and lower cost per part for production volumes above 10,000/year. The tradeoff: post-sintering surface finish (R_a = 5–10 μm), residual porosity (1–3%), and dimensional tolerance limitations.

8. Surface Roughness and Post-Processing

8.1 Surface Roughness Physics

As-built LPBF surface roughness arises from three superimposed mechanisms:

  1. Staircase effect: Layer discretization of inclined surfaces. For a surface at angle \theta from horizontal and layer thickness t:

At \theta = 30° and t = 40 μm: R_a \approx 8.7 μm.

  1. Partially melted powder adhesion: Powder at the melt pool boundary that sinters but doesn't fully melt, sticking to the surface. This is the dominant mechanism on down-facing surfaces (overhangs), where the laser melts into loose powder, and on vertical walls, where the melt pool edge contacts adjacent powder. Typical contribution: 5–15 μm Ra.
  1. Ball / spatter formation: Melt pool instabilities eject droplets that land on the powder bed and are incorporated into subsequent layers. Larger spatter (50–200 μm) becomes surface defects.

Total as-built roughness: R_a = 5–20 μm on upward-facing surfaces, 15–40 μm on downward-facing surfaces. For reference, a machined surface is R_a = 0.8–3.2 μm.

8.2 Post-Processing Chain

The standard post-processing workflow for production LPBF parts:

  1. Stress relief (annealing): 600–800°C for steels, 700–800°C for Ti-6Al-4V, 1,000–1,100°C for Inconel 718. Duration: 1–4 hours depending on section thickness. Inert atmosphere (argon or vacuum) to prevent oxidation.
  1. Wire EDM part removal: Cut from build plate. Leave 0.5–1.0 mm stock for machining.
  1. Support removal: Manual (pliers, chisel, bandsaw) or CNC. Fragile lattice structures require care.
  1. HIP (Hot Isostatic Pressing): 900–1,200°C at 100–200 MPa argon pressure for 2–4 hours. Collapses internal porosity (keyhole and lack-of-fusion), density → 99.9%+. The pressure P_{\text{HIP}} required to collapse a pore of radius r in a material with yield strength \sigma_y is approximately:

For Ti-6Al-4V at 920°C (\sigma_y \approx 50 MPa), collapsing a 50 μm pore: P \approx 25 MPa — within typical HIP capability.

  1. Machining: CNC milling/turning of functional surfaces. 0.2–0.5 mm material removal.
  1. Surface finishing: Abrasive flow machining (AFM) for internal channels, shot peening for fatigue improvement, electropolishing for medical implants.

8.3 HIP Parameter Closure

HIP simultaneously eliminates porosity and modifies microstructure. The Ashby HIP maps for a given alloy define the combination of temperature, pressure, and time needed to achieve full densification — and whether that combination also causes undesirable grain growth. For Ti-6Al-4V, the standard aerospace HIP cycle (920°C / 100 MPa / 2 hr) was chosen to balance pore closure with avoiding excessive \beta grain growth.

9. Fatigue Behavior and Defect Tolerance

Fatigue is the Achilles' heel of as-built LPBF. Unlike wrought material, where the fatigue limit is controlled by the intrinsic microstructure, LPBF fatigue is dominated by process-induced defects.

9.1 Kitagawa-Takahashi Diagram for AM

The Kitagawa-Takahashi diagram plots fatigue strength \sigma_f vs. defect size \sqrt{\text{area}}. For short cracks/defects, the fatigue limit is controlled by the intrinsic material fatigue limit \sigma_{w0} (the slip-band-controlled endurance limit). For long cracks, the threshold stress intensity factor \Delta K_{th} controls:

where Y is the geometry factor (~0.65 for surface defects, ~0.5 for internal defects). The transition occurs at the critical defect size:

For Ti-6Al-4V (\Delta K_{th} \approx 4 MPa√m at R = 0.1, \sigma_{w0} \approx 550 MPa for HIP'd material): \sqrt{\text{area}_c} \approx 50 μm. Defects larger than this — and as-built LPBF typically has lack-of-fusion defects 50–200 μm — reduce the fatigue limit proportionally.

The post-processing strategy is clear: HIP to close internal defects, then surface finishing (machining + shot peening) to remove surface defects and introduce compressive residual stress. Together, these can bring LPBF Ti-6Al-4V HCF strength from 200–350 MPa to 500–650 MPa — approaching wrought values.

9.2 NDT for AM Parts

Industrial CT (X-ray computed tomography) is the primary NDT method for production AM parts. Resolution limits:

The detectability threshold is ~3× the voxel size, so a 15 μm voxel scan can reliably detect pores ≥45 μm. This misses some fatigue-critical defects (which can be 20–40 μm), so process qualification relies on statistical sampling — batch builds with witness coupons that are destructively tested.

10. Powder Management and Reuse

Gas-atomized powder for LPBF costs 50–300/kg depending on alloy (AlSi10Mg at the low end, Ti-6Al-4V at 150–250, Inconel 718 at 80–150, CoCr at 80–120). With typical build efficiencies of 5–20% (only 5–20% of the powder in the build volume is actually melted; the rest is recovered), powder reuse economics dominate variable cost.

10.1 Powder Degradation Mechanisms

With each reuse cycle, powder degrades:

  1. Oxygen pickup: Spatter particles oxidize in the build chamber and mix back into the recovered powder. Ti-6Al-4V oxygen content can increase from 0.08% (virgin) to 0.15–0.20% after 20 reuses — embrittling the material.
  2. Particle size distribution shift: Fine particles are preferentially consumed (higher surface area, faster melting) and also preferentially lost during sieving. D50 can increase from 30 μm (virgin) to 35–40 μm after 15 reuses.
  3. Satelliting: Small particles fuse to larger ones during the thermal spray of nearby melting, creating irregular shapes that reduce flowability.
  4. Morphology degradation: Spherical particles become increasingly irregular, reducing apparent density and flowability.

10.2 Reuse Strategies

Most production shops use a "top-up" strategy: sieve recovered powder (63 μm mesh for LPBF), blend with 20–40% virgin powder to maintain chemistry and PSD targets, and run chemical analysis every 5 builds. Critical alloys (Ti-6Al-4V for aerospace, Inconel for rotating parts) are typically restricted to 5–15 reuses. Non-critical applications (tooling, prototyping) can stretch to 30+ reuses with proportionally more virgin top-up.

11. Design for Metal AM (DfAM)

11.1 Topology Optimization Constraints

Topology optimization for LPBF must respect manufacturing constraints that traditional TO algorithms ignore:

Commercial TO software (nTopology, Altair OptiStruct, Autodesk Fusion Generative Design) now supports these constraints natively.

11.2 Lattice Structures

LPBF enables engineered lattice structures for lightweighting, energy absorption, and biomedical osseointegration. Key lattice types:

for stretching-dominated behavior at relative densities above 0.15. Below that, bending dominates and the exponent approaches 3.

11.3 Self-Supporting Internal Channels

Circular channels require supports above ~8 mm diameter for horizontal orientation. Tear-drop or diamond cross-sections eliminate the flat roof and can be built unsupported at any diameter. Conformal cooling channels in injection mold inserts use this principle — 5–10 mm teardrop channels follow the cavity contour, impossible to achieve with drilled straight lines. The thermal benefit is real: conformal cooling reduces cycle time by 20–40% and improves part quality by maintaining uniform mold temperature.

12. Production Economics

12.1 Cost per Part Model

The total cost per LPBF part breaks down as:

where:

For a typical Ti-6Al-4V aerospace bracket (200 g, 15-hour build, 4 parts per build plate):

Cost Element · Per Part

Material (Ti-6Al-4V, 200/kg, 15% powder loss) · 46

Machine amortization (EOS M290, 80/hr) · 300

Labor (setup + support removal) · $60

Post-processing (stress relief + WEDM + machining) · $120

QA (CT scan + tensile coupons, amortized) · $150

Total · $676

The same bracket machined from billet: 350–500. LPBF is more expensive in this case — but if the bracket was topology-optimized to 120 g (40% lighter, saving 80 g × 2,000/kg fuel cost over the aircraft's life = $160), the lifecycle economics flip. For complex geometries that CAN'T be machined (conformal-cooled inserts, lattice implants), LPBF is the only option regardless of cost.

12.2 Build Rate Economics

LPBF build rate scales approximately as:

For an EOS M300-4 (4 × 1,000 W lasers): \dot{V} \approx 4 \times 1,000 / 60 \approx 67 mm³/s = 240 cm³/hr. A single-laser M290: \dot{V} \approx 370 / 60 \approx 6.2 mm³/s = 22 cm³/hr. The 4-laser system is 11× faster — but with a 1.5M price tag vs. 600K for the single-laser system, the crossover depends on utilization.

The economic break-even for multi-laser systems is around 4,000–5,000 build hours/year (50–60% utilization for a 24/7 operation). Below that, a fleet of single-laser machines provides better flexibility and redundancy.

13. The Indian Metal AM Ecosystem

India's metal AM installed base has grown from ~15 systems in 2018 to ~80+ in 2025, concentrated in:

The Indian powder supply chain is also maturing. While most shops import gas-atomized powder from AP&C (Canada), Carpenter (US), or TLS (Germany), Indian atomizers (Intech Additive, Objectify) now produce 316L, maraging steel, and CoCr powder at 60–70% of import cost.

13.1 Cost Comparison: Indian vs. Global

Service · India (₹/cm³) · USA (₹/cm³ equivalent) · Europe (₹/cm³ eq.)

LPBF Ti-6Al-4V · ₹80–150 · ₹200–500 · ₹180–450

LPBF AlSi10Mg · ₹40–80 · ₹100–200 · ₹90–180

LPBF 316L · ₹35–60 · ₹90–180 · ₹80–160

DED Ti-6Al-4V · ₹50–100 · ₹150–300 · ₹120–250

Binder Jet 316L (sintered) · ₹20–40 · ₹60–120 · ₹50–100

Indian metal AM is globally competitive on price, propelled by lower labor costs for support removal and post-processing, which constitute 30–50% of total part cost. The bottleneck is capacity — the total Indian LPBF installed base is still smaller than a single GE Additive factory in Ohio.

14. Future Directions

Several developments will reshape metal AM in the 2026–2030 timeframe:

Multi-laser, large-format LPBF: The EOS M400-4 (4 × 1,000 W, 400 × 400 × 400 mm) and SLM Solutions NXG XII 600 (12 × 1,000 W, 600 × 600 × 600 mm) bring build rates of 300–1,000 cm³/hr. At these rates, metal AM becomes cost-competitive with casting for production volumes of 5,000–20,000 parts/year.

In-situ monitoring and closed-loop control: Melt pool monitoring via photodiodes and high-speed cameras, combined with machine learning classifiers that detect conduction-to-keyhole transition in real time, will reduce the need for post-build CT scanning. GE's "process compensated" system already adjusts laser power scan-by-scan based on melt pool emission.

Copper and refractory metals: The shift from 1,064 nm fiber lasers to 532 nm (green) and 450 nm (blue) wavelengths dramatically improves copper absorptivity (from 3% to 40–50%), enabling LPBF of pure copper for heat exchangers, induction coils, and rocket combustion chambers. Launcher's E-2 engine uses a copper alloy LPBF combustion chamber with integral cooling channels.

Multi-material LPBF: Systems with multiple powder hoppers and selective deposition (Aerosint, now part of HP) enable functionally graded transitions between alloys — e.g., a rocket nozzle with a copper liner (high conductivity) and Inconel jacket (high strength at temperature), printed as a single monolithic part.

Conclusion

Metal additive manufacturing sits at the intersection of laser physics, metallurgy, and industrial economics. The Rosenthal equation governs the melt pool. Marangoni convection governs the track shape. The cooling rate — 10^6 K/s — governs the microstructure, producing an \alpha' lath 500 nm wide where wrought material would show 10 μm equiaxed grains. The inherent strain from those thermal gradients produces 400 MPa residual stress, and your cantilever deflection is approximately \sigma_r L^2 / (Et).

None of this is black magic. It is physics — precise, predictable, and increasingly controllable. The manufacturers who understand the physics will be the ones who turn metal AM from a prototyping curiosity into a competitive production advantage.

References (selected):

  1. DebRoy, T., et al. "Additive manufacturing of metallic components — Process, structure and properties." Progress in Materials Science, 92, 112–224 (2018).
  2. Zhao, C., et al. "Real-time monitoring of laser powder bed fusion process using high-speed X-ray imaging and diffraction." Scientific Reports, 7, 3602 (2017).
  3. Thijs, L., et al. "A study of the microstructural evolution during selective laser melting of Ti-6Al-4V." Acta Materialia, 58(9), 3303–3312 (2010).
  4. Cunningham, R., et al. "Keyhole threshold and morphology in laser melting revealed by ultrahigh-speed x-ray imaging." Science, 363(6429), 849–852 (2019).
  5. Wang, Y.M., et al. "Additively manufactured hierarchical stainless steels with high strength and ductility." Nature Materials, 17, 63–71 (2018).
  6. Herzog, D., et al. "Additive manufacturing of metals." Acta Materialia, 117, 371–392 (2016).
  7. ISO/ASTM 52900:2021 — Additive manufacturing — General principles — Fundamentals and vocabulary.
  8. EOS GmbH. "Material Data Sheet: EOS Titanium Ti64." (2024).
  9. GE Additive. "Direct Metal Laser Melting: Process Qualification for Aerospace." (2023).
  10. Indian 3D Printing Network. "India AM Market Report 2025." (2025).

More FabFlow blog posts