Additive manufacturing (3D printing)

In short

Additive manufacturing (AM) builds a component directly from its 3D CAD model by adding material layer upon layer, instead of cutting it away (subtractive) or filling a cavity (formative). Because no part-specific tooling is needed, it is unbeatable for one-offs, prototypes, patient-specific implants and highly complex shapes such as internal cooling channels and lattices. Standard Indian syllabi (AICTE "Additive Manufacturing"/Rapid Prototyping electives) and GATE-ME ask the seven ISO/ASTM process families, the CAD-to-part chain, STL and slicing, support and orientation decisions, the staircase effect, and build-time and cost estimation. It matters commercially in India for aerospace and space (ISRO, HAL, Godrej Aerospace, Wipro 3D, Agnikul), automotive prototyping, tooling and dental/orthopaedic implants, and is backed by MeitY's National Strategy on Additive Manufacturing (2022).

Explained

What AM is, and the CAD-to-part chain

• Definition (ISO/ASTM 52900): the process of joining materials to make parts from 3D model data, usually layer upon layer, as opposed to subtractive and formative manufacturing.
• Older names: rapid prototyping (when output was only a visual/fit model), rapid tooling (moulds, dies, fixtures), rapid manufacturing / direct digital manufacturing (end-use parts).
• Eight generic steps: (1) 3D CAD model → (2) convert to a tessellated file (STL, or AMF/3MF which also carry colour, material and units) → (3) transfer and repair the file (fix non-manifold edges, flipped normals, gaps) → (4) choose build orientation and generate supports → (5) slice into layers and generate machine code/scan paths → (6) build on the machine → (7) remove part, de-powder/de-support → (8) post-process (cure, stress-relieve, machine, finish) and inspect.
• STL file: the surface is approximated by triangular facets, each stored as three vertices plus an outward unit normal; curved surfaces therefore carry a chordal (faceting) error. Finer tessellation = smaller error but much larger files.
• Slicing fixes layer thickness, contour (shell) paths and infill/hatch paths; slicing software also sets raster angle, shell count, infill pattern and density.

The seven process families of ISO/ASTM 52900

1. Material extrusion (MEX / FDM, FFF): a heated nozzle extrudes thermoplastic filament (PLA, ABS, PETG, PC, nylon, PEEK, ULTEM) road by road onto a heated bed. Cheapest; layer 0.1–0.3 mm; anisotropic because adjacent roads are only partly fused; needs support for overhangs.
2. Vat photopolymerisation (VPP / SLA, DLP, LCD-MSLA): a UV laser (SLA) or a projected/masked image (DLP, LCD) cures liquid photopolymer layer by layer. Best surface finish and fine features (layer 25–100 µm); parts need draining, IPA washing and UV post-curing; resins are brittle and UV-ageing.
3. Powder bed fusion (PBF): a recoater spreads a thin powder layer; a heat source fuses the cross-section.
   • Polymer: SLS (laser, nylon PA12), MJF (fusing agent + IR lamp) — self-supporting in the un-sintered powder cake, so no support structures.
   • Metal: LPBF (also SLM/DMLS), layer 20–60 µm, under argon or nitrogen; EBM, electron beam in vacuum at 600–1,000 °C, layer 50–100 µm, lower residual stress, rougher surface.
4. Binder jetting (BJT): an inkjet head prints liquid binder onto powder (metal, ceramic, or sand for foundry moulds and cores). Green part is fragile; metal/ceramic parts are de-bound and sintered, which gives large, direction-dependent shrinkage; sand moulds are used as-printed.
5. Material jetting (MJT / PolyJet, NanoParticle Jetting): print-head jets droplets of photopolymer (or wax, or metal nanoparticle ink) that are UV-cured instantly; multi-material and full colour, 16–32 µm layers, needs a removable support resin.
6. Directed energy deposition (DED / LENS, LMD, laser cladding, WAAM): powder or wire is fed into a melt pool created by a laser, electron beam or arc, usually on a 3–5 axis head or robot. Poor resolution (beads of 0.5–3 mm) but very high deposition rate — used for large near-net shapes, cladding and repair of turbine blades, dies and shafts; always needs finish machining.
7. Sheet lamination (SHL / LOM, UAM): sheets of paper, polymer or metal foil are bonded (adhesive or ultrasonic welding) and the contour cut by laser or blade; little used today except ultrasonic AM for dissimilar metals.

Process physics and the parameters you are examined on

• Layer thickness t: the master trade-off — halving t halves the staircase error and roughly doubles the number of layers and the build time.
• For FDM: nozzle diameter, road width w, layer height h, print speed v, extrusion temperature, bed temperature, raster/infill angle (±45° alternating is common), shell count, infill density, cooling.
• For LPBF/SLS: laser power P, scan speed v, hatch spacing hd, layer t, scan strategy (stripes, chequerboard, island, rotated 67° per layer), spot size, preheat, oxygen level (<0.1 % for Ti), powder PSD (typically 15–45 µm) and flowability.
• Volumetric energy density combines the four main laser parameters; too low gives lack-of-fusion porosity, too high gives keyhole porosity, spatter and evaporation of volatile alloying elements (Zn, Mg).
• Melt-pool/balling: insufficient wetting breaks the track into droplets; excess scan speed causes humping and poor layer bonding.
• Residual stress: steep thermal gradients make each layer contract against the solid below, causing warping, part–plate separation, cracking in crack-prone alloys and distortion after cutting off the plate. Reduced by plate preheat, island scanning, short vectors, and mandatory stress-relief heat treatment before removing the part from the build plate.
• Anisotropy: Z-direction (build-direction) strength and elongation are lower than X-Y in nearly every process, because inter-layer bonds, lack-of-fusion defects and columnar grains align with Z. Specimen orientation is reported using the ISO/ASTM 52921 nomenclature (e.g. XY, ZX).

Design for AM (DfAM), orientation and supports

• Freedoms AM gives: conformal/internal cooling channels, topology-optimised organic shapes, lattice and honeycomb infill for stiffness-to-weight, part consolidation (an assembly of many parts printed as one, removing joints, fasteners and leak paths), mass customisation at no tooling penalty, graded and multi-material structures.
• Constraints to design to: minimum wall thickness (about 0.4–0.8 mm in LPBF, ~2 perimeters ≈ 0.8 mm in FDM); unsupported overhangs; trapped powder/resin (give drain and de-powdering holes); small holes print undersize and are better drilled; threads are usually machined or inserted.
• Self-supporting rule: surfaces inclined more than about 45° to the build platform generally need no support; shallower overhangs, bridges and down-facing holes do. Chamfer or fillet instead of printing a sharp overhang.
• Supports do four jobs: hold overhangs, anchor the part against warping, conduct heat out of the melt region, and resist recoater forces. They cost material, time and a post-processing operation, and leave witness marks — so orientation is chosen to minimise them.
• Orientation decides build height (hence time), staircase error on critical surfaces, support volume, where the best-finish surfaces lie, and the direction of the weakest (Z) strength — never put the main tensile load along Z.
• Cylindrical features: a hole axis parallel to Z prints round; parallel to the platform prints as an ellipse with a sagging roof.

Accuracy, defects and post-processing

• Staircase (stair-step) effect: inclined and curved surfaces appear as steps of the layer height — the dominant geometric error. Mitigated by thinner layers, adaptive slicing (thin layers only where the surface is shallow) and machining.
• Other error sources: STL chordal error, shrinkage and curl, beam/road width compensation, support witness marks, powder adhesion on down-skins, and Z-axis "first-layer" error.
• Typical as-built roughness: FDM Ra 10–25 µm; LPBF Ra 5–20 µm (down-skins worse); EBM Ra 20–35 µm; SLA Ra 1–5 µm. Any sealing, bearing or fatigue surface must be finish machined or polished.
• Common defects: porosity (gas, lack-of-fusion, keyhole), un-melted particles, delamination, warping, cracking, swelling/elephant foot, inclusions from contaminated or over-recycled powder.
• Post-processing chain (metal LPBF): de-powder and sieve recovered powder → stress relieve on the plate → cut from plate (wire EDM/bandsaw) → remove supports → HIP (hot isostatic pressing) to close internal porosity → solution/age heat treatment → machine datums and critical surfaces → shot peen/tumble/abrasive-flow or electro-polish → NDT (CT scan, dye penetrant) and CMM inspection.
• Polymer post-processing: support/powder removal, bead blasting, vapour smoothing (ABS in acetone), UV post-cure, dyeing, epoxy infiltration.

Economics and comparison with conventional routes

• Cost per part = machine cost (hourly rate × build time ÷ parts per build) + material (part + support + lost powder) + pre-processing/labour + post-processing + inspection. AM cost is almost flat with quantity; moulding/casting/forging costs fall with quantity because tooling is amortised — so a break-even quantity exists.
• AM wins on: very low volume, high geometric complexity, short lead time (days, not the weeks a die takes), high buy-to-fly material saving, spares for obsolete parts, and customisation.
• AM loses on: unit cost at volume, size of build envelope, surface finish and tolerance straight off the machine, material cost (metal powder is many times the price of bar stock), limited qualified material list, and anisotropic, scatter-prone properties needing qualification.
• Complexity is "free", but volume and height are not — build time scales with the part's bounding height (layers) and its volume (scan/extrusion), not with how intricate it is.

Standards, safety and Indian practice

• Terminology and framework: ISO/ASTM 52900 (terms and classification), 52901 (requirements for purchased AM parts), 52902 (test artefacts), 52910/52911 (design for AM), 52920/52930 (qualification of producers), 52921 (coordinate systems and test-specimen orientation); material/process specs in the ASTM F3049, F3055 (IN718), F3056, F3301, F2924 (Ti-6Al-4V) family. BIS has an additive manufacturing sectional committee and has been adopting the ISO/ASTM 52900 series as Indian Standards — check the current BIS catalogue, as adoptions are still being added.
• Medical AM in India: patient-specific implants and surgical guides are regulated by CDSCO under the Medical Devices Rules, 2017, with device classification and QMS (ISO 13485) requirements; aerospace parts additionally need customer/AS9100 and Nadcap-style process qualification.
• Safety (OSH Code, 2020 — in force 21 November 2025, replacing the Factories Act, 1948 — and the state factory rules saved under it; ISO/ASTM 52931/52932 guidance): fine Ti, Al and Mg powders are pyrophoric and form explosive dust clouds — handle in inert atmosphere, use earthed conductive equipment, wet/HEPA vacuum (never a normal shop vacuum), P3/respirator and nitrile gloves, no compressed-air blow-down; photopolymer resins are skin sensitisers; laser class 1 enclosure interlocks; fume and ultrafine-particle extraction for FDM of ABS/PC.
• Policy: MeitY's National Strategy on Additive Manufacturing (February 2022) sets indicative targets — roughly a 5 % share of the global AM market, 50 India-specific materials/technologies, 100 new AM startups, 500 new products and about 1 lakh trained workers — supported by AM centres of excellence (e.g. the National Centre for Additive Manufacturing, Hyderabad, CMTI, CIPET, IITs). Treat the numbers as policy targets, which may have been revised.

Key points

• Additive manufacturing - joining material layer upon layer from 3D model data (ISO/ASTM 52900); opposite of subtractive machining.
• Seven process families - material extrusion, vat photopolymerisation, powder bed fusion, binder jetting, material jetting, directed energy deposition, sheet lamination.
• STL - triangle-facet surface file; no units, no colour, no tolerance; introduces chordal error on curves.
• Slicing - converts the oriented solid into layer contours and hatch/raster paths plus machine code.
• Layer thickness - controls staircase error, Z-strength and build time; the single most important parameter.
• Staircase effect - stepped approximation of inclined surfaces; cusp height = t·cos θ for a surface at θ to the platform.
• 45° rule - overhangs steeper than ~45° from the platform are self-supporting; shallower ones need support.
• Supports - anchor, support, conduct heat; cost time, material and a removal operation, and leave witness marks.
• SLS/MJF need no supports - the loose powder cake supports the part; metal PBF still needs them for heat and stress.
• Volumetric energy density - E = P/(v·hd·t); too low → lack-of-fusion porosity, too high → keyhole porosity.
• Residual stress - stress-relieve before cutting the part off the build plate, or it will distort.
• Anisotropy - Z (build direction) strength and ductility are lowest; never align the main tensile load with Z.
• HIP - hot isostatic pressing closes internal porosity and raises fatigue life of metal AM parts.
• Binder jetting shrinks - green parts sinter with ~12–20 % linear shrinkage; CAD must be scaled up.
• DED/WAAM - highest deposition rate, worst resolution; the route for large near-net shapes and repair/cladding.
• Break-even quantity - AM is cheaper below it, tooling-based processes above it; AM cost is nearly volume-independent.
• Complexity is free, height and volume are not - build time depends on layer count and material deposited.
• Buy-to-fly - ratio of input material to finished part mass; AM slashes it versus machining a forging.

Formulas and examples

Example 1 - FDM build time, filament and cost. A PLA bracket of solid volume 60 cm³ and height 80 mm is printed with road width w = 0.5 mm, layer height h = 0.2 mm, speed v = 50 mm/s, non-printing time 3 s per layer. PLA density 1.24 g/cm³, filament 1.75 mm, spool ₹1,200/kg.

• Q = 0.5 × 0.2 × 50 = 5 mm³/s; V = 60,000 mm³ → extrusion time = 60,000/5 = 12,000 s = 3.33 h.
• N = 80/0.2 = 400 layers → non-print time = 400 × 3 = 1,200 s = 0.33 h. Total ≈ 3.67 h (3 h 40 min).
• Filament area = 0.7854 × 1.75² = 2.405 mm² → length = 60,000/2.405 ≈ 24.9 m.
• Mass = 60 × 1.24 = 74.4 g → material cost = 0.0744 × 1,200 ≈ ₹89. (At a machine rate of ₹150/h, machine cost ≈ 3.67 × 150 ≈ ₹550, so time, not material, dominates.)

Example 2 - LPBF energy density and build time. SS 316L on LPBF: P = 200 W, v = 800 mm/s, hatch hd = 0.12 mm, layer t = 0.03 mm. Part volume 50 cm³, height 40 mm, recoat 9 s/layer.

• Build rate Q = 800 × 0.12 × 0.03 = 2.88 mm³/s = 2.88 × 3,600 = 10,368 mm³/h ≈ 10.4 cm³/h.
• E = 200/2.88 = 69.4 J/mm³; linear energy = 200/800 = 0.25 J/mm.
• Scan time = 50,000/2.88 = 17,361 s = 4.82 h; N = 40/0.03 = 1,334 layers → recoat = 1,334 × 9 = 12,006 s = 3.34 h. Total ≈ 8.2 h, of which 41 % is recoating — which is why parts are nested flat and built in batches.
• If v is doubled to 1,600 mm/s to halve the time, E falls to 34.7 J/mm³ — likely lack-of-fusion porosity unless P is also raised.

Example 3 - staircase effect and STL error. Layer thickness 0.10 mm; a conical surface meets the platform at θ = 30°.

• c = 0.10 × cos 30° = 0.10 × 0.866 = 0.0866 mm (86.6 µm); step width s = 0.10/tan 30° = 0.10/0.5774 = 0.173 mm; approximate Ra ≈ 86.6/4 ≈ 22 µm.
• Reducing t to 0.05 mm gives c = 43.3 µm but doubles the layer count and roughly the build time.
• For a Ø50 mm (R = 25 mm) cylinder tessellated to δ = 0.01 mm: n ≈ π√(25/0.02) = π × 35.36 ≈ 111, rounded up to 112 facets around the circle.

Example 4 - break-even against injection moulding. A 40 g nylon housing: mould ₹6,00,000 plus ₹35 per moulded part; SLS cost ₹520 per part with no tooling.

• N* = 6,00,000 / (520 − 35) = 6,00,000/485 ≈ 1,237 parts.
• Below ~1,237 pieces SLS is cheaper (and available in days); above it, moulding wins and the gap widens fast — at 10,000 parts moulding costs ₹9.5 lakh against ₹52 lakh for SLS.

Example 5 - binder jetting sinter shrinkage. A 316L green part is at 60 % of theoretical density and sinters to 98 %.

• k = (0.60/0.98)^1/3 = (0.6122)^1/3 = 0.8492 → linear shrinkage ≈ 15.1 %.
• For a final 100.0 mm length, the green (printed) length must be 100/0.8492 = 117.8 mm, i.e. the CAD model is scaled by 1.178 in that direction. Shrinkage differs with build direction and with support of the part during sintering, so setters and test coupons are used.

Where it is used

• Space and aerospace: ISRO's PS4 rocket engine built by LPBF at Wipro 3D, Bengaluru (part count and lead time cut sharply); Agnikul Cosmos' Agnilet single-piece 3D-printed semi-cryogenic engine, flown on Agnibaan SOrTeD in 2024; HAL and Godrej Aerospace for brackets, ducts and engine hardware.
• Automotive and heavy engineering: prototype and validation parts, assembly jigs and fixtures, conformally cooled injection-mould inserts, low-volume spares at Tata Motors, Maruti Suzuki, Mahindra, Bharat Forge, Ashok Leyland and tier-1 suppliers; Indian Railways uses printed spares and sand-printed moulds.
• Foundry: binder-jetted sand moulds and cores, and SLA/wax patterns for investment casting — removes pattern-making entirely for one-off castings.
• Medical and dental: titanium cranioplasty plates and patient-specific orthopaedic implants, surgical guides, aligners and crowns (service bureaus such as Anatomiz3D, Imaginarium, Osteo3d), regulated by CDSCO under the Medical Devices Rules, 2017.
• Industrial maintenance: laser cladding and DED repair of dies, shafts, turbine blades and mill rolls — restores worn geometry at a fraction of new-part cost.
• Job roles: AM/DfAM design engineer, process (build preparation) engineer, materials and powder quality engineer, post-processing and NDT engineer, applications engineer with machine builders (Intech Additive, EOS, Wipro 3D, Altem, Divide By Zero) and service bureaus; architecture, jewellery, footwear and consumer products also recruit.

Interview questions

• Q: What exactly does "additive manufacturing" mean, and how does it differ from rapid prototyping? A: It is joining material layer upon layer from 3D model data; rapid prototyping is the historical, narrower use of the same processes to make non-functional models, whereas AM now includes end-use production parts.
• Q: Why is the STL format still used, and what are its limitations? A: It is simple and universal — triangular facets with normals — but it carries no units, tolerance, colour or material, approximates curves with chordal error, and is easily corrupted (flipped normals, gaps); AMF and 3MF fix most of this.
• Q: How do you choose build orientation? A: Balance four things: keep the main tensile load out of the Z direction, put critical surfaces where the staircase error is smallest, minimise support volume on cosmetic or inaccessible faces, and minimise the build height to cut layer count and time.
• Q: What is the 45° rule? A: Surfaces inclined more than about 45° to the build platform are generally self-supporting; shallower overhangs sag or dross and need support or a redesign with a chamfer or fillet.
• Q: Why do metal powder bed parts need supports even though the powder bed surrounds them? A: Loose metal powder cannot carry the load or conduct heat; supports anchor the part against large residual-stress warping and conduct heat from the melt pool to the plate.
• Q: Why are AM metal parts stress-relieved before being cut off the build plate? A: Layer-by-layer solidification locks in high residual stresses; releasing the part first lets it distort or crack, and the distortion cannot be undone.
• Q: What causes porosity in LPBF and how is it controlled? A: Too little energy density gives lack-of-fusion between tracks and layers; too much gives keyhole collapse and trapped gas. It is controlled by the P–v–hd–t parameter window, powder quality and shield-gas purity, and residual internal porosity is closed by HIP.
• Q: Why are 3D-printed parts anisotropic? A: Inter-layer bonds are weaker than the bulk material, defects and un-fused regions lie in the layer planes, and grains grow columnar along the build direction — so Z strength and elongation are lowest.
• Q: Which process would you pick for 50,000 identical plastic clips, and why? A: Injection moulding — AM cost is almost independent of quantity, so beyond the break-even quantity (typically a few hundred to a few thousand parts) tooling is far cheaper per piece.
• Q: How is build time estimated? A: Sum the deposition/scan time (total deposited volume divided by the volumetric build rate) and the per-layer overhead (recoating, exposure delays) multiplied by the number of layers, which equals part height divided by layer thickness.
• Q: What is DfAM, and name three things it lets you do that machining cannot? A: Design for additive manufacturing — designing to AM's freedoms and limits; it allows conformal internal cooling channels, topology-optimised and lattice structures, and consolidation of a multi-part assembly into one printed part.
• Q: What safety precautions apply to a metal AM shop? A: Treat fine Ti/Al powders as pyrophoric and explosive — handle under inert gas, earth all equipment, use wet or HEPA ATEX vacuums and respirators, never compressed air; plus laser-safe interlocked enclosures and fume extraction, under the OSH Code, 2020 (which replaced the Factories Act) and state factory rules.

Common mistakes

• Saying AM has "no design constraints" — overhang angle, minimum wall, trapped powder, build envelope, support access and anisotropy are all hard constraints.
• Forgetting that build time depends on part height (layer count × recoat time), not only on volume — a tall thin part can take longer than a short bulky one.
• Using N = H/t without rounding up, or mixing mm³/s with cm³/h in build-rate problems.
• Writing the cusp height as t·sin θ or t·tan θ; it is t·cos θ for a surface at θ to the platform (zero for a vertical wall, maximum for a near-horizontal one).
• Quoting handbook (wrought) properties for an as-built AM part, or ignoring the Z-direction knock-down and the need for heat treatment/HIP.
• Modelling a binder-jetted or MIM part at final size and forgetting the 12–20 % sinter shrinkage scale factor.
• Treating brochure tolerance claims as universal — machine brochure accuracy is for a calibration artefact, not for your part; and assuming SLS/MJF rules (no supports) apply to metal PBF.

Related topics

Powder metallurgy · CNC machining and CAM · Plastics processing and injection moulding · Casting · Metrology and quality control · Welding and joining