ACP Fusion (DFM) cheat sheet

Minimum feature size is bounded by the machine's spot/laser diameter and layer thickness; a feature narrower than the melt pool will not resolve, so keep walls at least 2-3x the beam diameter.

Self-supporting overhang rule for powder-bed fusion: surfaces angled more than about 45 degrees from horizontal (i.e., at least 45 degrees from the build plate) generally print without support.

Support-free design reduces material waste, post-processing labor, and residual stress; orient the part so critical surfaces are either vertical or upward-facing.

Trapped powder and trapped liquid escape paths must be designed in: add drain/vent holes of at least 1-2 mm to hollow volumes before printing.

Anisotropy: FDM parts are weakest along the Z (layer) axis; design load paths to run in-plane, and specify build orientation in the drawing.

Dimensional tolerance for metal AM is typically +/-0.1 mm to +/-0.5 mm depending on process and size; do not assign tight tolerances to as-built surfaces that will not be machined.

Thermal distortion scales with section thickness change; use uniform wall thickness and gradual transitions to avoid warping and residual stress concentration.

Design for post-processing: leave machining stock (0.5-1.5 mm) on critical mating surfaces and design access for support removal tools.

Volumetric energy density (VED) = laser power P / (scan speed v x hatch spacing h x layer thickness t); it is a comparative index, not a material property.

Laser powder-bed fusion key parameters: laser power, scan speed, hatch spacing, layer thickness, spot size, and scan strategy (stripe, island, chessboard).

Nesting/packing efficiency = total part cross-sectional area in a layer / build platform area; higher packing reduces cost per part but can raise thermal interaction between parts.

Build time estimate = number of layers x (layer scan time + recoating time); layer count = build height / layer thickness.

Support removal, heat treatment, and surface finishing are separate cost drivers; include them in the routing, not only in the print step.

Powder reuse: track reuse cycles and sieve powder between builds; oxygen content and particle size distribution are the key acceptance metrics.

Post-build heat treatment relieves residual stress and defines microstructure; the stress-relief cycle must occur before the part is removed from the build plate for thin geometries.

Machine calibration checks: laser power at the build plane, spot size, galvo accuracy, and recoater levelness; drift in any of these changes melt pool behavior.

Thermo-mechanical simulation couples a transient thermal solve (moving heat source) to a quasi-static mechanical solve to predict distortion and residual stress.

Inherent strain method replaces the full thermal history with an equivalent anisotropic strain applied per layer; it is much faster and used for full-part distortion prediction.

Mesh convergence check: refine the element size until predicted peak temperature or distortion changes by less than about 5 percent.

Thermal boundary conditions: convection and radiation to the chamber, plus conduction into the build plate; omitting plate conduction overpredicts part temperature.

Melt pool metrics from simulation: pool length/width ratio, cooling rate, and solidification gradient; these link process parameters to porosity and microstructure.

Support structures are modeled as effective thermal conductors, not just mechanical anchors; ignoring them biases distortion predictions near overhangs.

Validation loop: compare predicted distortion to measured coordinates on a test artifact (e.g., a cantilever or bridge) before trusting a production model.

Material property inputs must match the process: temperature-dependent conductivity, specific heat, and yield strength for the alloy and build direction.

AM data chain: CAD model -> STL or 3MF mesh -> sliced build file (e.g., machine-specific format) -> machine; each handoff is a versioned artifact.

STL stores only triangulated surface geometry with no units, material, or build orientation; 3MF carries units, color, material, and metadata and is preferred for collaboration.

File naming and revision control: every build file must trace to a released CAD revision and a specific machine and parameter set.

Build traveler/job ticket records part ID, material lot, powder reuse count, machine ID, operator, and post-processing steps for traceability.

Intellectual property: mesh files are easily copied; access control, encryption, and audit logging on the repository are standard controls.

Change management: a parameter or orientation change is a design change and requires re-validation of the affected part, not just a file overwrite.

Data volume: a full-resolution mesh or simulation result can be gigabytes; store compressed meshes plus the source CAD, and archive simulation inputs separately from results.

Digital thread linkage: connect the as-designed model, simulation results, build parameters, and inspection data by a common part and serial number.

Down-facing surfaces printed over powder or support have higher roughness (Ra often 10-20 micrometers or worse) than up-facing surfaces; mark which faces need machining.

Stair-stepping error depends on layer thickness and surface angle: a surface at angle theta from horizontal shows step height t / tan(theta).

Build orientation determines both support volume and anisotropy direction; it is a design decision that must be recorded, not left to the operator.

Cylindrical holes printed in the horizontal plane become elliptical due to gravity and melt-pool sag; design undersize or plan to ream them.

Minimum hole diameter for powder removal is roughly 1 mm for metal LPBF and larger for FDM; smaller channels risk trapping powder or failing to clear support.

Sharp internal corners concentrate residual stress and are hard to clean; use fillets and radii where geometry allows.

Threaded features are usually printed undersize and tapped afterward; printed threads below about M6 are unreliable in metal AM.

Cost per part = (machine rate x build time + material cost + post-processing cost + labor) / number of good parts per build.

Yield loss sources: porosity, delamination, warpage, support failure, and powder contamination; each has a distinct inspection signature.

In-process monitoring (melt pool imaging, layer photography) generates data per layer; define acceptance thresholds before the build, not after.

CT scanning detects internal porosity and trapped powder; it is slow and expensive, so reserve it for critical or first-article parts.

Dimensional inspection should reference the same datum scheme used in the CAD model; as-built surfaces may need to be machined before they can serve as datums.

Parameter qualification: a process parameter set is qualified for a material and machine combination; transferring it to another machine requires re-verification.

First-article inspection compares measured geometry to the nominal model and documents deviations; it is the gate before production release.

Distortion prediction compares simulated node positions after cooling and plate removal to nominal CAD; the difference field is the compensation input.

Pre-compensation: invert the predicted distortion and apply it to the nominal geometry before slicing; iterate until residual error is within tolerance.

Residual stress magnitude scales with thermal gradient and constraint; simulating plate removal releases constraint and reveals the final warped shape.

Support generation tools use overhang angle thresholds; simulation should check that chosen supports carry the thermal and mechanical load without fracturing.

Porosity prediction models use energy density and melt pool geometry (lack-of-fusion vs keyhole regimes) rather than a single scalar threshold.

Sensitivity studies vary laser power, scan speed, and hatch spacing to map a process window; the window edges are where defects appear.

Simulation credibility is documented by comparison to experiment: report the measured vs predicted error, not just a pass/fail statement.

Role separation: designer releases CAD, manufacturing engineer owns build files and parameters, quality owns inspection records; permissions should reflect these roles.

A build package includes the released CAD, mesh, slice file, parameter set, material certificate, and job ticket; missing any element breaks traceability.

Version control rules: never overwrite a released build file; issue a new revision with a change note referencing the affected part numbers.

Material traceability links powder lot and reuse count to each serialized part; this is required for aerospace and medical qualification.

Simulation data management: store input decks, solver version, and boundary conditions with results so a prediction can be reproduced.

Collaboration handoffs should be asynchronous and documented, with a defined approval state (draft, in review, released, obsolete) for each artifact.

Security controls for AM data include access logs, encryption at rest and in transit, and controlled distribution of machine parameter files.

Archive policy: retain build records for the product's service life plus the regulatory retention period; simulation results may be archived at reduced resolution.

Lattice relative density is the strut volume fraction; stiffness scales roughly with relative density for bending-dominated lattices and more steeply for stretch-dominated ones.

Minimum strut diameter is set by the process resolution; struts below about 2-3x the beam diameter may not form or may contain porosity.

Lattice cells need powder-escape paths; closed cells trap powder and are not manufacturable in powder-bed processes without vents.

Lattice-to-solid transitions should be graded to avoid stress concentration and thermal gradients at the interface.

Design lattices for the load case: bending-dominated (e.g., body-centered cubic) for energy absorption, stretch-dominated (e.g., octet) for stiffness.

Lattice geometry multiplies file size; use implicit or procedural representations rather than explicit triangle meshes where possible.

Powder particle size distribution controls flowability and layer quality; typical LPBF metal powders are 15-45 micrometers or 20-63 micrometers depending on the machine.

Recoating defects (streaks, short feeds) trace to powder condition, recoater wear, or platform leveling; check these before blaming the laser.

Inert gas atmosphere with controlled oxygen (often below 1000 ppm for reactive alloys) prevents oxidation and changes melt pool wetting.

Build plate adhesion: the first layers must be firmly attached; plate material and preheating affect first-layer success.

Powder handling is a safety issue: fine metal powders are combustible; use grounded equipment and inert handling per the safety data sheet.

Machine setup checklist: plate leveling, recoater condition, gas flow, filter condition, laser calibration, and parameter file verification.

Cooling rate during solidification controls grain size; faster cooling gives finer microstructure and higher strength in many alloys.

Thermal gradient G and solidification velocity V determine morphology: a high G/V ratio favors columnar growth, low ratio favors equiaxed grains.

Layer-by-layer thermal cycling causes in-situ tempering; the simulated thermal history must include reheating from subsequent layers.

Peak temperature above the alloy's liquidus confirms full melting; lack-of-fusion shows as regions below the melting point between tracks.

Heat source models: a volumetric (Goldak) source is more accurate than a point source for deep keyhole or conduction regimes.

Simulation time step must resolve the moving heat source; too coarse a step smears the thermal gradient and understates residual stress.

Neutral CAD exchange formats (STEP) preserve solid geometry; meshes (STL, 3MF) are lossy and should be regenerated from the released CAD, not edited downstream.

Units mismatch is a common failure: STL has no units, so the receiving system must be told whether the model is in millimeters or inches.

Parameter sets should be exported in a machine-readable file with units and a checksum, not transcribed by hand.

API or MES integration passes job status and serial numbers automatically; manual entry is the main source of traceability errors.

Metadata schema should include part number, revision, material, machine, parameter set ID, and operator for every build record.

Cross-site collaboration requires a common naming convention and a shared definition of release states to avoid building from draft files.

AM allows assemblies to be consolidated into one part; evaluate each joint removed for function, sealing, and serviceability before consolidating.

Consolidation eliminates fasteners and joints but can remove access for maintenance; check that wear parts remain replaceable.

Internal channels enabled by consolidation need support-free cross-sections (e.g., teardrop or diamond) to print without internal support.

Consolidation changes the inspection plan: an assembly of simple parts becomes one complex part requiring internal inspection such as CT.

Weight and stiffness trade-offs: removing joints reduces stress concentrators but may increase the volume that must be support-free.

Stress relief is performed before removing the part from the plate for thin or crack-sensitive geometries; the cycle depends on alloy and build.

Support removal methods: manual tools, CNC machining, EDM, or chemical dissolution for soluble supports; choose based on access and surface requirements.

HIP (hot isostatic pressing) closes internal porosity; it changes mechanical properties and must be reflected in the qualified parameter set.

Surface finishing options include machining, bead blasting, vibratory finishing, and electrochemical polishing; each changes final dimensions.

Heat treat and HIP are separate operations with their own records; link each to the part serial number in the traveler.

Inspection after post-processing must confirm that finishing did not remove functional material or close required channels.

Symmetry can reduce model size only when geometry, scan pattern, and thermal boundary conditions are all symmetric; scan strategy often breaks symmetry.

Element type matters: thermal elements need temperature degrees of freedom; structural elements need displacement; coupled analysis requires both.

Contact between part and build plate should be modeled with a thermal contact conductance, not a perfect bond, unless the plate is machined flat and preheated.

Initial condition is typically the preheat temperature of the plate and chamber, not room temperature.

Material model for the solid-liquid transition needs temperature-dependent properties through the melting range; a step change causes numerical instability.

Result interpretation: distinguish transient distortion during the build from final distortion after plate removal and cooling.

Inspection records must reference the nominal model revision and the datum scheme used; otherwise measurements cannot be compared across builds.

Nonconformance reports link a deviation to the affected serial numbers and the disposition (rework, accept, scrap).

Statistical process control on key dimensions and porosity detects drift before parts fail; define control limits from qualified builds.

Certificates of conformance and material certificates are part of the build record and must be retrievable by serial number.

Audit trail requirements: who changed what, when, and why, for CAD, parameters, and inspection results.

Retention and access rules should be defined per program; regulated industries require longer retention and controlled access.

As-built AM surfaces are not suitable as primary datums without machining; define datums on machined features where precision is required.

Tolerance stack-up must account for build orientation, shrinkage, and post-processing material removal.

Geometric tolerances (flatness, parallelism) on as-built surfaces should be loose; specify machining for tight callouts.

Shrinkage compensation is applied in the slice file, not the CAD; the CAD model should represent the final nominal geometry.

Hole and shaft fits: printed holes tend to be undersize; specify a machining allowance or design an interference-free fit for as-built.

Orientation choice trades build height (time) against support volume and surface quality on critical faces.

Support types: block, line, point, and tree; tree supports reduce material and contact area but may be less stiff.

Support contact points leave marks; place them on non-critical surfaces or add machining stock.

Overhang angle threshold is machine and material dependent; verify with a test artifact rather than assuming 45 degrees universally.

Nested parts should be spaced to allow recoater clearance and to limit thermal interaction; too-close parts can fuse or warp together.

Calibrate the heat source absorptivity and effective conductivity against measured melt pool dimensions from single-track or bulk samples.

Uncertainty in material properties (conductivity, absorptivity) often dominates prediction error; report a range, not a single value.

Mesh and time-step sensitivity together define numerical error; separate it from model-form error before comparing to experiment.

Calibration on one geometry may not transfer to another; validate on a second artifact with different overhang and section changes.

Document solver version and settings with results; a prediction without its inputs cannot be reproduced or defended.

Link each simulation run to the CAD revision, parameter set, and material lot it represents; otherwise the prediction cannot be applied to a specific build.

Store simulation inputs and outputs together; results alone cannot be re-run or audited.

Build files should carry a hash or checksum so the file used on the machine can be verified against the released version.

When a parameter changes, flag all linked simulations as stale and require re-run before release.

Shared repositories should separate released data from working data to prevent accidental production use of draft files.

Access control should restrict parameter file editing to qualified manufacturing engineers, with changes logged.

Aluminum alloys (e.g., AlSi10Mg) have high thermal conductivity and reflectivity; design for laser absorptivity and preheat differences versus steel.

Titanium alloys are reactive and require low-oxygen atmosphere; design features that avoid trapped powder because removal is difficult and hazardous.

Nickel superalloys are crack-sensitive; avoid sharp corners and abrupt section changes that concentrate residual stress.

Polymer FDM parts creep under sustained load; design for stiffness over time rather than short-term strength.

Material choice drives minimum feature size and tolerance; confirm the process capability for the specific alloy before finalizing dimensions.

LPBF (laser powder-bed fusion) gives fine features and good mechanical properties but has limited build volume and high cost per part.

EBM (electron beam melting) runs hotter, reduces residual stress, and needs sintered powder for charge dissipation; surfaces are rougher.

DED (directed energy deposition) builds large parts and can repair or add features, but has coarser resolution and needs machining.

Binder jetting builds fast and cheaply but requires sintering, which introduces shrinkage that must be compensated.

FDM is suited to prototypes and fixtures; strength is limited along the layer direction and surface finish is coarse.

Process selection criteria: size, resolution, material, mechanical requirements, post-processing, and quantity.

Step 1: simulate the build and cooling to get the distortion field relative to nominal geometry.

Step 2: invert the distortion field and apply it to the nominal mesh to create a compensated geometry.

Step 3: re-slice and re-simulate the compensated geometry; iterate until residual error is within tolerance.

Compensation must respect support and overhang constraints; a compensated shape may create new overhangs that need support.

Verify compensation on a physical build and measure the result; simulation-only compensation is not a validated process.

Record the compensation iteration and final scale factors with the part revision for traceability.

Design review for AM should include design, manufacturing, quality, and simulation representatives before build file release.

Review checklist: orientation, support strategy, drain paths, tolerance and datum scheme, material, post-processing, and inspection plan.

Decisions and action items must be recorded with owners and due dates; verbal approvals are not traceable.

Sign-off states should be explicit: design released, manufacturability approved, quality plan approved, build authorized.

Post-build review compares actual to predicted results and feeds lessons back into the design rules and simulation calibration.

Layer thickness sets vertical resolution; thinner layers improve surface finish and Z accuracy but increase build time linearly.

Beam diameter sets horizontal resolution; features smaller than the beam cannot be resolved reliably.

Minimum wall thickness for metal LPBF is often around 0.2-0.4 mm for self-supporting walls, but thin walls may warp; verify with test builds.

Text and fine detail should be oriented to minimize stair-stepping or added as a post-processing operation.

Feature size below the process capability must be redesigned or moved to a different process; do not specify them on the drawing.

Parameter development uses designed experiments varying power, speed, hatch spacing, and layer thickness to find a defect-free window.

Response variables include density, porosity, melt pool dimensions, surface roughness, and mechanical properties.

A process window is the parameter region with acceptable density and properties; operating near the edge increases defect risk.

Single-track and multi-track tests establish melt pool behavior before full-part builds.

Qualification builds use the production geometry and machine to confirm the parameter set transfers.

Document the experiment design and results; parameter sets without supporting data are not qualified.

One-way coupling applies the thermal result as a load to the mechanical model; two-way coupling is rarely needed for AM distortion.

Elastic-plastic material models are required to capture residual stress; purely elastic models overpredict stress.

Temperature-dependent yield strength changes the stress state as the part cools; a constant yield strength is a poor approximation.

Plate removal simulation releases constraint and allows the part to spring; this step is essential for final distortion prediction.

Support removal changes the constraint path; simulate it if supports carry significant load or the part is thin.

Report distortion in the part coordinate system with the datum scheme defined, so it can be compared to inspection data.

In-process monitoring data can be linked to the part serial number to form a per-part digital record.

Feedback from inspection and monitoring updates simulation calibration and design rules for the next build.

A digital twin of the process requires consistent inputs: machine state, parameters, material, and environment.

Data quality matters more than volume; inconsistent metadata makes monitoring data unusable for feedback.

Close the loop by comparing predicted distortion to measured distortion on every qualified build, not just the first.

Teardrop and diamond cross-sections allow internal horizontal channels to print without internal support.

Chamfers and fillets at overhang transitions reduce the unsupported area and lower support requirements.

Orienting holes and channels at 45 degrees or steeper avoids flat overhanging roofs.

Bridging is possible over short spans but sags; limit unsupported spans to a tested maximum for the process.

Self-supporting lattice cells (e.g., with inclined struts) can eliminate internal support entirely.

Build plate flatness and leveling directly affect first-layer adhesion and recoating uniformity.

First-layer thickness is set by the gap between the plate and the recoater; verify with a test print after any plate change.

Plate preheating reduces thermal gradient and warping but must match the material and machine capability.

Plate material should be compatible with the part alloy to avoid contamination or poor wetting.

Record plate ID and leveling results with the build record for traceability.

Lack-of-fusion defects occur when energy density is too low; simulation shows unfused regions between tracks or layers.

Keyhole porosity occurs when energy density is too high; simulation shows deep, unstable melt pools and voids at the bottom.

Balling is driven by low wettability and high scan speed; melt pool aspect ratio is an indicator.

Residual stress cracking risk rises with high thermal gradients and constrained sections; simulation flags these regions.

Defect prediction is probabilistic; report a risk map rather than a single pass/fail for the whole part.

AM quality standards define requirements for feedstock, process qualification, and inspection; map each build record to the applicable standard.

Material certificates must be linked to the powder lot and the parts built from it.

Process qualification typically requires a documented parameter set, machine qualification, and mechanical test data.

Operator training and certification records are part of the quality system and should be retained.

Audit readiness requires that any part's full history be retrievable from the data system on request.

Conformal cooling channels follow the part surface and improve heat removal; design them support-free and with drain paths.

Channel cross-section and spacing determine cooling uniformity; too-close channels reduce structural integrity.

Thermal mass distribution affects warping; keep section thickness uniform and avoid isolated thick regions.

Heat sinks and fins printed in AM should be oriented for printability and surface area, not just thermal performance.

Simulation of thermal performance should use the as-built surface roughness, which reduces effective heat transfer.

Recoater collision from warped parts is a common failure; mitigate with stress relief, support, and orientation.

Delamination between layers results from insufficient energy or contamination; check parameter set and powder condition.

Support fracture during the build causes distortion or part loss; support design must carry thermal and mechanical load.

Gas flow non-uniformity causes spatter accumulation and porosity; verify flow before production builds.

Build interruption recovery requires a documented restart procedure and inspection of the interrupted layer.