Choosing a process: a decision guide
📍 Where we are: Part IV · Making It Real — Chapter 15 of 15, the last chapter of Book 2. You now know how each process works, how to design and prepare a part, how to finish it, how to inspect it and how it is qualified. This chapter pulls it all into one decision procedure.
Choosing an AM process is like choosing how to travel. A bicycle, a train and a plane are all good, but not for the same trip. You do not start by asking "which vehicle do I like?" You start with where you are going, how far, how many people and how fast. In AM, the "trip" is what the part must do, the material it needs, its size and shape, how many you need and what proof is required. Only then does one process, or a short list, come out ahead.
What this chapter covers
We give a requirements-first procedure in seven questions and apply it to the running example, the lightweight mounting bracket. We then compare every process family from this book in one table, show where each is strong or weak, and include an honest check on whether AM is the right choice at all. ISO/ASTM 52900 defines seven process categories: material extrusion, vat photopolymerization, powder bed fusion, material jetting, binder jetting, directed energy deposition and sheet lamination [1]. For broader overviews of AM processes, materials, applications and open challenges, see [2] [3]. The table below uses those families and adds the material-led groups from Ceramics and other materials.
The procedure: seven questions
Ask them in this order. A common mistake is to start with a machine instead of a requirement, so the order puts the requirement first.
- Function. What loads, temperatures and environment must the part survive? Is it a look-and-fit model, a functional prototype or an end-use part?
- Material. Polymer, metal, ceramic or composite? This one answer removes most processes at once.
- Size and geometry. How big is it? Does it have internal channels, thin walls or lattices (Topology optimization and lattices)? Can loose powder or resin be removed from it?
- Surface and tolerance. Which faces must be accurate? As-built metal surfaces are markedly rougher than machined ones [4], so machining or finishing those faces is possible, but only if the design leaves material and access (Post-processing).
- Quantity and cost. How many, how often? Cost analyses of AM split cost into machine, labour and material [5], and a model of metal AM production cost finds machine productivity and high specific cost to be central constraints [6], so include post-processing labour, not just print time.
- Evidence. What proof is required? ISO/ASTM 52920 adds AM-specific qualification requirements on top of an existing quality management system [7], and ISO/ASTM 52904 covers metal powder bed fusion where failure leads to immediate threats [8]. See Quality, inspection and non-destructive testing and Standards, qualification and certification.
- Alternatives. Would casting, machining or moulding do the job more cheaply? AM wins when geometry, customisation or low volume dominate, not by default.
Comparing the processes
This table is qualitative and summarises the earlier chapters; each chapter carries the detail and sources. "Typical" means widely used, not the only option.
| Process family | Typical materials | Strong at | Weak at |
|---|---|---|---|
| Material extrusion | Thermoplastics, some composites | Low-cost, accessible, tough functional parts | Visible layers, directional (anisotropic) strength |
| Vat photopolymerization | Photopolymer resins | Fine detail, smooth surface | Resin handling, post-cure, brittle or limited materials |
| Polymer powder bed fusion | Nylon-type powders | Support-free nesting, functional end-use polymer parts | Powder handling, grainy surface |
| Material and binder jetting | Photopolymers; sand, polymer | Multi-material, colour, sand moulds | Limits on mechanical performance |
| Laser powder bed fusion | Steels, Al, Ti, Ni alloys | Complex dense metal parts | Cost, supports, residual stress, heavy post-processing |
| Electron beam and DED | Metals (powder or wire) | Large parts, repair, high deposition | Coarser features and surfaces |
| Metal binder jetting and bound metal | Metal powders with binder, then sinter | Volume production, no laser | Shrinkage on sintering, density limits |
| Ceramics and other materials | Ceramics, fibre composites, sheet, concrete | Materials others cannot take | Specialised, narrower supply |
Applying it to the bracket
The same bracket gives different answers depending on its answers to the seven questions. These are illustrative paths, not rules.
- A look-and-fit model for a design review. Function is low, polymer is fine, quantity is one. Material extrusion or vat photopolymerization is usually the fast, cheap answer.
- A functional polymer bracket, small batch. Needs real strength, no tooling, many parts per build. Polymer powder bed fusion fits, because parts can be nested without supports.
- An end-use metal bracket with a topology-optimised shape. Complex geometry in metal, modest quantity, demanding performance. Laser powder bed fusion is the usual choice, with stress relief, support removal, perhaps HIP and machining of the mounting faces, and the inspection and qualification layers that go with them.
- A large or repair-type metal part. Directed energy deposition, which can build large and repair worn parts.
- A simple metal bracket in volume. Metal binder jetting or bound metal, which trade the laser for sintering; or, honestly, machining or casting.
Start from requirements, let the material remove most options, then check surface and tolerance, post-processing, qualification and cost.
Original diagram by the authors, created with AI assistance.
A tiny cost sketch (illustrative, not measured)
Break-even logic is the same in every process. Suppose conventional production of the bracket needs a one-time tooling cost and then a low cost per part, while AM needs no tooling but a higher cost per part. The crossover quantity is tooling cost divided by the per-part difference. With invented numbers, a 10,000-unit tooling cost and a per-part difference of 40 units gives 10,000 / 40 = 250 parts. Below that, AM is cheaper; above it, tooling wins. The numbers are illustrative and say nothing about real prices, but the shape holds: AM has no tooling cost and favours low volume or high complexity, and the crossover moves with design, material and finishing effort.
Common traps
- Picking the machine first. Start from requirements, then process.
- Counting only print time. Depowdering, supports, heat treatment, machining, finishing and inspection are real cost and lead time.
- Ignoring qualification. A process that suits the part technically may still lack the evidence your sector requires; check early (Standards, qualification and certification).
- Forgetting design. The right process with a part designed for machining wastes AM's strengths; see Design for AM.
- Treating one answer as final. Prototype in one process and produce in another is common; just remember that properties change when the process changes.
Why it matters
A good choice makes the part cheaper, lighter or faster to deliver. A bad one produces a beautiful part that cannot be qualified, or an expensive one that a conventional route would have made better. A short, disciplined procedure beats enthusiasm for a technology.
In the real world
Teams usually go through this procedure informally and revisit it as requirements change. Part of the skill is knowing when the honest answer is a combination: AM for the complex core, machining for the interfaces, a conventional process for the simple remainder.
Key terms
- Requirements-first selection — choosing a process by what the part needs rather than by which machine you own.
- Process family — one of the broad groups of AM processes, such as the seven ISO/ASTM 52900 categories.
- Anisotropy — a property that differs with direction, such as strength along versus across layers.
- Break-even quantity — the volume at which two production routes cost the same.
- Tooling — moulds, dies and fixtures needed by conventional production, which AM generally avoids.
- Nesting — packing many parts into one build volume to use the machine efficiently.
- Near-net shape — a part built close to its final form, then finished by machining.
- Hand-off — the move from choosing and making a part to working with open tools or tackling open problems.
Where this leads
This ends Book 2. Two paths continue from here. If you want to run, modify and build the machines and software yourself, go to Book 3, Open-Source Additive Manufacturing. If you want to see the research problems and industrial challenges that remain unsolved, go to Book 4, Research Frontiers & Industrial Challenges.