3D printing has a special place in mechanical design. It is the only manufacturing process where the designer can hold a part in their hands hours after sketching it. That speed is intoxicating, and it hides a trap: the printed part looks like a production part, but it is not one. The plastic is anisotropic, the surface finish lies, the dimensions drift, and the part that looked perfect on the build plate fails in the first week of service.
This guide covers the design rules for 3D-printed mechanical parts that actually matter: process selection, wall and infill thinking, orientation, tolerances, and the habits that separate printed prototypes from printed failures.
First, Which Process Are You Designing For?
“3D printing” is a family of processes with different strengths and different rules. The same geometry that prints beautifully in one process is a nightmare in another.
Process · Strengths · Design realities
FDM / FFF · Cheap, tough, large parts · Layer lines, weak Z direction, needs support
SLA / resin · Fine detail, smooth finish · Brittle, UV-sensitive, needs support and cleaning
SLS / nylon · Strong, no supports, good detail · Powder, grainy surface, sealed hollows
Metal (DMLS/SLM) · Real metal properties · Supports everywhere, heat stress, expensive
The design rule changes with the process. A snap-fit that works in SLS nylon will snap in brittle resin. A part designed for FDM with walls aligned to the layer direction will delaminate under load.
The Layer Direction Is the Load Direction
The most important rule in printed part design is also the most ignored: printed parts are not isotropic. The material is deposited in layers, and the layers are the weak planes.
In FDM, the strength between layers is typically 50 to 70 percent of the in-plane strength. A tensile load pulling layers apart will fail at a fraction of the load that the same part carries in-plane.
The design response:
• Orient the build so the highest loads run in-plane. A hook that hangs a load should print with the hook axis in the layer plane, not pointing out of the build plate.
• Watch for bending across layers. A cantilever that bends will peel between layers if the layers run across the bend direction.
• Test in the weak direction. If the part must survive a load that cannot be aligned with the layers, test that exact load case, on the actual print, at the actual infill. Do not trust the data sheet.
• Use the layer lines as information. The visible striations tell the mechanic which way the layers run. Design the part so the loaded direction is obvious and correct.
Walls, Shells, and Infill: The Internal Skeleton
A printed part is mostly air and thin walls, and the designer controls both.
Walls and Shells
The wall thickness should be a multiple of the nozzle width or the process’s minimum feature size. For FDM with a 0.4 mm nozzle, walls of 0.8, 1.2, or 1.6 mm print cleanly; a 1.0 mm wall prints with gaps and weaknesses.
Thin walls fail in two ways: they tear at the interlayer bonds, and they warp because each deposited line shrinks as it cools. The rule is to design walls at least 1.2 mm for cosmetic features and 2 to 3 mm for load-carrying features, and to verify the slicer actually fills the wall without gaps.
Infill: The Structural Compromise
Infill is where the designer trades strength, weight, and time. The percentage matters less than the pattern and the duty:
• Light duty, cosmetic: 15–20 percent grid or gyroid. The part holds shape but carries little.
• Medium duty, brackets and housings: 30–50 percent. The part carries real load.
• High duty, structural: 80–100 percent, or switch to a stronger process. High infill in FDM is still a plastic part.
The gyroid pattern deserves a mention: it is isotropic in its strength response and prints fast, and for most mechanical parts it beats the classic grid.
The Shell-Infill Interface
Loads travel through the shell; the infill supports the shell. A part that needs stiffness benefits more from thicker walls than from higher infill. The bending stiffness of a thin-walled box scales with the cube of the wall offset, so a small wall increase is a large stiffness gain.
Support: Design It Away Before You Print It
Supports are the tax on overhanging geometry. They cost material, they cost time, and they leave scars on the surfaces they touch. The design should avoid them wherever possible.
The classic rules:
• Keep overhangs under 45 degrees. Above that, most processes need support. Chamfers beat overhangs.
• Self-supporting angles. Design the geometry with a built-in draft or chamfer so the slicer has nothing to support.
• Bridge the gaps. A short bridge between two walls prints cleanly if it is short. Split long bridges with a rib or a dog-bone slot.
• Put supports where the scars do not matter. If support is unavoidable, place it on a hidden face or a face that will be machined or assembled against.
Every support scar is a surface finish and a dimensional reality. The design that avoids supports is the design that prints overnight instead of failing at 3 a.m.
Tolerances and Shrinkage: The Numbers That Drift
Printed parts do not hold the tolerances that a drawing implies. The honest number depends on the process and the printer, but the reality is:
• FDM: ±0.3 to 0.5 mm on simple dimensions, worse across the build height.
• SLA: ±0.1 to 0.2 mm with good calibration.
• SLS: ±0.1 to 0.3 mm, with the powder acting as a natural support.
• Metal: ±0.1 to 0.2 mm before post-processing.
The design response is not to tighten the printer. It is to design around the reality:
• Design clearance, not fit. A printed hole that must slide over a shaft should be sized for the loose end of the tolerance. Test the first article and adjust.
• Print holes undersized and ream, or add the shrinkage allowance. A critical bore gets reamed or bored after printing, with stock left in the design.
• Watch the shrink. Every material shrinks as it cools from the nozzle or the laser. The printer’s firmware compensates for the nominal value, but the compensation is not perfect, and the error grows with the part.
• Keep critical features in one build orientation. The tolerance story changes with the direction: in-plane dimensions behave differently from build-height dimensions.
The practical habit: print the critical feature as a small test coupon first, measure it, and feed the measured values into the design. One hour of calibration beats a week of scrapped parts.
Threads, Inserts, and Fasteners
Printed threads are a beginner’s trap. The printed thread looks right and strips at half the torque of a machined thread, and the fix is usually not more thread.
The dependable options:
• Heat-set inserts. The brass insert is pressed into a printed boss with a soldering iron, and it takes the fastener torque like a machined part. Design the boss with the insert’s recommended wall thickness, typically 1.5 to 2.5 mm around the insert, and a counterbore for the insert flange.
• Trapped nuts. The hex pocket printed in the part catches a standard nut. Design the pocket with a small clearance and an anti-rotation tab, and the nut seats before the screw is driven.
• Through-holes with bolts and captive hardware. The bolt passes through, and the nut or the threaded insert lives on the far side.
• Tapping the printed hole as a last resort. It works for low-torque, one-time assembly, and it must be tested. It is not a production fastening method.
The rule: the plastic part should not carry the thread. The hardware carries the thread, and the plastic holds the hardware.
Design for the Process, Not Against It
The most productive mindset is designing with the process’s strengths instead of fighting its weaknesses:
• Let the part be hollow. The process rewards shells and ribs over solid lumps. A printed bracket with internal ribs is stiffer than a solid bracket of the same weight.
• Use the print to consolidate. The process can print a hinge, a clip, and a latch in one piece. Design the single-piece assembly where service allows.
• Print the fixtures too. The jig that holds the printed part for post-machining prints in the same batch. The process that made the part can make its own tooling.
• Post-process where the process ends. Machining, sanding, and chemical smoothing take the printed part to the tolerance and finish the application needs. Design with the post-processing step in mind.
When to Use a Printed Part, and When Not To
The printed part has a job description:
Use it when:
• The design is still moving and the geometry is the question.
• The quantity is one, two, or ten.
• The geometry is impossible for machining: internal channels, complex lattices, consolidated assemblies.
• The tooling cost of the production process cannot be justified yet.
Do not use it when:
• The part carries fatigue or impact load in production. Printed plastics are not fatigue materials.
• The temperature is high. PLA softens around 60 C, PETG around 80 C, and even nylon and PC drift well below the metal’s comfort zone.
• The tolerance and finish demand a controlled process.
• The quantity and duty justify a mold or a machining run. At quantity, the printed part stops being cheap.
The professional habit is to ask: what is the printed part for? If the answer is “to learn something,” print it. If the answer is “to be the production part,” have a serious conversation with the process engineer first.
A Printed-Part Design Checklist
Before sending a part to the printer:
• [ ] The process is chosen, and the design follows its rules, not another process’s.
• [ ] The highest loads run in-plane with the layers.
• [ ] Wall thicknesses are multiples of the nozzle or minimum feature size.
• [ ] Infill matches the duty: cosmetic, medium, or structural.
• [ ] Overhangs stay under 45 degrees, or supports are designed onto hidden faces.
• [ ] Critical bores carry stock for reaming or are sized from a test coupon.
• [ ] Fasteners use inserts or trapped nuts, not printed threads.
• [ ] The orientation is chosen for strength and surface quality, not for the prettiest screenshot.
• [ ] The first article is measured before the design is trusted.
Post-Processing: From Print to Part
The printer does not deliver a finished part. It delivers a near-net shape with layer lines, support scars, and a dimensional story that depends on the process. Post-processing is where the printed part becomes the part the drawing describes, and the design should anticipate it.
The Post-Processing Toolbox by Process
Each process has its natural finishing steps:
• FDM: Remove supports, sand the layer lines on the visible faces, and optionally vapor-smooth with acetone for ABS or use a chemical smoother for the specialized filaments. The sanded and painted part hides the layer story entirely, which matters when the customer sees the prototype as the preview of the production part.
• SLA: Wash in IPA, remove the supports, and post-cure under UV. The resin part is then sanded and primed for paint. The SLA surface is smooth enough for cosmetic prototypes straight off the build, but the strength story needs the full cure cycle.
• SLS: The powder blasting removes the loose powder from the surface and the internal channels. The surface is grainy, and the sealed hollows need a vent or the powder stays inside forever.
• Metal: The support removal, the heat treatment, and the machining of the critical features. The metal printed part is a forging, not a finished component, and the drawing must say which features are as-printed and which are machined.
The Machining That Makes It a Part
The critical features should be machined, not printed to tolerance. The design that leaves stock on the bore, the face, and the thread location is a design that can be brought to specification in the post-process.
The workflow: print near-net, machine the functional features, and inspect the result. The printed part provides the shape and the near-net geometry; the machining provides the tolerance and the finish that the application demands.
The Surface That Speaks
The surface finish of a printed part is a language. The layer lines say “prototype” to every observer, which is sometimes exactly the message the design wants to send. A prototype that is meant to be handled, shown, and discussed benefits from a finish that invites the touch. A part that is meant to test a fit benefits from a finish that does not add friction where there should be none.
The design decision is to choose the finish deliberately: as-printed for the test, sanded for the presentation, painted for the customer review, machined for the functional test.
The Inspection That Closes the Loop
The post-process ends with measurement. The critical dimensions from the design are measured on the finished part, and the measurements feed back into the design and the process settings.
The first-article inspection is the moment of truth: the part either meets the drawing, or the drawing, the orientation, the material, or the post-process needs an adjustment. The team that measures the first article and adjusts before the second is the team that learns from the process instead of fighting it.
The designing parts for 3D printing and additive manufacturing design guidelines in this article do not end at the build plate. The printed part is the beginning of the finishing story, and the design that anticipates the post-process, the machining, and the inspection is the design that turns a print into a part.
Conclusion
Designing parts for 3D printing is not harder than designing for machining. It is different, and the difference is in the physics: anisotropic layers, drifting tolerances, support scars, and materials that soften with heat. The design rules for 3D printing in this article, from layer orientation to insert bosses, are the translation layer between the CAD screen and the printed reality.
The additive manufacturing design guidelines here do not make the printer perfect. They make the designer honest: orient for the load, thicken the wall, avoid the support, measure the coupon, and let the hardware carry the thread. Do that, and the printed part becomes what it should be: the fastest way to hold the answer in your hands.
