3D Printing and Prefabrication: How Each Process Actually Works

• 1
  What 3D printing and prefabrication each do to a part
• 2
  Where 3D printing holds tolerance and where it drifts
• 3
  The joint is the real tolerance problem in prefabrication
• 4
  Matching process to part count, size, and finish
• 5
  3D printing vs prefabrication: choosing by requirement

3D printing and prefabrication get grouped together because both shorten the path from drawing to physical object. They work in opposite directions. Additive printing builds a part by depositing or curing material point by point, layer by layer, until the geometry is complete. Prefabrication builds a larger assembly by making sub-units somewhere controlled, then joining them on site. One is a forming process. The other is a logistics and fit-up strategy.

That difference decides almost everything downstream. With printing, the constraint is the machine envelope and the minimum feature the nozzle or laser can resolve. With prefabrication, the constraint is the joint: how two modules meet, how much gap the connection tolerates, and whether the interface can be machined after assembly. A printed bracket can be a single part with no joint at all. A prefabricated frame always has at least one.

Engineers often ask which is faster. The honest answer is that printing wins on geometry and lead time for one-offs, and prefabrication wins on repeat count and unit cost once the fixture or mold exists. Neither replaces the other. Most real programs use both: printed or machined prototypes to validate the interface, then prefabricated or molded production parts built to the validated design.

One more distinction matters for quoting. Printing is priced per part and per machine hour, so part count barely changes the unit price curve. Prefabrication is priced per setup plus per unit, so the first unit is expensive and the tenth thousand is cheap. Send both numbers to your supplier early. The crossover point is usually somewhere between 50 and 500 units, depending on size and finish.

Printers are sold on layer height, but layer height is not tolerance. A 0.1 mm layer does not mean a 0.1 mm accurate part. Thermal shrinkage, support removal, and post-cure movement all push the final dimensions around. On a well-tuned machine you can expect roughly ±0.1 mm on small features, and worse as the part grows. A 300 mm long printed beam can easily drift 0.5 mm over its length simply because the material cools at different rates along the build.

This is why printed parts are rarely used as final locating surfaces. If a printed housing must hold a bearing, machine the bore after printing. We routinely print near-net and then clean up critical features on a CNC. That combination gets you the freeform geometry of additive plus the ±0.005 mm tolerance of subtractive work on the two or three surfaces that actually matter.

Material choice sets the ceiling. Standard resins and PLA are fine for form and fit checks, but they creep under load and soften with temperature. Engineering plastics such as PEEK, PA, and PC hold up better, and they cost more per kilogram and per hour. For anything carrying structural load, printed metal is an option, but it needs stress relief and often HIP treatment before it behaves like wrought stock.

Support strategy is the other quiet failure mode. Overhangs need support, and support leaves witness marks. If your part has internal channels or a downward-facing sealing face, plan the build orientation around it. Rotating a part 30° to move a critical face off the support contact often costs nothing and removes a whole finishing step.

Prefabrication moves work from the field to a controlled shop, which is why it is faster and cleaner. The trade is that every module boundary becomes a stack-up. If six modules each carry ±0.3 mm, the assembled frame can be off by more than a millimeter before anyone touches a bolt. The fix is not tighter modules across the board. It is one designated datum module, with every other module toleranced relative to it.

Connections decide whether the assembly can be adjusted. Slotted holes, shim packs, and oversized bolt holes with hardened washers let crews absorb stack-up on site. Welded joints do not. If a design is fully welded, the fit must be right in the shop, which means a fixture. Fixtures cost money once and then hold the tolerance for the whole run, so welded prefabrication favors higher volumes.

Material movement during joining is the second issue. Welding pulls. Bolting with a torque sequence pulls less but still moves thin sections. On a 4,000 mm frame, a weld can close a gap by a millimeter or more on one side. Machining the critical interfaces after joining, not before, is the standard way to escape this. It costs a second setup and saves a field rework.

Finally, shipping sets hard limits that no drawing shows. A module that fits the shop may not fit the truck or the site gate. Break lines should follow shipping limits first and structural logic second. Where those two conflict, add a bolted splice rather than shrink the module and multiply the joints.

Start with size. Printed parts are bounded by the build chamber, and long thin parts warp. Prefabricated modules can be far larger because they are made from stock and joined, but each joint adds cost and a tolerance risk. If a part is under roughly 300 mm and has internal channels or lattice, printing is usually the cleaner route. Over 1,000 mm, subtractive or fabricated construction with a bolted interface is almost always better.

Then look at surface finish. As-machined aluminum sits around Ra 1.6–3.2 μm, fine finishing reaches Ra 0.2–0.8 μm, and printed surfaces are rough by comparison and need sanding or coating to match. If a face is a sealing surface, a sliding surface, or cosmetic, plan for machining or finishing after printing. If it is a bracket, a duct, or an enclosure wall, printed texture is acceptable and the finishing step can be dropped.

Part count drives the economics more than any other variable. Below about 50 units, printing or CNC from solid is normally cheaper because there is no tooling. Above a few hundred, molded, cast, or fixture-based prefabrication spreads the setup cost thin. In the middle, the answer depends on how much post-processing the printed route needs. A printed part that needs three hours of hand finishing is rarely cheaper than a machined one.

Certification can override all of it. Medical and automotive programs need documented material traceability and inspection records. Both printed and machined routes can supply that, but the supplier must be set up for it. Ask for the inspection report with the first article, not after the production run. If the supplier cannot show raw material certificates and in-process records, the process choice is already wrong.