Brim vs Raft: 5 3D Printing Secrets to Avoid Failed Prints

A brim is extra perimeter loops printed on the same layer as the part, directly on the build plate. It touches the model along its outline, usually 0.1–0.3 mm away, so it anchors the corners without sitting under the part. There is no support surface to peel, and the bottom face of the model keeps whatever texture the plate gives it.

A raft is a horizontal platform printed underneath the part, typically 2–4 layers of dense lines with a looser top layer. The model starts on top of that platform, so its first layer never contacts the plate. That is the whole point: the raft absorbs plate flatness errors and gives the part a consistent base, at the cost of extra material, extra time, and a rough bottom face.

Neither option changes what happens above layer one. If the part warps because of thermal shrinkage in the upper walls, no brim will hold it forever. That is why the choice has to follow geometry and polymer, not habit.

When we review a failed print in our additive lab, the slicer settings are usually the last thing we look at. First we ask what the part was supposed to be: a tall bracket with sharp corners, a flat cover panel, a thin-walled duct. The answer narrows the choice before any number is typed.

Tall parts with small footprints are the classic brim case. A 150 mm tall ABS bracket with a 30 × 30 mm base has no adhesion area and a lot of bending force at the top. Every pass of the nozzle drags on the part, and one knock ends the run. Five to ten brim loops widen that footprint just enough to resist the drag, and the brim is cheap to remove.

Large flat parts are the opposite case. A 200 × 200 mm PLA cover panel already grips the plate across its whole face. Adding a brim there does almost nothing, and a raft would only add a rough bottom and thirty minutes of print time. If this part curls, the cause is usually plate temperature or an uneven plate, not missing adhesion area.

Thin, flexible, or high-shrinkage materials shift the answer. TPU grips a raft more predictably than it grips glass or PEI, so a raft is often the practical choice for a soft part with a small base. Nylon and polycarbonate shrink hard and warp early; a brim holds the outline while the part cools, but a draft shield or an enclosed chamber does more for those polymers than either option.

There is also a geometric limit. A brim only helps where the part meets the plate. If the first layer is a small circle in the middle of a wide, thin flange, the loops spread the load but the flange still lifts. Sometimes the right move is to reorient the part so the widest face sits on the plate.

The gap between brim and part controls how hard removal will be. Zero gap fuses the brim to the wall and you will cut it with a knife, leaving marks. A gap of 0.1–0.2 mm keeps the two connected through the first layer squish but lets you peel the brim with a fingernail. On a 0.4 mm nozzle we start at 0.15 mm and adjust from there.

Brim width matters more than most people expect. Three loops on a 0.4 mm nozzle is about 1.2 mm of extra grip, which is close to nothing. Eight to twelve loops gives 3–5 mm of skirt around the part, and that is the range where corner lift actually drops. Beyond about 15 mm the returns are flat and you are just printing scrap.

Raft settings work the same way in reverse. The air gap between the raft and the part decides how easily the part releases. Too small and the bottom fuses into the raft; too large and the part slides during printing. Start around 0.2 mm and check the bottom face after the first part. If the base looks torn, the gap is too tight.

Treat these numbers as starting points, not rules. Nozzle diameter, layer height, and first-layer squish all shift the result. Print a small test block with the same polymer and plate before committing a long run. Ten minutes of testing beats a four-hour failure.

Adhesion is a temperature problem as much as a chemistry problem. PLA sticks well to a clean plate at 55–65 °C and often needs no brim on a textured surface. ABS wants 95–110 °C and an enclosed space; below that, the corners pull up no matter how wide the brim is. PETG sits around 70–85 °C and can stick too well on smooth PEI, so a release agent is worth having.

The plate itself changes the answer. Smooth PEI grips hard and releases hard. Textured PEI holds less but forgives more. Glass with a glue stick is predictable but needs reapplication. A raft is often used to paper over a plate that is worn or coated unevenly, and that works until the raft itself stops sticking.

Clean the plate before blaming the settings. Finger oils on a PEI sheet will kill first-layer adhesion on a part that printed fine yesterday. Warm water and a mild detergent, then a wipe with isopropyl alcohol, is usually enough. Do this before you change the brim width or the plate temperature.

First-layer speed and squish belong in the same conversation. A slow first layer at 15–20 mm/s with proper squish does more for adhesion than any brim. If the first layer is glossy and thin, the nozzle is too close. If the lines are round and separate, it is too far. Fix that before adding loops.

A raft will mask an uneven plate. The part prints flat on top of the raft and looks fine, which is exactly the problem. The error is still there, and it shows up later as a tilted base, a mismatched split part, or a first layer that varies across the plate. Two parts from the same file can end up different heights.

The same applies to a brim. If the nozzle is 0.2 mm too far from the plate on one side, the brim there is barely touching and adds no grip. You have paid for the loops without getting the benefit. Level or mesh the plate, set the Z offset, then decide whether you still need a brim.

Some printers handle this with mesh bed leveling, which compensates for small deviations. That helps with a warp of a few hundredths of a millimeter, not with a plate that is visibly out of plane or a gantry that is not square. Check the mechanical side first.

The test is simple. Print a single-layer square across the plate and look at it. Consistent width and slight squish everywhere means the plate is ready. Lines that fade on one side mean no brim width will save a tall part. Fix the machine, then print the part.

Printing is the right process when the geometry is complex, the quantity is low, and the material is a polymer that machines badly or wastes a lot of stock. Internal channels, lattice, and organic shapes are cheap to print and expensive to cut. A bracket with a few holes and flat faces is the reverse: a 5-axis mill makes it in one setup with a ±0.005 mm tolerance and a known surface finish.

The decision usually comes down to tolerance and surface. FDM layers leave a stair-step on curved faces and a rough bottom where a raft was. If the part is a functional mount, a sealing face, or anything that mates with a bearing, the print will need post-machining anyway. At that point, cutting it from 6061-T6 or 17-4PH is often faster than printing plus finishing.

Rafts and brims also add handling. Every part needs the brim peeled or the raft pried off, and that step is manual, variable, and hard to automate. On a run of 200 parts, the removal time adds up. Machined parts come off the machine ready for inspection, which fits a production schedule better than a bench full of scrapers.

Use printing where it wins. Use machining where the drawing demands a number. Mixed programs are normal here: print the prototype housing to check fit, then cut the production version in aluminum. Both processes have a place, and the part tells you which one.