FDM 3D printing: what it does well and where it falls short

Most of what leaves our workshop in Dubai is printed on FDM. A spool of plastic feeds into a hot nozzle, the nozzle lays the material down line by line, and some hours later there is a part on the plate that someone can hold, bolt to something, or stand on a plinth. The process is old by 3D printing standards and it is still the first one we reach for on anything larger than a shoebox.

It also has a shape it cannot make. Lettering under half a millimetre wide, a crisp 0.3 mm fillet, a clear lens, a snap fit with a tight clearance: those either go to resin or they come back from the printer as a disappointment. Deciding which side of that line a part sits on takes about a minute with the file open, and that minute saves a day.

If you are still choosing a process rather than working inside one, our piece on FDM, SLA or SLS covers that decision. Here we stay inside FDM and look at how the machine actually builds a part and where its ceiling sits.

How an FDM printer builds a part

Fused deposition modeling, also called FFF, works by melting a thermoplastic filament and depositing it in a controlled path. A drive gear pushes the filament into a heated block, the melt leaves through a nozzle that is usually 0.4 mm across, and the machine draws the cross-section of the object at that height. Then the build plate drops, or the head rises, and the next layer goes on top of the one below while it is still hot enough to fuse.

That last point carries most of the consequences. The bond between layers happens in a narrow window of time and temperature. Print too cold, too fast, or with too much fan on the wrong material, and the layers sit on each other like stacked ribbons instead of welding into one wall.

The slicer decides everything the printer does: how many perimeters form the outer wall, how much lattice goes inside as infill, where support structures hold up overhanging geometry, what layer height to use. Common layer heights run from about 0.1 mm for careful work to 0.3 mm and above on large surfaces where speed matters more than the finish.

The first layer is its own small event. If it does not stick cleanly across the whole footprint, the print either lifts a corner or fails outright three hours later, and a large flat base on a cold plate is the usual culprit.

The filaments we actually use, and what Dubai does to them

PLA is the default for indoor work. It is stiff, dimensionally well behaved, it prints without an enclosure and it sands reasonably. Filament datasheets put its glass transition around 60 degrees Celsius, which is the number that matters here more than any strength figure.

PETG trades a little stiffness for toughness and a higher softening point, generally quoted in the high seventies to low eighties. It tolerates a knock, it tends to leave fine strings between features, and it prefers a slightly slower print than PLA. For anything that gets handled by the public, it is often the better choice.

ABS softens somewhere around 105 degrees and sands and solvent-smooths very well, which makes it useful for painted work. It also shrinks as it cools, so large flat ABS panels want a warm enclosure or they split along a layer halfway up.

TPU is the flexible one: gaskets, feet, grips, anything that needs to bend and come back. It prints slowly because a soft filament resists being pushed through a nozzle, and sharp detail is not what it is for.

Then there is the local part of the question. Inside an air-conditioned showroom or office none of this matters much. A part that lives in a car, in a window, on a balcony or anywhere in direct July sun is a different brief. Dark surfaces in Dubai sun run far hotter than the air around them, and PLA that has gone soft does not snap, it sags quietly and keeps its new shape. For outdoor or in-vehicle parts we move to PETG, or to ASA, which behaves like ABS but holds up better under UV. Resin parts also yellow and get brittle in sun, so that is not an escape route either.

Humidity is the quieter problem. Filament picks up moisture from the air, and an open spool left on the bench through a coastal summer prints with popping sounds, rough walls and weak layer bonds. Spools live sealed with desiccant, and anything that has been sitting out goes through a dryer before a job that matters.

Where FDM earns its place

Size is the honest answer. FDM is the cheapest way to put a large volume of plastic into a shape, because the part does not have to be solid. A 2 to 4 mm shell with internal ribs gives a surprising amount of stiffness for very little material, and a big hollow object can print faster than a small dense one.

Anything larger than the build chamber gets cut into sections and bonded, which is how we go from 5 cm figures up to builds around 15 m. The logic of splitting, registration and assembly is a job in itself, covered in our piece on large-scale printing in sections.

Beyond scale, FDM covers the everyday work well. Enclosures and housings where you need to check that a board, a cable gland and a screw boss all coexist. Fit checks against an existing assembly. Jigs and fixtures nobody will photograph. Exhibition props and stand elements that get painted anyway, so the raw surface quality matters less than the geometry and the weight.

Cost follows the same logic. The material itself is the least expensive plastic we print, so what moves a quotation is the volume of extruded material, machine hours, the amount of support that has to be removed, how many sections need joining, and how far the finish goes. A rough functional part and a show-finished exhibition piece of the same geometry are different jobs with the same print time.

Where FDM runs out of road

Fine relief is the first wall. A feature narrower than the nozzle cannot be drawn as a wall. Engraved text at 0.3 mm wide with a 0.4 mm nozzle simply does not appear; open it to roughly double that width and give it real depth and it survives both printing and sanding.

Layer lines are visible, and how visible depends on geometry more than on the machine. On a near-vertical wall the layers stack almost directly above each other and the surface reads as fine ribbing. On a shallow slope the same layers form a staircase: at 0.2 mm layer height a surface 10 degrees off horizontal gives a tread about 1.1 mm wide, while at 30 degrees that tread drops to roughly 0.35 mm. The flattest parts of a dome always look the worst, which catches people out.

Transparency is not available. Clear filaments exist and they come out translucent and cloudy, because light scatters at every bead boundary. Anything that has to be genuinely clear goes to resin.

Then there is anisotropy. A printed part is strong along the beads and weaker across the layer bonds, so a thin pin or a tab printed standing up in Z will break at a layer line under load that the same geometry laid flat would shrug off. We orient parts around this constantly, and when the load direction fights the surface quality we say so before printing rather than after.

Threads and tight clearances round out the list. Printed threads in fine pitches are fragile and inaccurate. Holes come out slightly undersized because of the extruded bead geometry, so functional holes get drilled or reamed afterwards, and threads get a tapped hole, a heat-set insert or a captive nut pocket instead of printed teeth.

What gets decided in the model, not on the machine

Orientation decides which axis carries load, which faces get support scars, and where the staircase effect lands. We choose it so that supports touch surfaces nobody will see, and so that the weak direction is not the direction being pushed.

Wall thickness does more work than infill. Adding perimeters stiffens a part far more efficiently than filling the middle, and on large pieces we would rather have a 3 mm shell with ribs than a thin shell packed with lattice. More infill mostly buys weight and print hours.

Overhangs steeper than roughly 45 degrees from vertical need support, and support always leaves a mark. Sometimes the fix is a small chamfer in the model so the geometry supports itself. Redesigning an overhang out of a part costs five minutes and saves an hour of scraping.

Files arrive with problems of their own: non-manifold edges, flipped normals, walls that exist visually but have no thickness. Those stop a slicer before any of the above matters.

The part is not finished when the printer stops

Support removal comes first, and it leaves the surface rough wherever it touched. From there the work is mechanical. Sanding through progressively finer grits, filling the seams on sectioned pieces, then filler primer.

Primer is also a diagnostic tool. A flat grey coat kills all the highlights that were hiding the defects and shows exactly what the sanding missed, which is why an exhibition piece typically goes through sand, prime, inspect and sand again three or four times before any colour goes on. On a large sectioned object, that finishing cycle routinely takes longer than the printing did.

For stand and showroom work we default to matte or satin topcoats, because exhibition lighting is harsh and gloss turns every remaining layer ridge into a bright line.

When we move a part to SLA instead

Resin goes on the list when the part is small and the detail is fine, when surfaces need to look moulded straight off the machine, when a part has to be clear, or when a tight assembly fit matters more than toughness. Layer heights there are an order of magnitude finer than on FDM, and the trade is size, cost per volume and brittleness under sunlight.

Plenty of projects use both: a resin master or a detailed insert sitting inside a large printed shell. If a file is sitting on your desk and you want a straight answer about which way it should go, send it over and we will tell you what we would print it on and why.

FAQ

What is FDM 3D printing in simple terms? A machine melts a plastic filament and squeezes it out through a fine nozzle, drawing the object one thin layer at a time and fusing each layer to the one below. The same process is also called fused deposition modeling or FFF. It is the most common 3D printing technology in use, and it is what most people picture when they picture a 3D printer.

Which FDM material should I choose for the UAE? For anything that lives in air conditioning, PLA is fine and prints predictably. For parts that get handled, knocked, or left in a hot car, PETG. For outdoor exposure, ASA, with the honest caveat that no plastic finish outdoors is permanent here. TPU is for flexible parts only. PLA is the one to avoid anywhere sun or a closed vehicle is involved.

Can FDM print large objects? Yes, by splitting them. Anything bigger than the build chamber is divided into sections, printed as hollow shells with internal ribs, then bonded, filled, sanded and painted into a single object. We build this way from small figures up to constructions in the region of 15 m. The design of the split lines and the assembly sequence is part of the job, not an afterthought.

Will the layer lines show? On a raw part, yes. How much depends on layer height and on the angle of the surface: steep walls show fine ribbing, shallow slopes show a visible staircase. Sanding and filler primer remove most of it, and a matte topcoat hides what remains. Parts that need to look moulded straight off the printer are better suited to resin.

When is FDM the wrong choice? Fine sculptural detail, sub-millimetre features, transparent parts, precise threads, thin pins loaded across the layers, and anything expected to stay rigid above roughly 60 degrees in PLA. Those jobs go to SLA, get redesigned, or get a different material. Telling you that before we print is cheaper for everyone than finding out afterwards.