Running a multi-printer print farm: queue and scheduling basics

Adding a second printer doesn’t double your output. It should, on paper. In practice most two-printer shops run at maybe 1.5× the throughput of one, because the thing that was easy with one machine — deciding what prints next — is now a scheduling problem, and it’s usually being solved badly with sticky notes.

Scheduling a farm comes down to three questions: what runs where, in what order, and how do you keep beds from sitting idle. Get those right and the machines earn their keep.

Idle time is the real enemy

A printer that finishes a 6-hour job at 2 a.m. and sits until you walk in at 9 has lost 7 hours. Across a week, one machine idling overnight is easily 40+ printing hours gone — the output of a whole extra printer you already own but aren’t using.

The fix is to always have the next job ready to start the moment a bed clears, not to decide what’s next when you happen to be standing there. That means a queue: an ordered list of jobs waiting, each already assigned to a printer or ready to drop onto whichever one frees up first. The question “what prints next?” should have an answer before the current print ends.

Assign by capability, not by whichever’s closest

Not every job runs on every printer. One machine has the 0.6 mm nozzle for fast, chunky parts. One is dialed in for PETG and holds temperature better. One has the textured plate your customers like. Assigning a job means matching it to a printer that can actually run it well, not just the nearest free bed.

Material loaded is the most common constraint. If an order needs matte black and only one of your six machines has it loaded, that job is pinned to that machine no matter how busy it is. Trying to route around this in your head is where color mix-ups happen — the part comes out in the wrong black and has to be reprinted.

Batch by material and color to cut changeovers

Every filament swap costs time: unload, load, purge until the color runs clean, sometimes re-level. Call it 10–15 minutes and 15–20 g of purge per change. Do that six times a day across a farm and you’ve lost an hour and a spool’s worth of purge to changeovers alone.

Batching is the answer. Group the jobs that use the same material and color and run them back to back on the same machine, so you change filament twice instead of eight times. Dedicating a machine to one material for a whole day — the “always black PLA” printer — eliminates its changeovers entirely. You trade a little scheduling flexibility for a real cut in wasted time and purge.

Estimate throughput from print hours, not job count

“We can do 60 orders a week” is meaningless without hours behind it. A farm’s real capacity is machine-hours available minus the time lost to changeovers, failures, and idle beds.

Six printers running 16 usable hours a day is 96 machine-hours daily. Knock off maybe 15% for changeovers, failed prints, and beds that sat idle before someone cleared them, and you’ve got roughly 80 productive hours a day to sell. If your average order is 4 print-hours, that’s about 20 orders a day at full tilt — and that ceiling only holds if the queue keeps every machine fed.

Most farms never hit their theoretical ceiling, and the gap is almost always idle time and changeovers, not raw machine speed.

Manual where it matters, automatic where it helps

There’s a temptation to fully automate scheduling. Resist it for the decisions that need judgment — which rush job jumps the queue, whether to interrupt a batch for a hot order, when a finicky part needs the good machine. Those are yours to make.

What should be automatic is the mechanical part: showing every bed’s status at a glance, advancing an order as its prints finish, and surfacing the next job so a free printer never waits on you. Keep the judgment calls, hand off the bookkeeping.

That’s the split SpoolDeck’s print-job queue is built around — assign jobs across your printers, see every station at once, and let orders advance as prints complete, while you keep the final say on what runs where.