In manufacturing, nothing is as important as uptime. To reduce downtime by a single percent is a huge win. But in additive manufacturing - more specifically 3D printing - it’s normal for even the most efficient operations to experience huge amounts of downtime (~30% - 50%). Truly competitive 3D printing operations hone in on this and find ways to eliminate it. But how do they do this?

What Causes Downtime?

Before we discuss how to maximize uptime, we must first discuss what the main contributors to downtime on a print farm are.

There are three main reasons a printer might be sitting idle at any given moment: (1) it’s waiting for a human to do a scheduled task, (2) it has a gap in its schedule, or (3) it stopped printing because of an error and nobody knows yet. In order to pinpoint where inefficiencies crop up, let’s go through each of these scenarios to find the underlying reasons they occur.

Waiting For A Scheduled Task

The most common reason a printer sits idle is simply waiting for its operator to get to it. 3D print farms often agree that their printers likely sit idle around 30% of the time due to filament changeover and print removal/start. Essentially, the printers are waiting for their humans to attend to them - kind of like an inefficient employee lingering at the water cooler because they can’t get work done at their desk until their boss is free to answer their very important question.

That’s not to say these scheduled tasks aren’t critical, they absolutely are, but as we’ll discuss later what’s not critical is that a human performs them.

Schedule Gaps

When looking to maximize uptime, most print farms will first turn to schedules. The logic is, if you can plan ahead of time to do long prints overnight, short ones between 9am and 5pm, plan when filament will be switched, and add additional staff hours where needed, you can increase your uptime. Compared with a first-come-first-serve approach (as is often employed by small businesses and universities), this is certainly efficient.

There are problems with this workflow, however. First, there’s the complexity of setting up the schedule in the first place, and the more printers you have, the more this complexity increases. Each new printer adds a disproportionate amount of work in the scheduling process. Going from 10 to 20 printers isn’t like trying to go from solving a 100-piece puzzle to a 200-piece puzzle. It’s more like trying to solve two 100-piece puzzles simultaneously with all the pieces mixed together. The time and brainpower required to create such a meticulous schedule can be hard to come by when you’re also running the farm in real-time.

Second, there is the issue of scheduled downtime. It’s uncommon that your print times add up to exactly 24 hours, so in between prints downtime is to be expected. Part of the idea with having a schedule is that you can control exactly where this downtime is and how long it lasts.

The other problem is, crafting this schedule only helps insofar as it’s followed. The problem with 3D printing (and all manufacturing), is that things don’t go as planned 100% of the time. Something as small as a first layer fail could lead to cascading delays that push your whole print farm off schedule. Then, someone has to go in and adjust the schedule to reflect what has actually happened and (hopefully) compensate for the time lost.

Not only does this create a lot of work for one unlucky farm operator, but it risks creating more downtime by splintering their attention and taking their time away from more productive work.

Unexpected Errors

It’s not uncommon for issues to crop up during the manufacturing process - that’s why defect rates are tracked. However, 3D printers have no way of recovering after a fail on their own. If something goes wrong in a 12-hour print at 2AM and nobody is there to respond to it, that printer will either stop on its own or keep printing until it damages itself to the point where it is forced to stop. Perhaps a more common scenario is a spool running out of filament at 5am when a 12-hour print is 75% done, causing the print to stop at that point.

If the goal is to maximize uptime, the last thing you need is a printer sitting idle for hours before someone is in to see it. Plus, the risk of damaging your printer beyond immediate repair is very much present - a poor first layer can lead to a detached print that leads to a nozzle blob that forces you to buy a new hotend. These replacements can be costly as they take time to do and result in your operation being down a printer until you can get a replacement piece and fix it.

To reiterate, there’s also the cost of rescheduling everything across your print farm when this situation occurs. If your schedule relied on having 28 operational printers, and you suddenly have 27, everything has to shift in order to hit those same deadlines.

Eliminating Downtime

If our main causes of downtime are waiting on human tasks, schedule gaps, and unexpected printer-stopping issues, it follows that addressing these three areas of inefficiency will increase our uptime. Our 3D print farm management and automation software, AutoFarm^3D™, was designed to do just this so that print farms can run at maximum capacity.

Automating Scheduled Tasks

The best way to prevent a printer from sitting idle while humans complete scheduled tasks, is to eliminate those human tasks. We specifically referenced part removal, print start, and filament changeover, the last of which has the least obvious solution, so let’s start with it.

Filament changeover is an unavoidably manual task, but that does not mean it has to interrupt production. As we often mention, we recommend print farms use 5kg spools or load all 4 AMS slots (even on single-colour printers, since it will automatically roll over) to reduce the frequency of required filament changes. When changes are necessary, however, they can be scheduled ahead of time. Many operations already try to do this, but part of the problem is knowing exactly how much filament has been used already. AutoFarm^3D has filament tracking built in; it finds the weight from the .gcode or .gcode.3mf file and uses this to calculate drawdown in real time on each printer. When filament is getting low, the spool turns red and the operator knows to schedule a change. All the operator has to do is update the printer’s filament weight after switching spools.

Part removal and print start are slightly more straightforward. We offer auto ejection kits, which include minor hardware additions to the printer and allow AutoFarm^3D to autonomously eject prints 24/7. Once a print is removed and the system has confirmed the bed is clear, the printer starts printing the next compatible print job in the queue. As long as there are print jobs sliced for that printer (and with matching material/colour), it will not sit idle - even when no staff are onsite.

Creating Dynamic Schedules

With the above tasks automated, the need for scheduling changes. A fixed schedule isn’t really helpful, because now you really just want the next print to start as soon as a printer has removed the previous one. What you need is a dynamic schedule.

AutoFarm^3D prints everything in order, as you set in the queue, but is able to adapt in real time as things change. If Printer A was supposed to finish Job A then start Job B, but Printer B has the same filament and is available first, it will take Job B. Instead of an operator having to look at each individual printer when considering the full farm’s schedule, the printers are able to simply take the next print on as soon as they are available. Sort of like pulling tasks from a To-Do list instead of asking your boss what your next assignment is.

This makes production more flexible and resilient, because if a printer does need to be pulled offline for maintenance, and you go from 28 to 27, it only impacts that one printer and not the overall workflow.

Catching Unexpected Issues

The last big contributor to downtime is unexpected errors, especially at inconvenient times. Filament runout is one that we already addressed, AutoFarm^3D reports how much filament a job will use as soon as it is accepted by a printer, and it will very soon be able to block jobs from being sent to printers that have less filament loaded than is required.

The other main cause of unexpected errors is print fails. QuinlyVision AI failure detection is built in to AutoFarm^3D and detects multiple failure types. This allows AutoFarm^3D to respond accordingly. If a print starts turning into spaghetti, instead of letting it go until it’s a big mess, AutoFarm^3D is able to alert you and automatically respond by halting the print, cooling the bed, and auto ejecting the failed print. Then, the printer simply moves on to the next print in the queue. Downtime is minimized, the operator is kept in the loop, and the printer’s schedule is minimally impacted.

Maximizing Uptime

We’ve discussed the main sources of downtime and how they can be reduced or eliminated, for a more productive operation. Not only does this increase your print farm’s uptime, it also gives operators a jumping off point for improving efficiency across the entire production.

With small, frequent tasks automated, operators are freed up to craft their own schedules; filament can be changed once or twice a week at a pre-scheduled time, prints can be picked up once or twice a day at pre-scheduled times, and unexpected interruptions, when they do occur, can be quickly addressed. Workarounds put in place for the old workflow can also be rethought - for example weekend-long batches which risk multiple prints failing if one thing goes wrong can be replaced with regular production, even in the absence of an operator. The farm can be checked in on it remotely if needed, via a secure encrypted tunnel that goes directly from your device to your farm (no cloud).

On top of the day-to-day workflow improvements, ongoing but infrequent tasks can be scheduled without dramatically impacting output. Printer maintenance can be easily planned in advance since pulling a single printer offline doesn’t require the entire farm schedule to change. Each printer also gets used evenly, as print jobs are load balanced across the entire farm, so their overall lifespan is longer and the likelihood of unexpected maintenance is decreased with more consistent use across the board.

Automation is the key to unlocking more uptime and efficiency in your 3D print farm’s operations. If you’re ready to take your operation to the next level, book a personalized consultation to discuss how AutoFarm^3D can benefit your print farm’s workflow and specific needs.