A failed print is not an event you absorb, it is a rate you carry. Anyone who prints regularly fails regularly, and the useful question is not "what did that one cost" but "what percentage should be on every job to pay for the ones that do not finish". This is the method for working that out, and it takes about a week of record-keeping.
The four costs, only two of which people count
Material. The plastic in the bin. Everyone counts this and it is usually the smallest of the four.
Machine time. The hours the printer spent producing nothing. If your machine is busy, those hours had an alternative use, and if it is not, they are still wear on a belt, a nozzle and a fan. This is the term that dominates a long failure.
Your attention. Noticing, clearing the plate, picking plastic off a nozzle, re-slicing, restarting. Fifteen minutes of your time is worth more than the spaghetti it produced.
The schedule. A failure discovered at eight in the morning does not just cost the print; it costs the day the part was needed in. On a commercial job this is the expensive one and it never appears in a cost model.
The first two are computable and the calculator handles them. The second two you should carry as a judgement, and being explicit that you are doing so is better than pretending they are zero.
Turning "one in eight" into a number
Count your last twenty prints and note how many finished as intended. One failure in eight is a 12% failure rate, and that is not a bad figure for someone printing varied geometry in mixed materials.
Now apply it to a real job: a 320 g part, 14 hours, a machine drawing 120 W, filament at $25/kg. The printer is valued at 50 cents for every hour it is tied up, and the meter charges the 17.5-cent United States average.
The base cost before any allowance is $15.29 — $8.00 of filament, 29 cents of electricity, $7.00 of machine time. At a 12% failure rate the surcharge is $2.09, giving $17.38 per part that actually gets delivered.
That surcharge is not an estimate of what failures feel like. It is the exact figure that makes your expected revenue cover your expected costs: you need 1.1364 attempts on average to get one good part, so each good part carries the cost of that fraction of a wasted one. Doubling the failure rate does not double the surcharge, either: attempts per good part is one divided by the success rate, so it climbs faster than the rate itself does. The cost calculator therefore derives that term rather than adding a flat percentage to the total.
The lever that beats reducing the failure rate
Here is the same job again, same 12% failure rate, but with one assumption changed. Instead of assuming a failed attempt runs to the end and wastes everything, assume you catch failures roughly a third of the way in — a camera, a first-layer check, or simply being in the room for the first twenty minutes.
The surcharge drops from $2.09 to $0.73, and the delivered cost from $17.38 to $16.02.
Read that carefully, because it is the most useful thing on this page. You did not fail less often. You failed cheaper. The same reduction in surcharge would have required cutting the failure rate by more than half, which is far harder than watching the first layer go down.
That is why first-layer monitoring, a webcam, a filament runout sensor and the habit of not walking away until the skirt is drawn are worth more than any settings change. Most failures announce themselves in the first few minutes — first layer not sticking and spaghetti failures both have signatures visible long before the print is over.
Which failure rate to use, honestly
Your rate is not one number, and quoting one average across everything will overcharge the easy jobs and undercharge the hard ones. Keep three:
- Familiar material, geometry you have printed before, short job. A few percent. This is most hobby printing and most repeat commercial work.
- New geometry, or a material you are still learning. Ten to twenty percent, and be honest that a first attempt at a large flat ABS part on a marginal enclosure can be worse than that.
- Long overnight prints of tall or delicate parts. Higher than you want, because the exposure is the number of hours multiplied by everything that can go wrong in them.
If you are quoting for someone else, use the rate that matches the job rather than your flattering average. If you are printing for yourself, use the real number, because the point of measuring it is to find out whether a habit is expensive.
When to abandon a print in progress
There is a real decision here and people usually get it backwards, letting a doomed print run because "it might be fine" and stopping a salvageable one out of irritation.
The money already spent is gone either way. The only question is whether the remaining hours are more likely to produce something usable than a restart would be. So:
- Stop immediately if the first layer is wrong, if a corner has lifted, or if the part has detached. None of those recover.
- Stop if the defect is in a feature that matters — a mating face, a thread, a visible surface.
- Let it run if the defect is cosmetic, in a region you can sand or hide, or in support material.
- Let it run if you would learn something from seeing where it goes wrong, which on a new design is often worth an hour of machine time on its own.
Recording it so the number stays true
Keep a line per failed print: date, material, what failed, and roughly how far through it was. Twenty lines is enough to see your pattern, and the pattern is nearly always narrower than it feels — one material, one plate, or one kind of geometry accounting for most of the losses.
Once you know that, the surcharge stops being a tax you pay and becomes a thing you can spend deliberately: on an enclosure, on a new build plate, on drying, or on the first twenty minutes of attention that turns a $2.09 problem into a 73-cent one. Pricing that surcharge into what you charge customers is covered on pricing printed parts for sale.