3D printing failure rate: what is normal, and what it costs you

What a failure rate does to your unit cost

A failed print still consumed filament, electricity and machine hours. To deliver one good unit at failure rate f you must start 1 ÷ (1 − f) prints. The table applies that to the 100 g PLA part from the cost breakdown, whose subtotal before failures is $1.79 on an entry-level printer at US electricity prices, and shows the two things a failure rate actually changes: what the part costs you, and the minimum price you can quote at a 30% margin.

The curve is gentle at first and steep later: going from 0% to 10% adds $0.20 per unit, while going from 20% to 30% adds $0.32 for the same ten points. At 30% failure you have to charge $3.66 to keep a 30% margin on that part, and 21% of that price is there to cover prints nobody buys. That is why a workshop measuring 25% should stop quoting and fix the process first — the failure line is not overhead you can hide, it is a fifth of your revenue.

Put your own part in

The table above fixes the material and the printer. Yours will differ: a 500 g PETG bracket on a P1S in Germany is the same arithmetic with four different inputs. Change them here and the result updates as you type. Failure rate is the field worth moving — drag it from 5% to 25% and watch the sellable cost follow.

What one failed attempt costs, by material and part size

The number people underestimate is not the failure rate, it is the bill attached to a single failure. Below is what one failed attempt costs on the entry printer at US electricity prices — the full loss, filament plus electricity plus machine time, for the same part in eleven materials and three sizes.

The loss is linear in mass — all three lines scale with the grams — so a failed 500 g part costs five times a failed 100 g one, whatever the material. That is what makes the failure rate bite on big parts: at $21.42 for a single failed 500 g PVA print, a 25% rate costs you $7.14 of pure waste for every part you deliver. Material also takes a larger share of the loss the dearer it gets: 84% of a failed 100 g PLA attempt is filament, against 93% for PVA. Between the cheapest and dearest material the spread is 2.4× at the same mass — $1.79 for PLA, $4.28 for PVA.

One honest caveat: this is the cost of a failure discovered at the end. Most failures announce themselves in the first two minutes — a first layer that will not stick, a support that is not there — and a print you stop at minute two costs a gram and a minute. The gap between those two cases is the entire value of watching the first layer. Which of the eight symptoms caused the failure, and what repeating that attempt costs, is on the eight-symptom fix order.

Measure your own rate in twenty prints

Twenty is enough to know whether you are a 5% operator or a 25% one, which is the only distinction that changes a quote. Log every print for a month: model, material, spool age, outcome, and the reason if it failed. Count failures as prints that did not produce a sellable part — a print you stopped at hour two because the first layer was wrong is a failure, not a cancellation.

Then set the failure rate in the cost calculator to what you measured, not to what this page says. A measured 18% is worth more than an assumed 10%.

The four things that actually move the number

1. First layer. Most failures are decided in the first two minutes. Clean the sheet, re-run the mesh levelling when you change a nozzle, and print a skirt wide enough to show you the flow is right before the part starts. A brim costs a gram and saves whole prints.
2. Supports. Support that fuses to the part, or that the nozzle knocks over at hour six, is a process error rather than bad luck. Increase the Z distance, use a support interface, and orient the part so that large overhangs are supported by the bed instead.
3. Wet filament. PETG, nylon, TPU and any spool that has sat open for months will pop, foam and delaminate. A dry box or a heated dryer is the cheapest reliability upgrade in the hobby, and it is the one most people skip.
4. Broken meshes. A model with holes or inverted normals slices into something the printer cannot build. Check the file before you print it: the STL checker reports triangle count, dimensions, volume, watertightness and non-manifold edges in the browser, without uploading the model anywhere.

The honest summary: failure rate is a property of your process, not of 3D printing. The gap between 30% and 5% is mostly bed adhesion, moisture and mesh quality — three things you can fix this week, and three things that are worth 43% on the cost of every part you sell.

Where this number belongs in a quote

In the denominator, never in a vague “add 10% for waste”. Divide by 1 − f, show it as its own line, and let the customer see it. The pricing guide does exactly that, and the arithmetic is the same one the calculator uses. If you want the underlying parameters with their check dates and source links — filament prices, print power draw, five countries of electricity prices — they are published as a citable table on 3D printing cost data.

Sources, dates, and what this page will not publish

Every figure on this page is computed at build time from the site's parameter dataset, checked 2026-10-23. Nothing is typed in by hand, so the tables cannot drift away from the data behind the calculators.

• Failure-rate bands — an estimate, flagged as one in the dataset: a 2019 Filamentive survey of more than 200 users, plus self-reported threads in r/3Dprinting. Self-selected sample, direction not benchmark.
• Filament prices — 1 kg spool listings scraped 2026-10-23 from us.elegoo.com, overture3d.com, bambulab-us.myshopify.com and printedsolid.com; USD, excluding tax and shipping. The mid price used here is the median of that sample, and bulk 5–10 kg spools are excluded.
• Filament density — manufacturer technical data sheets (ISO 1183). Per-material source links are listed on the cost data page.
• Printer draw — measured average while printing PLA:
  • Entry FDM (Elegoo Centauri 2) — 125 W average while printing PLA: 3dsolved.com
  • Mid FDM (Bambu Lab P1S) — 105 W average while printing PLA: wiki.bambulab.com
  • High-end FDM (Bambu Lab X1 Carbon) — 105 W average while printing PLA: wiki.bambulab.com
• Electricity — household retail rates, not industrial:
  • US — 18.31 c/kWh ($0.1831), as of 2026-07: www.eia.gov
  • UK — 26.32 p/kWh ($0.3521), as of 2026-Q4: www.ofgem.gov.uk
  • DE — 37.0 ct/kWh ($0.4146), as of 2026-YTD: bdew.de
  • AU — 33.1364 c/kWh ($0.2323), as of 2026-07: www.originenergy.com.au
  • CA — 11.2 c/kWh ($0.0798), as of 2026-11: nugget.ca
  Converted to USD at ECB reference rates of 2026-10-22. The countries are not like-for-like: the US and German figures are national household averages, the UK figure is the Ofgem price cap, the Australian figure is one NSW retailer's standing offer and the Canadian figure is Ontario's first tier.
• Machine-hour rate — purchase price ÷ an assumed service life, plus 10% of the purchase price for maintenance. An estimate, and labelled one wherever it appears.

What this page will not publish: a failure rate per material. No filament maker, standards body or slicer vendor publishes one, and the community numbers that exist are self-selected anecdotes. A tidy table of invented per-material rates would look more authoritative than the evidence behind it, so the calculator asks you for your own measured figure instead. That is the number that changes your prices.

Frequently asked questions