3D printing failure rate: how to price in failed prints
The two ways to do the math
Every good part also has to pay for the attempts that went wrong. There are two ways to work that out:
Method 1, a percentage of production cost: reserve = production cost × failure rate. This is what the calculator does with the number you type in the field.
Method 2, dividing by the success rate: cost per good part = production cost ÷ (1 − failure rate). The equivalent reserve is rate ÷ (1 − rate).
Method 2 is the exact one. If 1 attempt in 10 fails, the 9 good parts pay for all 10 attempts: each carries 1/9 of an extra attempt, or 11.1%. With a 10% reserve, the 9 parts only collect 0.9 of an attempt, and something is missing.
The two methods agree when failures are rare and drift apart when they are frequent:
| Measured failure rate | Reserve with method 1 | Exact reserve (method 2) | Shortfall with method 1, per $100 of production |
|---|---|---|---|
| 5% | 5% | 5.3% | $0.26 |
| 10% | 10% | 11.1% | $1.11 |
| 15% | 15% | 17.6% | $2.65 |
| 20% | 20% | 25.0% | $5.00 |
| 30% | 30% | 42.9% | $12.86 |
| 50% | 50% | 100% | $50.00 |
Exact math: reserve = rate ÷ (1 − rate). The rates in the table are examples for the math, not industry averages.
To use the exact method in the calculator, enter the reserve from the third column: with a 10% failure rate, type 11.1.
What the reserve covers and what it leaves out
The reserve is added to the production cost of the part, meaning material + electricity + machine. In the MakerPrice 3D calculator, the "Reserve for failures" field applies to exactly those three lines. It leaves out:
- Your time to start over. Clearing the bed, pulling off the spaghetti and sending the job again takes minutes. If that happens often, add those minutes to your prep time.
- Supplies and packaging. Usually only the good part gets painted and packed, so these are not lost in a failure.
- Defects found after the sale. That is a return: it is a cost of the order, not of the print.
Example with the 85 g part
Take the part from the guide to how much to charge for 3D printing: $1.70 of material, $0.10 of electricity and $0.75 of machine time, a production cost of $2.55 per attempt. Suppose your history for the month shows 18 good parts and 2 failed ones: 2 ÷ 20 = 10%.
- Method 1, a 10% reserve: $2.55 × 0.10 = $0.26.
- Method 2: $2.55 ÷ 0.90 = $2.83 per good part, a reserve of $0.28.
- With a new material that loses 1 part in 4 (25%, an assumption): $0.64 with method 1 and $0.85 with method 2 ($2.55 ÷ 0.75 = $3.40).
On a small PLA part, the reserve is a matter of cents, because at $20 a kilo the plastic is cheap. What costs more is your time: 10 minutes spent clearing the bed and restarting is $5.00 at $30 an hour, which is why it belongs in your prep time. The reserve also gets the markup and fees on top, because it is part of the cost.
A mid-print failure only wastes part of the job
A print that comes loose on the first layer wastes a few grams and a few minutes. A layer shift at 90% of the height wastes almost everything. If you know roughly where your failures happen, the reserve can be smaller:
reserve = rate ÷ (1 − rate) × average fraction already printed when the failure happens
In the example, with failures that stop on average 40% of the way through: $0.28 × 0.4 = $0.11 per good part, instead of $0.28.
Two caveats:
- A failure nobody sees wastes everything. Unless someone is watching or a sensor stops the printer, it keeps extruding into the air until the end of the file. The result is spaghetti that used the material and electricity of the whole part.
- In the calculator, fix the grams. When you confirm a print with failed parts, the preview counts full attempts; if a failure happened before the end, correct the weight, as the screen itself reminds you.
How to measure your own failure rate
failure rate = lost parts ÷ (good parts + lost parts), counted over a period.
In MakerPrice 3D, when a print is done, the "Confirm print" block asks for "Parts ready" and "Failed parts". Only the good ones go into stock, and the "Confirmed prints" list shows how many failed in each production run. Add up the good and failed parts for a month and divide. For example, 46 good and 4 failed gives 4 ÷ 50 = 8%.
Four things to watch:
- Count parts, not plates. A plate with 4 parts where 1 is lost is 4 attempts and 1 failure.
- Split by material and by printer when they behave differently. Mixing a PLA you know well with a new material hides both numbers.
- Small samples mislead. With 5 attempts, one failure is 20%; with 50, it is 2%. The more prints in your history, the more the number means.
- Calibration prints are not production failures. Test prints for a new material are the cost of learning, not a reserve on the parts you sell.
Until you have a history, the field's hint suggests starting at 8% (12% for resin). That is a starting guess, not a measured average: replace it as soon as you have your own number.
Situations that change the rate
| Situation | Why the rate changes | How to handle it |
|---|---|---|
| Tuned printer, material and profile you already use | fewer new variables | use the history for that combination |
| New material, color or profile | the first prints are tuning | keep calibration separate and measure again afterwards |
| Large, long prints | more hours for something to go wrong, and a failure loses more | measure the rate on long prints only |
| Tall, thin parts, heavy supports, materials that warp | adhesion and stability are harder | keep a separate reserve for that kind of part |
| Several parts on one plate | one failure can lose one part or the whole plate | count per part and note what happens to the neighbors |
No numbers on purpose: there is no published standard rate that applies to your printer. What changes from row to row is the direction; the value comes from your history.
On large parts, one failure weighs more than any percentage suggests. Picture a 600 g, 30-hour print with the same reference values: $12.00 of material, $0.61 of electricity and $4.50 of machine time, a production cost of $17.11. A failure 80% of the way through throws away $13.69, plus a day of printer time. In cases like this, lowering the risk is worth more than raising the reserve: a good first layer, dry filament, watching the start and, where possible, splitting the part. The guide to beginner mistakes covers the most common causes, and the one on batch cost per part shows how to split the costs of a full plate.
Frequently asked questions
Does the reserve become profit if nothing fails?
Yes. It is part of the cost, so it gets the markup on top like any other line. If you failed less than planned by the end of the month, the difference stays with you; if you failed more, it comes out of profit.
Does a part with a small defect count as a failure?
If you will not sell it at full price, yes. Another option is to record it as good and sell it as a second, for less. What matters is that the month's math matches what really happened.
Should filament and resin use the same reserve?
Not necessarily. They are different processes, and each has its own history. In resin, a failure can lose the whole plate at once, and a part stuck to the FEP film can ruin the film. Measure each technology separately.
Does a failure from wet filament or a clogged nozzle count?
It does, because it happened. But if a cause keeps repeating, fixing it is cheaper than carrying a high reserve forever: drying a spool costs less than reprinting.
Do I need a reserve if I reprint for free when something fails?
Yes, because the reprint is not free: it uses material, electricity and machine hours. The reserve is the money that pays for that reprint. Without it, your profit pays for the lost part.