B10 life calculator
B10 life is the age at which 10% of units have failed. Reliability engineers use it for warranty, bearing catalogs, and early-life risk — it is much more conservative than MTTF or characteristic life η. Iron Lake Reliability evaluates Bx life from the current workbook fit; this page shows the identity so you can check the arithmetic.
Open the Iron Lake Reliability workbook Open Weibull Calculator Compute BX% from a fit
Definition
For any life distribution, Bx is the quantile at unreliability x/100. B10 ⇒ F = 0.10. On a two-parameter Weibull,
tx = η [−ln(1 − x/100)]1/β
Because −ln(0.90) ≈ 0.10536 is far below 1, B10 sits well below η whenever β is not tiny. Do not substitute MTTF or η for B10 on a drawing: they answer different questions.
Worked example (β = 2, η = 1 000 h)
Take a wear-out Weibull with shape 2 and characteristic life 1 000 hours — round numbers you can redo on paper.
B10 = 1 000 × [−ln(0.90)]1/2 = 1 000 × √0.105361 ≈ 325 h
| Life | Unreliability | Time |
|---|---|---|
| B1 | 1% | 100 h |
| B10 | 10% | 325 h |
| B50 (median) | 50% | 833 h |
| Characteristic life η | 63.2% | 1 000 h |
| MTTF | mean, not a quantile | 886 h |
Why B10 is not “about the same” as η
−ln(0.90) = 0.105360516, well below 1, so B10 is η times a number smaller than 1, raised to 1/β. How much smaller depends on the shape. The table holds η at 1 000 h and changes only β. Each B10 is 1 000 × [−ln(0.90)]1/β, and each median is 1 000 × [−ln(0.50)]1/β.
| β | B10 | B10 / η | B50 |
|---|---|---|---|
| 0.5 | 11.10 h | 0.0111 | 480.5 h |
| 1 | 105.4 h | 0.1054 | 693.1 h |
| 2 | 324.6 h | 0.3246 | 832.6 h |
| 3 | 472.3 h | 0.4723 | 885.0 h |
| 4 | 569.7 h | 0.5697 | 912.4 h |
At β = 0.5 the 10% life is 11 h while η is still 1 000 h. At β = 4 it has moved up to 570 h, because a large shape piles the failures into a narrow band around the scale. Quoting η when the requirement is a 10% life overstates that life at every row, and the overstatement is worst for infant mortality. The MTTF page shows the matching means: for β < 1 the mean is longer than η even while B10 is tiny.
Hand check for β = 2
Shape 2 makes 1/β a square root. −ln(0.99) = 0.010050336 and its square root is 0.100251, so B1 = 100.3 h (the summary table rounds to 100 h). −ln(0.50) = 0.693147 and its square root is 0.832555, so B50 = 832.6 h (rounded to 833 h in that table). The unrounded B10, 324.593 h, rounds to 325 h. Reliability at that B10 time is 0.90 by construction, which is the check that BX% life and reliable life at 90% are the same quantile.
When to use B10
Use B10 when the decision is an early failed fraction: a warranty allowance, a bearing rating, a spares plan that cannot wait for the average unit. Bearing catalogs publish L10, the running life at which 10% of bearings are expected to have failed under a stated load. That is this same quantile. The catalog figure is the application’s B10 only when load, speed, and lubrication match the rating basis. Once you have your own β and η, recompute it.
Use B1 when 10% failed is already too many. Use B50 for the typical unit, the median, knowing that half the population fails sooner. Use MTTF only when you want the average of every life, including the long tail. The Weibull analysis page is where β and η come from. This page only evaluates the quantile.
Common mistakes
- Writing MTTF or η where the drawing asked for B10. On the round example those three times are 886 h, 1 000 h, and 325 h.
- Treating B10 as “we failed 10% of the sample.” B10 is a population quantile from a fitted model. Ten failures out of a hundred on test is a binomial count, not the Weibull quantile, until you fit the distribution and evaluate it.
- Changing the time unit in only one place. If η is in hours, B10 is in hours. Cycles and hours need a duty cycle you measured. The formula does not invent one.
- Ignoring suspensions and then trusting the quantile. The formula is exact for the (β, η) you give it. It cannot repair a fit that threw away the survivors. Fit the censored sample first, then set BX% = 10.
Enter those β and η on Weibull Calculator (or fit your own grid on the workbook) and set BX% = 10. The bearing-sample fit on the Weibull analysis page gives B10 ≈ 254 h for β ≈ 2.17, η ≈ 716 h — same formula, different LifeFit.
FAQ
What is B10 life?
B10 is the time at which 10% of the population is expected to have failed (F = 0.10, R = 0.90). It is a quantile of the life distribution, not a test duration and not MTTF.
How is B10 calculated for a 2P Weibull?
t_0.10 = η [−ln(1 − 0.10)]^(1/β) = η [−ln(0.90)]^(1/β). The same formula with 0.01 or 0.50 gives B1 and B50. Characteristic life η is the 63.2% point, not B63 as a named Bx in most handbooks.
Is B10 the same as 90% reliable life?
Yes. Reliable life t_R at R = 0.90 is the same quantile as B10. Iron Lake Reliability reports BX% life and reliable life from the same fitted LifeFit so the two cannot drift apart.
Where do I compute B10 in Iron Lake Reliability?
Fit life data on the Weibull Workbook, then open Weibull Calculator → Life → BX% life with 10. Test vs use also lists B1, B10, and B50 on the use-case curve after an ALT projection.
Open the Iron Lake Reliability workbook Open Weibull Calculator