Calculator overview
Inputs and outputs
This summary comes from the calculator's published input and output contract.
Inputs
- Part 7 stage (0 = not used) (required)
-
Default 0 Range 0 to 4
About this input
The stage of part 7: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 8 stage (0 = not used) (required)
-
Default 0 Range 0 to 4
About this input
The stage of part 8: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Seed (required)
-
Default 2026 Range 1 to 2147483646
About this input
The seed of the random stream, a whole number from 1 to 2147483646; the same seed always gives the same trials, and each part always keeps its own draw.
- Part 4 stage (0 = not used) (required)
-
Default 0 Range 0 to 4
About this input
The stage of part 4: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 5 stage (0 = not used) (required)
-
Default 0 Range 0 to 4
About this input
The stage of part 5: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 6 stage (0 = not used) (required)
-
Default 0 Range 0 to 4
About this input
The stage of part 6: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Your prediction (required)
-
Default Not yet predicted Allowed Not yet predicted, Below 0.9, 0.9 up to 0.99, 0.99 up to 0.999, 0.999 or above
About this input
Challenge 1: your prediction of the system reliability, one of Below 0.9, 0.9 up to 0.99, 0.99 up to 0.999 or 0.999 or above; Not yet predicted keeps the results hidden. Not used outside challenge 1.
- Stage 4 working parts needed (required)
-
Default 1 Range 1 to 4
About this input
The working parts stage 4 needs (k of n): 1 is a parallel stage, all its parts a series stage, between a vote. Ignored with Bridge and for a stage not in use. Set by the k-of-n badge on the canvas.
- Structure (required)
-
Default Stages in series Allowed Stages in series, Bridge
About this input
The structure: Stages in series (up to four stages, each working while at least k of its parts work) or Bridge (parts 1 to 5 in the classic bridge; the stage inputs are then ignored but kept). Chosen on the canvas.
- Trials (required)
-
Default 500 Range 50 to 500
About this input
The number of simulated systems used, a multiple of 50 from 50 to 500; more trials extend the same run.
- Stage 1 working parts needed (required)
-
Default 1 Range 1 to 4
About this input
The working parts stage 1 needs (k of n): 1 is a parallel stage, all its parts a series stage, between a vote. Ignored with Bridge and for a stage not in use. Set by the k-of-n badge on the canvas.
- Stage 2 working parts needed (required)
-
Default 1 Range 1 to 4
About this input
The working parts stage 2 needs (k of n): 1 is a parallel stage, all its parts a series stage, between a vote. Ignored with Bridge and for a stage not in use. Set by the k-of-n badge on the canvas.
- Stage 3 working parts needed (required)
-
Default 1 Range 1 to 4
About this input
The working parts stage 3 needs (k of n): 1 is a parallel stage, all its parts a series stage, between a vote. Ignored with Bridge and for a stage not in use. Set by the k-of-n badge on the canvas.
- Part 3 stage (0 = not used) (required)
-
Default 2 Range 0 to 4
About this input
The stage of part 3: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 1 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 1 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 2 reliability (required)
-
Default 0.8 Range 0.01 to 0.9999
About this input
The reliability of part 2 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 3 reliability (required)
-
Default 0.8 Range 0.01 to 0.9999
About this input
The reliability of part 3 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Challenge (required)
-
Default 0 Range 0 to 5
About this input
Selects a challenge: 0 for none, 1 to 5 for a predefined challenge, which fixes the design except the inputs it leaves to you.
- Confidence level (required)
-
Default 0.95 Range 0.8 to 0.99
About this input
The confidence level of the Wilson band around the sample share, from 0.8 to 0.99.
- Mission time (h) (required)
-
Unit h Default 1000 Range 1 to 100000
About this input
The mission time in hours, from 1 to 100,000. Each part's reliability is given at this time; it only rescales every time (frames, survivors and causes do not change with it).
- Part 4 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 4 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 8 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 8 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 1 stage (0 = not used) (required)
-
Default 1 Range 0 to 4
About this input
The stage of part 1: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 2 stage (0 = not used) (required)
-
Default 2 Range 0 to 4
About this input
The stage of part 2: 1 to 4, or 0 when the part is not used (ignored with Bridge). Set on the canvas by adding, moving or removing its block.
- Part 5 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 5 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 6 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 6 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
- Part 7 reliability (required)
-
Default 0.9 Range 0.01 to 0.9999
About this input
The reliability of part 7 at the mission time, from 0.01 to 0.9999: the probability that it still works then (lives are exponential). Checked even when the part is not used.
Outputs
- Sample minus exact reliability
-
No unit declared
About this output
The sample share minus the exact reliability. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Sample band, low end
-
No unit declared
About this output
The low end of the Wilson band around the sample share at the chosen level. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Sample reliability
-
No unit declared
About this output
The share of the simulated systems surviving the mission. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Sample mean system life
-
Unit h
About this output
The mean life of the simulated systems, hours (a sample value). Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Sample band, high end
-
No unit declared
About this output
The high end of the Wilson band around the sample share at the chosen level. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Reliability over time at the 41 frame times from 0 to twice the mission: exact against the sample share alive
-
No unit declared
About this output
At the 41 frame times from 0 to twice the mission: the time, the exact reliability and the sample share alive. Exact and Sample are blank while challenge 1 awaits a prediction.
- The parts: stage, reliability at the mission, mean life, importance and how often each failed the system
-
No unit declared
About this output
One row a part (1 to 8): the part number, its stage (blank with Bridge), its reliability, its mean life in hours, its importance and how many of trials 1 to T it ended. For an unused part only the part number shows. Importance and Failed the system are blank while challenge 1 awaits a prediction.
- Running estimate every 10 trials: the share surviving the mission with its Wilson band
-
No unit declared
About this output
Every 10 trials up to T: the trials so far, the share surviving the mission, the Wilson band low and high, and 1 when the exact value lies inside the band, else 0. Rows past T/10 are blank; every cell is blank while challenge 1 awaits a prediction.
- Rule of three
-
No unit declared
About this output
With no failure among the T trials, the rule of three 3/T (the 95% rule of thumb). Blank when a failure is seen, when the inputs are not valid or while challenge 1 awaits a prediction.
- Trials used
-
No unit declared
About this output
The number of trials used. Blank when the inputs are not valid.
- Exact system reliability at the mission
-
No unit declared
About this output
The exact probability that the system still works at the mission time. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Zero-failure upper bound on the failure chance
-
No unit declared
About this output
With no failure among the T trials, the exact upper bound on the failure chance at the chosen level, 1 - (1 - level)^(1/T). Blank when a failure is seen, when the inputs are not valid or while challenge 1 awaits a prediction.
- The simulated systems, one row a trial: its life, the frame in which it fails (1 to 41) and the part whose failure ended it
-
No unit declared
About this output
One row a trial (1 to 500): the system life in hours, the fail frame (1 to 41: the system is alive at frame f while f is below it) and the part whose failure ended it. Rows past T are blank; every cell is blank while challenge 1 awaits a prediction.
- Simulated systems surviving the mission
-
No unit declared
About this output
The simulated systems (trials 1 to T) still working at the mission time. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Stages in use
-
No unit declared
About this output
The stages holding at least one part. Blank with Bridge and when the inputs are not valid.
- Seed used
-
No unit declared
About this output
The seed used. Blank when the inputs are not valid.
- Structure used
-
No unit declared
About this output
The structure used (a challenge sets its own). Blank when the inputs are not valid.
- The stages: parts, working parts needed and the reliability at the mission (stages in series only)
-
No unit declared
About this output
One row a stage (1 to 4), stages in series only (blank with Bridge): the stage number, its parts, the working parts it needs and its reliability at the mission (both blank for a stage not in use). Reliability is blank while challenge 1 awaits a prediction.
- Most important part
-
No unit declared
About this output
The part the system depends on most (the largest Birnbaum importance; the lowest part number wins a tie within one part in a billion). Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Confidence level used
-
No unit declared
About this output
The confidence level used. Blank when the inputs are not valid.
- Parts used
-
No unit declared
About this output
The parts used (5 with Bridge). Blank when the inputs are not valid.
- Decision margin
-
No unit declared
About this output
The smallest gap between a simulated system life and a frame time, relative to the mission time; below one in a billion the status warns. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Critical value of the band
-
No unit declared
About this output
The standard normal critical value of the band at the chosen level. Blank when the inputs are not valid.
- Challenge result
-
No unit declared
About this output
No challenge selected, MET or NOT MET.
- Challenge goal
-
No unit declared
About this output
The goal of the selected challenge (blank with none).
- Challenge feedback
-
No unit declared
About this output
Feedback on the selected challenge: the prediction prompt, the met text or the not-met hint (blank with none).
- Challenge score
-
No unit declared
About this output
The challenge score: 100 when met, else 0.
- Challenge: the value scored
-
No unit declared
About this output
The value the selected challenge scores: the exact system reliability (challenges 1 to 4) or the number of failures (challenge 5); blank with none, when not valid or while challenge 1 awaits a prediction.
- Model Status
-
No unit declared
About this output
NOT VALID with the reason when an input or the design cannot be used; CHECK when a simulated system life lies within a billionth of the mission time of a frame time (rounding could then move a frame or survivor count by one); otherwise OK.
- Mission time used
-
Unit h
About this output
The mission time used, hours. Blank when the inputs are not valid.
- Formula module versions
-
No unit declared
About this output
The formula modules the model is built from, with their versions.
- Model version
-
No unit declared
About this output
The model version.
- System lives in 20 bins of a tenth of the mission up to twice the mission, plus the longer lives, against the exact expected counts
-
No unit declared
About this output
The system lives in 20 bins of a tenth of the mission over (0, 2 t_m] plus one bin for longer lives: the bin edges in hours, the count of trials 1 to T and the exact expected count. Count and Expected are blank while challenge 1 awaits a prediction.
- Expected failed systems
-
No unit declared
About this output
The exact expected number of failed systems among the trials, T x (1 - R). Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Exact value inside the band
-
No unit declared
About this output
Yes when the exact reliability lies inside the band, else No (a miss happens in about one run in twenty at 95%, by chance). Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Its importance
-
No unit declared
About this output
The largest importance: the system reliability with that part always working minus that with it never working. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
- Simulated systems failed by the mission
-
No unit declared
About this output
The simulated systems failed by the mission time. Blank when the inputs are not valid or while challenge 1 awaits a prediction.
LogicCommons is in beta. If a result, label, or reference looks wrong, tell us here; we read every message.
Methodology
Purpose and model boundary
This lab models the reliability of a system built from up to eight parts, in one of two structures: up to four k-out-of-n stages in series, or the five-part bridge. Each part has a reliability at the mission time, and its life is exponential. The model returns the exact system reliability at the mission time and over time, each part's importance, and a seeded sample of up to 500 simulated systems: the life of each, the frame in which it fails and the part whose failure ended it, the share surviving the mission with its Wilson confidence band, a running estimate, a life histogram against exact expected counts, and, for a test with no failure, the exact zero-failure bound beside the rule of three. Five scored challenges use the same model.
It is a lab for learning system reliability: why parallel parts raise reliability and series stages lower it, why a stricter vote costs reliability, which part is worth upgrading, and what a test campaign with no failure can and cannot prove. It is not a design or certification tool. Parts are independent with constant failure rates, with no repair, no standby switching and no common-cause failures.
The model is the calculation authority. The page sends the values you set to the model and draws only what it returns: the block diagram of the used design with its k-of-n badges, the field of simulated systems frame by frame, the exact and sample reliability curves, the running estimate with its band and the life histogram. The page computes no reliability, life, count or band itself.
Inputs and units
Reliabilities and the confidence level are probabilities from 0 to 1; the mission time is in hours.
| Input | Meaning and unit |
|---|---|
| Structure | Stages in series (default: up to four stages, each working while at least k of its parts work) or Bridge (parts 1 to 5 in the classic bridge; the stage inputs are then ignored but kept). |
| Part 1 to 8 stage | The stage of each part, 1 to 4, or 0 when the part is not used; ignored with Bridge. Defaults: part 1 in stage 1, parts 2 and 3 in stage 2, parts 4 to 8 not used. |
| Part 1 to 8 reliability | rⱼ, the probability that part j still works at the mission time, from 0.01 to 0.9999. Defaults 0.9, 0.8, 0.8, then 0.9 for parts 4 to 8. Checked even when the part is not used. |
| Stage 1 to 4 working parts needed | k, the working parts the stage needs, a whole number from 1 to 4 (default 1): 1 is a parallel stage, all its parts a series stage, anything between a vote. Ignored with Bridge and for a stage not in use. |
| Mission time | t_m, from 1 to 100,000 hours (default 1,000). Each part's reliability is given at this time. It only rescales every time: frames, survivors and causes do not change with it. |
| Trials | T, the number of simulated systems, a multiple of 50 from 50 to 500 (default 500). More trials extend the same run. |
| Seed | The start of the random stream, a whole number from 1 to 2,147,483,646 (default 2026). Each part always keeps its own draw. |
| Confidence level | The level of the Wilson band, from 0.8 to 0.99 (default 0.95). |
| Challenge | 0 for none, or 1 to 5 to load a challenge. |
| Your prediction | Challenge 1 only: Not yet predicted, Below 0.9, 0.9 up to 0.99, 0.99 up to 0.999, or 0.999 or above. |
Where each input is set:
- On the canvas only: the structure (its chip), the stage of each part (add a block to a stage, drag it to another stage, or remove it), the working parts each stage needs (the k-of-n badge on the stage), and your prediction for challenge 1.
- In the input panel: the part reliabilities (the fields of unused parts are hidden but still sent), the mission time, the trials, the seed, the confidence level and the challenge number.
Governing relationships
Parts. With exponential lives, part j works at time t with probability rⱼ^(t/t_m), that is e^(−λⱼt) with failure rate λⱼ = −ln(rⱼ)/t_m. Its mean life is t_m / (−ln rⱼ).
A k-out-of-n stage. The probabilities of 0 to 4 working parts are built slot by slot: with qᵢ the probability that i of the first parts work, adding a part of reliability p gives qᵢ(1 − p) + qᵢ₋₁·p. The stage reliability is the probability that at least k parts work, the sum of the counts from k to n. For identical parts this is the binomial sum Σᵢ₌ₖⁿ C(n, i) pⁱ(1 − p)ⁿ⁻ⁱ. A stage not in use has reliability 1.
The system. Stages in series: R = R₁ · R₂ · R₃ · R₄. Bridge, by pivoting on part 5 (the cross link):
R = r₅ · (1 − (1 − r₁)(1 − r₂)) · (1 − (1 − r₃)(1 − r₄)) + (1 − r₅) · (1 − (1 − r₁r₃)(1 − r₂r₄)).
With part 5 working the system is (1 or 2) in series with (3 or 4); with part 5 failed it is the path 1 to 3 in parallel with the path 2 to 4. The exact curve R(t) evaluates the same formulas at 41 frame times t_f = t_m · f/20, f = 0 to 40 (0 to twice the mission), with R(0) = 1.
Importance. The Birnbaum importance of part j is
Iⱼ = R(system | part j always works) − R(system | part j never works).
R is linear in each rⱼ, so raising rⱼ by d raises the system reliability by exactly Iⱼ · d. The most important part is the one with the largest importance; within one part in a billion, the lowest part number wins.
The generator. The model implements Park and Miller's minimal standard generator (MINSTD) with the multiplier 48271. Starting from s₀ = the seed, each state is
sₖ = MOD(48271 · sₖ₋₁, 2³¹ − 1), and each draw is u = sₖ / (2³¹ − 1),
which lies strictly between 0 and 1. Trial i gives part j the draw from state s₈₍ᵢ₋₁₎₊ⱼ, whatever its stage and structure, so adding, removing or moving a part never changes another part's life (common random numbers). All 500 trials are computed; the sample uses the first T.
Simulated lives, by the inverse-transform method: part j's life is L = t_m · ln(u) / ln(rⱼ). A part outlives the mission exactly when u < rⱼ. A k-out-of-n stage lives as long as its k-th longest-lived part; a series of stages lives as long as its shortest-lived stage; the bridge lives as long as the best of its four minimal paths (1 and 3, 2 and 4, 1, 5 and 4, 2, 5 and 3), each path living as long as its shortest-lived part. The system life is therefore exactly one part's life, and that part is the one whose failure ended the system.
Frames and survivors. A system is alive at time t exactly when its life is greater than t. Its fail frame is the number of frame times strictly below its life, from 1 to 41; it is alive at frame f while f is below its fail frame, and it survives the mission when its fail frame is above 20.
Sample estimates. The sample reliability is survivors / T. The Wilson score band for x survivors of n trials, with z = NORM.S.INV(1 − (1 − level)/2), p̂ = x/n and d = 1 + z²/n, has centre (p̂ + z²/(2n)) / d and half-width z·√(p̂(1 − p̂)/n + z²/(4n²)) / d; its ends are exactly 0 when x = 0 and exactly 1 when x = n. The running estimate repeats this every 10 trials. The expected number of failed systems is T(1 − R).
Life histogram. Twenty bins of a tenth of the mission over 0 to twice the mission, and one bin for longer lives. The expected count of a bin is T times the drop in the exact R(t) across it, and T · R(2t_m) for the last bin.
No failure. When all T trials survive, the exact one-sided upper bound on the failure chance at the chosen level is
1 − (1 − level)^(1/T),
shown beside the rule of three, 3/T, its 95% rule of thumb.
Calculation sequence
The model works through the relationships above in this order:
- It checks the entered values in the order of the status list under Errors and warnings, items 1 to 9.
- It fixes the values used: your inputs, or in a challenge the challenge's own design, except the inputs it leaves to you. It then checks the design used: at least one part, at most four parts a stage, and no stage needing more parts than it holds.
- It computes the exact stage and system reliability at the 41 frame times, and each part's importance.
- It generates the draws, the part lives, each trial's system life, its fail frame and the part that ended it.
- It counts the survivors and computes the sample share, the Wilson band, the running estimate, the sample curve, the life histogram, the mean life and, with no failure, the zero-failure bound.
- It finds the smallest gap between a system life and a frame time.
- It sets the status and, in a challenge, scores the challenge.
Outputs and interpretation
- Exact system reliability at the mission, sample reliability and failures: the primary results. The sample reliability wanders around the exact value from seed to seed.
- Survivors, expected failed systems, and the sample minus exact reliability.
- Sample band, low and high ends, the critical value of the band, and exact value inside the band, Yes or No. At 95%, No happens by chance in about one run in twenty.
- Most important part and its importance.
- Sample mean system life in hours, a sample value.
- Zero-failure upper bound and rule of three: shown only when no simulated system failed.
- Decision margin: the smallest gap between a simulated system life and a frame time, relative to the mission time.
- Parts used (5 with Bridge), stages in use (stages in series only), and the values used: the structure, mission time, trials, seed and confidence level of the run, which in a challenge are the challenge's own; the model version and formula module versions.
- Tables: the parts (stage, reliability, mean life in hours, importance, and how many trials each part ended), the stages (parts, working parts needed and reliability at the mission), the trials (life in hours, fail frame and failed part), the exact and sample reliability at the 41 frame times, the running estimate every 10 trials with its band and an exact-in-band flag, and the life histogram with its bin edges in hours, counts and exact expected counts.
Reading the canvas: the block diagram shows the design used, with each stage's k-of-n badge; no number is printed on it. The field of simulated systems, one square a trial, plays frames 0 to 40: a square goes dark in its fail frame, and clicking it shows its life, fail frame and failed part. The curves set the exact R(t) beside the sample share alive, with the mission marked; the running estimate narrows its band as trials accumulate.
What to take from the lab: redundancy pays most where a part is weak and the stage is a bottleneck; a part's importance times the rise in its reliability is the gain from upgrading it; and a clean test of T trials bounds the failure chance only to about 3/T at 95%, so a rare failure can easily go unseen.
Challenges
Choosing a challenge number from 1 to 5 loads that challenge. Every challenge uses stages in series, a mission of 1,000 hours, 500 trials and a 95% level, fixes its own design, and leaves you only the inputs it names. The result reads MET (score 100) or NOT MET (score 0), with the goal, the value the challenge is scored on and a short explanation or hint. A reliability target counts as met within one part in a billion below it.
- Predict redundancy. Three parts that each work 80% of the time are placed in parallel. Predict the system's reliability before you run it. Open: your prediction. Met when your range (below 0.9, 0.9 up to 0.99, 0.99 up to 0.999, or 0.999 or above) holds the exact system reliability.
- Fewest parts for 0.99. Two stages in series, each working while any one of its parts works, every part 80% reliable. Reach a system reliability of at least 0.99 with the fewest parts. Open: the stage of each part. Met with exactly stages 1 and 2 in use, at most 7 parts, and an exact reliability of at least 0.99.
- The strictest vote. Three parts of 90% each form a voting stage. Choose the strictest vote that still keeps the reliability at 0.97 or more. Open: the working parts stage 1 needs. Met with at least 2 needed and an exact reliability of at least 0.97.
- Upgrade the right part. Raise the reliability of just one part, to at most 0.99, so the system reaches a reliability of at least 0.93. Open: the reliabilities of parts 1 to 5. Met when at most one of them differs from the challenge's own value, none is above 0.99, and the exact reliability is at least 0.93.
- A clean test. One part with a reliability of 0.998 is tested 500 times. Find a seed whose test shows no failure at all. Open: the seed. Scored on the number of failed simulated systems, met at 0.
Challenges 1 to 4 are scored on the exact reliability, so the seed does not decide them; challenge 5 is about the sampling itself.
Challenge 1 holds back the answer until you predict: while your prediction reads Not yet predicted, the exact and sample reliabilities, the band, the counts of survivors and failures, the importances, the bounds, the mean life, the margin, the value scored, the trial table and the running estimate are blank, as are the reliability columns of the curve and the stage table, the importance and failure columns of the parts table, and the histogram's counts. The result reads NOT MET with the prompt, and the status reads OK. The design, the bin edges, the critical value and the values used still show. Your prediction changes nothing outside challenge 1.
Validation and status logic
The status reads the first match in the list under Errors and warnings. Items 1 to 9 apply to the entered values, inside a challenge too; the three design checks and the CHECK item apply to the values used, so a challenge's own design decides them. The design checks apply only to stages in series. A CHECK does not stop the calculation: every result is still shown.
Assumptions and limitations
- Independent parts with exponential lives (constant failure rates); no repair, no standby switching, no common-cause failures.
- Two structures only: up to four k-out-of-n stages in series with at most four parts a stage, or the five-part bridge.
- The sample is at most 500 trials of 8 draws. At a reliability of 0.999 a 500-trial sample usually shows no failure; that is the lesson of the rare event, and the challenges that ask about reliability are scored on exact values.
- The generator is a small linear congruential generator, deterministic and reproducible, fine for teaching and wrong for serious simulation.
- The field and the sample curve resolve time in steps of a twentieth of the mission; the trial table carries each life in full.
- Exact values are exact in the mathematical sense; in floating point they hold to about 1×10⁻¹⁵.
- A simulated life within 1×10⁻⁹ of the mission time of a frame time could fall in a neighbouring frame in another computation; the model reports the smallest such gap and warns.
Restrictions and non-computing states
Every input is required and is checked on its entered value, even when the structure does not use it. A challenge number that is not a whole number from 0 to 5 counts as no challenge selected.
Under NOT VALID every result and every table is blank, except the model and module versions, the challenge result (No challenge selected, or NOT MET with a challenge selected), a score of 0, and the goal and hint of a selected challenge.
With Bridge, parts 6 to 8 are not used, and the stage column, the stage table and the stages in use are blank. For an unused part, only the part number shows in the parts table. The zero-failure bound and the rule of three are blank whenever a simulated system failed.
In challenge 1, until you predict, the results named under Challenges are held back.
Errors and warnings
The status messages, first match wins (s, k and n are whole numbers):
- NOT VALID: choose Stages in series or Bridge from the list
- NOT VALID: each part's stage must be a whole number from 0 (not used) to 4
- NOT VALID: each part's reliability must be from 0.01 to 0.9999
- NOT VALID: each stage's number of working parts needed must be a whole number from 1 to 4
- NOT VALID: the mission time must be from 1 to 100,000 hours
- NOT VALID: the number of trials must be a multiple of 50 from 50 to 500
- NOT VALID: the seed must be a whole number from 1 to 2147483646
- NOT VALID: the confidence level must be from 0.8 to 0.99
- NOT VALID: the challenge number must be a whole number from 0 to 5, and the prediction one of the listed choices
- NOT VALID: put at least one part into a stage: stages in series with every part unused.
- NOT VALID: stage s holds more than 4 parts; move a part to another stage: the lowest such stage.
- NOT VALID: stage s needs k working parts but holds only n: the lowest stage in use that needs more parts than it holds.
- CHECK: a simulated system life falls within a billionth of the mission time of a frame time, so a frame or survivor count could differ between spreadsheet engines; try another seed: the decision margin is below 1×10⁻⁹.
- OK: the run is computed with nothing to flag.
References
This laboratory is an educational model of system reliability. Parts have exponential lives with constant failure rates; a k-out-of-n stage works while at least k of its parts work, stages in series multiply, and the bridge is solved by pivoting on its cross link. Each part's Birnbaum importance is the change in system reliability between that part always working and never working. The seeded sample draws part lives by the inverse-transform method from the Park-Miller minimal standard (MINSTD) generator with the multiplier 48271, gives the share surviving the mission a Wilson score band, and, for a test with no failure, reports the exact zero-failure bound beside the rule of three.
Sources and further reading
- The exponential life model, its constant failure rate and mean time to fail: NIST/SEMATECH, e-Handbook of Statistical Methods, section 8.1.6.1, Exponential. NIST
- Series, parallel and r-out-of-n system models built from component reliabilities: NIST/SEMATECH, e-Handbook of Statistical Methods, section 8.1.8, How can you evaluate reliability from the bottom-up, and section 8.1.8.4, R out of N model. NIST
- Zero-failure test plans for the exponential model: NIST/SEMATECH, e-Handbook of Statistical Methods, section 8.3.1.1, Exponential life distribution (or HPP model) tests. NIST
- The rule of three for a sample with no events: J. A. Hanley and A. Lippman-Hand, "If nothing goes wrong, is everything all right? Interpreting zero numerators", JAMA 249(13), 1983. doi:10.1001/jama.1983.03330370053031
- The score interval for a proportion: E. B. Wilson, "Probable inference, the law of succession, and statistical inference", Journal of the American Statistical Association 22(158), 1927, pages 209-212. doi:10.1080/01621459.1927.10502953
- The minimal standard generator: S. K. Park and K. W. Miller, "Random number generators: good ones are hard to find", Communications of the ACM 31(10), 1988, pages 1192-1201. doi:10.1145/63039.63042
- Structure functions, the bridge by pivotal decomposition, and Birnbaum's measure of importance: M. Rausand and A. Høyland, System Reliability Theory: Models, Statistical Methods, and Applications, 2nd edition, Wiley, 2004.
No table or text from any of these publications is reproduced, and none of their authors or publishers has reviewed or endorses this laboratory.
This laboratory is for learning. Its results hold only within its stated ranges; parts are independent with constant failure rates and no repair, which real equipment rarely satisfies.
Links were current at build time. If one moves, search the document's title rather than reporting a broken link as a defect.
Found a problem, or have an idea?
Tell us if a result looks wrong, a label is unclear, or something is missing. We read every message.