Reliability engineering studies how likely a system is to keep working for a required time under stated conditions. It matters in products such as aircraft, medical devices, power systems, servers, and cars because failures can be costly or dangerous. Engineers use measurements such as failure rate, reliability, availability, and MTBF to compare designs and plan maintenance.
A central idea is that failure behavior often changes over a product lifetime rather than staying constant forever.
The bathtub curve shows failure rate versus time with three main regions: early failures, useful life, and wear-out. Early failures often come from manufacturing defects or installation problems, while the useful life region is often modeled with an approximately constant failure rate. In the wear-out region, aging, fatigue, corrosion, or accumulated damage cause the failure rate to rise.
Reliability can be improved by better quality control, derating, preventive maintenance, redundancy, and choosing series or parallel architectures wisely.
Key Facts
- Reliability is the probability that a system works for a specified time: R(t) = P(T > t).
- For a constant failure rate, reliability follows R(t) = e^(-λt).
- Mean time between failures for a repairable system with constant failure rate is MTBF = 1/λ.
- For independent components in series, system reliability is R_series = R1 R2 R3 ... Rn.
- For independent components in parallel with one required to work, reliability is R_parallel = 1 - (1 - R1)(1 - R2)...(1 - Rn).
- Availability for a repairable system is often approximated by A = MTBF / (MTBF + MTTR).
Vocabulary
- Reliability
- Reliability is the probability that a device or system performs its required function without failure for a specified time under stated conditions.
- Failure Rate
- Failure rate λ(t) is the rate at which failures occur at a particular time among items that have survived up to that time.
- MTBF
- Mean time between failures is the average operating time between failures for a repairable system, often equal to 1/λ when the failure rate is constant.
- Bathtub Curve
- The bathtub curve is a graph of failure rate over time with early failure, useful life, and wear-out regions.
- Redundancy
- Redundancy is the use of extra components or paths so a system can continue operating if one part fails.
Common Mistakes to Avoid
- Treating MTBF as a guaranteed lifetime is wrong because MTBF is an average, not the time when every unit will fail.
- Using R(t) = e^(-λt) for the entire bathtub curve is wrong because that formula assumes a constant failure rate, which usually fits only the useful life region.
- Multiplying failure probabilities instead of reliabilities in a series system is wrong because a series system works only if every required component works, so the component reliabilities are multiplied.
- Assuming redundancy always improves safety is wrong because redundant components can share common causes of failure such as heat, software errors, bad power, or poor maintenance.
Practice Questions
- 1 A server has a constant failure rate of λ = 0.0002 failures per hour. Find its MTBF and its reliability for 1000 hours using R(t) = e^(-λt).
- 2 Three independent components in series have reliabilities 0.98, 0.95, and 0.90 for a mission. Find the total system reliability. Then find the reliability if two identical 0.90 components are placed in parallel and only one must work.
- 3 A product shows many failures in the first month, few failures during the next two years, and then increasing failures after heavy use. Identify the three bathtub curve regions and explain one engineering action that could reduce failures in each region.