Design and Development for Reliability
Design for reliability (DFR)
DFR builds reliability in rather than testing it in. Key tools: derating (operating parts below rated stress), robust design (Taguchi methods, parameter and tolerance design), design of experiments, and redundancy. Candidates should know that derating reduces stress-driven failure rates, and that robust design reduces sensitivity to noise factors.
Failure modes and stress-strength
Stress-strength interference analysis compares the distribution of applied stress with the distribution of strength; reliability is the probability that strength exceeds stress. Increasing margin or reducing variance both improve reliability. Traps include treating stress and strength as single values rather than distributions.
Redundancy types
- Active (parallel): all units operate; system works if at least one works.
- Standby: unit switched in on failure; requires a reliable switch.
- Voting (m-of-n): system works if at least m of n agree. Know that redundancy improves reliability but adds complexity, weight, and common-cause vulnerability. Common-cause failures defeat independence assumptions.
Reliability prediction
Parts-count and part-stress methods (e.g., MIL-HDBK-217 style approaches, Bellcore/Telcordia) estimate failure rate from component data, environment, and quality factors. Predictions are comparative tools, not guarantees; field data should validate them.
Worked example
A component rated 10,000 hours at 100 C derated to operate at 70 C gains life per an Arrhenius-type relationship. If activation energy gives a factor of about 2 per 10 C reduction, a 30 C drop yields roughly 2^3 = 8x life. Use this as a rule of thumb, not a precise claim, and confirm with vendor data.