Design for Reliability
Design for Reliability
This domain tests how reliability is designed in: allocation, prediction, derating, redundancy, tolerance and stress-strength interference, and removal of failure modes.
Core rules
- Series reliability R_s=∏R_i. Active parallel reliability R_p=1-∏(1-R_i). Redundancy improves reliability only if failures are independent and switching is reliable.
- Derating lowers stress and often lowers failure rate; it cannot compensate for a wearout mechanism with a hard life limit.
- Stress-strength interference: failure occurs when stress exceeds strength. Increase margin or reduce variance to reduce overlap.
- Reliability allocation splits a system goal among subsystems, often by complexity or criticality. Parts-count prediction is a rough early estimate; parts-stress prediction is more detailed.
Worked example
Three series blocks have reliabilities 0.98, 0.97, and 0.99. R_s=0.98×0.97×0.99=0.9411. Add one active parallel unit to the weakest block: R=1-(1-0.97)^2=0.9991. The new system is 0.98×0.9991×0.99=0.9693. The added unit raises system reliability by about 2.8 points; duplicating the whole system would cost far more.
Traps
- Ignoring common-cause failure in redundant designs.
- Treating a calculated prediction as demonstrated reliability.
- Using redundancy to fix a dominant failure mode that also disables the standby path.
- Forgetting thermal, interface, and manufacturing stresses.