Tools › Reliability › Safety Analysis › SIL Verification and LOPA › Example
Standby diesel generators tested monthly: average probability of failure on demand of one and of two (NUREG/CR-6823)
| Tool | SIL Verification and LOPA |
| Source | Atwood, C. L., LaChance, J. L., Martz, H. F., Anderson, D. J., Englehardt, M., Whitehead, D., and Wheeler, T. (2003). Handbook of Parameter Estimation for Probabilistic Risk Assessment. NUREG/CR-6823, U.S. Nuclear Regulatory Commission, Washington, DC. Section 2.3.3.1.2, equations 2.4 and 2.5 (p. 2-11) and Sections 2.3.4.2 and 2.3.4.3 (p. 2-12). |
| Level | Graduate |
Problem
An emergency diesel generator is tested monthly by starting it; in 100 monthly tests 2 failures were seen, so the standby failure rate is estimated as 0.02 per month. Find the probability that it fails to start when demanded at a random time, and the probability that two such generators, tested at essentially the same time, both fail to start.
Inputs to Enter
- Analysis: SIL verification only; required RRF blank; two subsystem rows (clear the third)
- Row 1: voting 1oo1, λ_DU 27,397.26 FIT (0.02 per month of 730 h), λ_DD 0, λ_S 0, test interval 0.083333 yr (one month), coverage 100 %, mission time 0.083333 yr, β 0
- Row 2: the same with voting 1oo2 (two generators, either one sufficient)
Answers
- One generator, simplified: λ t / 2 = 0.02 / 2 = 0.01000 (book: 0.01, equation 2.4)
- One generator, exact time average 1 − (1 − e^−λt) / (λt) = 0.009934 (the more accurate expression printed above equation 2.4)
- Two generators, simplified: (λ t)² / 3 = (0.02)² / 3 = 1.333 × 10⁻⁴ (book: (0.02)²/3, equation 2.5), against 4 × 10⁻⁴ if the two were treated as independent demands of probability 0.02
- Two generators, exact time average of (1 − e^−λs)²: 1.314 × 10⁻⁴
Notes
Not the tool's default case: enter the rows as listed. The tool prints the two rows as subsystems of one function; read each row of the PFD table by itself. The simplified column reproduces the handbook's figures and the exact column is the integral the handbook approximates. Both are asserted in test_sil_verification_lopa.php, which also checks the exact averages against an independent Markov-chain model and a Monte Carlo simulation. The handbook excludes common-cause failure (β = 0 here).